The Hong Kong College of Orthopaedic Surgeons (HKCOS) is the official professional body responsible for organising the training of orthopaedic surgeons and conducting specialist examinations in orthopaedics in Hong Kong. The College also plays a key role in setting standards of care in orthopaedic surgery, which includes, but is not limited to, the investigation, preservation, and restoration of the form and function of the extremities, spine, and associated structures. The aim is to bring relief to patients of all ages affected by injury or disease of the musculoskeletal system.1

Knee osteoarthritis (knee OA) is common among middle-aged and older individuals. Its incidence and severity increase with age.2 Hong Kong, one of the cities with the highest life expectancy globally,3 faces a growing demand for knee OA treatment.4 In the public healthcare sector, patients often experience a long waiting time for medical consultation, physiotherapy, and surgical interventions related to knee OA.4 With an ageing population, the number of individuals affected by knee OA is expected to further increase over the next two to three decades,5 highlighting the urgent need for effective management strategies in Hong Kong.

To provide evidence-based guidance for the public and healthcare professionals on the management of knee OA, the HKCOS published a position statement in August 2022 (Table 1).6 This position statement was developed based on a consensus survey conducted in June 2022, along with a review of high-quality evidence in the literature. A total of 26 treatment recommendations were formulated, reflecting the consensus of expert orthopaedic surgeons in Hong Kong. The methodology used to develop this position statement is available on the College’s website.6

To gather updated expert opinions on the management of knee OA, the College conducted a second survey in January 2024. This article presents the findings from the 2022 and 2024 surveys.

This study was conducted under the supervision and ethical oversight of the Council of HKCOS. The requirement to obtain informed consent from participants was waived by the Council. Two cross-sectional surveys were conducted online in June 2022 and January 2024, respectively. The same online questionnaire was used for both study periods to ensure consistency in data collection.

The online questionnaire was developed by the HKCOS Osteoarthritis of the Knee Working Group (HKCOS-OAKWG).6 A total of 26 interventions for knee OA were selected based on the practices of orthopaedic surgeons in Hong Kong and guidelines from international recognised organisations.7 8 An independent literature review was conducted, and the quality of evidence was graded according to the Oxford Centre for Evidence-Based Medicine levels of evidence.9 Available Level 1 and Level 2 evidence up to March 2022 was summarised for each intervention.

Fellows of the HKCOS listed in the Specialist Register (Orthopaedics and Traumatology) of the Hong Kong Medical Council were invited via email to participate in the surveys. Each survey remained open for 4 weeks. The surveys collected opinions on the 26 interventions. For each item, the questionnaire provided: (1) a brief description of the intervention; (2) a concise summary of the supporting Level 1 and Level 2 evidence; and (3) a single question asking whether the expert would recommend the intervention. Expert panellists were instructed to select one of three response options: ‘Recommend’, ‘Do not recommend’, or ‘Abstain’.

Descriptive statistics were reported for the two cross-sectional surveys conducted in 2022 and 2024. These included the number of expert panellists who participated in each survey with their recommendations regarding the 26 interventions for knee OA. The number of panellists who participated in both the 2022 and 2024 surveys was also recorded. The experience of the expert panel was described based on the number of years each panellist had been registered as a specialist with the Hong Kong Medical Council. Consensus was defined as a minimum of 70% agreement among the expert panellists. The resulting recommendations formed the basis of the HKCOS position statement. No analytical statistical tests were performed.

A total of 106 expert panellists participated in the 2022 survey, while 28 participated in the 2024 survey. At the time of survey distribution in June 2022 and January 2024, the College had 528 and 567 fellows, respectively. In June 2022, the numbers of active fellows, overseas fellows, and inactive fellows were 495, 4, and 27, respectively. The corresponding numbers in January 2024 were 526, 7, and 34.10 The response rates for the surveys were 20% in 2022 and 5% in 2024, respectively. Twenty expert panellists participated in both the 2022 and 2024 surveys. The mean duration of experience as registered orthopaedic specialists with the Hong Kong Medical Council (±standard deviation) was 15.6±9 years for the 2022 cohort and 19.2±7 years for the 2024 cohort (Fig 1). The distribution of expert panellists according to their affiliations is presented in Figure 2.

The results of the expert recommendations from the 2022 and 2024 surveys are summarised in Table 2. Consensus was reached among the expert panellists for 14 interventions in 2022 and 13 interventions in 2024. As consensus had already been achieved for most treatment options in the first-round survey, no changes were made to the HKCOS position statement on the management of knee OA, and the content of the second survey remained unchanged. In both surveys, a supermajority of panellists (≥70%) recommended 11 interventions: patient education; land-based exercise; water-based exercise; self-management programme; weight reduction; use of a cane; oral paracetamol; topical non-steroidal anti-inflammatory drugs (NSAIDs); oral NSAIDs; high tibial osteotomy; and joint replacement surgery. Conversely, at least 70% of respondents in both surveys did not recommend the use of lateral wedge insoles or denervation therapy. Consensus was not achieved for 12 interventions in 2022 and 13 interventions in 2024.

A consensus of 70% or greater agreement was reached among the expert panellists for 14 interventions in the 2022 survey and 13 interventions in the 2024 survey. Consensus remained consistent across both cross-sectional surveys for 13 interventions, including: recommendations in favour of patient education, land-based and water-based exercise, self-management programmes, weight reduction, use of a cane, oral paracetamol, topical and oral NSAIDs, high tibial osteotomy and joint replacement surgery, as well as recommendations against the use of lateral wedge insoles and denervation therapy.

There are two primary objectives in the management of knee OA: (1) control of disease progression; and (2) control of symptoms. Regarding disease control, it is rare to reverse the underlying pathological changes, such as joint space narrowing and varus malalignment of the lower limb, through non-operative treatment once the condition is established. The aim of management is to slow the progression of osteoarthritis by promptly identifying and treating ‘joint-threatening’ pathologies (such as symptomatic loose bodies, full-thickness, full-width root tears of the meniscus, and spontaneous osteonecrosis of the knee)6 11 and by controlling risk factors associated with disease progression (eg, through weight reduction and prevention of knee injury).12 Regarding symptom control, most non-operative treatments focus on pain management13 and improving quality of life.14 Pain is the most common presenting symptom in knee OA.15 However, there is only a weak correlation between radiological abnormalities and patient-reported symptoms.16 The severity of pain experienced is influenced by sensitisation mechanisms—including both peripheral and central sensitisation—rather than by anatomical changes in the arthritic joint alone.15 It is increasingly recognised that inflammation plays a prominent role in the pathogenesis and symptoms of OA.17 Inflammatory cytokines (eg, interleukin 6 and tumour necrosis factor alpha) and inflammatory mediators (eg, prostaglandin and bradykinin) contribute to peripheral sensitisation of nociceptors in the subchondral plate and joint capsule by activating G protein–coupled receptors, ionotropic receptors, and tyrosine kinase receptors located on nerve terminals and cell bodies.18 Accordingly, most treatments for knee OA aim to control inflammation in order to alleviate symptoms.7 The cell bodies of nociceptors are located in the dorsal root ganglia of the spinal cord, where nociceptive input is transmitted to the brain and brainstem via the spinothalamic tract, spinoreticular tract, spinomesencephalic tract, and spinohypothalamic tract.13 Persistent nociceptive stimulation leads to central sensitisation, characterised by hyperactivity and hyperexcitability of neurons in the brain and spinal cord.19 Inflammation also contributes to central sensitisation by increasing the production of neurotransmitters and neuromodulators, including glutamate and substance P.18 Furthermore, there is a loss of inhibitory control due to a reduction in inhibitory neurotransmitters, primarily gamma-aminobutyric acid (GABA), resulting in heightened pain sensitivity.18 The perception of nociceptive stimuli as pain is further influenced by psychological factors (such as depression, anxiety, and poor coping skills) and social factors (such as workers’ compensation claims and lack of social support). The interaction among biological, psychological, and social factors is described by the biopsychosocial model, which is considered important in the development of chronic pain.19 As a result, education, self-management programmes, and cognitive behavioural therapy are also important treatment strategies for knee OA.7

Surgical treatments, such as joint replacement surgery and high tibial osteotomy, are typically reserved for patients with end-stage knee arthritis or those with persistent symptoms despite adequate non-operative management.14

Most patient education programmes provide guidance on: (1) the nature, presenting symptoms, and treatment options for knee OA; (2) the importance of adherence to treatment and the adoption of preventive lifestyle behaviours; and (3) the expected natural disease course and treatment outcomes.20 Patient education operates through a theory of planned behaviour-based intervention.20 By modifying patients’ attitudes and the subjective norms of family members, both symptoms and quality of life can be improved, even in the absence of changes in physical condition.20

There is high-quality evidence in the literature supporting the effectiveness of exercise—both land-based and water-based—along with weight reduction, use of a cane, and self-management programmes in reducing pain and improving quality of life in patients with knee OA.21

Exercise alleviates symptoms of knee OA by increasing lower limb muscle strength and improving overall physical fitness.21 Improved muscle strength alters joint biomechanics, thereby reducing joint loading and pain. Aerobic exercise increases peak oxygen uptake by muscles and enhances the patient’s overall fitness, facilitating the performance of daily activities and enhancing functional ability.21 Exercise should be performed regularly, with sufficient—but not excessive—intensity.14 Recommended land-based activities include lower limb strengthening exercises, static cycling, walking programmes, Tai Chi, and Baduanjin,21 while water-based activities include swimming and pool therapy.22

Obesity is a known risk factor for the development of knee OA12 and the subsequent need for joint replacement surgery.23 Beyond increased mechanical loading, obesity is associated with metabolic syndrome—a cluster of metabolic disorders including abdominal obesity, hyperlipidaemia, hypertension, and elevated fasting serum glucose.24 Metabolic syndrome is increasingly recognised as an important risk factor for OA due to its association with systemic inflammation.25 Among obese individuals with OA, disability may be reduced with weight loss exceeding 5%.26 Evidence supports the use of intensive low-energy diet programmes for achieving weight loss and reducing pain in this population.27

The use of a cane reduces weight-bearing on the arthritic joint, thereby alleviating pain. A randomised controlled trial has demonstrated significant improvements in pain and function among patients who use a cane for walking.28

Self-management programmes have been shown to reduce pain and improve function in patients with knee OA.29 30 31 In addition to education and exercise, these programmes equip patients with a range of coping and management strategies, including pain coping techniques,29 joint protection and injury prevention strategies,30 and stress management skills.31

Pain may arise when nociceptors in the knee joint are stimulated.13 Although articular cartilage and the meniscus are commonly damaged in knee OA, they are aneural and therefore not direct sources of pain. In contrast, nociceptors are present in the subchondral bone, periosteum, joint capsule, and infrapatellar fat pad.13 In a healthy joint, these nociceptors are not stimulated during physiological loading, meaning that individuals without OA do not experience pain during daily activities such as walking, squatting, or running. Inflammatory cytokines, including interleukin 6, tumour necrosis factor alpha, and other pro-inflammatory molecules (eg, prostaglandin and bradykinin), increase in response to joint injury or during OA flare-ups. These molecules bind to nociceptors on the cell surface, inducing hypersensitivity and hyperexcitability. Consequently, pain may be experienced even during normal joint loading in previously painless knees—a phenomenon known as peripheral sensitisation.18 Non-steroidal anti-inflammatory drugs are effective in alleviating pain in knee OA32 by reducing intra-articular and circulating levels of inflammatory molecules, including prostaglandins.

While oral NSAIDs are effective in relieving OA-related pain, their side-effects warrant caution.33 Systemic NSAIDs should be used judiciously in older patients and in individuals with peptic ulcer disease, renal impairment, or a history of asthma or ischaemic heart disease.34 Topical NSAIDs, by contrast, have demonstrated efficacy in reducing knee pain while offering a much safer side-effect profile.35 As such, topical NSAIDs should be considered before initiating oral formulations.34

Paracetamol has been shown to improve pain and function in patients with knee OA.36 It is considered a safe medication, even in patients at higher risk of complications.7 However, overdose can result in hepatotoxicity and death, necessitating proper patient education regarding safe dosing practices.8

Both joint replacement surgery and high tibial osteotomy are salvage operations for symptomatic end-stage knee arthritis following unsuccessful non-operative treatment. Both procedures have demonstrated effectiveness in relieving pain in patients with knee OA.37 38 Joint replacement surgery has been shown to improve function regardless of the extent of joint involvement.39 In contrast, high tibial osteotomy is effective for symptom relief only in a selected group of patients whose symptomatic OA is confined to the medial compartment of the knee.38

The recommendations of the HKCOS position statement on the management of knee OA,6 alongside those of the National Institute for Health and Care Excellence (NICE) guideline,14 the American Academy of Orthopaedic Surgeons (AAOS) guideline,7 and the Osteoarthritis Research Society International (OARSI) guideline,33 are summarised in Table 3.

All of these guidelines recommend patient education, exercise, self-management programmes, weight reduction, and the use of a cane as appropriate management strategies for patients with knee OA.6 7 14 33 With respect to knee biomechanical interventions, such as knee braces and sleeves, the AAOS guideline is unique in suggesting that these “may be used” in knee OA.7 Concerning foot biomechanical interventions, the HKCOS, NICE, and AAOS guidelines advise against the use of lateral wedge insoles.6 7 14

For pharmacological treatment, all reviewed guidelines recommend the use of NSAIDs, either in topical or oral formulations.6 7 14 33 However, owing to the increased risk of adverse effects associated with oral NSAIDs in patients with co-morbidities, both HKCOS and NICE recommend topical NSAIDs as first-line treatment, reserving oral NSAIDs for second-line use.6 14

There is variation in guidance concerning the use of oral paracetamol. While HKCOS and AAOS support its use as a first-line treatment in knee OA,6 7 NICE and OARSI do not recommend its routine use.14 33 None of these guidelines recommend opioids for the treatment of knee OA.6 7 14 33 Regarding intra-articular steroid injections, the NICE, AAOS, and OARSI guidelines either recommend or conditionally recommend their use for short-term symptom relief.7 14 33

In contrast, intra-articular hyaluronic acid injections are not recommended in the NICE and AAOS guidelines.7 14 The only reviewed guideline that conditionally supports their use is the OARSI guideline.33

Based on the comparison outlined above, substantial differences are observed among the various guidelines (Table 3), despite all having evaluated the same set of treatment options for knee OA and drawing upon similar English-language literature. These differences may reflect variations in cultural context and healthcare systems across different geographical regions. Hong Kong is distinct from North America, Europe, and the United Kingdom due to the considerable influence of traditional Chinese medicine (TCM) within its healthcare system. Many patients in Hong Kong seeking orthopaedic care for knee OA—whether through the Hospital Authority or private healthcare providers—also receive TCM treatment. Some doctors registered with the Medical Council of Hong Kong have received training in TCM, and basic education in TCM is included in the Bachelor of Medicine and Bachelor of Surgery curriculum in Hong Kong. It is thus crucial for the HKCOS to develop a position statement tailored to the local population. The comparison of various guidelines highlights differences between the HKCOS position statement and guidelines issued by other reputable international organisations. The recommendations put forward by HKCOS may be better suited to the needs of patients in Hong Kong. This position statement also serves a valuable reference for healthcare systems in which TCM plays a prominent role, including those in Mainland China, Taiwan, Korea, and Japan.

The present study has several limitations. First, its descriptive design precluded the use of analytical statistics; as a result, we were unable to identify whether there were significant shifts in opinion regarding interventions for knee OA within the Hong Kong orthopaedic community between 2022 and 2024. Second, only 19% of the expert panellists who participated in the 2022 survey also took part in the 2024 study. Consequently, we could not assess potential changes in opinion among those who did not respond in 2024. Third, the overall response rate for the 2024 survey was low, which may limit the representativeness of the findings and reduce their generalisability to the wider orthopaedic specialist population in Hong Kong. Lastly, the threshold for consensus, defined as a minimum of 70% agreement among the expert panellists, was arbitrary and may not be universally accepted.

