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 >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 ≤7 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

Over the past few decades, the widespread adoption of robotic surgical systems has revolutionised surgical management, particularly for radical prostatectomy.1 Enhanced visualisation, superior dexterity, and tremor filtration enable surgeons to maintain precision in the deep pelvis with better ergonomics. These advantages contribute to lower rates of positive surgical margins and biochemical recurrence, thereby reducing the need for salvage therapy.2 Evidence also shows that robotic surgery can significantly reduce postoperative complications and shorten the time to regain continence and potency.3

For the past two decades, the global market for robotic surgery has been dominated by the da Vinci Surgical System (Intuitive Surgical, Sunnyvale [CA], US). However, worldwide adoption of robotic surgery remains limited by the high costs associated with device acquisition, maintenance, and disposables. Consequently, access disparities persist worldwide, particularly exacerbating inequities in surgical care4 in low- and middle-income countries.5 Surgeons and trainees are also deprived of opportunities to acquire robotic skills, creating a bottleneck in the clinical application of advanced surgical technologies. To address this problem, a novel, high-performing yet affordable robotic surgical system is needed to expand access to the highest standards of surgical care.

The Sentire Surgical System (C1000) is a novel robotic surgical system developed by Cornerstone Robotics Limited (Hong Kong, China). It is the first locally developed surgical robot designed for multispecialty use. The system consists of three interconnected components: a Patient-Side Robot (PSR), a Surgeon Console, and a Vision Cart. The PSR has four robotic arms, on which an endoscope and up to three robotic surgical instruments can be mounted. The endoscope provides a high-resolution, three-dimensional image for the console surgeon. An array of instruments, including graspers, needle drivers, clip appliers, and scissors, can be installed depending on the operational requirements. The Surgeon Console includes two hand controls and a set of foot pedals, enabling surgeons to operate the surgical instruments and endoscope on the PSR, while applying energy through the surgical instruments. The Vision Cart includes a touchscreen display showing the endoscopic view for the assistant surgeon. It houses the energy source for monopolar and bipolar instruments, as well as the light and camera source for the endoscope.

This study aimed to evaluate the clinical safety and efficacy of the Sentire Surgical System by reporting outcomes from the first clinical trial for radical prostatectomy, as part of a multispecialty clinical study. Comparable safety and patient outcomes will provide supporting evidence for the continued development of this robotic surgical technology, facilitating its implementation and evaluation in further clinical trials.

This was a prospective, single-centre, single-arm study aligned with Stage 1 (Innovation) of the IDEAL (Innovation, Development, Exploration, Assessment, Long-term Study) framework.6 The study formed part of a multi-speciality clinical investigation to evaluate the safety and efficacy of the Sentire Surgical System for robot-assisted colorectal, upper gastrointestinal, and urological surgery (radical prostatectomy).

From August 2022 to September 2023, 20 adult men with clinically localised prostate cancer and planned radical prostatectomy were recruited. The inclusion criteria were: (1) age between 50 and 80 years; (2) clinical diagnosis of non–metastatic prostate cancer; (3) body mass index <35 kg/m²; (4) deemed suitable for minimally invasive treatment; and (5) provision of informed consent. Exclusion criteria were: (1) contraindication to general anaesthesia; (2) prior history of prostatic surgery; (3) untreated active infection; (4) uncorrected coagulopathy; (5) presence of other malignancies or distant metastases; and (6) membership in a vulnerable population.

All 20 robot-assisted radical prostatectomies were performed by four practising urological surgeons, each with extensive experience (>50 prior cases as chief surgeon). All participating surgeons had been trained in the use of the system and had prior experience performing procedures with the system on cadaveric and live porcine models.

Patients were placed under general anaesthesia in the lithotomy position. A total of four robotic ports were used. First, a 10-mm endoscope port was placed supraumbilically through an open approach. After pneumoperitoneum was established, three additional 8-mm robotic ports were inserted: one on either side of the supraumbilical port, adjacent to the left mid-clavicular line at approximately the vertical level of the umbilicus, and one along the right or left anterior axillary line, about 2 cm above the anterior iliac spine. A 12-mm assistant port was inserted opposite the most lateral robotic port, mirrored across the midline. Finally, a 5-mm assistant port was inserted between the midline and either the right or left mid-clavicular line (on the same side as the 12-mm assistant port) for suction and retraction (Fig 1).

Following port placement, the patient was placed in the Trendelenburg position. The PSR of the Sentire Surgical System, which consists of four robotic arms extending from a central column, was positioned between the patient’s legs. Figure 2 shows photos of the operating room during radical prostatectomy.

For most radical prostatectomy procedures, a 0° endoscope was used throughout. One surgeon alternated among the 0°, 30° up (for initial bladder detachment), and 30° down endoscopes (for bladder neck dissection) according to their preference. Bipolar Maryland Forceps (Cornerstone Robotics Limited, Hong Kong SAR, China) were used in the left hand, Monopolar Curved Scissors (Cornerstone Robotics Limited, Hong Kong SAR, China) in the right hand, and ExtraGrasp forceps (Cornerstone Robotics Limited, Hong Kong SAR, China) on the fourth robotic arm for traction. Left-hand and right-hand instruments were exchanged for the Large Needle Driver during vesicourethral anastomosis and other suturing tasks. The right-hand instrument could also be replaced with the Large Clip Applier for vessel ligation.

