Elderly patients have multiple co-morbidities and they are consequently prone to multiple medication use. Inappropriate medication use is common among hospitalised older adults. The number of drugs taken is one of the important determinants of risk for receiving an inappropriate medication.1 There is a high prevalence of unnecessary drug use in frail older people. In one hospital study, 44% of patients were prescribed at least one unnecessary drug, with the most common reason being lack of indication.2 The most commonly prescribed unnecessary drug classes were gastrointestinal, central nervous system, and therapeutic nutrients/minerals.2 Appropriate use of drugs is particularly important in the frail older people who are especially at risk of adverse drug reactions.3 It has been shown that implementation of a clinical pharmacist service has a positive effect on medication use and health care service utilisation among hospitalised patients.4 5 A local study in a geriatric hospital demonstrated the effectiveness of a drug rationalisation programme with involvement of a clinical pharmacist in reducing the incidence of polypharmacy and inappropriate medications.6 Interacting with the health care team on patient rounds, interviewing patients, reconciling medications, and providing patient discharge counselling and follow-up all resulted in improved outcomes.7 It is for this reason that patient safety strategies encourage the use of medication reconciliation and clinical pharmacists in health care systems to reduce adverse drug events.8 9 10

There is not much information about the effectiveness of a medical review programme among hospitalised elderly patients in Hong Kong. Two recent local reports that examined the effects of a clinical pharmacist–led medication review on hospital readmissions showed conflicting results and did not specifically address elderly patients.11 12 We therefore conducted a prospective controlled study to investigate the effectiveness of a comprehensive pharmacist intervention on medication use and hospital readmission among a group of geriatric in-patients in Hong Kong.

This prospective controlled study was conducted in the geriatric unit of a regional hospital in Hong Kong. The unit has 38 in-patient beds and admits older people aged 65 years or above who are transferred from an acute hospital after initial stabilisation of medical and/or geriatric problems. The unit admits more than 1000 patients per year and provides medical treatment, rehabilitation, and discharge planning services by a multidisciplinary team composed of a geriatrician, residents, nurses, physiotherapists, occupational therapists, and medical social workers. All patients admitted to the unit during December 2013 to September 2014 were included. Patients were excluded if they refused to participate, were terminally ill with a life expectancy of less than 3 months, or if they had already received pharmacist intervention in another hospital prior to this admission. Eligible subjects were assigned to an intervention or control group according to the admission day of the week. Those who were admitted on Monday through Thursday were assigned to the intervention group, and those admitted on Friday through Sunday to the control group. This arrangement was to ensure that pharmacist intervention could be initiated promptly within 48 hours of patient admission. Demographic data, functional status, co-morbidities, and number of drugs on admission were collected at admission.

The intervention was conducted by a pharmacist who was present in the unit from Monday to Saturday. The pharmacist provided pharmaceutical care from admission to discharge. Interventions performed by the pharmacist consisted of the following:

(1) Medication reconciliation on admission to identify unintended discrepancies between medications prescribed on admission and the usual medications prior to admission—sources to assist medication reconciliation included: electronic patient record; patient’s ward case notes; interview with patient and/or patient carer. The number and type of unidentified discrepancies were recorded.

(2) Medication review to check for medication appropriateness on admission and also at discharge—medication appropriateness was assessed by the Medication Appropriateness Index (MAI).13 There are 10 criteria to assess for appropriateness, namely indication, effectiveness, dosage, correct direction, practical direction, drug-drug interaction, drug-disease interaction, duplication, duration, and expense. For a drug item coded as ‘inappropriate’, relative weights for each criterion would apply. A sum of MAI scores could then be calculated to give a score ranging from 0 to 18. The higher the score, the more inappropriate the drug. Recommendations from the pharmacist after the reconciliation and medication review in the intervention group were then communicated to the in-charge doctor via a written note in the medical records. Recommendations were reinforced verbally if deemed appropriate by the pharmacist.

(3) Pharmacist counselling on admission and also at discharge was provided to improve patients’ drug knowledge to ensure proper use of drugs and compliance after discharge. A discharge counselling service was provided for all patients who returned home. The counselling included any changes to drug regimen; an explanation of each drug’s indication; any untoward effects that might occur and when to seek medical advice; and drug storage and administration instructions. To ensure patient understanding, written information such as patient information leaflets were given to patients and their carers to remind them of the correct drug regimen. If the patient was illiterate, a simple diagram was drawn on drug labels to demonstrate the time of day and number of tablets to be taken. If necessary, individualised pictorial schedules with drug images and administration instructions could be produced for patients and their carers. The assistance of a family member or external care services such as a community nurse was enlisted if the patient was found to have compliance issues.

The control group received routine clinical services. Records of the control group were retrospectively reviewed by the pharmacist after patient discharge to check for medication appropriateness on admission and also at discharge. The primary outcome measure was the appropriateness of prescription as measured by the MAI. Secondary outcomes included the acceptance rate by physicians, number of subjects with unintended discrepancies, patient satisfaction with the programme (for those home-living only), and unplanned hospitalisations 1 and 3 months after discharge.

A sample size of 98 patients per group was required to have 85% power to detect an effect size of 0.9 on the MAI. Our sample size was finally set at 210 patients to account for loss of participants due to dropout or death. This sample size was comparable with a study by Spinewine et al14 in which MAI was used as one of the tools to assess appropriateness of prescribing in an acute geriatric care unit and 203 patients were recruited. Our study included 212 patients and was expected to have adequate statistical power to detect differences between groups. Descriptive analyses were performed and included the number and types of unintended discrepancies, MAI score upon admission and at discharge, types of drug-related problems, number of interventions made by pharmacists, and number of recommendations accepted by doctors and implemented. Outcomes for the two groups before and after the programme were compared using the t test and Chi squared test. The Statistical Package for the Social Sciences (Windows version 17.0; SPSS Inc, Chicago [IL], United States) was used and a P value of <0.05 was regarded as statistically significant. The study was approved by the Cluster Research Ethics Committee of the Hospital Authority Hong Kong West Cluster. Written consent was obtained from the patient or their caregiver. The absence of pharmacist intervention in the control group was considered acceptable because a pharmacy service was not a part of routine care at the institution.