Findings from both the 2022 and 2024 surveys demonstrated consistent expert consensus on key recommended and non-recommended interventions for knee OA. The recommendations provide evidence-based guidance to the public and medical practitioners regarding the effectiveness of various management strategies for knee OA.

The author solely contributed to the concept or design of the study, acquisition of the data, analysis or interpretation of the data, drafting of the manuscript, and critical revision of the manuscript for important intellectual content.

The author had full access to the data, contributed to the study, approved the final version for publication, and takes responsibility for its accuracy and integrity.

The author has disclosed no conflicts of interest.

The author would like to acknowledge the contributions of the following Fellows of the Hong Kong College of Orthopaedic Surgeons, who served as members of the expert panel and provided recommendations on interventions for the management of knee osteoarthritis in the survey conducted in 2022 (in alphabetical order):
Chan Baldwin, Chan Ka-wah, Chan Sai-keung, Chan Shiuwai, Chan Tsang-tung, Chan Tze-wang, Chan Wai-kwan Vincent, Chan Wai-lam, Chan Wai-sing, Chang His-tse Joseph Jeremy, Chang Shao, Chang Yun-po Robert, Cheng Hung-on, Cheong Peng-meng, Cheung Chi-nok, Cheung Ho-man, Cheung Man-hong, Cheung Pui-yin Jason, Cheung Wai-yuen, Cheung Wang-yan Warren, Cheung Yim-ling Amy, Chien Ping Eric, Chiu Chi-kit, Chiu Kwong-yuen Peter, Choi Shing-hing, Choi Sum-hung, Chow Kai-pun, Chu Kai-man, Chun Siu-yeung, Chung Kwong-yin, Chung Man-ting Marvin, Fan Chi-ho Jason, Fang Christian Xinshuo, Fu Chun-him Henry, Fu Wai-kee, Fung Kwai-yau, Hau Vincent, Ho Ki-wai, Ho Sheung-tung, Hung Leong-pan, Kwok Hau-yan, Lai Kam-keung, Lau Chi-yuen, Lau Chi-yuk, Lau Hoi-kuen, Lau Tak-wing, Lau Yip-kwong Francis, Law Sheung-wai, Law Ying-kan, Lee Qunn-jid, Lee Sung-yee, Lee Yeung-fai, Leung Ka-hei, Leung Yum-kwong, Ling Ka-kin Samuel, Loong Tak-wan, Lui Ming-yan, Man Lok-pong, Ng Fu-yuen, Ng Cheukkee Kenneth, Ng Pak-lin Eugene, Ng Wai-kit Raymond, Ng Weng-io, Ngai Wai-kee, Ong Tim-yun Michael, Sin Cheuk-hang, Siu Kam-to, Sun Lun-kit, Sun Kin-wai Kelvin, Tam Ka-ki, Tam Kwok-wai Kelvin, Tang Yan-ho Bruce, Tang Yuk-kwan, Tong Hoi-yiu Sara, Tse Lung-fung, Tsoi Chi-wah Danny, Tung Kam-lung, Wong Hok-leung, Wong Kwok-shing Patrick, Wong Man-kwan, Wong Nang-man Raymond, Wong Sze-hung, Wu Daniel Yiang, Wu Tsz-kit, Yang Isaac Bruce, Yeung Lok-yin Michael, Yeung Sze-tsun Eric, Yeung Yip-kan, Yuen Shiu-him Jonathan, Yung Shing-yat Colin.

The author would like to acknowledge the contributions of the following Fellows of the Hong Kong College of Orthopaedic Surgeons, who served as members of the expert panel that provided recommendations on interventions for the management of knee osteoarthritis in the 2024 survey (in alphabetical order):
Chan Shiu-wai, Chan Wai-lam, Cheung Chi-nok, Cheung Wang-yan Warren, Chiu Kwong-yuen Peter, Choi Shing-hing, Fu Chun-him Henry, Ho Sheung-tung, Lam Chi-keung Johnson, Lau Pui-yau, Lee Qunn-jid, Leung Ka-hei, Liem Man-shing, Ng Fu-yuen, Wong Hok-leung, Yeung Sai-hung.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

This research was approved by the Council of the Hong Kong College of Orthopaedic Surgeons. The requirement for informed consent was waived by the Council.

1. The Hong Kong College of Orthopaedic Surgeons. Constitution. Available from: https://www.hkcos.org.hk/aboutus_constitution.asp. Accessed 4 Jul 2022.

2. Tang X, Wang SF, Zhan SY, et al. The prevalence of symptomatic knee osteoarthritis in China. Results from the China health and retirement longitudinal study. Arthritis Rheumatol 2016;68:648-53. Crossref

3. Woo J. Age and Ageing journal 50th anniversary commentary series. Healthcare for older people in Asia. Age Ageing 2022;51:afab189. Crossref

4. Hospital Authority, Hong Kong SAR Government. Elective total joint replacement surgery. 2023. Available from: www.ha.org.hk/visitor/ha_visitor_index.asp?Parent_ID=214172&Content_ID=221223#. Accessed 26 Feb 2024.

5. GBD 2021 Osteoarthritis Collaborators. Global, regional, and national burden of osteoarthritis, 1990-2020 and projections to 2050: a systematic analysis for the Global Burden of Disease Study 2021. Lancet Rheumatol 2023;5:e508-22. Crossref

6. Yau WP, Chiu CK, Cheung S, Cheung A. The Hong Kong College of Orthopaedic Surgeons position statement in management of osteoarthritis of knee. Available from: https://www.hkcos.org.hk/Position_Statement/HKCOS_Position_Statement_in_Management_of_Osteoarthritis_of_Knee.pdf. Accessed 26 Feb 2024.

7. American Academy of Orthopaedic Surgeons. Management of osteoarthritis of the knee (nonarthroplasty). Evidence-based clinical practice guideline. Available from: https://www.aaos.org/globalassets/quality-and-practice-resources/osteoarthritis-of-the-knee/oak3cpg.pdf. Accessed 26 Feb 2024.

8. McAlindon TE, Bannuru RR, Sullivan MC, et al. OARSI guidelines for the non-surgical management of knee osteoarthritis. Osteoarthritis Cartilage 2014;22:363-88. Crossref

9. Oxford Centre for Evidence-Based Medicine. OCEBM levels of evidence. Available from: http://www.cebm.ox.ac.uk/resources/levels-of-evidence/ocebm-levels-of-evidence. Accessed 26 Feb 2024.

10. The Hong Kong College of Orthopaedic Surgeons. Statistics. Number of Fellows. Available from: https://www.hkcos.org.hk/aboutus_statistics.asp. Accessed 16 Sep 2024.

11. Driban JB, Harkey MS, Barbe MF, et al. Risk factors and the natural history of accelerated knee osteoarthritis: a narrative review. BMC Musculoskeletal Disord 2020;21:332. Crossref

12. Glyn-Jones S, Palmer AJ, Agricola R, et al. Osteoarthritis. Lancet 2015;386:376-87. Crossref

13. Gwilym SE, Pollard TC, Carr AJ. Understanding pain in osteoarthritis. J Bone Joint Surg Br 2008;90:280-7. Crossref

14. National Institute for Health and Care Excellence. Osteoarthritis in over 16s: diagnosis and management. Available from: http://www.nice.org.uk/guidance/ng226. Accessed 26 Feb 2024.

15. Lane NE, Brandt K, Hawker G, et al. OARSI-FDA initiative: defining the disease state of osteoarthritis. Osteoarthritis Cartilage 2011;19:478-82. Crossref

16. Courties A, Kouki I, Soliman N, Mathieu S, Sellam J. Osteoarthritis year in review 2024: epidemiology and therapy. Osteoarthritis Cartilage 2024;32:1397-404. Crossref

17. Jenei-Lanzl Z, Zaucke F. Osteoarthritis year in review 2024: biology. Osteoarthritis Cartilage 2025;33:58-66. Crossref

18. Matsuda M, Huh Y, Ji RR. Roles of inflammation, neurogenic inflammation, and neuroinflammation in pain. J Anesth 2019;33:131-9. Crossref

19. Cohen SP, Vase L, Hooten WM. Chronic pain: an update on burden, best practices, and new advances. Lancet 2021;397:2082-97. Crossref

20. Saffari M, Emami Meybodi MK, Sanaeinasab H, Karami A, Pakpour AH, Koenig HG. A theory of planned behavior–based intervention to improve quality of life in patients with knee/hip osteoarthritis: a randomized controlled trial. Clin Rheumatol 2018;37:2505-15. Crossref

21. Fransen M, McConnell S, Harmer AR, Van der Esch M, Simic M, Bennell KL. Exercise for osteoarthritis of the knee. Cochrane Database Syst Rev 2015;2015:CD004376. Crossref

22. Silva LE, Valim V, Pessanha AP, et al. Hydrotherapy versus conventional land-based exercise for the management of patients with osteoarthritis of the knee: a randomized clinical trial. Phys Ther 2008;88:12-21. Crossref

23. Chen L, Yao F, Wang T, et al. Horizontal fissuring at the osteochondral interface: a novel and unique pathological feature in patients with obesity-related osteoarthritis. Ann Rheum Dis 2020;79:811-8. Crossref

24. Courties A, Gualillo O, Berenbaum F, Sellam J. Metabolic stress-induced joint inflammation and osteoarthritis. Osteoarthritis Cartilage 2015;23:1955-65. Crossref

25. Puenpatom RA, Victor TW. Increased prevalence of metabolic syndrome in individuals with osteoarthritis: an analysis of NHANES III data. Postgrad Med 2009;121:9-20. Crossref

26. Christensen R, Bartels EM, Astrup A, Bliddal H. Effect of weight reduction in obese patients diagnosed with knee osteoarthritis: a systematic review and meta-analysis. Ann Rheum Dis 2007;66:433-9. Crossref

27. Bliddal H, Leeds AR, Stigsgaard L, Astrup A, Christensen R. Weight loss as treatment for knee osteoarthritis in obese patients: 1-year results from a randomised controlled trial. Ann Rheum Dis 2011;70:1798-803. Crossref

28. Jones A, Silva PG, Silva AC, et al. Impact of cane use on pain, function, general health and energy expenditure during gait in patients with knee osteoarthritis: a randomised controlled trial. Ann Rheum Dis 2012;71:172-9. Crossref

29. Somers TJ, Blumenthal JA, Guilak F, et al. Pain coping skills and lifestyle behavioural weight management in patients with knee osteoarthritis: a randomized controlled study. Pain 2012;153:1199-209. Crossref

30. Hurley MV, Walsh NE, Mitchell HL, et al. Clinical effectiveness of a rehabilitation program integrating exercise, self-management, and active coping strategies for chronic knee pain: a cluster randomized trial. Arthritis Rheum 2007;57:1211-9. Crossref

31. Helminen EE, Sinikallio SH, Valjakka AL, Väisänen-Rouvali RH, Arokoski JP. Effectiveness of a cognitive-behaviour group intervention for knee osteoarthritis pain: a randomized controlled trial. Clin Rehabil 2015;29:868-81. Crossref

32. Schnitzer TJ, Kivitz A, Frayssinet H, Duquesroix B. Efficacy and safety of naproxcinod in the treatment of patients with osteoarthritis of the knee: a 13-week prospective, randomized, multicenter study. Osteoarthritis Cartilage 2010;18:629-39. Crossref

33. Bannuru RR, Osani MC, Vaysbrot EE, et al. OARSI guidelines for the non-surgical management of knee, hip, and polyarticular osteoarthritis. Osteoarthritis Cartilage 2019;27:1578-89. Crossref

34. Kolasinski SL, Neogi T, Hochberg MC, et al. 2019 American College of Rheumatology/Arthritis Foundation guideline for the management of osteoarthritis of the hand, hip, and knee. Arthritis Rheumatol 2020;72:220-3. Crossref

35. Dehghan M, Asgharian S, Khalesi E, Ahmadi A, Lorigooini Z. Comparative study of the effect of Thymus daenensis gel 5% and diclofenac in patients with knee osteoarthritis. Biomedicine (Taipei) 2019;9:9. Crossref

36. Prior MJ, Harrison DD, Frustaci ME. A randomized, double-blind, placebo-controlled 12-week trial of acetaminophen extended release for the treatment of signs and symptoms of osteoarthritis. Curr Med Res Opin 2014;30:2377-87. Crossref

37. Heath EL, Ackerman IN, Cashman K, Lorimer M, Graves SE, Harris IA. Patient-reported outcomes after hip and knee arthroplasty: results from a large national registry. Bone Jt Open 2021;2:422-32. Crossref

38. van Outeren MV, Waarsing JH, Brouwer RW, Verhaar JA, Reijman M, Bierma-Zeinstra SM. Is a high tibial osteotomy (HTO) superior to non-surgical treatment in patients with varus malaligned medial knee osteoarthritis (OA)? A propensity matched study using 2 randomized controlled trial (RCT) datasets. Osteoarthritis Cartilage 2017;25:1988-93. Crossref

39. Wilson HA, Middleton R, Abram SG, et al. Patient relevant outcomes of unicompartmental versus total knee replacement: systematic review and meta-analysis. BMJ 2019;364:I352. Crossref

Prostate cancer is one of the most common cancers in men. In 2020, there were 2315 new cases of prostate cancer diagnosed in Hong Kong, with an age-standardised incidence rate of 30.5 per 100 000 population.1 Globally, >1.4 million new prostate cancer cases and more than 370 000 related deaths were reported in 2020.2 Robot-assisted radical prostatectomy (RARP) is one of the most common procedures used to eradicate localised prostate cancer. However, erectile dysfunction (ED) and urinary incontinence (UI) are common side-effects of RARP.3

The RARP is typically performed using robotic surgical platforms, such as the da Vinci Surgical System4 which facilitates minimally invasive prostatectomy. Keyholes are created through which high-resolution, illuminated cameras and robotic arms are inserted into the peritoneal cavity, which is inflated with carbon dioxide to provide adequate space for surgery. If the tumour is small and likelihood of extracapsular extension (ECE) is low, bilateral or unilateral nerve sparing (NS) may be performed to preserve postoperative erectile5 and lower urinary tract function,6 7 while taking oncological outcomes into consideration. The decision to use an NS technique is made by the surgeon, who carefully assesses the patient’s disease characteristics, drawing on personal experience and current research evidence.8 The prostate is then dissected from the bladder and urethra, and a re-anastomosis is performed between the bladder neck and the urethra.

A meta-analysis of NS techniques in radical prostatectomy (including RARP) has shown that the use of NS techniques results in lower risks of ED and UI at 3- and 12-month follow-ups.9 Nerve sparing cases demonstrate superior erectile function, urinary continence, and oncological outcomes compared with non-NS cases.9 Further analyses indicated that NS is associated with fewer complications than non-NS.10 11 They also suggest that the use of NS techniques does not lead to inferior oncological outcomes.10 11 Therefore, we hypothesised that bilateral NS in RARP improves erectile function and urinary continence after surgery. Our study aimed to investigate the effects of NS RARP on the aforementioned side-effects of ED and UI in Hong Kong and provide suggestions for enhancing patient’s quality of life.

We retrospectively recruited 431 patients who underwent RARP in a university-based teaching hospital (our institution) between January 2018 and April 2023. We retrieved their basic demographics, relevant surgical parameters (NS approach, positive surgical margin [PSM], ECE and Gleason score), postoperative 1-hour pad test results (at 1, 2, 3, 6, and 12 months postoperatively), and pre- and postoperative International Index of Erectile Function–5 (IIEF-5) scores12 at the same time points from electronic medical records. The IIEF-5 assesses erectile function using a 5-point scale across several domains, including erectile function, orgasmic function, sexual desire, intercourse satisfaction, and overall sexual satisfaction.12 All patients attended follow-up at our institution’s urology nurse clinic and received guidance on postoperative management, including pelvic floor strengthening exercises.

The primary outcome of the study was to evaluate the effect of NS in RARP on postoperative ED and UI. Secondary outcomes included correlations between other factors (PSM and ECE) and functional outcomes (ED and UI); correlations between NS and PSM or ECE; and postoperative trends in ED and UI beyond 1 year.