Each operation was performed using the standard transperitoneal anterior approach. Briefly, bladder dissection was carried out in the areolar tissue plane between the peritoneum and transversalis fascia after the peritoneum had been incised at the lateral borders of the urinary bladder. Dissection continued caudally to develop the retropubic space. Preprostatic fat was removed for enhanced anatomical localisation. Dissection of the endopelvic fascia was then performed. The dorsal venous complex was ligated according to surgeon preference. Bladder neck dissection followed, then dissection of the vas deferens and seminal vesicles. The posterior plane of the prostate was then dissected, continuing to the prostatic apex. Lateral dissection was performed next, with the degree of nerve sparing determined by preoperative erectile function, tumour location, and tumour grade. Apical dissection was performed and the urethra was then divided.

Haemostasis was secured by further suturing of the dorsal venous complex, if not previously completed. If indicated, pelvic lymph node dissection was performed at this stage. Posterior Rocco’s stitch reconstruction was carried out, followed by vesicourethral anastomosis with continuous sutures. After confirming watertightness, a pelvic drain was placed via the most lateral robotic port. The robot was then undocked and specimens were removed in a specimen bag through an extended camera port incision. Wounds were closed using standard techniques.

Primary endpoints included the conversion rate and the incidence of perioperative complications within 30 days. Conversion was defined as an emergent change to a conventional laparoscopic or open approach. Intraoperative events and postoperative complications during the hospital stay and within 30 days of discharge were recorded. The severity of complications was graded according to the Clavien–Dindo Classification. Whether complications were anticipated was determined based on their consistency with the current investigational plan or consent form. In the event of postoperative complications, the relationship to the Sentire Surgical System, the surgical procedure, and the patient’s underlying condition was documented, along with actions taken and outcomes.

Secondary endpoints included perioperative and pathological outcomes. Perioperative outcomes included total operative time, docking time, total console time, estimated blood loss, time to resume regular activity, length of hospital stay, time of drain and removal, time of urethral catheter removal, visual analogue scale (VAS) pain scores at 14 and 30 days, and use of pads for urinary incontinence at 14 days and 30 days. Pathological outcomes included surgical margins, lymph node metastasis, tumour involvement, pathological tumour stage, number of lymph nodes harvested, and postoperative prostate-specific antigen (PSA) level.

Data are presented in a descriptive manner. Continuous variables are reported as means with standard deviations (SDs) or medians with interquartile ranges, whereas categorical variables are presented as percentages. All analyses were performed using SPSS (Windows version 25.0; IBM Corp, Armonk [NY], US).

Twenty patients were enrolled in the trial, and all successfully underwent robot-assisted radical prostatectomy using the Sentire Surgical System. The mean age was 68.5 years (SD=4.1) and the mean body mass index was 24.2 kg/m² (SD=2.8). Two patients had a history of abdominal surgery: Case 3 had undergone open appendicectomy, and Case 11 had undergone laparoscopic cholecystectomy. Seventeen patients underwent radical prostatectomy only, while three also underwent lymphadenectomy. The mean prostate volume was 40.7 cc (SD=17.8). Nine patients had a Gleason score of 6, nine had a score of 7, and two had a score of 8. Five patients were classified as low-risk, 11 as intermediate-risk, and four as high-risk. Two patients received neoadjuvant hormonal therapy due to concerns about prolonged waiting times during the coronavirus disease 2019 period; their mean initial PSA level dropped from 16.2 ng/mL to 2.19 ng/mL. The mean preoperative PSA level for patients who did not receive neoadjuvant therapy was 15.2 ng/mL (SD=20.8) [Table 1].

All operations were completed without conversion. There were no intraoperative complications related to the robotic device. Minor surgical complications (Grade I-II) occurred in seven patients, including four cases of wound infection, one case of urinary retention, one case of prolonged anastomotic leakage, one case of bilateral lower limb rash, and one case of left loin pain. Only five of these complications—namely wound infections and prolonged anastomotic leakage—were considered related to the surgery (Table 2).

The mean total operative time (from skin incision to closure) was 184.5 minutes, including a mean docking time of 4.2 minutes and a mean total console time of 149.7 minutes. The median estimated blood loss was 175.0 mL. Postoperatively, the median time to resumption of regular activity was 7.0 days, the median duration of drainage was 2.0 days and the median duration of urethral catheterisation was 7.0 days. The median length of hospital stay was 3.0 days. The median VAS pain score at the 14-day follow-up was 3.0, whereas at the 30-day follow-up it was 0. Pad usage for urinary incontinence was 3.5 at the 14-day follow-up and 2.0 at the 30-day follow-up (Table 3).

Regarding pathological outcomes, 13 of the 20 patients (65%) had negative surgical margins, including the two who received neoadjuvant hormonal therapy. Seventeen patients (85%) had undetectable PSA levels 1 month after surgery, while two other patients (95%) achieved undetectable levels at 3 months. Among the three patients who underwent lymphadenectomy, the mean number of lymph nodes harvested was 9 (range=5-11) [Table 4].

This prospective study reports the results of the first-in-human clinical trial of the Sentire Surgical System—the first locally developed multispecialty surgical robot—used for radical prostatectomy. All 20 procedures were completed successfully, without conversions and device-related intraoperative or postoperative complications, demonstrating that this novel surgical system is safe and effective for robot-assisted radical prostatectomy.

Since the introduction of robotic surgery in Hong Kong in 2005, its use—particularly in urology—has steadily expanded.7 According to the latest Surgical Outcome Monitoring and Improvement Programme report (2022-2023), almost all radical prostatectomies (409 of 410 patients) were performed using robotic surgery.8 Improved outcomes have been a key factor supporting its development.9 However, the high cost of robotic systems considerably limits further expansion and popularisation. With the expiration of certain technological patents, new robotic systems are rapidly emerging worldwide.10 The development of a locally based robotic system in Hong Kong represents an important milestone for the future of robotic surgery in the region.