Figure 1 summarises the patient flow from recruitment to hospital discharge, the components of the medication review programme, and the planned outcome measures. A total of 212 patients were recruited. There were 108 subjects in the intervention group and 104 in the control group (Fig 2). There were no statistical differences in the baseline characteristics of patients (Table 1).

On admission, 51.9% (56/108) of the intervention group and 58.7% (61/104) of the control group had at least one drug classified as inappropriate (P=0.319). Overall, 1996 drug items were reviewed by a pharmacist on admission of which 1020 were from the intervention group and 976 from the control group. Among them, 9.3% and 11.1% of the drugs, respectively, were classified as inappropriate (P=0.282). In the intervention group, 93 recommendations were made by the pharmacist of which 60 (64.5%) were accepted by the physicians and implemented. The mean (± standard deviation) MAI score per patient was 2.19 ± 3.03 in the intervention group and 2.28 ± 3.09 in the control group (P=0.841). The mean MAI score per drug was 0.23 ± 0.30 in the intervention group and 0.25 ± 0.31 in the control group (P=0.628) [Table 2].

After the program and at discharge, the proportion of subjects with inappropriate medications in the intervention group was significantly lower than that in the control group (28.0% vs 56.4%; P<0.001). Among the 1999 drug items reviewed by the pharmacist on patient discharge, 3.5% (37 of 1048) of the intervention group and 9.7% (92 of 951) of the control group were classified as inappropriate (P<0.001). The intervention group also had a significantly lower MAI score per patient (0.95 ± 2.02 vs 2.02 ± 2.53; P<0.001) and MAI score per drug (0.09 ± 0.17 vs 0.24 ± 0.30; P<0.001) implying a significant reduction in medication inappropriateness after the pharmacist medication review (Table 2).

Types of inappropriateness according to the MAI in the two groups are illustrated in Figure 3. In both the intervention and control groups, the common causes were indication, effectiveness, dosage, practical direction, duration, and expense. After the programme, there was a significant reduction in the number of these drug-related problems in the intervention group.

Among the 108 subjects in the intervention group, 19.4% had at least one unintended discrepancy in medications, involving a total of 43 drug factors. The majority (90.7%) of these factors were omission of drugs, and 4.6% were due to inappropriate dosages. Of all the drug factors involved, 69.8% involved prescribed drugs from hospitals, 25.6% were from a private clinic, and 4.6% were over-the-counter drugs. Overall, 41 recommendations were made, of which 32 (78.0%) were accepted by physicians and implemented.

Contact was made with 50 of the 90 non-institutionalised subjects 1 month after discharge to assess satisfaction with the programme. Of those contacted, 98.0% were satisfied with the programme and only one (2.0%) patient expressed a neutral opinion.

There were no statistical differences in the length of hospital stay, in-patient mortality, or mortality at 1 month or 3 months after discharge. There was also no statistical difference in the number of attendances at the accident and emergency department 1 month or 3 months after discharge or in the unplanned hospital readmission rate at 3 months after discharge. The unplanned hospital readmission rate 1 month after discharge, however, was significantly lower in the intervention group than that in the control group (13.2% vs 29.1%; P=0.005) [Table 3].

To the best of our knowledge, this is the first local prospective controlled study to investigate the effectiveness of a pharmacist-led medication review programme on medication appropriateness and clinical outcomes among geriatric in-patients in Hong Kong. This study has demonstrated superior outcomes that favour a pharmacist-led intervention. There was a substantial reduction in the use of inappropriate medications and all-cause unscheduled readmissions 1 month after hospital discharge. Nonetheless, analysis of length of hospital stay, number of all-cause emergency department visits, and mortality rate favoured neither the intervention nor the usual pharmacist care.

This study showed that one in five geriatric in-patients had an unintended medication discrepancy on admission. This figure was slightly higher than that found in a group of 3317 hospitalised medical patients (13%) over 1 year in an acute hospital in Hong Kong by Kwok et al.15 Subjects in our study were all elderly patients, whereas those in Kwok et al’s study were adults of all ages. Elderly subjects tend to have polypharmacy and thus are more vulnerable to unintended medication discrepancy when they move in and out of hospital or are transferred to another health care unit for further care. Unjustifiable medication discrepancies account for more than half of the medication errors that occur during transition of care and up to one third have the potential to cause harm.16 17 This does not bode well for our elderly patients with multiple co-morbidities.

Up to 30% of the discrepancies in our study involved medications that had been prescribed by private practitioners or purchased over-the-counter. Unlike medications prescribed from the Hospital Authority, these medications might be overlooked unless the admitting doctor specifically asks for a detailed drug history from the patient. Knowing the medication history and hence resuming these medications are important if new health problems are to be prevented. Pharmacist-led medication reconciliation is therefore a critical process that can enhance patient medication safety by compiling a complete and accurate medication list for patients in hospital.

This study revealed that more than half of the subjects (55.2%) received inappropriate medications. The majority of reasons for inappropriateness related to effectiveness, dosage, practical directions, and expense as reflected by the MAI. The inappropriate dosage and the questionable effectiveness might lead to not only failed pharmacological effects, but also potentially an untoward adverse drug reaction, especially in elderly individuals with pre-existing organ dysfunction.18 When a medication is not used according to the practical directions, it may lead to patient non-compliance. Optimising outcomes while reducing costs are the keys for medication management in today’s health care environment.19 Often there are several choices of drugs available to treat a disease or health condition and some are more expensive than others. The involvement of a ward-based pharmacist to review medication can enhance the use of appropriate medications in hospitalised patients and potentially reduce medication costs.