Inclusion criteria comprised all patients who had undergone RARP in our institution with follow-up in our nurse clinic. Treatment via bilateral NS, unilateral NS, or non-NS RARP was performed at the surgeon’s discretion and the patient’s preferences. Exclusion criteria included incomplete data. For the ED outcome, patients with an IIEF-5 score of 7 or below were excluded because this score indicates severe ED,13 and improvement beyond the preoperative baseline was not expected after NS RARP. Patient selection flowcharts for ED and UI are provided in online supplementary Figures 1 and 2, respectively.

R (R Foundation for Statistical Computing, Vienna, Austria) and RStudio software were used for data analysis.14 All statistical tests were two-sided and incorporated a 5% significance threshold. Cases with missing data due to loss to follow-up were excluded from analysis. Patients selected for ED and UI analysis had comparable age profiles (online supplementary Tables 1 and 2, respectively). Categorical variables were analysed using Chi squared tests or Fisher’s exact tests, depending on the observed frequencies. Continuous variables were analysed using independent sample t tests and Pearson correlations were utilised.

Among the 431 eligible patients included in the analysis, the mean age was 67.67 years. Regarding ED, the mean ages (±standard deviation) in the non-NS, unilateral NS, and bilateral NS groups were 65.20±4.95, 64.84±6.51, and 66.64±5.35 years, respectively, with no statistically significant differences observed (online supplementary Table 3). Concerning UI, the mean ages (±standard deviation) in the non-NS, unilateral NS, and bilateral NS groups were 66.82±5.48, 67.90±5.49, and 67.94±5.14 years, respectively, with no statistically significant differences observed (online supplementary Table 4). The mean tumour percentage was 12.85% of the total prostate volume and the mean resected prostate volume was 54.39 g. The distribution of pathological Gleason scores is shown in online supplementary Table 5. The majority of patients had a Gleason score of 7: 41.0% had a score of 3+4 (International Society of Urological Pathology [ISUP] Grade Group 2), and 23.9% had a score of 4+3 (ISUP Grade Group 3). Patients with low to intermediate risk prostate cancer, based on the National Comprehensive Cancer Network risk classification, were more likely to undergo bilateral NS operations (Table 1). Further details of NS approaches are presented in Table 1.

The mean preoperative IIEF-5 score was 10.22 (n=75; online supplementary Fig 1). For the primary outcome, patients with bilateral NS had a higher mean postoperative IIEF-5 score than those without NS at 2 months (non-NS vs bilateral NS=3.19 vs 7.60, t=-2.35; P=0.037). Bilateral NS patients also had a higher mean postoperative IIEF-5 score than unilateral NS patients at 2 months (2.50 vs 7.60, t=-2.69; P=0.020) and 3 months (2.06 vs 7.40, t=-2.61; P=0.027) [Table 2]. Differences in IIEF-5 scores at 1, 6, and 12 months postoperatively were not significant among any of the groups (Table 2). Concerning the secondary outcome, younger age was associated with a higher postoperative IIEF-5 score among non-NS patients (m=-0.42; P=0.024).

With respect to postoperative penile rehabilitation, the use of phosphodiesterase type 5 inhibitors (PDE5i) was reported by 38.3%, 77.8% and 60.0% in non-NS, unilateral NS, and bilateral NS patients, respectively. Among those who did not take PDE5i, reasons included financial considerations—patients must pay out of pocket for PDE5i in Hong Kong, and a perceived lack of efficacy. Even among those who utilised PDE5i, only 38.9%, 35.7%, and 66.7% of non-NS, unilateral NS, and bilateral NS patients, respectively, reported subjective improvement in erectile function. Objective changes in IIEF-5 scores (difference between preoperative and 12-month postoperative scores) were -10.20, -17.03, and -7.67 in non-NS, unilateral NS, and bilateral NS patients, respectively. Further details can be found in online supplementary Tables 6 to 8.

After initial screening and the exclusion of records with missing follow-up data, 264 patients were included in the analysis of UI following RARP (online supplementary Fig 2).

For the primary outcome, patients with bilateral NS had lower mean urinary leakage volume in the 1-hour pad test relative to patients without NS after 1 month (non-NS vs bilateral NS=49.44 g vs 16.40 g, t=3.92; P<0.001) and 2 months (35.45 g vs 13.60 g, t=2.67; P=0.009). Patients with bilateral NS also had lower mean urinary leakage volume than unilateral NS patients after 1 month (50.82 g vs 16.40 g, t=2.61; P=0.01). Differences in UI at 3, 6, and 12 months were not significant among the groups (Table 3). Further details comparing non-NS and unilateral NS can be found in online supplementary Tables 9 and 10.

Tumour recurrence is an adverse surgical outcome of RARP that requires further oncological management. We identified patients who underwent adjuvant radiotherapy, salvage radiotherapy, or experienced cancer-related death, then stratified them according to NS group, adjusted for age and total Gleason score. Statistical analysis showed no significant differences in oncological outcomes between non-NS and bilateral NS patients (odds ratio [OR]=0.75, 95% confidence interval [95% CI]=0.39-1.27; P=0.321), as well as unilateral NS and bilateral NS patients (OR=0.78, 95% CI=0.18-2.82; P=0.720). These findings indicated that bilateral NS was neither superior nor inferior in oncological outcomes compared with unilateral and non-NS groups, consistent with literature reports.15 16

Older patients had lower postoperative IIEF-5 scores at 6 months (Pearson correlation=-0.18; P=0.013) and 12 months (Pearson correlation=-0.22; P=0.014). However, the correlations were not statistically significant at 1, 2, or 3 months postoperatively. There also was no statistically significant correlation between age and postoperative UI.

Patients who underwent non-NS RARP were more likely to have ≥1 positive surgical margin (Chi squared=4.2673, P=0.039, OR=0.46; 95% CI=0.22-0.91). This result was likely attributable to disease-related factors. A larger proportion of patients in the non-NS group had a higher National Comprehensive Cancer Network risk score, indicating more aggressive tumours. Distributions of patients’ NS status, tumour grading, and PSMs are shown in Table 1.

Comparisons of postoperative IIEF-5 scores among NS groups revealed significant differences. Patients with bilateral NS exhibited higher postoperative IIEF-5 scores than those without NS at 2 months, highlighting the positive impact of bilateral NS on early erectile function recovery. Similar trends were observed when comparing bilateral NS with unilateral NS at both 2 and 3 months postoperatively. However, no significant differences in IIEF-5 scores were noted at 6 or 12 months, suggesting a convergence of outcomes beyond the initial recovery phase. A meta-analysis Nguyen et al17 on NS techniques in radical prostatectomy (including RARP) showed that NS cases had lower risks of ED at 3 and 12 months (risk ratio [RR] at 3 months=0.77; 95% CI=0.70-0.85; RR at 12 months=0.53; 95% CI=0.39-0.71). Some differences between our study (Table 2) and that of Nguyen et al17 might be attributable to the small sample size and loss to follow-up in our cohort, which may have introduced selection bias.

Our study also demonstrated that the bilateral NS technique was linked to better erectile function outcomes than unilateral NS, consistent with previous findings. In the study by Berg et al,18 the proportion of patients who were alive, continent, and potent was significantly greater among those with bilateral NS (67.6%) compared to unilateral NS (31.3%) [P<0.001]. Other studies also showed that bilateral NS was associated with lower risks of ED at 3 months (RR=0.80; 95% CI=0.70-0.90) and 1 year (RR=0.80; 95% CI=0.72-0.88) relative to unilateral NS.17

However, our study did not identify statistically significant benefits from unilateral NS in terms of ED and UI compared with non-NS. This may be due to a combination of factors, including variation in surgeons’ techniques for unilateral NS, small sample size, and relatively small absolute differences in outcomes between the unilateral and non-NS groups.

Further analysis also found that PDE5i therapy yielded only modest improvements in erectile function among individuals who underwent non-NS or unilateral NS procedures. Changes in IIEF-5 scores from preoperative to 12 months postoperative after PDE5i use were -10.20, -17.03 and -7.67 in non-NS, unilateral NS, and bilateral NS patients, respectively. Although approximately two-thirds of bilateral NS patients experienced improved erectile function, the limited number of patients who received PDE5i warrants caution when interpreting effects in this subgroup.

The relationship between age and erectile function in patients without NS was notable. Younger age was associated with higher postoperative IIEF-5 scores, emphasising the potential influence of age on postoperative erectile function recovery, consistent with international literature.19 Among older patients, postoperative ED remains an important complication. Thus, age is a key consideration when balancing quality of life and oncological control in planning surgical approaches. Surgeons should thoroughly discuss potential side-effects of ED with older patients who have concerns about sexual function prior to surgery involving NS.

In terms of UI, patients with bilateral NS demonstrated lower urinary leakage volumes in the 1-hour pad test compared with those without NS at 1 and 2 months postoperatively. Similarly, bilateral NS was associated with lower urinary leakage volumes than unilateral NS at 1 month. These findings suggest that NS techniques enhance short-term UI recovery, although the differences tend to diminish over time. The benefit of NS on long-term UI recovery beyond 1 year remains uncertain.9 17 20 Choi et al20 reported findings similar to ours, indicating that bilateral NS was associated with a higher continence rate than non-NS at 4 months (P=0.043), although the differences at 12 and 24 months were not significant. Conversely, a meta-analysis by Nguyen et al17 on NS techniques in radical prostatectomy (including RARP) found that NS was associated with lower risks of UI at both 3 months (RR =0.75, 95% CI=0.65-0.85) and 12 months (RR=0.61, 95% CI=0.44-0.84). These discrepancies may be attributed to differences in study design, sample size, and baseline patient characteristics.

Other studies have also shown that bilateral NS is associated with a lower risk of UI at 1 year (RR=0.70, 95% CI=0.50-0.98) and lower risks of ED at 3 months (RR=0.80, 95% CI=0.70-0.90) and 1 year (RR=0.80, 95% CI=0.72-0.88) compared with unilateral NS.17 In our study, while we observed a lower risk of UI at 1 and 2 months after surgery, this difference was not statistically significant at 3 months.

Our findings have implications for surgical decision-making and patient counselling. Surgeons should consider the potential benefits of bilateral NS for early postoperative recovery of erectile function and urinary continence.21 Although our study did not demonstrate statistically significant long-term differences in these outcomes, bilateral NS has been reported to play a key role in early recovery. Contributing to improved quality of life and patient satisfaction.21 Moreover, age should be considered when evaluating erectile function outcomes in patients undergoing non-NS RARP.

Future research could investigate the long-term trajectories of erectile function and urinary continence, exploring factors that contribute to outcome convergence over time. Additionally, efforts to evaluate the impact of NS techniques on quality of life and patient satisfaction could offer a more comprehensive understanding of the clinical implications of these findings.

We also intend to further examine why patients with unilateral NS did not demonstrate better erectile function outcomes than those in the non-NS group. Notably, there was a significant difference in outcomes between the bilateral NS and non-NS groups, despite the small number of patients with bilateral NS (n=10). Therefore, we plan to conduct a detailed review of surgical records for patients with unilateral NS to determine why erectile function outcomes were not superior to those in the non-NS group.

This study also highlights the need for better public and patient education regarding sexual health. In Hong Kong, sexual function currently remains a major taboo topic among older individuals. Many are reluctant to discuss ED and are even more reluctant to seek medical treatment. In our study, 121 patients displayed severe preoperative ED, but none had sought medical attention. This problem is compounded by the financial barriers to treatment. In public hospitals, PDE5i is entirely self-financed, and even government employees are unable to reclaim their costs. These factors collectively create a substantial barrier for older men in Hong Kong to recognise ED as a treatable clinical condition that could improve their sexual health.

To our knowledge, this is the first local retrospective cohort study in Hong Kong comparing the efficacy of NS approaches in RARP for reducing ED and UI. These findings provide valuable insights into current local RARP practices and serve as a foundation for future prospective studies.

However, there were several limitations, including the retrospective design, potential selection biases, single-centre setting, and inconsistency in follow-up intervals, as some patients might have attended follow-up appointments earlier or later than scheduled due to personal reasons. These factors could affect the interpretation and generalisability of the results. Furthermore, the IIEF-5 score used in this study was based on patients’ subjective self-assessment and may not accurately reflect changes in erectile function, particularly among patients without preoperative sexual activity.22

The bilateral NS technique during prostatectomy demonstrated a significant positive impact on the recovery of erectile function and urinary continence within the first 6 months postoperatively, without compromising oncological outcomes. However, the extent of this benefit appears to diminish over time, indicating the need for longer-term assessment. These findings contribute valuable insights into the role of NS in prostate cancer surgery and may inform clinical decision-making in prostate cancer management. To validate and expand upon these observations, further prospective, randomised studies with extended follow-up are warranted.

Concept or design: OWK Tsui, S Chun, BSH Ho.
Acquisition of data: OWK Tsui, KCH Shing.
Analysis or interpretation of data: OWK Tsui, KCH Shing, APM Lam, S Chun.
Drafting of the manuscript: OWK Tsui.
Critical revision of the manuscript for important intellectual content: All authors.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

As an editor of the journal, BSH Ho was not involved in the peer review process. Other authors have disclosed no conflicts of interest.

This research was presented at The Hong Kong Urological Association 28th Annual Scientific Meeting held in Hong Kong on 19 November 2023.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

This research was approved by the Institutional Review Board of The University of Hong Kong/Hospital Authority Hong Kong West Cluster, Hong Kong (Ref No.: UW-24-099). The requirement for informed patient consent was waived by the Board due to the retrospective nature of the research.

The supplementary material was provided by the authors and some information may not have been peer reviewed. Accepted supplementary material will be published as submitted by the authors, without any editing or formatting. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by the Hong Kong Academy of Medicine and the Hong Kong Medical Association. The Hong Kong Academy of Medicine and the Hong Kong Medical Association disclaim all liability and responsibility arising from any reliance placed on the content.

1. Centre for Health Protection, Department of Health, Hong Kong SAR Government. Prostate cancer. 2025 Feb 24. Available from: https://www.chp.gov.hk/en/healthtopics/content/25/5781.html#:~:text=Prostate%20cancer%20was%20the%20fourth,per%20100%20000%20male%20population. Accessed 24 Feb 2025.