Perioperative outcomes in this trial were favourable, with a mean total operative time of 184.5 minutes, comparable to reported times in the existing literature on both laparoscopic and robotic radical prostatectomy.11 The short docking time reflected a smooth docking process. Most cases had acceptable blood loss, and no patients required transfusion, again comparable to the literature.11 Despite the higher-than-expected wound infection rate, most infections were minor and managed with simple dressing. No specific cause was identified, and these events are unlikely to be related to the robotic system. Postoperative hospital stay was also short, with satisfactorily low VAS pain scores. Early functional outcomes, specifically pad usage for urinary incontinence, were also satisfactory (Table 3).

Short-term oncological outcomes also support the system’s efficacy, such that 65% of patients achieved a negative surgical margin and 85% and 95% of patients showed no PSA persistence (PSA level ≥0.1 ng/mL) at 30 days and 3 months after surgery, respectively—results that are comparable to the literature.12 In the three cases where lymphadenectomy was performed, the mean lymph node yield was 9, which is considered optimal for balancing the biochemical recurrence-free rate and complication risk.13 These findings indicate that the system can achieve oncological outcomes in radical prostatectomy comparable to those of established robotic systems.

Compared with our previously reported outcomes using other robotic systems,14 the performance of the current system was highly comparable. The mean operative time and hospital stay were 184.5 minutes/3.1 days (Sentire C1000), 225.8 minutes/3.3 days,14 (da Vinci S), and 223.7 minutes/3.0 days,14 (da Vinci SP). These results not only demonstrate that the Sentire C1000 offers performance comparable to existing robotic systems but also confirm the ease of surgical transition. A key factor enabling this transition is the similarity of the robotic control interface—including hand controls and foot pedals, allowing surgeons to adopt the new system more readily and apply their existing robotic experience. This is akin to driving different brands of cars, where consistent interfaces such as the steering wheel and pedals enable smooth adaptation. As a result, most new robotic systems adopt similar interface designs to support seamless integration by experienced surgeons. Moreover, the PSR has a similar arm configuration and setup to the da Vinci system, which also helps both surgeons and nursing staff adapt to the setup.

As this is the initial report of the Sentire Surgical System’s clinical use in radical prostatectomy, further clinical studies are warranted to evaluate its performance across a broader range of procedures.

Our findings indicate that the Sentire Surgical System (C1000) is a safe and effective robotic platform for radical prostatectomy. Its successful performance in this first-in-human experience supports ongoing development and broader application of this novel surgical robotic system.

Concept or design: CF Ng.
Acquisition of data: CH Yee, PKF Chiu, MHM Tam, FPT Lai.
Analysis or interpretation of data: FPF Lai.
Drafting of the manuscript: CF Ng, FPF Lai.
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, CF Ng was not involved in the peer review process. Other authors have disclosed no conflicts of interest.

The authors thank the support from the operating theatre staff in Prince of Wales Hospital.

The research was supported by Cornerstone Robotics Limited for equipment and research fund. The funder had no role in the study design, data collection/analysis/interpretation, or manuscript preparation.

The research was approved by the Joint Chinese University of Hong Kong—New Territories East Cluster Clinical Research Ethics Committee, Hong Kong (Ref No.: CREC 2021.472). It was registered on www.ClinicalTrials.gov (Ref No.: NCT05151835). Written informed consent was obtained from all patients for the publication of this research and the accompanying images.

1. Ng AT, Tam PC. Current status of robot-assisted surgery. Hong Kong Med J 2014;20:241-50. Crossref

2. Okegawa T, Omura S, Samejima M, et al. Laparoscopic radical prostatectomy versus robot-assisted radical prostatectomy: comparison of oncological outcomes at a single center. Prostate Int 2020;8:16-21. Crossref

3. Allan C, Ilic D. Laparoscopic versus robotic-assisted radical prostatectomy for the treatment of localised prostate cancer: a systematic review. Urol Int 2016;96:373-8. Crossref

4. Bansal E, Kunaprayoon S, Zhang LP. Opportunities for global health diplomacy in transnational robotic telesurgery. AMA J Ethics 2023;25:E624-36. Crossref

5. Mehta A, Cheng Ng J, Andrew Awuah W, et al. Embracing robotic surgery in low- and middle-income countries: potential benefits, challenges, and scope in the future. Ann Med Surg (Lond) 2022;84:104803. Crossref

6. IDEAL Collaboration. The IDEAL Framework. Available from: https://www.ideal-collaboration.net/the-ideal-framework/. Accessed 26 May 2025.

7. Chan SY, Hou SM, Wong WS, Ng CF. Robotic urological surgery: prospects for Hong Kong. Surg Pract 2007;11:154-8. Crossref

8. Surgical Outcomes Monitoring and Improvement Programme (SOMIP) report. Volume Fifteen: July 2022—June 2023. Hospital Authority, Hoong Kong SAR Government; 2024.

9. Lo KL, Ng CF, Lam CN, Hou SS, To KF, Yip SK. Short-term outcome of patients with robot-assisted versus open radical prostatectomy: for localised carcinoma of prostate. Hong Kong Med J 2010;16:31-5.