Following the medication review (60/93) and medication reconciliation (32/41), there were 92 recommendations that were accepted by the physician. The overall acceptance rate by physicians and the implementation of pharmacist recommendations in our study was 68.7% (92/134). This figure ranged from 39% to 100% in a previous systematic review of 32 studies.4 Our study did not specifically record recommendations accepted by physicians but not implemented. Hence the acceptance rate in our study may be underestimated. Nevertheless, the clinical pharmacist is encouraged to discuss the medication-related problems in person with the physician as well as contacting the patient in order to enhance the implementation rate.4

Pharmacy departments within the public hospital system in Hong Kong have strived to implement the aforementioned patient safety strategies in different specialties. Nonetheless, this is not a standard practice simply because of insufficient pharmacist staffing resources. In some hospitals, a pharmacist service is provided in wards, but this does not apply in all cases and is not standardised. Experience of a pharmacist-led medication reconciliation service from an acute teaching hospital in Hong Kong showed promising results over a 1-year trial run with high acceptance and recognition by other health care professionals.16 It is hoped that the clinical role of clinical pharmacists in patient medication management in hospitals can be encouraged. Another local study has demonstrated a positive impact on medication safety in patients with diabetes by pharmacists’ intervention in collaboration with a multidisciplinary team.20 The feasibility of incorporating a pharmacist as part of a multidisciplinary team of health care professionals must be explored in geriatric wards in Hong Kong. With increasing life expectancy, the expanding elderly population will equate to an increase in morbidity and mortality owing to drug-related problems where the need for trained health care professionals to perform medication reviews will be in even greater demand. To enhance safe drug use with limited resources, a systematic approach must be adopted to cover all aspects that affect drug therapy.

In terms of the impact on health care services utilisation, a recent systematic review and meta-analysis of the effectiveness of a pharmacist-led medication reconciliation programme revealed a substantial reduction in the rate of all-cause readmissions (19%), all-cause emergency department visits (28%), and adverse drug event–related hospital visits (67%).21 Our study revealed a significant reduction in unplanned hospital admissions (all-cause admission) at 1 month but not at 3 months. This implication might be due to an inadequate sample size to show the difference at 3 months. Alternatively, it might also imply that pharmacist intervention needs to be continued after patient discharge in order to have a sustained effect. This is supported by a study by Schnipper et al22 in which pharmacist intervention after patient discharge was associated with a lower rate of preventable adverse drug events 30 days after hospital discharge.

During the pharmacist review, cost-effectiveness of drug use was assessed through MAI. Alternative options such as less-expensive formulations or drugs but of the same quality would be recommended to the doctor in-charge. This was a means of encouraging cost-effective use of drugs in a hospital. Furthermore, the reduction in unscheduled hospital readmissions in the intervention group implies a potential saving in hospital costs. Although a detailed analysis was not performed in this study, a rough estimation is that nearly HK$2 million may be saved annually as a result of lower drug costs and reduced hospital admissions, even after considering the cost of employing a pharmacist. The estimation was based on the following calculations. The estimated drug saving as a result of a switch to a more cost-effective alternative was HK$7500 among the 108 patients in the intervention group. This can be projected to a saving of about HK$69 500 in a unit that admits 1000 patients annually. In the current study, there were 16 fewer readmissions in the intervention group compared with the control group. Assuming a daily cost of an acute hospital bed is HK$4680 and the mean length of hospital stay is 3 days, this equates to a potential saving of about HK$2 080 000 per year in a unit that admits 1000 patients annually. If it is assumed that a pharmacist spends 30 minutes for each patient at an hourly salary of HK$433, the projected cost of an additional pharmacist to run the intervention would be HK$216 500 per year. The net annual saving of this programme to serve 1000 patients in this unit would thus still be close to HK$2 million.

This study had several limitations. First, a substantial proportion (35%) of all the admitted patients were not screened by a pharmacist on admission. This was due to a temporary pause in subject recruitment when patients were admitted on public holidays, when the pharmacist was on holiday or when she had to relieve another pharmacist in the hospital. Moreover, a substantial proportion (44%) of eligible subjects were not included owing to no consent or refusal. These factors might have resulted in selection or self-selection bias. Second, subject recruitment was not randomised, but done according to the day of admission. This might be a source of bias. Nevertheless, this would have minimal influence on the outcomes, as the baseline characteristics of the intervention and control groups were comparable. Third, the pharmacist who carried out the review and data extraction was not blinded to the study hypothesis and the group status of the subjects. This could potentially lead to information bias, although this might be partially offset by the fact that the majority of the information or data on the outcome measures were taken with reference to a well-established and validated tool. Fourth, this study was performed in a single unit, so generalisation to other settings is not possible. Fifth, MAI is an implicit tool that is subjective. A single pharmacist as the rater might limit the reliability of the assessment results. Nevertheless, the more explicit tools of STOPP/START criteria23 had also been referred to in addition to the MAI during the review process. Sixth, this study only addressed appropriateness of drug use, whereas underuse of drugs was not investigated. Finally, this study could not conclude a causal relationship between the reduction in inappropriate medications and the reduction in unscheduled hospital readmissions because there were several components in the intervention that included a medication review, medication reconciliation, and discharge counselling. It is difficult to be certain which of these components alone or in combination gave rise to the positive outcome of this study.

On the other hand, there were several strengths in this study. This was the first prospective controlled study of the effect of a pharmacist-led medication review programme on medication use and health services utilisation involving over 200 Chinese elderly patients in Hong Kong. Second, a well-validated tool was used to assess medication appropriateness. The use of the MAI tool focused on the patient and the entire medication regimen. Third, there was a comprehensive review of outcomes including quality of prescribing, health services utilisation, mortality, length of hospital stay, and patient satisfaction.

This study supported the role of a hospital-based clinical pharmacist to enhance appropriate medication use among elderly Chinese in-patients. A systematic medication review programme in a geriatric unit resulted in a reduced number of drug omissions and fewer inappropriate medications. The service provided by the clinical pharmacist and supported by geriatricians was welcomed by patients and their carers. Together with the potential to reduce hospital readmissions and their associated cost, it is hoped that an in-hospital pharmacist-led medication review programme can be recognised as one of the important strategies to enhance the safety and quality of prescription among elderly patients in hospitals. It is strongly recommended that these programmes be standardised and implemented in all medical and geriatric wards in Hong Kong. Future studies should recruit a larger sample size in a randomised controlled design in other geriatric hospital settings to reiterate our findings. Furthermore, these studies might consider including adverse drug event–related hospital visits as one of the outcome measures.

All authors have disclosed no conflicts of interest.