2. Wang L, Lu B, He M, Wang Y, Wang Z, Du L. Prostate cancer incidence and mortality: global status and temporal trends in 89 countries from 2000 to 2019. Front Public Health 2022;10:811044. Crossref

3. Kesch C, Heidegger I, Kasivisvanathan V, et al. Radical prostatectomy: sequelae in the course of time. Front Surg 2021;8:684088. Crossref

4. Deligiannis D, Anastasiou I, Mygdalis V, Fragkiadis E, Stravodimos K. Change of practice patterns in urology with the introduction of the Da Vinci surgical system: the Greek NHS experience in debt crisis era. Arch Ital Urol Androl 2015;87:56-61. Crossref

5. Harris CR, Punnen S, Carroll PR. Men with low preoperative sexual function may benefit from nerve sparing radical prostatectomy. J Urol 2013;190:981-6. Crossref

6. Kübler HR, Tseng TY, Sun L, Vieweg J, Harris MJ, Dahm P. Impact of nerve sparing technique on patient self-assessed outcomes after radical perineal prostatectomy. J Urol 2007;178:488-92; discussion 492. Crossref

7. Avulova S, Zhao Z, Lee D, et al. The effect of nerve sparing status on sexual and urinary function: 3-year results from the CEASAR study. J Urol 2018;199:1202-9. Crossref

8. Vis AN, van den Bergh RC, van der Poel HG, et al. Selection of patients for nerve sparing surgery in robot-assisted radical prostatectomy. BJUI Compass 2021;3:6-18. Crossref

9. Liu Y, Deng XZ, Qin J, et al. Erectile function, urinary continence and oncologic outcomes of neurovascular bundle sparing robot-assisted radical prostatectomy for high-risk prostate cancer: a systematic review and meta-analysis. Front Oncol 2023;13:1161544. Crossref

10. Furrer MA, Sathianathen N, Gahl B, et al. Oncological outcomes after attempted nerve-sparing radical prostatectomy (NSRP) in patients with high-risk prostate cancer are comparable to standard non-NSRP: a longitudinal long-term propensity-matched single-centre study. BJU Int 2024;133:53-62. Crossref

11. Takahara K, Sumitomo M, Fukaya K, et al. Clinical and oncological outcomes of robot-assisted radical prostatectomy with nerve sparing vs. non-nerve sparing for high-risk prostate cancer cases. Oncol Lett 2019;18:3896-902. Crossref

12. Díaz-Mohedo E, Meldaña Sánchez A, Cabello Santamaría F, Molina García E, Hernández Hernández S, Hita-Contreras F. The Spanish version of the International Index of Erectile Function: adaptation and validation. Int J Environ Res Public Health 2023;20:1830. Crossref

13. Yin Y, Wang K, Xu Y, et al. The impact of using donor sperm after ICSI failure in severe oligozoospermia on male mental health and erectile function. J Multidiscip Healthc 2024;17:21-8. Crossref

14. Gasparini A. comoRbidity: an R package for computing comorbidity scores. J Open Source Softw 2018;3:648. Crossref

15. Kumar A, Samavedi S, Bates AS, et al. Safety of selective nerve sparing in high-risk prostate cancer during robotassisted radical prostatectomy. J Robot Surg 2017;11:129-38. Crossref

16. Yuh B, Artibani W, Heidenreich A, et al. The role of robot-assisted radical prostatectomy and pelvic lymph node dissection in the management of high-risk prostate cancer: a systematic review. Eur Urol 2014;65:918-27. Crossref

17. Nguyen LN, Head L, Witiuk K, et al. The risks and benefits of cavernous neurovascular bundle sparing during radical prostatectomy: a systematic review and meta-analysis. J Urol 2017;198:760-9. Crossref

18. Berg KD, Thomsen FB, Hvarness H, Christensen IJ, Iversen P. Early biochemical recurrence, urinary continence and potency outcomes following robot-assisted radical prostatectomy. Scand J Urol 2014;48:356-66. Crossref

19. Tal R, Alphs HH, Krebs P, Nelson CJ, Mulhall JP. Erectile function recovery rate after radical prostatectomy: a metaanalysis. J Sex Med 2009;6:2538-46. Crossref

20. Choi WW, Freire MP, Soukup JR, et al. Nerve-sparing technique and urinary control after robot-assisted laparoscopic prostatectomy. World J Urol 2011;29:21-7. Crossref

21. Ngoo KS, Honda M, Kimura Y, et al. Longitudinal study on the impact of urinary continence and sexual function on health-related quality of life among Japanese men after robot-assisted radical prostatectomy. Int J Med Robot 2019;15:e2018. Crossref

22. Tang Z, Li D, Zhang X, et al. Comparison of the simplified International Index of Erectile Function (IIEF-5) in patients of erectile dysfunction with different pathophysiologies. BMC Urol 2014;14:52. Crossref

Fragile X syndrome (FXS; OMIM #300624), an X-linked dominant condition, is one of the most common inherited causes of intellectual disability (ID)1 2 3 and autism spectrum disorder (ASD).2 3 4 5 The prevalence of FXS is most widely regarded as 1 in 4000 for males and 1 in 8000 for females.6 7 8 9 Fragile X syndrome is within the spectrum of FMR1-related disorders,10 caused by pathogenic variants in the FMR1 (fragile X messenger ribonucleoprotein 1) gene (OMIM #309550) mapped to the chromosome Xq27.3 region, which encodes the fragile X mental retardation protein.

Fragile X syndrome is the first genetic disorder known to be caused by trinucleotide repeat expansions—specifically, cytosine-guanine-guanine (CGG) repeats in the 5’ untranslated region of the FMR1 gene. FMR1 alleles are categorised as normal (<45), intermediate (45-54), premutation (PM, 55-200), and full mutation (FM, >200) based on repeat size. Premutation alleles are associated with elevated levels of FMR1 messenger ribonucleic acid,10 leading to ribonucleic acid toxicity that can result in fragile X–associated tremor/ataxia syndrome, fragile X–associated primary ovarian insufficiency, or fragile X–associated neuropsychiatric disorders.10 Conversely, FXS typically results from FM with promoter region hypermethylation and histone protein deacetylation,11 12 causing transcriptional silencing.13 14 Most individuals inherit the FM from their mothers, who are PM carriers. Stability upon maternal transmission depends on the size of the PM.15

Characteristic signs of FXS, including prominent ears, elongated face, protruding ears, and macroorchidism, tend to evolve with age.1 4 Facial dysmorphism can vary depending on ethnic background,4 and females exhibit greater clinical variability.16 17 Most patients are not diagnosed until the age of 3 years.18 19 Fragile X syndrome is also associated with multiple medical co-morbidities, such as recurrent otitis media, mitral valve prolapse, and connective tissue problems.3

Clinical presentation can be further complicated by either size mosaicism or methylation mosaicism.20 Size mosaicism refers to cell populations with variably sized CGG repeats—typically the presence of PM or intermediate/normal alleles in addition to FMs. Methylation mosaicism involves both methylated and unmethylated cell populations at the FMR1 locus. Mosaicism in males with FXS has been reported in 12% to 41% of cases.21 22 23

While the epidemiology of FXS has been extensively studied in Western populations,6 7 8 9 the reported prevalence of FXS among Chinese patients with developmental delay or ID showed variability (ranging from 0.43% to 12.9%).24 25 Furthermore, the prevalence of medical co-morbidities remains understudied in the Chinese population.

In this single-centre retrospective study, we aimed to: (1) review the clinical features of FXS patients referred to the Department of Clinical Genetics of the Hospital Authority (formerly the Clinical Genetic Service of the Department of Health); (2) evaluate parameters regarding growth, medical co-morbidities, and neurobehavioural features in the Hong Kong Chinese patient population with FXS; (3) assess the diagnostic yield of FXS testing in patients with unexplained developmental delay or ID; and (4) review the diagnostic journey of such patients.

Neurodevelopmental delay, ID, or ASD are the main reasons for ordering FXS testing. Over the 30-year period from 1993 to 2022, 7291 patients referred for such neurodevelopmental conditions underwent FXS molecular testing after clinical genetic evaluation. Maternal testing and further cascade testing were considered upon diagnosis in index cases.

Patients with FMR1 FMs were included in the initial analysis, and a retrospective chart review of printed and electronic records was performed. For analysis of clinical features among Chinese patients with FXS, individuals who self-identified as non-Chinese or had co-existing copy number variants or chromosomal structural abnormalities were excluded.

Genomic DNA was extracted from peripheral blood leucocytes using standardised methods, in accordance with the manufacturer’s instructions. Prior to 2014, polymerase chain reaction (PCR) followed by Southern blot analysis was used to identify individuals with FXS. This approach was subsequently replaced by conventional PCR that can detect (CGG)_n alleles up to 90 repeats, followed by triplet-primed PCR and methylation-specific PCR using the AmplideX kit (Asuragen, Austin [TX], US), if necessary.

Baseline demographic characteristics were descriptively summarised. Continuous variables were reported as means and standard errors for normally distributed data, and as medians and ranges/interquartile ranges (IQRs) for non-parametrically distributed data. To assess the association between age at diagnosis and year of assessment, correlation analysis was performed using the Pearson correlation coefficient (r), with a statistical significance threshold of 5%. Prevalence proportions were used to evaluate categorical clinical characteristics. Comparisons between males and females were made using the Chi squared test or Fisher’s exact test. Statistical analysis was performed using SPSS (Windows version 26.0; IBM Corp, Armonk [NY], US).

Overall, 103 individuals from 61 families were confirmed to have an FM in the FMR1 gene. Index cases were defined as patients referred from their parent institution for their condition. Among the index cases, eight individuals came from four families, with two affected members referred separately in each family. In six other families, the consultand was an unaffected member referred due to a positive family history. Family screening identified 44 additional cases in 29 families, comprising 13 males (29.5%) and 31 females (70.5%) [Table 1].

Details of family history for 55 unrelated index cases and six consultands are presented in Table 2. Overall, 41 (67.2%) had a positive family history in one or more aspects.

Of 7291 patients underwent testing, 59 index cases were identified, yielding an overall diagnostic rate of 0.8%. The sex-specific diagnostic yields were 1.0% for males and 0.3% for females. Additionally, one male and one female patient had PMs. There was an upward trend in the number of FXS tests performed (unpublished data). The median ages at diagnosis were 6.73 years (range, 1.17-52.36) among all FXS patients (including those identified through family screening) and 4.10 years (range, 1.72-26.95) when considering index cases alone. The median diagnostic lag time for index cases, defined as the time elapsed between referral and diagnosis, was 11.0 months (IQR=6.53-20.0, n=54).

The temporal trends in diagnosis are shown in Table 3. A weak negative correlation between age and assessment year was observed for all cases (r=-0.267, n=103; P=0.006). Regarding index cases, a moderate negative correlation was observed (r=-0.396, n=59; P=0.0019), suggesting a trend towards earlier diagnosis over time.

The mosaicism statuses of our patients are summarised in Table 4.

Among the 61 families, 54 underwent maternal testing—44 were PM carriers and 10 were FM carriers. Cascade testing was conducted in other family members in 45 families (73.8%). Twenty siblings were identified as affected individuals, and maternal second-/third-degree relatives constituted another 13 cases. In 27 families (44.2%), more than one FM carrier was identified—15 families (24.6%) had two affected members, nine (14.8%) had three affected members, and three (4.9%) had four affected members. Nonetheless, 16 families (26.2%) did not proceed with further cascade testing after maternal testing. Four families (6.6%) did not undergo any family testing at all.

Prenatal diagnosis was arranged for 11 families (18%), involving 10 PM carriers and two FM carriers. Two male fetuses were affected by FM, and these pregnancies were terminated. One FM carrier opted for termination of pregnancy at 10 weeks of gestation despite counselling regarding the availability of prenatal diagnosis.

Seventy Chinese patients with FMR1 FM from 55 different families were included in the analysis of clinical features (Fig); details are summarised in Table 5.

The presence and severity of ID, co-morbid ASD, or attention-deficit/hyperactivity disorder were determined based on clinician reports. More than half of the male patients (54.0%) had ID of moderate or greater severity. None of the female patients had severe ID; three females had borderline intelligence not supporting a diagnosis of ID.

Epilepsy was diagnosed in 12 patients (17.1%). One 10-year-old boy with refractory epilepsy had high-risk medulloblastoma and completed treatment at age 6 years. He developed spasm-like attacks and possible focal seizures at age 7 years. Among the remaining patients, eight had generalised seizures, two had a mixed semiology of generalised and focal seizures, and one patient had unclear seizure semiology. The age at seizure onset ranged from 2 to 19 years, with a median of 7.0 years (IQR=3.75-8.0). Three patients experienced convulsive status epilepticus triggered by infective episodes, one required intensive care unit admission.

Forty-three patients (61.4%) underwent neuroimaging (magnetic resonance imaging/computed tomography of the brain), and most results were unremarkable.

Eight patients (five males and three females) with mosaicism were eligible for analysis of clinical features after excluding individuals with inadequate data. These patients generally had less severe ID than non-mosaic patients, although proper comparison was hindered by the small sample size.

Gastrointestinal conditions and sleep problems were common co-morbidities, affecting 27.1% and 31.4% of patients, respectively. Seven patients underwent echocardiography at least once; two displayed transient aortic root dilatation. Congenital anomalies identified among our patients included Pierre Robin sequence, Klippel-Trenaunay syndrome, hemifacial asymmetry, microtia, and pigmentary mosaicism. These conditions were relatively rare in the literature.

Approximately 20% of our patients developed obesity in childhood or adolescence, which aligns with the general childhood overweight/obesity prevalence in Hong Kong (~20%).26 However, a US study27 examining 848 families with at least one child had FXS showed that 31% of male and 15% of female children were obese. With respect to obesity alone, the frequency may be higher among our patients than in the general population, which may be attributed to physical inactivity in individuals with ID, as well as the use of psychiatric medications.

Five male patients (10.0%) and one female patient (5.0%) exhibited macrocephaly, and a few had suspected overgrowth syndrome upon referral. A subset of FXS patients has been reported to present with Sotos- or Prader-Willi–like phenotypes.16 This feature may pose a diagnostic challenge.

In our study, the frequency of developmental delay and ID was consistent with findings in other populations. Female patients displayed milder phenotypes, which is compatible with the presentation of X-linked disorders. Additionally, 48.6% of patients had a clinician-reported diagnosis of ASD. The reported prevalence of co-morbid ASD in males with FXS varies widely across studies, from 30% to 60%.3 28 The use of different instruments has been reported to cause diagnostic inconsistency; this is further complicated by the intrinsic difficulty in diagnosing ASD among individuals with ID. The frequencies of hyperactivity and attention-deficit/hyperactivity disorder in our study are similar to rates in the literature (50%-60% and 12%-23%, respectively),29 but smaller percentages of our patients displayed inattention, anxiety problems, or depression compared to the literature (74%-84% for inattention, 58%-86% for anxiety problems, and 8%-12% for depression).29 The lower rates of such conditions in our study may be due to diagnostic overshadowing. Active research is underway to identify more accurate diagnostic measures for neurobehavioural co-morbidities.28

Overall, 17.1% of our patients displayed epilepsy, with a predilection towards generalised seizures. This is in agreement with the work of Berry-Kravis et al,30 who characterised seizures in the largest evaluated cohort of FXS patients, although earlier case series suggested that focal onset seizures with impaired awareness were the most common semiology.30 Notably, three patients presented with convulsive status epilepticus, which is uncommon among FXS patients.

The presence of co-morbidities such as gastrointestinal problems, sleep disturbances, joint laxity, and pes planus was consistent with commonly observed clinical patterns in individuals with FXS. Nonetheless, only a small percentage of patients in our cohort showed strabismus or refractive errors, scoliosis, or recurrent otitis media; none exhibited joint dislocations or mitral valve prolapse (Table 5). The true prevalence of mitral valve prolapse remains unclear. Loehr et al31 reported a prevalence as high as 55% in a series of FXS patients in 1986, whereas Kidd et al3 reported a prevalence of 0.8%; some Asian studies32 33 did not identify any individuals with mitral valve prolapse.

A systematic approach to health supervision for FXS has been recommended by the American Academy of Pediatrics1 28 across developmental stages. To our knowledge, there are no established surveillance guidelines in Hong Kong. Ultimately, FXS is more than a purely neurodevelopmental disorder; it is important to be aware of potential multisystemic approach and provide health supervision as needed.

Our diagnostic yield of 0.8% is consistent with a local study in 1999,34 which showed a diagnostic yield of 0.6% among 324 patients with mild ID of unspecified cause, and with a study by Chen et al (0.93%)35 that evaluated the diagnostic yield of FXS testing in 553 unrelated patients with moderate to severe ID of unknown cause in Beijing in 2015. Nonetheless, our yield is slightly lower than those reported by Mei et al (2.4%)32 and Zhong et al (2.8%),36 which were derived from relatively large-scale studies conducted in Chinese populations. Our results also revealed a slightly lower diagnostic yield compared with that of Western literature, which is around 1.5% to 2%.37 This may be explained by reported differences in the distribution of normal, PM, and FM alleles between Asian and non-Asian populations. Various studies have identified a lower prevalence of PM alleles in East Asians compared with Western populations. One study reported that the prevalence of PM and asymptomatic FM carriers in the Hong Kong Chinese pregnant population was 1 in 883,38 whereas another study showed a prevalence of 1 in 1113 among unaffected Chinese individuals.39 The reported prevalence of PM alleles in Western populations varies from 1 in 113 to 1 in 382, depending on ethnicity.39 Intriguingly, most FMR1 alleles contain 29 or 30 CGG repeats across different populations, including ours. Alternatively, the apparent difference in PM allele prevalence may be explained by the founder haplotype hypothesis, whereby various factors contribute to disparate rates of normal-to-PM transitions, including different AGG interruption patterns across populations.40 Although preliminary studies have explored an association between neurodevelopmental difficulties and PM status, findings have been inconclusive. In our cohort, only two patients referred for developmental delay exhibited PM status.