10. Marchegiani F, Siragusa L, Zadoroznyj A, et al. New robotic platforms in general surgery: what’s the current clinical scenario? Medicina (Kaunas) 2023;59:1264. Crossref

11. Novara G, Ficarra V, Rosen RC, et al. Systematic review and meta-analysis of perioperative outcomes and complications after robot-assisted radical prostatectomy. Eur Urol 2012;62:431-52. Crossref

12. Wu S, Lin SX, Cornejo KM, et al. Clinicopathological and oncological significance of persistent prostate-specific antigen after radical prostatectomy: a systematic review and meta-analysis. Asian J Urol 2023;10:317-28. Crossref

13. Zhang X, Zhang G, Wang J, Bi J. Different lymph node dissection ranges during radical prostatectomy for patients with prostate cancer: a systematic review and network meta-analysis. World J Surg Oncol 2023;21:80. Crossref

14. Ng CF, Teoh JY, Chiu PK, et al. Robot-assisted single-port radical prostatectomy: a phase 1 clinical study. Int J Urol 2019;26:878-83. 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

Child homicide represents a rare but important global issue with devastating consequences for families and communities. The global homicide rate among children aged 0 to 17 years was 1.6 per 100 000 population in 2016,1 and approximately 95 000 children are murdered annually.2 A 2017 review by Stöckl et al3 found that the majority of child homicides were committed by a family member; parents were responsible for over half of the cases involving child victims.3

Filicide refers to the act of killing one’s own child. Subcategories of filicide include neonaticide, a term introduced by Resnick4 to describe the murder of a child within the first 24 hours after birth, and infanticide, which applies when the victim is aged <1 year. Resnick4 identified various motives for filicide. In altruistic filicide, the parent believes that the act is in the child’s best interests. An acutely psychotic parent may kill a child under the influence of severe mental illness. In unwanted child filicide, a parent kills a child who is perceived as a hindrance. Accidental/fatal maltreatment describes the unintentional death of a child due to parental abuse or neglect. Spouse revenge filicide occurs when a child is killed as a means of exacting revenge upon the spouse or the other parent. Bourget and Bradford5 later emphasised the importance of the perpetrator’s gender by introducing paternal filicide as a distinct category.

Victim and perpetrator characteristics vary in cases of filicide. The first year of life is a critical period, and the highest risk of filicide occurs within the first 24 hours. Neonaticides are predominantly committed by mothers,6 and mothers are overrepresented across the entire spectrum of filicide.4 5 However, contradictory results have been reported.5 7 8 The gender distribution of victims also varies. Male children aged <1 year are at greater risk in high-income Western countries, such as the US9 and the UK10; the opposite trend has been observed in India and China.11 Some studies have shown that boys are overrepresented among victims,7 12 whereas others have identified comparable numbers of male and female filicide victims.13

Maternal and paternal perpetrators of filicide exhibit distinct characteristics.14 15 Maternal perpetrators tend to be younger and have younger victims compared with fathers.15 Younger maternal perpetrators are often poor, experience psychosocial stress, and lack family and community support, whereas older maternal perpetrators frequently have mental illnesses and lack criminal histories.13 14 16 In contrast, paternal perpetrators are more commonly driven by anger, jealousy, or marital and life discord.15 Fatal abuse and acts of retaliation are more prevalent among paternal perpetrators than among maternal perpetrators.17 Fathers are also more likely to attempt or die by suicide12 17 18 when committing filicide.14 18 Additionally, fathers typically use more violent methods to cause death.19

Pathological filicide, characterised by altruistic or actively psychotic motives, constitutes one of the most common categories of filicide.17 Psychiatric factors are involved in 36% to 85% of all filicide cases.5 16 20 21 22 Maternal perpetrators are more likely to have a history of mental illness and to exhibit symptoms at the time of the offence.22 The most frequent diagnosis among maternal perpetrators is major depressive disorder, followed by schizophrenia.5 16 20 23 Personality disorders and substance use are more often associated with paternal filicides.8

Filicide presents unique challenges for the criminal justice system. Societal attitudes regarding parents who kill their children are often ambivalent, balancing the need for justice due to loss of innocent life against calls for mercy towards offenders who may require care rather than punishment.

Legal systems worldwide acknowledge that filicide should be treated differently from other forms of homicide. The UK enacted the Infanticide Act in 1922 (amended in 1938)24 to recognise the biological vulnerability of women to psychiatric illnesses during the perinatal period. The Act mandated sentences of probation and psychiatric treatment for offenders.24 By the late 20th century, 29 countries had revised penalties for infanticide to consider unique biological and psychological changes associated with childbirth.25

In Hong Kong, perpetrators with mental illnesses can invoke psychiatric defences, including insanity or diminished responsibility. The insanity defence is based on the M’Naghten principles, which hold that it is unjust to punish an individual for an action performed without the mental capacity to control it. The defence of diminished responsibility applies when the offender demonstrates abnormal mental function arising from a recognised medical condition, which has substantially impaired their ability to either understand the nature of their conduct, form a rational judgement, or exercise self-control (or any combination of these impairments). Perpetrators with mental illnesses who are found not guilty by reason of insanity, or who successfully raise the partial defence of diminished responsibility—thereby reducing the charge from murder to manslaughter—may be sentenced to a Hospital Order at the Correctional Services Department Psychiatric Centre (Siu Lam Psychiatric Centre [SLPC]), under Section 75 of the Criminal Procedure Ordinance26 or Section 45 of the Mental Health Ordinance,27 respectively, for psychiatric observation and management.