1. Onder G, Landi F, Cesari M, Gambassi G, Carbonin P, Bernabei R; Investigators of the GIFA Study. Inappropriate medication use among hospitalized older adults in Italy: results from the Italian Group of Pharmacoepidemiology in the Elderly. Eur J Clin Pharmacol 2003;59:157-62. CrossRef

2. Hajjar ER, Hanlon JT, Sloane RJ, et al. Unnecessary drug use in frail older people at hospital discharge. J Am Geriatr Soc 2005;53:1518-23. CrossRef

3. Routledge PA, O’Mahony MS, Woodhouse KW. Adverse drug reactions in elderly patients. Br J Clin Pharmacol 2004;57:121-6. CrossRef

4. Graabaek T, Kjeldsen LJ. Medication reviews by clinical pharmacists at hospitals lead to improved patient outcomes: a systematic review. Basic Clin Pharmacol Toxicol 2013;112:359-73. CrossRef

5. Gillespie U, Alassaad A, Henrohn D, et al. A comprehensive pharmacist intervention to reduce morbidity in patients 80 years or older: a randomized controlled trial. Arch Intern Med 2009;169:894-900. CrossRef

6. Chan FH, Pei CK, Chiu KC, Tsang EK. Strategies against polypharmacy and inappropriate medication—are they effective? Australas J Ageing 2001;20:85-9. CrossRef

7. Kaboli PJ, Hoth AB, McClimon BJ, Schnipper JL. Clinical pharmacists and inpatient medical care: a systematic review. Arch Intern Med 2006;166:955-64. CrossRef

8. The ACHS EQuIP5 (5th edition of the ACHS Evaluation and Quality Improvement Program) clinical standard 1.5, criterion 1.5.1.: Medication are managed to ensure safe and effective consumer/patient outcomes. Sydney: The Australian Council on Health Care Standards; 2010.

9. Medication reconciliation. Patient Safety Network. Agency for Healthcare Research and Quality. US Department of Health & Human Services. Available from: http://psnet.ahrq.gov/primer.aspx?primerID=1. Accessed 11 Jun 2017.

10. Gurwitz J, Monane M, Monane S, Avorn J. Polypharmacy. In: Morris JN, Lipsitz LA, Murphy K, Bellville-Taylor P, editors. Quality care in the nursing home. St. Louis, MO: Mosby-Year Book; 1997: 13-25.

11. Wong CY, Lee TO, Lam MP. Pharmacist discharge intervention programme to reduce unplanned hospital use in patients with polypharmacy. Hong Kong Pharm J 2017;24(Suppl 1):S18.

12. Chan HH. To reduce avoidable readmission of patients who are categorized as high risk by 30-day hospital readmission model at medical ward of United Christian Hospital through medication reconciliation and discharge counseling. Hong Kong Pharm J 2017;24(Suppl 1):S21.

13. Hanlon JT, Schmader KE, Samsa GP, et al. A method for assessing drug therapy appropriateness. J Clin Epidemiol 1992;45:1045-51. CrossRef

14. Spinewine A, Swine C, Dhillon S, et al. Effect of a collaborative approach on the quality of prescribing for geriatric inpatients: a randomized, controlled trial. J Am Geriatr Soc 2007;55:658-65. CrossRef

15. Kwok CC, Mak WM, Chui CM. Implementational experience of medical reconciliation in Queen Mary Hospital. Hong Kong Pharm J 2009;16:50-2.

16. Rozich JD, Howard RJ, Justeson JM, Macken PD, Lindsay ME, Resar RK. Standardization as a mechanism to improve safety in health care. Jt Comm J Qual Saf 2004;30:5-14. CrossRef

17. Cornish PL, Knowles SR, Marchesano R, et al. Unintended medication discrepancies at the time of hospital admission. Arch Intern Med 2005;165:424-9. CrossRef

18. Cameron KA. Caregivers’ guide to medications and aging. National Center on Caregiving, Family Caregiver Alliance 2004. Available from: https://www.caregiver.org/caregiver%CA%BCs-guide-medications-and-aging. Accessed 11 Jun 2017.

19. Splawski J, Minger H. Value of the pharmacist in the medication reconciliation process. P T 2016;41:176-8.

20. Chung AY, Anand S, Wong IC, et al. Improving medication safety and diabetes management in Hong Kong: a multidisciplinary approach. Hong Kong Med J 2017;23:158-67. CrossRef

21. Mekonnen AB, McLachlan AJ, Brien JA. Effectiveness of pharmacist-led medication reconciliation programmes on clinical outcomes at hospital transitions: a systematic review and meta-analysis. BMJ Open 2016;6:e010003. CrossRef

22. Schnipper JL, Kirwin JL, Cotugno MC, et al. Role of pharmacist counseling in preventing adverse drug events after hospitalization. Arch Intern Med 2006;166:565-71. CrossRef

23. Lam MP, Cheung BM. The use of STOPP/START criteria as a screening tool for assessing the appropriateness of medications in the elderly population. Expert Rev Clin Pharmacol 2012;5:187-97. CrossRef

A variety of exotic animals are kept as ‘pets’ including fish, amphibians, reptiles, and arthropods. The keeping of exotic, and sometimes venomous pets, is becoming increasingly common worldwide. Some of these exotic pets are capable of causing injury to or even life-threatening envenomation of their owners.1

Reptiles are the most popular exotic pets worldwide. It has been estimated that 1.5 to 2.0 million households in the United States (US) own one or more pet reptiles. Snakes account for approximately 11% of the imported reptiles in the US, and up to 9% of these are venomous.2 Envenomation by exotic pets, particularly snakes, is an increasing cause for concern in both the US and Europe.3 In a study of exotic snake envenomation in the US, data from the National Poison Data System database revealed 258 cases of exotic snakebites involving at least 61 unique exotic venomous species between 2005 and 2011. Among these, 40% of bites occurred in a private residence.4 Another study of bites and stings by exotic pets in Europe reported 404 cases in four poison centres in Germany and France from 1996 to 2006. Exotic snakebites from rattlesnakes, cobras, mambas, and other venomous snakes were the cause of approximately 40% of envenomations.5 Another survey conducted in the United Kingdom reviewed the data from the National Health Service Health Episode Statistics from 2004 to 2010. A total of 709 hospital admissions associated with injuries from exotic pets were reported and approximately 300 hospital admissions were related to contact with scorpions, venomous snakes, and lizards.6 Nonetheless, no such epidemiological study has been conducted in Hong Kong. According to the thematic household survey report in 2006, 286 300 households in Hong Kong kept pets at home, of which 5% were pets other than dogs, cats, turtles, tortoises, birds, hamsters, and rabbits.7 The number of imported pet reptiles into Hong Kong has increased rapidly in recent years. In 2016, the Agriculture, Fisheries and Conservation Department (AFCD) recorded that almost 1 000 000 pet reptiles were imported into Hong Kong (Table 1).