Our study showed a weak but statistically significant trend towards an earlier age at diagnosis, which may be attributed to increased awareness of children’s developmental needs and, consequently, an earlier age at referral. The median age at diagnosis was 4.1 years for index cases alone, and 6.73 years for all cases in our study. These values are comparable to international data where the average age at diagnosis ranges from 2.9 to 6.3 years.18 33

There has been debate regarding whether FXS testing should be utilised as a first-line investigation to evaluate developmental delay. However, it is a simple and inexpensive test with a short turnaround time. The availability of such a test is crucial because it aids in prompt diagnosis, facilitating further cascade testing and reproductive planning. In our study, 44.2% of families had more than one affected member. Ten female PM carriers and two FM carriers from 11 families (18%) underwent prenatal diagnosis; two pregnancies were terminated after identification of FXS status. A diagnosis in one family member may influence others’ decisions regarding pregnancy and subsequently affect pregnancy outcomes. Fragile X PM carrier screening is recommended by organisations such as the American College of Obstetricians and Gynecologists41 and the American College of Medical Genetics and Genomics42 for women with a family history suggestive of fragile X—related disorders who are either considering pregnancy or currently pregnant. Although prenatal carrier testing is free for women of childbearing age in some countries, it is currently self-financed in Hong Kong and thus not widely implemented.

An expedited diagnosis can facilitate the timely implementation of medical interventions. For PM carriers who exhibit increased risks of fragile X—associated primary ovarian insufficiency and fragile X—associated tremor/ataxia syndrome, anticipatory guidance and timely referrals can be provided. Furthermore, multiple targeted therapeutic agents with the potential to reverse some neurobiological aspects of the disorder (eg, mavoglurant, metformin, cannabidiol transdermal gel, acamprosate, and lovastatin) are undergoing active evaluation. Should any of these candidates be approved in the future, early diagnosis would prove even more beneficial.

To our knowledge, this is the largest cohort of Chinese FXS patients reported to date. Because most FXS testing was performed at our centre, potential disease prevalence can be inferred. Our study offers a longitudinal perspective regarding the disease course and highlights areas for improvement in health supervision and management. Furthermore, we examined the landscape of cascade screening and prenatal diagnosis in our specific cultural setting.

However, this was a retrospective study and thus largely dependent on clinician-reported findings. The diagnostic yield may have been influenced by the secular trend of an increasing number of referrals for developmental delay. Furthermore, it was difficult to implement standardised diagnostic instruments for certain co-morbidities. Some patients had inadequate information or were lost to follow-up in the public sector. Finally, the lack of a standardised surveillance protocol for FXS contributed to potential confirmation bias.

In our study, we explored the diagnostic yield of FXS testing, as well as cascade testing and prenatal diagnosis in families with FXS in Hong Kong. Our study provides insights into the clinical features and co-morbidities of FXS in the largest cohort of Chinese patients reported to date. There has been improved awareness of children’s developmental needs, as demonstrated by a trend towards earlier diagnosis, but no local surveillance protocols exist for patients with FXS. The high prevalences of neurobehavioural and medical co-morbidities highlight the need for prompt diagnosis and structured health management. We recommend increased awareness of the multisystemic approach and targeted treatments currently under investigation, and we propose establishing rare disease registries to facilitate this process.

Considering the clinical utility of FXS testing in clinical and reproductive management, we believe it should continue to be included in the evaluation of patients with developmental delay or ID; its role in the diagnostic pathway should be determined by local resources.

Concept or design: CWM Au, HM Luk, IFM Lo.
Acquisition of data: CWM Au, S Ho.
Analysis or interpretation of data: CWM Au.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: All authors.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

All authors have disclosed no conflicts of interest.

The authors thank the patients and their families for contributing the clinical data used in this study.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

This research was approved by the Central Institutional Review Board of Hospital Authority, Hong Kong (Ref No.: PAED-2023-061). A waiver of informed patient consent was obtained from the Board due to the retrospective nature of the research.

1. Hersh JH, Saul RA; Committee on Genetics. Health supervision for children with fragile X syndrome. Pediatrics 2011;127:994-1006. Crossref

2. Jin X, Chen L. Fragile X syndrome as a rare disease in China—therapeutic challenges and opportunities. Intractable Rare Dis Res 2015;4:39-48. Crossref

3. Kidd SA, Lachiewicz A, Barbouth D, et al. Fragile X syndrome: a review of associated medical problems. Pediatrics 2014;134:995-1005. Crossref

4. Lubala TK, Lumaka A, Kanteng G, et al. Fragile X checklists: a meta-analysis and development of a simplified universal clinical checklist. Mol Genet Genomic Med 2018;6:526-32. Crossref

5. Tassone F, Pan R, Amiri K, Taylor AK, Hagerman PJ. A rapid polymerase chain reaction–based screening method for identification of all expanded alleles of the fragile X (FMR1) gene in newborn and high-risk populations. J Mol Diagn 2008;10:43-9. Crossref

6. Coffee B, Keith K, Albizua I, et al. Incidence of fragile X syndrome by newborn screening for methylated FMR1 DNA. Am J Hum Genet 2009;85:503-14. Crossref

7. Lévesque S, Dombrowski C, Morel ML, et al. Screening and instability of FMR1 alleles in a prospective sample of 24,449 mother–newborn pairs from the general population. Clin Genet 2009;76:511-23. Crossref

8. Monaghan KG, Lyon E, Spector EB. ACMG Standards and Guidelines for fragile X testing: a revision to the disease-specific supplements to the Standards and Guidelines for Clinical Genetics Laboratories of the American College of Medical Genetics and Genomics. Genet Med 2013;15:575-86. Crossref

9. Peprah E. Fragile X syndrome: the FMR1 CGG repeat distribution among world populations. Ann Hum Genet 2012;76:178-91. Crossref

10. Spector E, Behlmann A, Kronquist K, et al. Laboratory testing for fragile X, 2021 revision: a technical standard of the American College of Medical Genetics and Genomics (ACMG). Genet Med 2021;23:799-812. Crossref

11. Coffee B, Zhang F, Warren ST, Reines D. Acetylated histones are associated with FMR1 in normal but not fragile X–syndrome cells. Nat Genet 1999;22:98-101. Crossref

12. Crawford DC, Acuña JM, Sherman SL. FMR1 and the fragile X syndrome: human genome epidemiology review. Genet Med 2001;3:359-71. Crossref

13. Chiurazzi P, Pomponi MG, Willemsen R, Oostra BA, Neri G. In vitro reactivation of the FMR1 gene involved in fragile X syndrome. Hum Mol Genet 1998;7:109-13. Crossref

14. Sutcliffe JS, Nelson DL, Zhang F, et al. DNA methylation represses FMR-1 transcription in fragile X syndrome. Hum Mol Genet 1992;1:397-400. Crossref

15. Nolin SL, Brown WT, Glicksman A, et al. Expansion of the fragile X CGG repeat in females with premutation or intermediate alleles. Am J Hum Genet 2003;72:454-64. Crossref

16. de Vries BB, Halley DJ, Oostra BA, Niermeijer MF. The fragile X syndrome. J Med Genet 1998;35:579-89. Crossref

17. Finucane B, Abrams L, Cronister A, Archibald AD, Bennett RL, McConkie-Rosell A. Genetic counseling and testing for FMR1 gene mutations: practice guidelines of the National Society of Genetic Counselors. J Genet Couns 2012;21:752-60. Crossref

18. Bailey DB Jr, Raspa M, Bishop E, Holiday D. No change in the age of diagnosis for fragile X syndrome: findings from a national parent survey. Pediatrics 2009;124:527-33. Crossref

19. Carmichael B, Pembrey M, Turner G, Barnicoat A. Diagnosis of fragile-X syndrome: the experiences of parents. J Intellect Disabil Res 1999;43:47-53. Crossref

20. Kraan CM, Godler DE, Amor DJ. Epigenetics of fragile X syndrome and fragile X–related disorders. Dev Med Child Neurol 2019;61:121-7. Crossref

21. Baker EK, Arpone M, Vera SA, et al. Intellectual functioning and behavioural features associated with mosaicism in fragile X syndrome. J Neurodev Disord 2019;11:41. Crossref

22. Meng L, Kaufmann WE, Frye RE, et al. The association between mosaicism type and cognitive and behavioral functioning among males with fragile X syndrome. Am J Med Genet A 2022;188:858-66. Crossref

23. Nolin SL, Glicksman A, Houck GE Jr, Brown WT, Dobkin CS. Mosaicism in fragile X affected males. Am J Med Genet 1994;51:509-12. Crossref

24. Liu Y, Chen Y, Zhou X, et al. Mutational analysis of FMR1 gene in autistic children of Han ethnic [in Chinese]. Chin J Child Health Care 2011;19:701-3.

25. Zhang X, Zhong J, Huo X, et al. Screening and genetic diagnosis of fragile X syndrome among children with mental retardation of unknown cause [in Chinese]. Chin J Birth Health Hered 2010;18:38-9.

26. Centre for Health Protection, Department of Health, Hong Kong SAR Government. Overweight and obesity. 2021. Available from: https://www.chp.gov.hk/en/statistics/data/10/757/5513.html. Accessed 14 Jan 2024.

27. Raspa M, Bailey DB, Bishop E, Holiday D, Olmsted M. Obesity, food selectivity, and physical activity in individuals with fragile X syndrome. Am J Intellect Dev Disabil 2010;115:482-95. Crossref

28. Kidd SA, Berry-Kravis E, Choo TH, et al. Improving the diagnosis of autism spectrum disorder in fragile X syndrome by adapting the Social Communication Questionnaire and the Social Responsiveness Scale–2. J Autism Dev Disord 2020;50:3276-95. Crossref

29. Ciaccio C, Fontana L, Milani D, Tabano S, Miozzo M, Esposito S. Fragile X syndrome: a review of clinical and molecular diagnoses. Ital J Pediatr 2017;43:39. Crossref

30. Berry-Kravis E, Filipink RA, Frye RE, et al. Seizures in fragile X syndrome: associations and longitudinal analysis of a large clinic-based cohort. Front Pediatr 2021;9:736255. Crossref

31. Loehr JP, Synhorst DP, Wolfe RR, Hagerman RJ. Aortic root dilatation and mitral valve prolapse in the fragile X syndrome. Am J Med Genet 1986;23:189-94. Crossref

32. Mei L, Hu C, Li D, et al. The incidence and clinical characteristics of fragile X syndrome in China. Front Pediatr 2023;11:1064104. Crossref

33. Charalsawadi C, Wirojanan J, Jaruratanasirikul S, Ruangdaraganon N, Geater A, Limprasert P. Common clinical characteristics and rare medical problems of fragile X syndrome in Thai patients and review of the literature. Int J Pediatr 2017;2017:9318346. Crossref

34. Pang CP, Poon PM, Chen QL, et al. Trinucleotide CGG repeat in the FMR1 gene in Chinese mentally retarded patients. Am J Med Genet 1999;84:179-83. Crossref

35. Chen X, Wang J, Xie H, et al. Fragile X syndrome screening in Chinese children with unknown intellectual developmental disorder. BMC Pediatr 2015;15:77. Crossref

36. Zhong N, Ju W, Xu W, et al. Frequency of the fragile X syndrome in Chinese mentally retarded populations is similar to that in Caucasians. Am J Med Genet 1999;84:191-4. Crossref

37. Mullegama SV, Klein SD, Nguyen DC, et al. Is it time to retire fragile X testing as a first-tier test for developmental delay, intellectual disability, and autism spectrum disorder? Genet Med 2017;19:1380-1. Crossref

38. Cheng YK, Lin CS, Kwok YK, et al. Identification of fragile X pre-mutation carriers in the Chinese obstetric population using a robust FMR1 polymerase chain reaction assay: implications for screening and prenatal diagnosis. Hong Kong Med J 2017;23:110-6. Crossref

39. Huang W, Xia Q, Luo S, et al. Distribution of fragile X mental retardation 1 CGG repeat and flanking haplotypes in a large Chinese population. Mol Genet Genomic Med 2015;3:172-81. Crossref

40. Genereux DP, Laird CD. Why do fragile X carrier frequencies differ between Asian and non-Asian populations? Genes Genet Syst 2013;88:211-24. Crossref

41. The American College of Obstetricians and Gynecologists. Carrier Screening for Genetic Conditions. Committee Opinion. Number 691. March 2017. Available from: https://www.acog.org/clinical/clinical-guidance/committee-opinion/articles/2017/03/carrier-screening-for-genetic-conditions. Accessed 25 Apr 2024.

42. Gregg AR, Aarabi M, Klugman S, et al. Screening for autosomal recessive and X-linked conditions during pregnancy and preconception: a practice resource of the American College of Medical Genetics and Genomics (ACMG). Genet Med 2021;23:1793-806. Crossref

Adolescence marks a critical stage of physical growth, lean body mass development, and pubertal maturation. These biological and physiological changes increase the demand for micronutrients. In particular, iron deficiency (ID) remains a global public health concern.1 Iron deficiency is the most common nutritional deficiency and the leading cause of iron deficiency anaemia (IDA). Because dietary intake is the primary source of iron for most individuals, inadequate dietary iron intake is the main cause of IDA, particularly in adolescents, who are more likely to have poor dietary patterns.2 The Global Burden of Disease 2020 report estimated that approximately 60% of the total global burden of anaemia in 2019 arose from inadequate dietary iron intake.3 Consequently, ID was identified as the most important cause of anaemia-related disability.3 4

In addition to its essential role in haemoglobin synthesis, iron is a key element in brain metabolism and is vital for multiple cellular processes, including neurotransmitter synthesis, neuron myelination, and mitochondrial function.5 Studies in young children have demonstrated that ID during early life adversely affects psychomotor development, concentration, memory, and learning ability.6 7 Notably, the attention domain has received considerable research interest because iron plays a crucial role in the regulation of dopaminergic activity, which is implicated in the pathogenesis and symptoms of attention-deficit hyperactivity disorder (ADHD). Some studies have detected lower ferritin concentrations in children diagnosed with ADHD than in non-ADHD controls.8 9 However, many cognitive studies regarding ID have involved children aged ≤15 years.8 10 Few population-based studies have examined the effect of iron status on cognitive outcomes in adolescents and young adults, and no such studies have been conducted in Chinese populations.

Our previous study11 reported a prevalence of 11.1% for ID among Chinese school-aged adolescents aged 16 to 19 years in Hong Kong, with ID and IDA affecting 17.1% and 10.9% of girls, respectively, while no male participants were affected More than one-third of these adolescents reported regularly skipping at least one meal per day.11 Notably, lower serum ferritin concentrations were observed in adolescents who skipped meals, reported infrequent intake of iron-rich foods, or had heavy menstrual bleeding.11 Consistent with findings from other studies, poor iron reserves were associated with greater self-reported fatigue, reduced physical functioning, and worse school performance.11 Adolescence represents the second most critical period for the development of higher-order cognitive functions, including attention, self-control, and executive function. The adverse effects of low iron reserves on attention span and attentiveness are particularly relevant to upper secondary students in Hong Kong, who are expected to excel academically and prepare for the Hong Kong Diploma of Secondary Education Examination, the city’s university entrance examination. This study aimed to examine the association between iron status and attention outcomes among school-aged adolescents in Hong Kong.

This cross-sectional study recruited healthy adolescent students through the Hong Kong Red Cross Blood Transfusion Service blood donation campaigns at 16 secondary schools between October 2020 and December 2021. The detailed methodology was described in our previous report,11 which aimed to identify the risk factors of ID and IDA in this cohort to facilitate future association studies on health and functional outcomes. In the present study, the dataset was used to delineate the impact of iron reserves on performance-based attention functioning, which is distinct from the self-reported daily functioning outcomes presented in the previous report.11

Students eligible for this study were aged ≥16 years and had agreed to participate in blood donation screening. Students were excluded if they exhibited signs or symptoms of an active infection, reported a history of anaemia, or were receiving treatment for anaemia. Students who did not pass the blood donation screening were still permitted to participate in the study.