A separate legal provision exists for mothers who kill their children aged <1 year. Hong Kong has adopted the UK concept of infanticide, in which mothers experiencing vulnerability after childbirth are charged with infanticide rather than murder, under Section 47C of the Offences against the Person Ordinance.28

A study has shown that the local homicide rate in Hong Kong is lower than global averages (0.32 vs 6.1 victims per 100 000 population in 2017),29 but no filicide-specific data are available. The underlying hypothesis in this study was that the incidence of filicide would be lower in Hong Kong than in Western countries, consistent with the lower local homicide rate and the protective effects of cultural factors. The objectives of this study were to describe the epidemiology of filicide in Hong Kong, examine the characteristics of victims and perpetrators (including associated mental illnesses), and evaluate the local criminal justice system’s response to infanticide and other forms of filicide.

Data were obtained from the Hong Kong Police Force regarding child homicide cases that occurred from 2003 to 2017. These data included the age and gender of the victim, relationship of the perpetrator to the victim, mode of death, year of offence, and charges against the defendant along with corresponding outcomes and sentences. Medical records from the Hospital Authority and the SLPC of the Correctional Services Department were reviewed to determine any history of mental illness. Psychiatric diagnoses of the perpetrators, based on the International Classification of Diseases, Tenth Revision, were documented during forensic psychiatric assessments conducted by two psychiatrists, at least one of whom was a specialist. For the minority of defendants who were not sent to psychiatric hospitals or SLPC after the offences, the presence or absence of mental illness was cross-referenced using newspaper articles. Charges and sentences were verified through judgements available on the Judiciary’s official website.

All statistical analyses were performed using SPSS software (Windows version 21.0; IBM Corp, Armonk [NY], US). Data were analysed with descriptive statistics, including the mean, median, standard deviation, 95% confidence interval, and percentages for categorical variables. Differences between groups in demographic characteristics were assessed using t tests and univariate analysis of variance for continuous data. For nominal data, the Kruskal–Wallis and Chi squared tests were utilised.

From 2003 to 2017, 107 child homicide victims were recorded in Hong Kong, equating to approximately 0.70 death per 100 000 population, based on a population of 1 024 000 children aged <18 years in 2010.30 Among these victims, 81 (75.7%) were killed by their parents (Fig).

Among the filicide victims (n=81), 53.1% were female, 45.7% were male, and the gender of the remaining victim was unknown. There was no significant correlation between the gender of the victim and the gender of the perpetrator (χ²=0.13; P=0.82). The median age of the victims was 6 years (interquartile range [IQR]=0-8) [Table 1].

Of the 81 filicide victims, 54 (66.7%) were killed by their mothers, 16 (19.8%) by their fathers, and 11 (13.6%) by both parents. The median age of victims varied across perpetrator groups; the paternal group victims had a median age of 7.5 years (IQR=5-10.25), compared with 0 year (IQR=0-3.5) for the maternal group and 2 years (IQR=0-5) for the parental couple group (H²=14.31; P<0.001).

Forty victims aged <1 year were killed by 44 perpetrators, and 41 victims aged ≥1 year were killed by 40 perpetrators. No significant gender differences were observed among the victims (Table 1).

The median age of paternal perpetrators, 43.5 years, was significantly older than the median ages of maternal and parental couple perpetrators (H2=16.50; P<0.001). The median age of offenders in the infanticide group was younger than that of offenders in the other filicide group. In the infanticide group, nine mothers (26.5%) were <20 years, and all pregnancies had been concealed. These infants were killed immediately after birth. Single offenders were more prevalent in the infanticide group, whereas married offenders were more common in the other filicide group. Biological mothers were the main perpetrators in both groups; similar to paternal and couple perpetrators, maternal perpetrators were younger in the infanticide group (Table 2). The maternal group was responsible for 40% of victims aged <4 years, compared with 7.1% in the paternal group. A higher prevalence of mental illness was identified among perpetrators, particularly mothers, in the other filicide group. Among perpetrators in the infanticide group, depression (40%) was the most common diagnosis, followed by a psychotic disorder (20%), mental and behavioural disorders due to psychoactive substance use (20%), and mental retardation (20%). The only biological father in the infanticide group was diagnosed with harmful use of alcohol. In the other filicide group, among maternal perpetrators, 25.0% had a psychotic disorder, 18.8% had depression, 6.4% had bipolar affective disorder, and the remainder had unknown diagnoses. Among paternal perpetrators, 18.0% had depression, 9.1% had a psychotic disorder, and the remainder had undocumented diagnoses.

Suffocation or strangulation was the most common mode of death in infanticides, occurring in 95.7% of cases with maternal perpetrators. In contrast, paternal perpetrators (100%) and couples (50%) caused death mainly by bashing, throwing, or shaking the infants. The two most common modes of death across all filicides were drug overdose or poisoning (including charcoal burning) and stabbing. Drug overdose or poisoning was most frequently performed by maternal perpetrators (36.8%) and couples (57.1%), whereas paternal perpetrators most often engaged in stabbing (57.1%).

Excluding the four perpetrators who died by suicide, 80.0% of perpetrators in the infanticide group faced criminal charges and were convicted. The most common convictions were concealing the birth of a child, manslaughter, and infanticide (Table 2). In the other filicide group, excluding the 18 perpetrators who died by suicide, 95.5% of perpetrators were charged and convicted; manslaughter was the most common conviction, followed by murder. Sentences significantly differed between the infanticide and other filicide groups. Noncustodial sentences were more frequent in the infanticide group than in the other filicide group. Given the higher prevalence of mental illness in the other filicide group, 33.3% (5/15) of the perpetrators were convicted of manslaughter under diminished responsibility and sentenced to a Hospital Order, compared with 6.3% in the infanticide group (Table 2). Among paternal and couple perpetrators, 80% in the infanticide group and 92.3% in the other filicide group received prison sentences, ranging from 3 to 10 years and 18 months to life imprisonment, respectively. Similar proportions of maternal perpetrators in both groups—41.0% in the infanticide group and 42.9% in the other filicide group—were imprisoned. Among maternal perpetrators in the infanticide group, all but one received prison sentences of <1 year; the exception received an 8-year sentence. In the other filicide group, maternal perpetrators received sentences of 4 to 7 years.