The knowledge of local emergency physicians about the management of injuries by these exotic animals is limited. Since 2005, the Hong Kong Poison Information Centre (HKPIC) has provided a 24-hour telephone consultation service (tel: 2635 1111) for health care professionals in Hong Kong, offering poison information and clinical management advice. The objectives of this study were to use HKPIC records to describe the variety of reported exotic species and the clinical features and outcomes of injuries and envenomation caused by exotic pets.

This was a case series based on the database of the HKPIC. It included cases encountered by clinical frontline staff and surveillance data from routine reporting of poisoning cases by all accident and emergency departments (AEDs) under the Hospital Authority (HA). Cases of injuries and envenomation by exotic pets recorded by the HKPIC from 1 July 2008 to 31 March 2017 were retrospectively retrieved. Demographic data of the patients—including the involved species, clinical presentations, and outcomes—were reviewed from the patient electronic Health Record. This study was done in accordance with the principles outlined in the Declaration of Helsinki.

The severity of injuries and the effects of envenomation were defined as major (life-threatening or resulting in significant residual disability or disfigurement), moderate (pronounced, prolonged, or systemic signs and symptoms), or mild (minimal and rapidly resolving signs and symptoms).

During the study period, 15 cases of injuries and envenomation by exotic pets were reported to the HKPIC. Among the 15 patients, nine consulted the HKPIC for management advice and one was managed by the toxicology team of the AED. Local zoologists were consulted in five cases for species identification and opinion about the venomous nature of the species. All bites and stings were unintentional and occurred in a private household. The mean age of the exotic pet owners was 28.2 (range, 14-59) years and the majority (73%) were male. There were six cases of snakebite, four cases of fish sting, two cases of scorpion sting, two cases of lizard bite, and one case of turtle bite. The severity of injury and envenomation effect are summarised in Table 2.

All major effects occurred in patients with snakebite. A 16-year-old boy was bitten by a short-tailed mamushi (Gloydius blomhoffii brevicaudus; Fig 1a) on his left middle finger. The short-tailed mamushi is not native to Hong Kong but was being kept as a pet. The patient had a history of snakebite by a bamboo snake (Trimeresurus albolabris) that required antivenom treatment, sustained while attempting to catch the snake in the suburbs. Following the bite by the short-tailed mamushi, the patient developed severe local envenomation over his left hand and required admission to the intensive care unit for close observation of the rapidly progressing local envenomation. No systemic envenomation was observed. A local zoologist was consulted for snake identification. A total of three vials of antivenom for Agkistrodon halys were administered as treatment but ischaemia due to compartment syndrome developed in the left hand. Debridement and fasciotomy were eventually performed. The patient had a residual flexion contraction deformity of his left middle finger 2 months later. He recovered with full movement of the left middle finger 6 months after the injury. Another boy, aged 15 years, was bitten by a bamboo snake on his left thumb. The snake had been caught by the patient in the suburbs and kept as a pet. He developed severe local envenomation and was given three vials of antivenom for Agkistrodon halys and three vials of antivenom for green pit viper. The patient developed tenosynovitis of his left thumb and required emergency surgery for debridement. Another four patients were bitten by ‘nonvenomous’ snakes including a rainbow boa (Epicrates cenchria; Fig 1b), corn snake (Pantherophis guttatus), and eastern hognose snake (Heterodon platirhinos; Fig 1c). All snakes were kept as pets. An 18-year-old girl was accidentally bitten by a rainbow boa on her left hand but had no signs of local or systemic envenomation after the injury. Another patient developed a wound infection after being bitten by a corn snake 2 weeks previously (Fig 1d). She recovered after a course of antibiotic therapy. Two young men were bitten by hognose snakes. One developed local envenomation with progressive swelling over the injured hand (Fig 1e). The local envenomation resolved with conservative management. The HKPIC was consulted in all cases, of which three required consultation with a zoologist.

Injuries from reptiles other than snakes were also recorded. There were two cases of lizard bite. In one case, a 22-year-old man was bitten by a common iguana (Iguana iguana) on his left wrist. In the other case, a 41-year-old man presented to the AED approximately 2 hours after being bitten on his right hand by a Gila monster (Fig 2a). He developed intense pain and local swelling over the site of injury. The pain lasted for about 12 hours and then gradually improved. His haemodynamic state remained stable and no airway oedema or neurological symptoms were observed during his stay in the emergency medicine ward. He was eventually discharged. A young woman attended the AED because of a turtle bite over her left face with consequent minor physical injury.

Stings by aquarium fish were the second most common injuries by exotic pets. Four cases were recorded, including one sting by a blue tang fish and three by freshwater stingrays (Fig 2b). All patients developed severe pain over the site of injury that responded to immersion in hot water. One of the patients with a freshwater stingray sting developed a wound infection that required emergency surgery for wound exploration and irrigation.

Two male patients were stung by their pet scorpions: a thick-tailed scorpion (Parabuthus transvaalicus; Fig 2c) and a cave-claw scorpion (Pandinus cavimanus; Fig 2d). No systemic envenomation was observed. The patient with the cave-claw scorpion sting developed a local wound infection (Fig 2e) that recovered after a course of antibiotics.

The characteristics and management of the 15 cases are summarised in Table 3.