A serum ferritin concentration <15 μg/L was considered indicative of ID in both male and female participants, based on the World Health Organization definition.12 Iron deficiency anaemia was defined as the presence of both ID and anaemia. In accordance with the recommendations of the World Health Organization, anaemia was defined as a haemoglobin concentration <12 g/dL in female participants and <13 g/dL in male participants.13 All assays were conducted on the same day in the Department of Pathology Laboratory at Hong Kong Children’s Hospital. The specifications of the instruments and tests have been reported in our prior study.11

Before blood donation, participants completed the Conners Continuous Performance Test Third Edition (CPT-III), a validated assessment commonly used in clinical and research settings to evaluate attention.14 The CPT-III requires 14 minutes to complete and generates specific CPT attention measures (online supplementary Tables 1 to 3). Raw scores for each CPT measure were converted into T-scores based on normative samples (mean=50, standard deviation [SD]=10). Each CPT measure was classified as indicating no/mild (T-score within <1 SD), moderate (T-score within 1-2 SDs), or severe (T-score within >2 SDs) impairment.

Based on the CPT-III manual and the clinical discretion of a developmental specialist (the second author), attention measures were categorised into three clinically relevant attention domains of interest,14 namely, sustained attention impairment (inability to maintain attention), inattention (inability to focus or concentrate), and impulsivity (difficulty with response inhibition).

Fatigue, a recognised risk factor for diminished neurocognitive function, is associated with ID.11 15 Participants completed the PedsQL Multidimensional Fatigue Scale, which has been validated in young adults up to 25 years of age.16 Each item was scored on a 100-point reverse scale, where lower scores indicated more severe fatigue. The Traditional Chinese version of the PedsQL Multidimensional Fatigue Scale has demonstrated good internal consistency, reliability, and content validity in the Chinese population.17 18

We previously reported that dietary patterns are associated with iron reserves in Hong Kong adolescents.11 All participants self-reported their dietary patterns, including meal-skipping habits (breakfast, lunch, or dinner) and the frequency of consuming common iron-rich foods, namely, seafood, meat, iron-fortified cereal, leafy vegetables, beans, nuts, dried fruits, and eggs.11

The demographic and haematological characteristics of the cohort, along with their attention outcomes, were summarised using descriptive analysis.

The primary outcome was attention impairment. Serum ferritin concentration was used as the predictor of interest, rather than a comparison of attention outcomes between participants with and without ID or IDA, considering that clinical thresholds for diagnosing ID and IDA may not be applicable when evaluating the effect of iron on functional outcomes. Even if an adolescent is not clinically diagnosed with ID or IDA, a low-to-normal ferritin concentration may affect functional outcomes; previous studies have shown that the impact of ID on neurodevelopment may occur before ID manifests as clinical anaemia.19 20 The Mann-Whitney U test was utilised to compare serum ferritin concentrations between participants with normal attention function (ie, those who did not exhibit impairment in any of the three attention domains) and those with moderate or severe impairment in sustained attention, inattention, or impulsivity.

Multivariable analysis using a log-binomial regression model was conducted, with serum ferritin concentration, fatigue, dietary pattern, and dietary iron intake as predictors. Models were adjusted for age and sex. Risk ratios (RRs) and 95% confidence intervals (95% CIs) were calculated.

Given that previous studies have shown a positive association between iron reserves and functional outcomes regardless of sex,8 15 20 21 we first conducted all analyses in the overall cohort. Subsequently, analyses were performed separately for male and female participants.

The significance threshold was set at P<0.05. All statistical analyses were performed using SAS 9.4 (SAS Institute, Cary [NC], US) and were two-tailed.

As reported in our previous study,11 a total of 523 students were recruited (participation rate: 70%). Twenty-nine students were deferred from blood donation due to low haemoglobin concentrations but still completed the study procedures. Two-thirds of participants were female (n=340, 65.0%). The demographics of the study cohort, stratified by sex, are presented in Table 1.

The median ferritin concentration in male participants was 136.17 μg/L (interquartile range [IQR]=89.89-219.83; Fig a); no male participants were diagnosed with ID. Among female participants diagnosed with ID (n=58/340, 17.1%), the median haemoglobin concentration was 11.6 g/dL (IQR=11.1-12.2; Fig b). Among female participants with normal serum ferritin concentrations (n=282/340, 82.9%), the median serum ferritin concentration was 56.07 μg/L (IQR=33.82-84.11; Fig c).

Overall, 249 participants (47.6%) exhibited normal function in all three attention domains. Approximately one-quarter of the participants demonstrated moderate-to-severe impairment in sustained attention (n=131/523, 25.0%), inattention (n=145/523, 27.7%), and impulsivity (n=157/523, 30.0%).

Among female participants with ID, the rates of moderate-to-severe impairment in sustained attention, inattention, and impulsivity were 36.2% (n=21/58), 27.6% (n=16/58), and 37.9% (n=22/58), respectively. The rates of moderate-to-severe impairment in inattention and impulsivity among female participants with IDA were numerically higher at 43.5% (n=10/23 for both domains). Among male participants, the rates of moderate-to-severe impairment in sustained attention, inattention, and impulsivity were 18.0% (n=33/183), 23.5% (n=43/183), and 22.4% (n=41/183), respectively (Table 2).

In the overall cohort, participants with sustained attention impairment had significantly lower serum ferritin concentrations relative to those with intact attention function (median=51.2 μg/L, IQR=27.1-106.8 vs median=73.9 μg/L, IQR=37.8-138.0; P=0.020). Although the associations were not statistically significant, trends of lower serum ferritin concentrations were also observed in participants with impulsivity impairment (median=68.1 μg/L, IQR=29.0-114.8 vs median=73.9 μg/L, IQR=37.8-138.0; P=0.067) and inattention impairment (median=69.9 μg/L, IQR=32.0-110.8 vs median=73.9 μg/L, IQR=37.8-138.0; P=0.142) relative to those with intact attention function.

Pooled analysis of the overall cohort, adjusted for age and sex, showed a significant association between lower serum ferritin concentration and sustained attention impairment (RR=0.825, 95% CI=0.732-0.946; P=0.040), suggesting that each 10 μg/L increase in serum ferritin concentration was associated with a 17.6% decrease in the risk of sustained attention impairment. A higher level of fatigue was associated with impairment in sustained attention (RR=0.772, 95% CI=0.652-0.926; P=0.004), inattention (RR=0.824, 95% CI=0.733-0.942; P=0.016), and impulsivity (RR=0.792, 95% CI=0.683-0.922; P=0.004). Serum ferritin concentration was not significantly associated with risks of impairment in inattention or impulsivity (Table 3).

Female participants with sustained attention impairment had marginally lower serum ferritin concentrations relative to those with intact attention function (median=40.0 μg/L, IQR=18.8-52.1 vs median=48.5 μg/L, IQR=21.8-73.8; P=0.038). Although the associations were not statistically significant, trends for lower serum ferritin concentrations were also observed in participants with impulsivity impairment (median=43.0 μg/L, 95% CI=19.5-63.2 vs median=48.5 μg/L, IQR=21.8-73.8; P=0.071) relative to those with intact attention function. No significant difference was observed for inattention impairment. Additionally, no significant association was detected between iron reserves and attention impairment in male participants.

Multivariable analysis revealed that the association between iron reserves and sustained attention impairment in female participants was attenuated and not statistically significant (RR=0.954, 95% CI=0.904-1.005; P=0.073). A higher level of fatigue was associated with an increased risk of sustained attention impairment (RR=0.793, 95% CI=0.652-0.964; P=0.021). Among male participants, iron reserves did not affect attention outcomes, but fatigue was associated with impulsivity impairment (RR=0.712, 95% CI=0.548-0.942; P=0.018). Dietary patterns were not significantly associated with attention outcomes in either male or female participants (Table 3).

In the overall cohort, a lower serum ferritin concentration was associated with a higher risk of sustained attention impairment, consistent with previous reports that iron reserves play an essential role in functional performance in adolescents.6 7 8 21 When the analysis was stratified by sex, a similar but modest association between low iron reserves and sustained attention impairment was observed in female school-aged adolescents. This finding is supported by studies regarding the neurobiology of attention-related developmental disorders associated with ID.6 7 9 10 A meta-analysis of 10 studies, comprising 2191 healthy children and 1196 children with ADHD, showed that serum ferritin concentrations were 0.4-fold lower in children with ADHD than in those without developmental disorders.8 Iron deficiency may be associated with disruptions in monoamine synthesis and monoamine signal transduction, which manifest as attention deficits.10 22 Adequate iron intake and iron stores may, therefore, be important factors influencing the onset of attention problems in the developing brain. This finding should be prospectively validated in larger cohorts with a comprehensive assessment of cognitive domains beyond attention. However, from a developmental perspective, sustained attention is closely related to performance on targeted assessments, such as mathematical fluency and reading comprehension, as well as broader academic measures in national standardised examinations.23 24 This relationship is particularly relevant because the Hong Kong educational system is well known for its examination-dominated culture. Most examinations range from 2 to 3 hours, requiring students to maintain a high level of sustained attention. Therefore, these findings may have long-term implications for students’ academic success. Future research should investigate the effects of ID and IDA on subsequent academic achievement in Hong Kong adolescents.

Evidence regarding the effectiveness of iron supplementation in terms of improving neurocognitive function in children and adolescents has been inconclusive. Furthermore, iron supplements are associated with gastrointestinal symptoms and constipation, which contribute to non-adherence, particularly in adolescents.25 A systematic review of 14 randomised controlled trials indicated that iron supplementation improved attention and intelligence quotient in anaemic older children and adults.26 However, these effects were inconsistent across studies; they were influenced by socio-economic factors, participant age, and the clinical thresholds used to define ID and IDA.20 25 26 The benefits for cognitive development in older adolescents remain uncertain and warrant further investigation.26

In this study, we found that students who reported higher levels of fatigue were more likely to have worse attention outcomes. We also previously reported that lower serum ferritin concentrations are associated with self-reported fatigue in adolescents.11 Evidence supporting the role of iron supplementation in fatigue reduction is more consistent than its effects on cognitive function in young adults, particularly among non-anaemic menstruating women with low ferritin concentrations.21 27 Notably, iron supplementation has been associated with reductions in subjective measures of fatigue among non-anaemic iron-deficient adults.21 The present findings suggest that ID correction in adolescents could reduce fatigue levels, which may indirectly improve attention outcomes. Using a serum ferritin concentration threshold of 15 μg/L to diagnose clinical ID, some researchers have demonstrated that iron supplementation can improve fatigue and physical performance among individuals with serum ferritin concentrations at the lower end of the normal range (30-50 μg/L).21 Collectively, the known health risks of ID, including impaired physical growth, fatigue, and reduced fitness in adolescents, underscore the need to educate students about maintaining a balanced diet with adequate iron intake. Adolescents with low ferritin concentrations should receive counselling focused on the consumption of iron-rich foods and iron supplementation to alleviate fatigue, even in the absence of documented anaemia.

Dietary patterns and self-reported intake of iron-rich foods were not directly associated with attention outcomes in the multivariate analysis, likely because neurocognitive function is a multifactorial and complex phenotype influenced by both nutritional and non-nutritional factors. Additionally, we did not use a comprehensive measure of dietary iron intake. However, we previously showed that skipping at least one meal per day or exhibiting low dietary iron intake was associated with lower iron reserves.11 Iron deficiency prevention in adolescents requires effective management of knowledge gaps related to food nutrition, dieting, and body image. Collectively, these findings highlight the importance of developing nutrition education programmes to encourage proactive adoption of dietary and other nutrition-related behaviours that promote health and well-being.

Despite the relatively large cohort of school-aged adolescents and the well-characterised haematological assessments, this study had several important limitations. First, the participation rate in the blood donation programme was affected by the coronavirus disease 2019 pandemic and school closures. This change in participation rate may have introduced sampling bias because students with worse health statuses may have been more likely to decline blood donation. Second, we only assessed attention measures in this study. It was not feasible to administer a full neurocognitive test battery, which typically requires >1 hour, in a school-based environment with limited time, space, and supervisory personnel. Future studies should include a more comprehensive evaluation of neurocognitive function. Finally, we did not evaluate factors potentially associated with the causes of anaemia and cognitive function, such as markers of socio-economic status, family functioning, living environment, and physical activity.28 29 Nevertheless, our findings regarding the association between iron status and attention outcomes provide valuable local population data and guidance for future iron supplementation initiatives.

Lower serum ferritin concentrations and self-reported fatigue were associated with an increased risk of sustained attention impairment among school-aged adolescents in Hong Kong. The potential health consequences of ID without anaemia, particularly its effects on physical well-being and school performance, should be effectively communicated to the Hong Kong population, especially to female adolescents. Dietary interventions should target

Concept or design: All authors.
Acquisition of data: CK Lee, WC Tsoi, CW Lau, JNS Leung, STY Tsang, CLP Wong, YYL Chu, CK Li.
Analysis of data: YT Cheung, DFY Chan.
Interpretation of data: All authors.
Drafting of the manuscript: YT Cheung.
Critical revision of the manuscript for important intellectual content: All authors.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

All authors have disclosed no conflicts of interest.

The authors thank the principals and staff of the participating schools, as well as Mr Calvin Lam from Department of Paediatrics of The Chinese University of Hong Kong for assistance with data collection.

Part of the results was presented at the Joint Annual Scientific Meeting 2022 (hybrid meeting) of The Hong Kong Paediatric Society, Hong Kong College of Paediatricians, Hong Kong Paediatric Nurses Association, and Hong Kong College of Paediatric Nursing in Hong Kong on 26 September 2022.

This research was funded by the Health and Medical Research Fund, the former Food and Health Bureau, Hong Kong SAR Government (Ref No.: 17180441). The funder had no role in study design, data collection, analysis, interpretation, or manuscript preparation.

This research was approved by the Joint Chinese University of Hong Kong—New Territories East Cluster Clinical Research Ethics Committee, Hong Kong (Ref No.: 2019.107). Participants aged ≥18 years provided written informed consent, whereas those aged <18 years provided written assent along with informed consent from a parent or legal guardian.

The supplementary material was provided by the authors and some information may not have been peer reviewed. Accepted supplementary material will be published as submitted by the authors, without any editing or formatting. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by the Hong Kong Academy of Medicine and the Hong Kong Medical Association. The Hong Kong Academy of Medicine and the Hong Kong Medical Association disclaim all liability and responsibility arising from any reliance placed on the content.

1. Powers JM, O’Brien S, Berlan ED, Hoppin AG, editors. Iron requirements and iron deficiency in adolescents. UpToDate. Available from: https://www.uptodate.com/contents/iron-requirements-and-iron-deficiency-in-adolescents. Accessed 1 Apr 2025.

2. Camaschella C, Girelli D. The changing landscape of iron deficiency. Mol Aspects Med 2020;75:100861. Crossref

3. Safiri S, Kolahi AA, Noori M, et al. Burden of anemia and its underlying causes in 204 countries and territories, 1990-2019: results from the Global Burden of Disease Study 2019. J Hematol Oncol 2021;14:185. Crossref

4. Institute for Health Metrics and Evaluation, University of Washington. GBD results. 2020. Available from: https://vizhub.healthdata.org/gbd-results/. Accessed 6 May 2023.