Filicide-suicide is defined as the perpetrator dying by suicide within 24 hours of committing filicide. A significantly greater proportion of filicide-suicides occurred in the other filicide group. In the infanticide group, all perpetrators were biological mothers. In contrast, within the other filicide group, half of maternal perpetrators and 66.7% of paternal perpetrators had a diagnosed mental illness. The difference in mental illness prevalence between the two groups was not statistically significant (Table 3).

Of the 84 filicide perpetrators, 26 (31.0%) were diagnosed with mental illness. No mental illness was reported in the parental couple group. A higher prevalence of mental illness was observed among paternal perpetrators (58.3%) than among maternal perpetrators (38.0%), although the difference was not statistically significant. Depression was the most common diagnosis, followed by psychotic disorder. In cases of filicide-suicide, mental illness prevalence was higher among paternal perpetrators; this difference was not statistically significant (Table 4).

Excluding perpetrators who died by suicide, 41.7% of maternal perpetrators with mental illness received a Hospital Order for an unspecified period. Among the three paternal perpetrators with mental illness who did not die by suicide, only one (33.3%) was sentenced to a Hospital Order for an unspecified period.

The incidence of child homicide in Hong Kong, at 0.7 per 100 000 population, is lower than the global average (1.6 per 100 000 population)1 and lower than that of Asian countries with similar socio-economic status, such as South Korea (1.03 per 100 000 population).31 The protective influence of traditional Confucian cultural values may play a prominent role in Hong Kong.32 An idiom from the Sung dynasty, ‘even a vicious tiger would not eat its cubs’, continues to be taught in modern primary schools. This cultural ethos could explain why the incidence of child maltreatment in Hong Kong, at <0.14%,30 remains lower than the global rate of 0.3% to 0.4%.33 Consistent with studies worldwide,3 most child homicides in Hong Kong were perpetrated by parents. Mothers were the predominant perpetrators in filicides. The typical profile of an infanticidal perpetrator was a young, single mother who suffocated or strangled the infant. Some cases may represent neonaticides, as suggested by charges of concealing the birth of a child. Among cases involving the filicide of older children, perpetrator characteristics were more heterogeneous. Perpetrators tended to be older and married; they used methods such as overdosing, poisoning, or stabbing. The profiles of perpetrators and victims in this group also differed. The median age of maternal perpetrators was younger and their victims tended to be younger. Mothers most often caused death through overdosing or poisoning, whereas fathers were more likely to kill by stabbing.

In the present study, 31.0% of filicidal perpetrators had a diagnosed mental illness, a lower rate compared with other population studies.8 20 22 23 This discrepancy could be attributed to the lower prevalence of mental illness in Hong Kong. The Hong Kong Mental Morbidity Survey (2010-2013) revealed a 13.3% prevalence of mental disorders among Chinese adults,34 compared with 18.5% among adults in the US in 2013.35 It is also plausible that some perpetrators, especially those involved in filicide-suicide cases, had no prior contact with mental health services and may have had undiagnosed psychiatric illnesses. Mental illness prevalence was higher among paternal perpetrators than among maternal perpetrators in our filicide sample. This finding may be related to the small sample size or could reflect societal changes, such as fathers assuming greater childcare responsibilities.17 Consistent with some studies,20 22 depression was the most common diagnosis, followed by psychotic disorder.

Substantial proportions of filicide perpetrators (23.0% of maternal and 34.8% of paternal) died by suicide during or after committing the act. Charcoal burning was the most common method, comparable to the frequency of jumping from height. Charcoal burning is a relatively recent suicidal method,36 which has spread as a contagious phenomenon in other Asian countries; it is often portrayed as a ‘peaceful way of dying’ and has been used during >10% of suicides in the region.37 The proportion of filicide-suicides observed in this study was lower than that reported in other studies.17 23 This difference may be related to the lower prevalence of mental illness in our sample, the relatively lower lethality of charcoal burning in Hong Kong compared with firearm use in Western countries, or the possibility that attempted suicides not resulting in death were not captured in our data. Filicide-suicide events were more frequent in cases involving older children than in infanticides, potentially due to differences in underlying motives. Half of the filicide-suicide perpetrators in the present study had a history of mental illness, suggesting that altruistic motives were involved. Depression was the most frequently diagnosed condition in these cases.18 20

The majority of perpetrators with mental illness were convicted of manslaughter under diminished responsibility and sentenced to a Hospital Order at SLPC for an unspecified period under Section 45 of the Mental Health Ordinance.27 No insanity pleas were recorded in our sample. Consistent with international studies,38 maternal perpetrators in Hong Kong received more lenient outcomes relative to paternal perpetrators. Some young mothers who killed their children aged <1 year were released without charge; among those convicted, a few received noncustodial sentences. In contrast, all fathers who killed their children were imprisoned, with the exception of one who was sentenced to a Hospital Order at SLPC.