Injuries by a variety of exotic pets were encountered in this study. More than half of the injuries (9/15) were inflicted by reptiles. Reptiles are becoming increasingly popular to keep as pets in Hong Kong. According to the records of the AFCD over the past 5 years, the top 10 most common reptile species imported to Hong Kong are the European pond turtle (Emys orbicularis), razor-backed musk turtle (Sternotherus carinatus), common snapping turtle (Chelydra serpentina), red-bellied cooter (Pseudemys nelsoni), yellow-spotted Amazon River turtle (Podocnemis unifilis), Hermann’s tortoise (Testudo hermanni), African spurred tortoise (Geochelone sulcata), leopard tortoise (Stigmochelys pardalis), common iguana (Iguana iguana), and ball python (Python regius). Commonly imported pet snakes include the ball python (Python regius), king snake (Lampropeltis getula), corn snake (Pantherophis guttatus), rat snake (Elaphe obsoleta), milk snake (Lampropeltis triangulum), and western hognose snake (Heterodon nasicus). With the exception of the hognose snake, which is a mildly venomous species, they are all nonvenomous. Nonetheless, a much wider variety of species, including venomous reptiles, may be sold on the black market. Bites may occur during the care and handling of these exotic animals.3 Envenomation by exotic venomous species is an uncommon but often serious medical emergency.

The keeping of venomous snakes is common in the US.4 Amateur collectors are at risk of bites and envenomation and fatalities have been reported.8 Although envenomation from exotic snakes is rarely encountered in Hong Kong, it poses a great challenge to emergency physicians owing to their lack of experience and limited supplies of antivenom, as illustrated by our case of bite by a short-tailed mamushi. Currently, the HA stocks principally snake antivenom for local venomous species (Table 4). Bites by nonvenomous pet snakes may also result in local envenomation and complications; for instance, although the hognose snake is known as a nonvenomous species, one patient developed local envenomation after being bitten. Another patient developed a wound infection after being bitten by a corn snake.

As well as snakes, lizards are popular as pets. Bites by large species such as the common green iguana (Iguana iguana) can result in serious injury.9 Envenomation from lizard bites is rare in Hong Kong. Two lizards are well known to be venomous: the Gila monster (Heloderma suspectum)10 and the Mexican beaded lizard (Heloderma horridum).11 12 Both have venom-secreting glands and bites. The Gila monster is native to the southwestern US extending into Mexico, whereas the beaded lizard is native only to Mexico. The Gila monster is listed in the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) as a protected species.13 Captive-bred Gila monsters are traded in international pet markets. Venom of the Gila monster consists of a variety of proteins including gilatoxin, a kallikrein-like protease that can hydrolyse kininogen and produce bradykinin.11 14 The common envenomation effects are intense pain at the injured site, oedema, paraesthesia, weakness, dizziness, and nausea. Hypotension occurs in severe envenomation.15 The intense pain, oedema, and hypotension are likely due to the bradykinin-mediated effects. Airway oedema has been reported regardless of the site of bite and may occur up to 12 hours after the bite.14 Nevertheless, severe envenomation from the Gila monster occurs in only a minority of patients. In a retrospective study of all cases of Gila monster bite reported to the two Arizona poison control centres from 2000 to 2011, 105 cases of human exposure to Gila monsters were recorded and 70 cases were referred to health care facilities for medical treatment. Eleven cases required admission to hospital and five required care in an intensive care unit. Six patients developed airway oedema and three required emergent airway management including one cricothyrotomy.14 Treatment of Gila monster bites is mainly supportive. Intravenous crystalloid infusion and vasopressors may be required for treatment of hypotension in severe envenomation. Radiographic assessment is needed to look for retained teeth and subcutaneous air due to the chewing-like action during the bites.16 No antivenom to Gila monster is commercially available.17 Observation for at least 12 hours after the bite for delayed-onset airway oedema is recommended.14

Among all the reptiles, tortoises and turtles are the most popular in pet markets. All species of tortoises and turtles are nonvenomous although some, such as the alligator snapping turtle (Macrochelys temminckii) and the common snapping turtle (Chelydra serpentina), are aggressive and can grow to a very large size. Bites by these large species can result in severe limb injuries.18

Stings by aquarium fish contributed to the second largest group of injuries in our case series. The most commonly encountered aquarium fish was freshwater stingray. Freshwater stingrays (Potamotrygon species) are native to South America. They are regarded as dangerous by the native people of the Amazon and frequent sting during fishing season.19 Freshwater stingrays are not aggressive by nature; stings frequently occur when people step on them or handle them improperly. Different species of freshwater stingrays have different colour patterns on their body. They are popular aquarium fish as they are easy to keep although stings may result in severe envenomation.20 The most common feature of envenomation from freshwater stingrays is intense local pain. Systemic manifestations are rare. Skin necrosis is frequently observed in victims wounded by large freshwater stingrays in the wild.21 In addition, skin necrosis is more commonly observed in victims injured by freshwater stingrays than marine stingrays. A study of tissue extracts from the stingers of freshwater and marine stingrays showed that both tissue extracts had gelatinolytic, caseinolytic, and fibrinogenolytic activity but hyaluronidase activity was detected only in the extracts from freshwater stingrays.22 In our case series, no patient injured by a freshwater stingray developed skin necrosis. The risk of developing skin necrosis is likely related to the venom load. Larger stingrays possess a much larger venom load in their stingers. Small freshwater stingrays are commonly kept in an aquarium and skin necrosis as a result of their sting is uncommon. Hot water immersion is effective in controlling acute pain but does not prevent skin necrosis.21 Wounds caused by freshwater stingray stings such as the Aeromonas species can be complicated by severe secondary infection with virulent bacteria.23 Prophylactic antibiotic is often required.

Apart from freshwater stingrays, the stinging catfish (Heteropneustes fossilis) is another commonly reported freshwater aquarium fish that can cause injuries and envenomation. It possesses venom in the sting that is located in front of the soft-rayed portion of the pectoral and dorsal fins. Apart from intense local pain, systemic envenomation including weakness and hypotension can result from a sting.24 25 There was no case reported to the HKPIC of injury by this venomous catfish during the study period. Coral reef fish are also popular pets in Hong Kong. Some coral reef fish, such as the lionfish (Pterois volitans), are venomous.26 Nonetheless, injuries by aquarium coral reef fish were rarely encountered in the AED of Hong Kong.