5. Hare D, Ayton S, Bush A, Lei P. A delicate balance: iron metabolism and diseases of the brain. Front Aging Neurosci 2013;5:34. Crossref

6. Jáuregui-Lobera I. Iron deficiency and cognitive functions. Neuropsychiatr Dis Treat 2014;10:2087-95. Crossref

7. Pivina L, Semenova Y, Doşa MD, Dauletyarova M, Bjørklund G. Iron deficiency, cognitive functions, and neurobehavioral disorders in children. J Mol Neurosci 2019;68:1-10. Crossref

8. Wang Y, Huang L, Zhang L, Qu Y, Mu D. Iron status in attention-deficit/hyperactivity disorder: a systematic review and meta-analysis. PLoS One 2017;12:e0169145. Crossref

9. Bener A, Kamal M, Bener H, Bhugra D. Higher prevalence of iron deficiency as strong predictor of attention deficit hyperactivity disorder in children. Ann Med Health Sci Res 2014;4(Suppl 3):S291-7. Crossref

10. Tseng PT, Cheng YS, Yen CF, et al. Peripheral iron levels in children with attention-deficit hyperactivity disorder: a systematic review and meta-analysis. Sci Rep 2018;8:788. Crossref

11. Cheung YT, Chan DF, Lee CK, et al. Iron deficiency among school-aged adolescents in Hong Kong: prevalence, predictors, and effects on health-related quality of life. Int J Environ Res Public Health 2023;20:2578. Crossref

12. World Health Organization. WHO guideline on use of ferritin concentrations to assess iron status in individuals and populations. 2020. Available from: https://apps.who.int/iris/handle/10665/331505. Accessed 6 Oct 2023.

13. World Health Organization. Haemoglobin concentrations for the diagnosis of anaemia and assessment of severity. 2011 May 31. Available from: https://www.who.int/publications/i/item/WHO-NMH-NHD-MNM-11.1. Accessed 6 Oct 2023.

14. Conners CK, Sitarenios G. Conners’ Continuous Performance Test (CPT). In: Kreutzer JS, DeLuca J, Caplan B, editors. Encyclopedia of Clinical Neuropsychology. New York: Springer; 2011: 681-3. Crossref

15. Sulheim D, Fagermoen E, Sivertsen ØS, Winger A, Wyller VB, Øie MG. Cognitive dysfunction in adolescents with chronic fatigue: a cross-sectional study. Arch Dis Child 2015;100:838-44. Crossref

16. Varni JW, Limbers CA. The PedsQL Multidimensional Fatigue Scale in young adults: feasibility, reliability and validity in a university student population. Qual Life Res 2008;17:105-14. Crossref

17. Yeung NC, Lau JT, Yu X, et al. Psychometric properties of the Chinese version of the Pediatric Quality of Life Inventory 4.0 Generic Core Scales among pediatric cancer patients. Cancer Nurs 2013;36:463-73. Crossref

18. Hao Y, Tian Q, Lu Y, Chai Y, Rao S. Psychometric properties of the Chinese version of the Pediatric Quality of Life Inventory 4.0 Generic Core Scales. Qual Life Res 2010;19:1229-33. Crossref

19. Camaschella C. Iron deficiency. Blood 2019;133:30-9. Crossref

20. Hermoso M, Vucic V, Vollhardt C, et al. The effect of iron on cognitive development and function in infants, children and adolescents: a systematic review. Ann Nutr Metab 2011;59:154-65. Crossref

21. Houston BL, Hurrie D, Graham J, et al. Efficacy of iron supplementation on fatigue and physical capacity in non-anaemic iron-deficient adults: a systematic review of randomised controlled trials. BMJ Open 2018;8:e019240. Crossref

22. Kim J, Wessling-Resnick M. Iron and mechanisms of emotional behavior. J Nutr Biochem 2014;25:1101-7. Crossref

23. Gallen CL, Schaerlaeken S, Younger JW; Project iLEAD Consortium; Anguera JA, Gazzaley A. Contribution of sustained attention abilities to real-world academic skills in children. Sci Rep 2023;13:2673. Crossref

24. Schmengler H, Peeters M, Stevens GW, et al. Educational level, attention problems, and externalizing behaviour in adolescence and early adulthood: the role of social causation and health-related selection—the TRAILS study. Eur Child Adolesc Psychiatry 2023;32:809-24. Crossref

25. Finkelstein JL, Herman HS, Guetterman HM, Peña-Rosas JP, Mehta S. Daily iron supplementation for prevention or treatment of iron deficiency anaemia in infants, children, and adolescents. Cochrane Database Syst Rev 2018;2018:CD013227. Crossref

26. Falkingham M, Abdelhamid A, Curtis P, Fairweather-Tait S, Dye L, Hooper L. The effects of oral iron supplementation on cognition in older children and adults: a systematic review and meta-analysis. Nutr J 2010;9:4. Crossref

27. Vaucher P, Druais PL, Waldvogel S, Favrat B. Effect of iron supplementation on fatigue in nonanemic menstruating women with low ferritin: a randomized controlled trial. CMAJ 2012;184:1247-54. Crossref

28. Hess SY, Owais A, Jefferds ME, Young MF, Cahill A, Rogers LM. Accelerating action to reduce anemia: review of causes and risk factors and related data needs. Ann N Y Acad Sci 2023;1523:11-23. Crossref

29. Meredith WJ, Cardenas-Iniguez C, Berman MG, Rosenberg MD. Effects of the physical and social environment on youth cognitive performance. Dev Psychobiol 2022;64:e22258. Crossref

The coronavirus disease 2019 (COVID-19) pandemic has had extensive global social and health impacts. It triggered an international health and economic crisis that has profoundly altered people’s lives, perceptions, and behaviours. As of 13 March 2025, about 778 million confirmed cases of COVID-19 had caused around 7.1 million deaths worldwide.1 Various levels of non-pharmaceutical interventions, including frequent handwashing, mask-wearing, and social distancing, were implemented in most countries.2 These interventions played important roles in reducing community transmission of COVID-19.3 However, the stringent measures also led to negative consequences, such as economic slowdown, disrupted education, and increased social isolation and psychological stress.4 5 Many countries lifted non-pharmaceutical interventions while the number of cases was still increasing. In England, all COVID-19–related restrictions were lifted on 22 February 2022 under the ‘Living with COVID’ strategy,6 although the number of cases increased in subsequent months. Australia, Singapore, and Hong Kong adopted a ‘Zero-COVID’ strategy.7 In Australia, all mandatory mask-wearing requirements on public transport were lifted in mid-September 2022.8 Singapore also lifted such requirements on 9 February 2023.9 Hong Kong, a leading international business and financial centre, finally lifted all mandatory mask-wearing requirements on 1 March 2023,10 nearly 1000 days after the start of the pandemic in 2020. Since then, the city has been transitioning towards the post–COVID-19 era.

During the COVID-19 pandemic, many governments mandated mask-wearing in public areas. Mask-wearing behaviour was largely a response to legal restrictions and requirements. Obedience, as a form of social influence, played a role in mask adherence; individuals sought to avoid social punishment, including fines or imprisonment. Additionally, normative social influence emerged as a means of curbing the spread of COVID-19. A positive correlation was observed between social norms regarding mask-wearing and mask uptake, such that individuals were more likely to wear a mask if their friends and relatives did so.11 Furthermore, individuals’ beliefs about engaging in the right behaviour were associated with their behavioural intentions. Personal norms regarding mask-wearing were significantly associated with mask-wearing intention.12 In the post–COVID-19 era, individuals may continue mask-wearing even after governments have lifted mandatory requirements, potentially due to self-motivation for health protection. This study aimed to identify predictors of mask-wearing intentions and practices after the mandatory mask-wearing requirement had been lifted in Hong Kong by integrating the protection motivation theory (PMT) with the theory of planned behaviour (TPB). This integration provides a comprehensive framework for evaluating mask-wearing intentions by examining key factors influencing health behaviours, including perceived severity, perceived vulnerability, attitudes, and subjective norms. This approach may offer a nuanced understanding of predictors of mask-wearing intentions after the mandatory mask-wearing requirement had been lifted.

Protection motivation theory has been widely used as a framework for predicting the adoption of health-protective behaviours.13 This theory assumes that the adoption of protective behaviour against health threats depends on personal motivation for self-protection. Rooted in expectancy-value theory, PMT explains the social and cognitive processes underlying protective behaviours. The theory is based on the premise that the decision to counteract a health threat is determined by threat and coping appraisal processes.14 According to PMT, two primary processes—threat appraisal and coping appraisal—determine behavioural intention. Threat appraisal consists of three components: perceived vulnerability, perceived severity, and the perceived reward of maladaptive behaviours. Perceived vulnerability refers to an individual’s assessment of the likelihood of experiencing a health threat or developing a health condition. Perceived severity concerns the perceived seriousness of potential consequences associated with the condition. Therefore, perceptions of COVID-19 severity and vulnerability to disease would significantly predict adherence to protective measures.15 Perceived reward of maladaptive behaviours refers to beliefs regarding the benefits associated with engaging in risky behaviours. Patients with COVID-19 may experience long COVID symptoms, including increased fatigue, depressive symptoms, and reduced mental acuity.16 In this context, individuals may continue wearing masks due to concerns about long-COVID severity. Thus, perceived vulnerability and perceived severity are expected to be positively associated with the intention to continue wearing a mask in the post–COVID-19 era, whereas the perceived reward of maladaptive behaviours is expected to be negatively associated with this behaviour. Three hypotheses were proposed in relation to these elements (H1 to H3 in the online supplementary Table).

Coping appraisal comprises perceived response efficacy, perceived self-efficacy, and perceived response cost. Perceived response efficacy refers to belief in the effectiveness of the recommended behaviour with respect to mitigating or preventing potential harm.17 Perceived self-efficacy denotes an individual’s confidence in overcoming barriers to implementing the recommended behaviour.18 Perceived response cost refers to perceived costs associated with the behaviour. Perceived response efficacy has been positively associated with social distancing behaviours, a non-pharmaceutical intervention for COVID-19, among Hong Kong adults.19 Three hypotheses were derived in relation to these elements (H4 to H6 in the online supplementary Table).

The TPB is a well-established model for explaining health-related behavioural intentions, which are influenced by subjective norms (perceived expectations from significant others regarding the behaviour), attitude (personal feelings and beliefs about the behaviour), and perceived behavioural control (perceived ability to perform the behaviour). Individuals with a more positive attitude towards non-pharmaceutical interventions exhibit a greater intention to implement such interventions.20 Similarly, subjective norms and perceived behavioural control have demonstrated positive associations with the intention to adopt interventions against COVID-19.20 Five hypotheses were formulated in relation to these elements (H7 to H11 in the online supplementary Table).

The integration of PMT and TPB has been utilised to predict behavioural intention in various research contexts, such as adherence to COVID-19 behavioural guidelines,21 behavioural intention towards COVID-19 booster vaccination,22 and factors affecting preventive behaviours during the COVID-19 pandemic.23 In this study, the attitude component of TPB was used to assess an individual’s attitude towards continuing to wear a mask. Attitudes may be influenced by an individual’s protection motivation. A meta-analysis identified perceived importance, perceived benefits, perceived effectiveness, and perceived barriers to preventive behaviour as key attitudinal factors influencing such behaviour.24 Therefore, a conceptual model was developed to illustrate relationships between the constructs of PMT and TPB in predicting continued mask-wearing after the announcement that all mandatory mask-wearing requirements had been lifted. Fourteen hypotheses were formulated in relation to these elements (H12 to H25 in the online supplementary Table).

This cross-sectional study was conducted using an online questionnaire between 8 and 20 March 2023. Participants were recruited through a non-probability snowball sampling method that had been used in a previous study.3 The target sample size was determined based on the requirement that it should be 10 times the maximum number of measurement items associated with a single construct in the partial least squares path model.25 In this study, 37 items measured ten constructs, resulting in a target sample size of 370 (10 × 37). The online questionnaire was distributed via email and WhatsApp, a widely used social media platform in Hong Kong. Using the researchers’ personal social networks, eligible individuals of various ages and educational backgrounds were invited to participate. They were also encouraged to share the questionnaire link with suitable colleagues and friends. Additionally, the researchers contacted the heads of local community colleges to seek collaboration and support. Upon receiving approval from directors or presidents, the researchers sent the online questionnaire to those leaders for recruitment of eligible participants. Individuals were included in this study if they were Hong Kong residents aged ≥18 years and had access to the internet via a smartphone or computer. Participants read a statement on the survey’s background, anonymity, and participation agreement before providing consent. To prevent duplicate submissions, the prefix and first three digits of the Hong Kong Identity Card were collected and later removed prior to data analysis.

The questionnaire, consisting of four sections, was designed to assess perceived vulnerability, perceived severity, perceived reward of maladaptive behaviours, perceived response efficacy, perceived self-efficacy, perceived response cost, attitude, perceived behavioural control, subjective norms, and intention to continue wearing a mask after the mandatory mask-wearing requirement had been lifted. The first section included two questions focused on mask-wearing frequency after the mandatory requirement had been lifted and on verification of Hong Kong residency. The second section examined respondents’ adoption of health-protective behaviours, based on PMT.26 27 The third section measured variables related to respondents’ intention to continue wearing a mask, based on TPB.3 27 All items in the second and third sections were assessed using a five-point Likert scale (1=strongly disagree to 5=strongly agree). The final section collected demographic information, such as age, gender, education level, economic status, and self-reported health status, through close-ended questions.

Partial least squares structural equation modelling was utilised to examine the conceptual framework in this study. The SmartPLS 3.0 statistical software (SmartPLS GmbH, Bönningstedt, Germany) was used to assess both the reflective measurement model and the structural model. Study reliability and validity were evaluated by assessing internal consistency and convergent validity in the reflective measurement model.25 Convergent validity was considered acceptable if the outer loadings of the measurement items exceeded 0.5 and the average variance extracted for each construct was >0.5.25 28 Internal reliability was evaluated using composite reliability, which was recommended to exceed 0.708, and Cronbach’s alpha, which should be >0.6.25 Path coefficients were assessed within the structural model. A P value <0.05 was considered significant.

In total, 483 valid responses were included in the data analysis. Table 1 presents the participants’ demographic characteristics. The largest proportion of respondents belonged to the 18-25 age-group (28.2%), followed by the 56-65 (18.4%), the 66-75 (13.7%), and the 36-45 (13.0%) age-groups. The mean age was 43.56 years. Among the participants, 269 (55.7%) were men and 214 (44.3%) were women. Most respondents (59.0%) had attained a degree-level education or higher; more than two-fifths of respondents were employed. Additionally, approximately half of the respondents (46.6%) rated their health status as good. More than half of the respondents (53.6%) reported always wearing a mask after the mandatory mask-wearing requirement had been lifted. The median number of COVID-19 vaccine doses received was three (interquartile range=1).

Table 2 presents the model reliability. Loadings >0.7 indicate a satisfactory level of item reliability.25 29 The outer loadings of all items exceeded 0.7, except for one item related to perceived behavioural control; consequently, this item was removed. Internal consistency reliability was considered satisfactory because composite reliability and Cronbach’s alpha exceeded the threshold value of 0.7. The average variance extracted for all constructs was >0.5, suggesting good convergent validity after the removal of five items: one item each from perceived severity, perceived response efficacy, perceived self-efficacy, attitude, and behavioural intention. The variance inflation factor for each item was <5, indicating no critical levels of collinearity. Table 3 depicts the results of the assessment of discriminant validity. Given the adequacy of indicator reliability, internal consistency reliability, convergent validity, and discriminant validity, evaluation of the structural model could proceed.29

Table 4 displays the results of direct effects in the structural model. Of the 17 hypotheses, 10 were supported based on the results generated through a bootstrapping procedure with 5000 resamples. Four constructs—perceived severity, perceived self-efficacy, subjective norms, and attitude—had significant positive effects on the intention to continue wearing a mask. In contrast, perceived reward of maladaptive behaviours had a significant negative effect on mask-wearing intention. Consequently, hypotheses H2, H3, H5, H7, and H8 were supported. However, perceived vulnerability, perceived response efficacy, perceived response cost, and perceived behavioural control were not significantly associated with the intention to continue wearing a mask. Thus, hypotheses H1, H4, H6, and H9 were not supported.