Hong Kong developed its legislation based on the UK law, including the British Infanticide Act of 1922.21 24 Section 47C of the Offences against the Person Ordinance28 defines the offence of infanticide as follows: “Where a woman by any wilful act or omission causes the death of her child being a child under the age of 12 months but at the time of the act or omission the balance of her mind was disturbed by reason of her not having fully recovered from the effect of giving birth to the child or by reason of the effect of lactation consequent upon the birth of the child, then, notwithstanding that the circumstances were such that but for the provisions of this section the offence would have amounted to murder, she shall be guilty of infanticide, and shall be liable to be punished as if she were guilty of manslaughter.” In the present study, eight mothers who killed their children aged <1 year were convicted under the infanticide provision. There appears to be considerable application of this provision in Hong Kong; lenient noncustodial sentences are issued to mothers in such cases.

First, information provided by the Police was restricted to arrest cases; thus, the study may underreport the true incidence of filicides in Hong Kong. Second, although multiple sources of information were utilised, details regarding the perpetrators’ and victims’ abuse or victimisation histories, involvement with social services, or autopsy reports were unavailable. Third, the classification of neonaticides was challenging, although charges of concealing the birth of a child may indicate the death of a victim within 24 hours of birth. Fourth, although most diagnoses of offenders with mental illnesses were accessible, the availability of psychiatric records was limited. Information for a small number of cases (<5) was obtained from newspaper reports. Sixth, the absence of critical details, such as the onset of mental illness, symptomatology, and medication adherence, impeded a thorough exploration of the relationship between mental illness and filicides. A more comprehensive approach, such as conducting psychological autopsies—particularly in filicide-suicide cases—would provide deeper insights. Finally, the sample size was insufficient to allow for robust comparisons among perpetrators in maternal, paternal, parental couple, and stepparent filicide groups.

In this study, most child homicides were perpetrated by parents; mothers committed filicide more frequently than fathers. Maternal perpetrators and their victims were younger than their counterparts in the paternal perpetrator group. Mental illness was prevalent among filicidal perpetrators of both genders, with a higher prevalence in paternal perpetrators. Filicide-suicide is a substantial problem. Psychiatrists should remain vigilant in identifying depressed or psychotic parents and in eliciting self-harm or filicidal ideations among both mothers and fathers. Social support and child protection services should be actively offered to young single mothers. In Hong Kong, a comprehensive child development service has been established since 2005,39 with the aim of identifying and intervening early in cases that involve children and mothers in need; this service seeks to improve health outcomes for children and families. However, no local policies specifically address the needs of fathers. A multidisciplinary approach involving mental health professionals and social workers is recommended to screen fathers experiencing mental illness or distress and to identify early warning signs of risk. Finally, given the high prevalence of mental illness among filicidal perpetrators, forensic psychiatrists and related professionals should maintain a high index of suspicion for the presence of mental illness when evaluating filicidal offenders.

Concept or design: All authors.
Acquisition of data: YDY Tang.
Analysis or interpretation of data: YDY Tang, JPY Lam.
Drafting of the manuscript: YDY Tang.
Critical revision of the manuscript for important intellectual content: YDY Tang.

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 New Territories West Cluster Research Ethics Committee of the Hospital Authority, Hong Kong (Ref No.: NTWC/REC/19021). A waiver for informed patient consent was granted by the Committee due to the retrospective nature of the research.

1. United Nations Office on Drugs and Crime. Global Study on Homicide. Killing of Children and Young Adults. 2019. Available from: https://www.unodc.org/documents/data-and-analysis/gsh/Booklet_6new.pdf. Accessed 6 Apr 2023.

2. UNICEF. Hidden in plain sight: a statistical analysis of violence against children. New York: United Nations International Children’s Emergency Fund. 2014 Sep 4. Available from: https://data.unicef.org/resources/hidden-in-plain-sight-a-statistical-analysis-of-violence-against-children/. Accessed 6 Apr 2023.

3. Stöckl H, Dekel B, Morris-Gehring A, Watts C, Abrahams N. Child homicide perpetrators worldwide: a systematic review. BMJ Paediatr Open 2017;1:e000112. Crossref

4. Resnick PJ. Child murder by parents: a psychiatric review of filicide. Am J Psychiatry 1969;126:325-34. Crossref

5. Bourget D, Bradford JM. Homicidal parents. Can J Psychiatry 1990;35:233-8. Crossref

6. Wilson RF, Klevens J, Fortson B, Williams D, Xu L, Yuan K. Neonaticides in the United States—2008-2017. Acad Forensic Pathol 2022;12:3-14. Crossref

7. Vanamo T, Kauppi A, Karkola K, Merikanto J, Räsänen E. Intra-familial child homicide in Finland 1970-1994: incidence, causes of death and demographic characteristics. Forensic Sci Int 2001;117:199-204. Crossref

8. Bourget D, Gagné P. Paternal filicide in Québec. J Am Acad Psychiatry Law 2005;33:354-60.

9. Mariano TY, Chan HC, Myers WC. Toward a more holistic understanding of filicide: a multidisciplinary analysis of 32 years of U.S. arrest data. Forensic Sci Int 2014;236:46-53. Crossref

10. Brookman F, Nolan J. The dark figure of infanticide in England and Wales: complexities of diagnosis. J Interpers Violence 2006;21:869-89. Crossref

11. Sahni M, Verma N, Narula D, Varghese RM, Sreenivas V, Puliyel JM. Missing girls in India: infanticide, feticide and made-to-order pregnancies? Insights from hospital-based sex-ratio-at-birth over the last century. PLoS One 2008;3:e2224. Crossref