Exotic pet owners also enjoy keeping arthropods such as scorpions and spiders. There are approximately 2000 species of scorpion in the world but only a few (30 to 40) are highly venomous and able to cause severe envenomation in humans.27 Scorpion envenomation is reported throughout the world, mainly in subtropical and tropical regions.28 The majority of scorpion stings cause mild or no envenomation. Species that cause serious medical problems mainly belong to the Buthidae family. The genera of the Buthidae family include Centruroides, Tityus, Leiurus, Androctonus, Buthus and Parabuthus.29 Scorpions have a special venom apparatus, the telson, that produces venom. Scorpion venom comprises numerous toxins including several neurotoxins. Unlike snake venom, scorpion venom generally lacks enzyme activity. The main molecular targets of scorpion neurotoxins are the voltage-gated sodium channels and the voltage-gated potassium channels. Scorpion α-toxin, one of the most medically important neurotoxins in the scorpion venom, acts on the voltage-gated sodium channels. Once the toxin binds to voltage-gated sodium channels, it inhibits inactivation of the channel with consequent prolonged depolarisation and, hence, neuronal excitation. The autonomic centres, both sympathetic and parasympathetic, are stimulated. In most situations of scorpion envenomation, the sympathetic nerves are predominantly affected. Scorpion envenomation is characterised by relatively similar neurotoxic excitation syndromes, irrespective of the species. Parasympathetic effects tend to occur early and then sympathetic effects persist due to the release of catecholamines that are responsible for the severe envenomation. Parasympathetic (cholinergic) effects include hypersalivation, diaphoresis, lacrimation, miosis, diarrhoea, vomiting, bradycardia, hypotension, increased respiratory secretion, and priapism. Sympathetic (adrenergic) effects are manifested as tachycardia, hypertension, mydriasis, hyperthermia, hyperglycaemia, and agitation. Fatal effects of scorpion envenomation are largely due to cardiovascular effects. Various cardiac conduction abnormalities have been reported in patients with scorpion envenomation as well as catecholamine-induced cardiomyopathy, pulmonary oedema, and cardiogenic shock. Other manifestations of systemic envenomation include vomiting, abdominal pain, abnormal oculomotor movements, muscle fasciculation, and spasms of the face and limbs.29 Pancreatitis is also a well-reported complication of envenomation by certain species, such as Leiurus quinquestriatus.30 Nonetheless, severe local envenomation is generally uncommon. Differences in the clinical manifestations of systemic envenomation exist in some species. Delayed localised necrosis has been reported in patients stung by an Iranian scorpion (Hemiscorpius lepturus).31 Patients with envenomation from the thick-tailed scorpion (Parabuthus transvaalicus) in Zimbabwe have been reported to develop predominant symptoms from parasympathetic nerve system stimulation, including profuse sialorrhoea, sweating, and urinary retention, in the absence of sympathetic stimulation.32

Scorpion stings and envenomation are uncommon in Hong Kong. Most of the locally reported cases of scorpion sting occurred while patients were handling langsat, a type of tropical fruit from South-East Asia. The Chinese stropped bark scorpion (Lychas mucronatus) hides in the fruit and is subsequently imported into Hong Kong.33 Scorpions are also sold as fish food in aquarium shops in Hong Kong. People use scorpions to feed arowana, which are popular aquarium fish. Importation of endangered scorpion species (CITES-listed species) for commercial purposes is regulated by the Protection of Endangered Species of Animals and Plants Ordinance Cap. 586 in Hong Kong.34 According to the data from the AFCD for importation of CITES-listed scorpions, more than 1000 heads of emperor scorpion (Pandinus imperator) have been imported as pets to Hong Kong each year for the last 4 years. The emperor scorpion is a nonvenomous species and is native to the rainforests and savannas of West Africa. Most scorpions in the pet trade, such as the forest scorpion (Heterometrus species), have no potential for dangerous envenomation. Nonetheless venomous species may also be kept by hobbyists and severe envenomation may occur after stings.

Management of scorpion stings includes local wound care and supportive care for systemic envenomation. Expert opinion should be sought from a zoologist for species identification and to determine the venomous nature of the species. Patients with severe systemic envenomation may require antivenom therapy. Specific antivenom (Scorpifav; Sanofi Pasteur, France) for Androctonus australis, Buthus occitanus, and Leiurus quinquestriatus is currently available in the HKPIC.

Spiders, such as tarantulas, are popular exotic pets and are common in the pet trade in Hong Kong. Nonetheless, inexperienced owners may be unaware of the potential risk of ocular injury from the barbed urticating hairs on the abdomen of the tarantulas. Eye injuries occur when the barbed hairs come into contact with the eyes, either directly from the tarantula’s ejection or when the owners rub their eyes after handling the spider.35 Embedment of the hairs in the cornea can result in severe complications, including ophthalmia nodosa, iritis, and even permanent visual impairment.36 37

The diversity of pets is changing and keeping exotic animals is increasingly popular. Injuries from these exotic pets are expected to increase and envenomation may result from stings or bites from some species. In our case series, reptiles, scorpions, and fish were responsible for human injuries, and all cases of major envenomation were inflicted by snakes. Emergency physicians need to be aware of the appropriate management of injuries and envenomation by these exotic animals. The HKPIC plays an important role in the provision of expert advice about management of these special toxicological cases.

The authors would like to thank AFCD for providing the data of imported reptiles, scorpions, and spiders.

All authors have disclosed no conflicts of interest.

1. Bey TA, Boyer LV, Walter FG, McNally J, Desai H. Exotic snakebite: envenomation by an African puff adder (Bitis arietans). J Emerg Med 1997;15:827-31. Crossref

2. McNally J, Boesen K, Boyer L. Toxicologic information resources for reptile envenomations. Vet Clin North Am Exot Anim Pract 2008;11:389-401. Crossref

3. de Haro L, Pommier P. Envenomation: a real risk of keeping exotic house pets. Vet Hum Toxicol 2003;45:214-6.