Furthermore, subjective norms, perceived severity, perceived response efficacy, and perceived self-efficacy had significant positive effects on attitude, whereas perceived reward of maladaptive behaviours had a significant negative effect on attitude. Therefore, hypotheses H10, H13, H14, H15, and H16 were supported. However, no significant relationships were observed between perceived behavioural control and attitude, perceived vulnerability and attitude, or perceived response cost and attitude. These findings did not support hypotheses H11, H12, and H17 (Table 4). The results of the structural model are depicted in the Figure.

Table 5 shows the results of the mediation model. Attitude had a partial mediating effect on the relationships of perceived self-efficacy, perceived reward of maladaptive behaviours, subjective norms, and perceived severity with the intention to continue wearing a mask. These results partially supported hypotheses H19, H20, H21, and H22. Additionally, attitude had a full mediating effect on the relationship between perceived response efficacy and the intention to continue wearing a mask, supporting hypothesis H24. However, no mediating effect of attitude was observed in the relationships of perceived response cost, perceived vulnerability, and perceived behavioural control with continuous behavioural intention. These results did not support hypotheses H18, H23, and H25.

Most respondents continued wearing masks during the 3 weeks after the mandatory mask-wearing requirement had been lifted. Perceived severity, perceived self-efficacy, subjective norms, and attitude were positively associated with the intention to continue wearing a mask, whereas the perceived reward of maladaptive behaviours was negatively associated with this intention. Perceived severity suggests that individuals were concerned about the consequences of contracting COVID-19. Given that COVID-19 had influenced daily life and behaviour for 3 years, it is understandable that perceived severity remained a motivator for continued mask-wearing as a protective measure. Furthermore, some individuals may have experienced anxiety and sought to minimise the risk of infection. Thus, the pandemic itself may have outweighed their desire to return to pre-pandemic norms.30 Additionally, perceived self-efficacy indicates that individuals with confidence in their ability to wear a mask effectively were more likely to continue doing so. Personal protective measures can reduce the risk of infectious diseases31; mask-wearing is considered a feasible and acceptable method for preventing and reducing the spread of influenza-like illnesses.32 During the COVID-19 pandemic, some studies showed that perceived severity and perceived self-efficacy were significantly associated with intentions to comply with COVID-19 preventive behaviours.17 33 34 Individuals perceived that contracting COVID-19 posed a serious threat, whereas mask-wearing remained a feasible and effective strategy for preventing transmission, even after the mandatory mask-wearing requirement had been lifted.

Notably, the perceived reward of maladaptive behaviours had a significant negative effect on the intention to continue wearing a mask. This finding suggests that individuals who perceived benefits from not wearing a mask were less likely to express an intention to continue mask-wearing. The decision not to wear a mask may be attributed to various factors, including concerns about social judgement, the inconveniences associated with preventive measures, and daily hassles.35 36 The prolonged COVID-19 pandemic led to pandemic fatigue, which may have contributed to a perception among some individuals that the pandemic had ended once the mandatory mask-wearing requirement was lifted, thereby reducing their motivation to continue wearing a mask.

Attitudes and subjective norms had significant positive effects on the intention to continue wearing a mask. This observation indicates that individuals who held a favourable attitude towards mask-wearing and perceived social pressure or influence from others to wear a mask were more likely to express an intention to continue this practice. Attitudes and subjective norms were previously identified as predictors of mask-wearing intention during the COVID-19 pandemic.3 Before the pandemic, the local population in Hong Kong exhibited a positive attitude towards mask-wearing. For example, patients and caregivers in outpatient settings generally wore face masks; protecting others was a primary motivation for this approach.37 Individuals with a positive attitude towards mask-wearing may have been influenced by government-led promotion of preventive behaviours since the severe acute respiratory syndrome epidemic in 2003, which caused mask-wearing to become a social norm within the community.38 The present findings indicate that higher levels of perceived self-efficacy, perceived reward of maladaptive behaviours, subjective norms, and perceived severity not only directly increased the intention to wear a mask but also influenced individuals’ attitudes, leading to an increased intention to continue mask-wearing. These results provide empirical evidence supporting the role of attitude as a mediator in the intention to continue wearing a mask. Thus, the relationships among perceived self-efficacy, perceived reward of maladaptive behaviours, subjective norms, perceived severity, and the intention to continue mask-wearing can also be explained by individuals’ attitudes.

In the present study, perceived vulnerability did not directly predict the intention to continue wearing a mask. A study also showed no significant association between perceived vulnerability and the adoption of preventive behaviours.39 A possible explanation is that the prolonged COVID-19 pandemic led individuals to consider themselves less vulnerable compared with early stages of the pandemic. The removal of government restrictions may have further reinforced the perception of reduced vulnerability to COVID-19.40 Additionally, the results of this study did not demonstrate a statistically significant direct effect between perceived response efficacy and the intention to continue wearing a mask. However, a mediating role for attitude was identified in this relationship, indicating that perceived response efficacy influenced attitude, which then determined the intention to wear a mask.

This study highlights the importance of understanding the predictors of mask-wearing intention after the mandatory mask-wearing requirement was lifted. A high frequency of mask-wearing was observed after the removal of the requirement. This finding has implications for future research regarding the long-term effects of habitual mask use and its impact on public health. From a practical perspective, the findings indicate that attitude towards continued mask-wearing was the strongest predictor of mask-wearing intention, suggesting that citizens are prepared for future health crises. Policymakers can utilise these insights to develop guidelines encouraging mask use during influenza seasons.

This study had certain limitations. First, the sampling method relied on non-probability snowball sampling, which may affect the representativeness of the sample. Second, participation was limited to individuals with access to email and social media, leading to overrepresentation of younger and more educated individuals. Younger participants may consider themselves less likely to experience severe health consequences if they contract COVID-19. Consequently, the findings may not be generalisable to the entire population.

To our knowledge, this is one of the first studies to use an online questionnaire to identify the predictors of mask-wearing intention after the mandatory mask-wearing requirement in Hong Kong was lifted in March 2023. Attitude towards continued mask-wearing, subjective norms, and perceived self-efficacy exhibited strong positive effects on the intention to continue wearing a mask. Regarding research implications, this study provides new insights into the evaluation of Hong Kong citizens’ transition to a post-pandemic era. The high frequency of mask-wearing observed may be attributed to concerns about COVID-19 and the establishment of mask-wearing as an accepted and habitual behaviour within the local population. Furthermore, the findings suggest that Hong Kong citizens are well prepared for future health crises, such as severe acute respiratory syndrome and additional COVID-19 outbreaks. The positive attitude towards mask-wearing reflects recognition of its feasibility and effectiveness as a durable non-pharmaceutical public health intervention to reduce airborne disease transmission.

Concept or design: TKC Ng, BYF Fong.
Acquisition of data: All authors.
Analysis or interpretation of data: TKC Ng, BYF Fong.
Drafting of the manuscript: TKC Ng, BYF Fong.
Critical revision of the manuscript for important intellectual content: All authors.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

All authors have disclosed no conflicts of interest.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

This research was approved by the Research Committee of the College of Professional and Continuing Education of Hong Kong Polytechnic University, Hong Kong (Ref No.: RC/ETH/H/133). Informed consent was obtained from all participants prior to the study and for the publication of this research.

The supplementary material was provided by the authors and some information may not have been peer reviewed. Accepted supplementary material will be published as submitted by the authors, without any editing or formatting. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by the Hong Kong Academy of Medicine and the Hong Kong Medical Association. The Hong Kong Academy of Medicine and the Hong Kong Medical Association disclaim all liability and responsibility arising from any reliance placed on the content.

1. World Health Organization. WHO COVID-19 Dashboard. Available from: https://covid19.who.int/. Accessed 13 Mar 2025.

2. Lison A, Banholzer N, Sharma M, et al. Effectiveness assessment of non-pharmaceutical interventions: lessons learned from the COVID-19 pandemic. Lancet Public Health 2023;8:e311-7. Crossref

3. Duan Y, Shang B, Liang W, et al. Predicting hand washing, mask wearing and social distancing behaviors among older adults during the COVID-19 pandemic: an integrated social cognition model. BMC Geriatr 2022;22:91. Crossref

4. Diallo I, Ndejjo R, Leye MM, et al. Unintended consequences of implementing non-pharmaceutical interventions for the COVID-19 response in Africa: experiences from DRC, Nigeria, Senegal, and Uganda. Global Health 2023;19:36. Crossref

5. ÓhAiseadha C, Quinn GA, Connolly R, et al. Unintended consequences of COVID-19 non-pharmaceutical interventions (NPIs) for population health and health inequalities. Int J Environ Res Public Health 2023;20:5223. Crossref

6. Limb M. COVID-19: scientists and medics warn that it is too soon to lift all restrictions in England. BMJ 2022;376:o469. Crossref

7. Zhan Z, Li J, Cheng ZJ. Zero-COVID strategy: what’s next? Int J Health Policy Manag 2023;12:6757. Crossref

8. Dunstan J. Victoria to ease COVID-19 mask mandate on public transport from 11:59pm Thursday. ABC News [newspaper on the Internet]. 2022 Sep 21. Available from: https://www.abc.net.au/news/2022-09-21/victoria-mask-mandate-ends-trains-trams-buses-transport/101460606. Accessed 7 Jul 2024.

9. Lin C. Singapore relaxes COVID travel curbs, mask rules further. Reuters [newspaper on the Internet]. 2023 Feb 9. Available from: https://www.reuters.com/world/asia-pacific/singapore-relaxes-covid-travel-curbs-mask-rules-further-2023-02-09/. Accessed 7 Jul 2024.

10. Hong Kong SAR Government. Government lifts all mandatory mask-wearing requirements [press release]. 2023 Feb 28. Available from: https://www.info.gov.hk/gia/general/202302/28/P2023022800677.htm. Accessed 13 Mar 2025.

11. Barceló J, Sheen GC. Voluntary adoption of social welfare-enhancing behavior: mask-wearing in Spain during the COVID-19 outbreak. PloS One 2020;15:e0242764. Crossref

12. Lipsey NP, Losee JE. Social influences on mask-wearing intentions during the COVID-19 pandemic. Soc Pers Psychol Compass 2023;17:e12817. Crossref

13. Ezati Rad R, Mohseni S, Kamalzadeh Takhti H, et al. Application of the protection motivation theory for predicting COVID-19 preventive behaviors in Hormozgan, Iran: a cross-sectional study. BMC Public Health 2021;21:466. Crossref

14. Fischer-Preßler D, Bonaretti D, Fischbach K. A protection-motivation perspective to explain intention to use and continue to use mobile warning systems. Bus Inf Syst Eng 2022;64:167-82. Crossref

15. González-Castro JL, Ubillos-Landa S, Puente-Martínez A, Gracia-Leiva M. Perceived vulnerability and severity predict adherence to COVID-19 protection measures: the mediating role of instrumental coping. Front Psychol 2021;12:674032. Crossref

16. Bierbauer W, Lüscher J, Scholz U. Illness perceptions in long-COVID: a cross-sectional analysis in adults. Cogent Psychol 2022;9:2105007. Crossref

17. Lahiri A, Jha SS, Chakraborty A, Dobe M, Dey A. Role of threat and coping appraisal in protection motivation for adoption of preventive behavior during COVID-19 pandemic. Front Public Health 2021;9:678566. Crossref

18. Bandura A. The growing centrality of self-regulation in health promotion and disease prevention. Eur Health Psychol 2005;7:11-2.

19. Yu Y, Lau JT, Lau MM. Competing or interactive effect between perceived response efficacy of governmental social distancing behaviors and personal freedom on social distancing behaviors in the Chinese adult general population in Hong Kong. Int J Health Policy Manag 2022;11:498-507. Crossref

20. Ohnmacht T, Hüsser AP, Thao VT. Pointers to interventions for promoting COVID-19 protective measures in tourism: a modelling approach using domain-specific risk-taking scale, theory of planned behaviour, and health belief model. Front Psychol 2022;13:940090. Crossref

21. Nudelman G. Predicting adherence to COVID-19 behavioural guidelines: a comparison of protection motivation theory and the theory of planned behaviour. Psychol Health 2024;39:1689-705. Crossref

22. Zhou M, Liu L, Gu SY, et al. Behavioral intention and its predictors toward COVID-19 booster vaccination among Chinese parents: applying two behavioral theories. Int J Environ Res Public Health 2022;19:7520. Crossref

23. Khaday S, Li KW, Dorloh H. Factors affecting preventive behaviors for safety and health at work during the COVID-19 pandemic among Thai construction workers. Healthcare (Basel) 2023;11:426. Crossref

24. Liang W, Duan Y, Li F, et al. Psychosocial determinants of hand hygiene, facemask wearing, and physical distancing during the COVID-19 pandemic: a systematic review and meta-analysis. Ann Behav Med 2022;56:1174-87. Crossref

25. Hair Jr JF, Hult GT, Ringle CM, Sarstedt M. A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM). Los Angeles [CA]: Sage Publications; 2021.

26. Youn SY, Lee JE, Ha-Brookshire J. Fashion consumers’ channel switching behavior during the COVID-19: protection motivation theory in the extended planned behavior framework. Cloth Text Res J 2021;39:139-56. Crossref

27. Zhang X, Liu S, Wang L, Zhang Y, Wang J. Mobile health service adoption in China: integration of theory of planned behavior, protection motivation theory and personal health differences. Online Inf Rev 2020;44:1-23. Crossref

28. Ting MS, Goh YN, Isa SM. Determining consumer purchase intentions toward counterfeit luxury goods in Malaysia. Asia Pac Manage Rev 2016;21:219-30. Crossref

29. Sarstedt M, Ringle CM, Hair JF. Partial least squares structural equation modeling. In: Homburg C, Klarmann M, Vomberg AE, editors. Handbook of Market Research. Switzerland: Springer International Publishing; 2021: 587-632. Crossref

30. Mo PK, Yu Y, Lau MM, Ling RH, Lau JT. Time to lift up COVID-19 restrictions? Public support towards living with the virus policy and associated factors among Hong Kong general public. Int J Environ Res Public Health 2023;20:2989. Crossref

31. Masai AN, Akin L. Practice of COVID-19 preventive measures and risk of acute respiratory infections: a longitudinal study in students from 95 countries. Int J Infect Dis 2021;113:168-74. Crossref

32. Polonsky JA, Bhatia S, Fraser K, et al. Feasibility, acceptability, and effectiveness of non-pharmaceutical interventions against infectious diseases among crisis-affected populations: a scoping review. Infect Dis Poverty 2022;11:14. Crossref

33. Acar D, Kıcali ÜÖ. An integrated approach to COVID-19 preventive behaviour intentions: protection motivation theory, information acquisition, and trust. Soc Work Public Health 2022;37:419-34. Crossref

34. Kwok KO, Li KK, Chan HH, et al. Community responses during early phase of COVID-19 epidemic, Hong Kong. Emerg Infect Dis 2020;26:1575-9. Crossref

35. Lai DW, Jin J, Yan E, Lee VW. Predictors and moderators of COVID-19 pandemic fatigue in Hong Kong. J Infect Public Health 2023;16:645-50. Crossref

36. Rieger MO. To wear or not to wear? Factors influencing wearing face masks in Germany during the COVID-19 pandemic. Asian J Soc Health Behav 2020;3:50-4. Crossref

37. Ho HS. Use of face masks in a primary care outpatient setting in Hong Kong: knowledge, attitudes and practices. Public Health 2012;126:1001-6. Crossref

38. Mo PK, Lau JT. Illness representation on H1N1 influenza and preventive behaviors in the Hong Kong general population. J Health Psychol 2015;20:1523-33. Crossref

39. Zancu SA, Măirean C, Diaconu-Gherasim LR. The longitudinal relation between time perspective and preventive behaviors during the COVID-19 pandemic: the mediating role of risk perception. Curr Psychol 2024;43:12981-9. Crossref

40. Stefanczyk MM, Rokosz M, Białek M. Changes in perceived vulnerability to disease, resilience, and disgust sensitivity during the pandemic: a longitudinal study. Curr Psychol 2024;43:23412-24. Crossref