12. Dawson M. Canadian trends in filicide by gender of the accused, 1961–2011. Child Abuse Negl 2015;47:162-74. Crossref

13. Camperio Ciani AS, Fontanesi L. Mothers who kill their offspring: testing evolutionary hypothesis in a 110-case Italian sample. Child Abuse Negl 2012;36:519-27. Crossref

14. Harris GT, Hilton NZ, Rice ME, Eke AW. Children killed by genetic parents versus stepparents. Evol Hum Behav 2007;28:85-95. Crossref

15. West SG, Friedman SH, Resnick PJ. Fathers who kill their children: an analysis of the literature. J Forensic Sci 2009;54:463-8. Crossref

16. Friedman SH, Horwitz SM, Resnick PJ. Child murder by mothers: a critical analysis of the current state of knowledge and a research agenda. Am J Psychiatry 2005;162:1578-87. Crossref

17. Bourget D, Grace J, Whitehurst L. A review of maternal and paternal filicide. J Am Acad Psychiatry Law 2007;35:74-82.

18. Hatters Friedman S, Hrouda DR, Holden CE, Noffsinger SG, Resnick PJ. Filicide-suicide: common factors in parents who kill their children and themselves. J Am Acad Psychiatry Law 2005;33:496-504.

19. West SG, Hatters Friedman S. Filicide: a research update. In: Browne RC, editor. Forensic Psychiatry Research Trends. New York: Nova Science Publishers; 2007: 29-62.

20. Bourget D, Gagné P. Maternal filicide in Québec. J Am Acad Psychiatry Law 2002;30:345-51.

21. Hatters Friedman S, Resnick PJ. Child murder by mothers: patterns and prevention. World Psychiatry 2007;6:137-41.

22. Flynn SM, Shaw JJ, Abel KM. Filicide: mental illness in those who kill their children. PLoS One 2013;8:e58981. Crossref

23. Kauppi A, Kumpulainen K, Karkola K, Vanamo T, Merikanto J. Maternal and paternal filicides: a retrospective review of filicides in Finland. J Am Acad Psychiatry Law 2010;38:229-38.

24. Legislation.gov.uk. Infanticide Act 1938. Available from: https://www.legislation.gov.uk/ukpga/Geo6/1-2/36. Accessed 25 Mar 2025.

25. Oberman M. Mothers who kill: coming to terms with modern American infanticide. Am Crim L Rev 1996;34:1-110.

26. Hong Kong SAR Government. Criminal Procedure Ordinance (Cap 221). Available from: https://www.elegislation.gov.hk/hk/cap221. Accessed 17 Mar 2025.

27. Hong Kong SAR Government. Mental Health Ordinance (Cap 136). Available from: https://www.elegislation.gov.hk/hk/cap136. Accessed 6 Apr 2023.

28. Hong Kong SAR Government. Offences against the Person Ordinance (Cap 212). Available from: https://www.elegislation.gov.hk/hk/cap212. Accessed 6 Apr 2023.

29. Hong Kong Police Force. Crime statistics comparison. 2017. Available from: https://www.police.gov.hk/ppp_en/09_statistics/csc.html. Accessed 6 Apr 2023.

30. Child Fatality Review Panel, Social Welfare Department, Hong Kong SAR Government. Second report for child death cases in 2010-2011. July 2015. Available from: https://www.swd.gov.hk/storage/asset/section/655/en/fcw/CFRP2R-Eng.pdf. Accessed 6 Apr 2023.

31. Jung K, Kim H, Lee E, et al. Cluster analysis of child homicide in South Korea. Child Abuse Negl 2020;101:104322. Crossref

32. Lassi N. A Confucian theory of crime [dissertation]. University of North Dakota; 2018.

33. Stoltenborgh M, Bakermans-Kranenburg MJ, Alink LR, van IJzendoorn MH. The prevalence of child maltreatment across the globe: review of a series of meta-analyses. Child Abuse Rev 2015;24:37-50. Crossref

34. Lam LC, Wong CS, Wang MJ, et al. Prevalence, psychosocial correlates and service utilization of depressive and anxiety disorders in Hong Kong: the Hong Kong Mental Morbidity Survey (HKMMS). Soc Psychiatry Psychiatr Epidemiol 2015;50:1379-88. Crossref

35. Center for Behavioral Health Statistics and Quality, Substance Abuse and Mental Health Services Administration, US Department of Health and Human Services. Results from the 2013 National Survey on Drug Use and Health: Mental Health Findings. November 2014. Available from: https://www.samhsa.gov/data/sites/default/files/NSDUHmhfr2013/NSDUHmhfr2013.pdf. Accessed 6 Apr 2023.

36. The Hong Kong Jockey Club Centre for Suicide Research and Prevention, The University of Hong Kong. Statistics of suicide data in Hong Kong (by year). Distribution of method of suicide by age group in Hong Kong. 2020. Available from: https://www.csrp.hku.hk/statistics/. Accessed 6 Apr 2023.

37. Chang SS, Chen YY, Yip PS, Lee WJ, Hagihara A, Gunnell D. Regional changes in charcoal-burning suicide rates in East/ Southeast Asia from 1995 to 2011: a time trend analysis. PLoS Med 2014;11:e1001622. Crossref

38. Porter T, Gavin H. Infanticide and neonaticide: a review of 40 years of research literature on incidence and causes. Trauma Violence Abuse 2010;11:99-112. Crossref

39. Education Bureau, Hong Kong SAR Government. Comprehensive Child Development Service. 2021. Available from: https://www.edb.gov.hk/en/edu-system/preprimary-kindergarten/comprehensive-child-development-service/index.html. Accessed 18 Mar 2025.