4. Warrick BJ, Boyer LV, Seifert SA. Non-native (exotic) snake envenomations in the U.S., 2005-2011. Toxins (Basel) 2014;6:2899-911. Crossref

5. Schaper A, Desel H, Ebbecke M, et al. Bites and stings by exotic pets in Europe: an 11 year analysis of 404 cases from Northeastern Germany and Southeastern France. Clin Toxicol (Phila) 2009;47:39-43. Crossref

6. Warwick C, Steedman C. Injuries, envenomations and stings from exotic pets. J R Soc Med 2012;105:296-9. Crossref

7. Thematic household survey report 2006. Census and Statistics Department, HKSAR Government. Available from: http://www.statistics.gov.hk/pub/ B11302262006XXXXB0100.pdf. Accessed 1 Jun 2017.

8. Marsh N, DeRoos F, Touger M. Gaboon viper (Bitis gabonica) envenomation resulting from captive specimens—a review of five cases. Clin Toxicol (Phila) 2007;45:60-4. Crossref

9. Heffelfinger RN, Loftus P, Cabrera C, Pribitkin EA. Lizard bites of the head and neck. J Emerg Med 2012;43:627-9. Crossref

10. Hooker KR, Caravati EM, Hartsell SC. Gila monster envenomation. Ann Emerg Med 1994;24:731-5. Crossref

11. Cantrell FL. Envenomation by the Mexican beaded lizard: a case report. J Toxicol Clin Toxicol 2003;41:241-4.Crossref

12. Ariano-Sánchez D. Envenomation by a wild Guatemalan beaded lizard Heloderma horridum charlesbogerti. Clin Toxicol (Phila) 2008;46:897-9. Crossref

13. Wijnstekers W. The Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES)—35 years of global efforts to ensure that international trade in wild animals and plants is legal and sustainable. Forensic Sci Rev 2011;23:1-8.

14. French R, Brooks D, Ruha AM, et al. Gila monster (Heloderma suspectum) envenomation: descriptive analysis of calls to United States Poison Centers with focus on Arizona cases. Clin Toxicol (Phila) 2015;53:60-70. Crossref

15. Piacentine J, Curry SC, Ryan PJ. Life-threatening anaphylaxis following gila monster bite. Ann Emerg Med 1986;15:959-61. Crossref

16. French RN, Ash J, Brooks DE. Gila monster bite. Clin Toxicol (Phila) 2012;50:151-2. Crossref

17. Miller MF. Gila monster envenomation. Ann Emerg Med 1995;25:720. Crossref

18. Johnson RD, Nielsen CL. Traumatic amputation of finger from an alligator snapping turtle bite. Wilderness Environ Med 2016;27:277-81. Crossref

19. Junior VH, Cardoso JL, Neto DG. Injuries by marine and freshwater stingrays: history, clinical aspects of the envenomations and current status of a neglected problem in Brazil. J Venom Anim Toxins Incl Trop Dis 2013;19:16. Crossref

20. Brisset IB, Schaper A, Pommier P, de Haro L. Envenomation by Amazonian freshwater stingray Potamotrygon motoro: 2 cases reported in Europe. Toxicon 2006;47:32-4. Crossref

21. Haddad V Jr, Neto DG, de Paula Neto JB, de Luna Marques FP, Barbaro KC. Freshwater stingrays: study of epidemiologic, clinic and therapeutic aspects based on 84 envenomings in humans and some enzymatic activities of the venom. Toxicon 2004;43:287-94. Crossref

22. Barbaro KC, Lira MS, Malta MB, et al. Comparative study on extracts from the tissue covering the stingers of freshwater (Potamotrygon falkneri) and marine (Dasyatis guttata) stingrays. Toxicon 2007;50:676-87. Crossref

23. Polack FP, Coluccio M, Ruttimann R, Gaivironsky RA, Polack NR. Infected stingray injury. Pediatr Infect Dis J 1998;17:349,360.

24. Dorooshi G. Catfish stings: a report of two cases. J Res Med Sci 2012;17:578-81.

25. Satora L, Kuciel M, Gawlikowski T. Catfish stings and the venom apparatus of the African catfish Clarias gariepinus (Burchell, 1822), and stinging catfish Heteropneustes fossilis (Bloch, 1794). Ann Agric Environ Med 2008;15:163-6.

26. Chan HY, Chan YC, Tse ML, Lau FL. Venomous fish sting cases reported to Hong Kong Poison Information Centre: a three-year retrospective study on epidemiology and management. Hong Kong J Emerg Med 2010;17:40-4. Crossref

27. Santos MS, Silva CG, Neto BS, et al. Clinical and epidemiological aspects of scorpionism in the world: a systematic review. Wilderness Environ Med 2016;27:504-18. Crossref

28. Chippaux JP, Goyffon M. Epidemiology of scorpionism: a global appraisal. Acta Trop 2008;107:71-9. Crossref

29. Kounis NG, Soufras GD. Scorpion envenomation. N Engl J Med 2014;371:1558-9.

30. Sofer S, Shalev H, Weizman Z, Shahak E, Gueron M. Acute pancreatitis in children following envenomation by the yellow scorpion Leiurus quinquestriatus. Toxicon 1991;29:125-8.Crossref

31. Pipelzadeh MH, Jalali A, Taraz M, Pourabbas R, Zaremirakabadi A. An epidemiological and a clinical study on scorpionism by the Iranian scorpion Hemiscorpius lepturus. Toxicon 2007;50:984-92.Crossref

32. Bergman NJ. Scorpion sting in Zimbabwe. S Afr Med J 1997;87:163-7.

33. Centre for Food Safety, Hong Kong SAR Government. Scorpion stings and langsat. Food Safety Focus (16th Issue, Nov 2007). Available from: http://www.cfs.gov.hk/english/ multimedia/multimedia_pub/multimedia_pub_fsf_16_04. html. Accessed 1 Jun 2017.

34. Agriculture, Fisheries and Conservation Department, Hong Kong SAR Government. Protection of Endangered Species of Animals and Plants Ordinance. Available from: http://www.afcd.gov.hk/english/conservation/con_end/ con_end_reg/con_end_reg_ord/con_end_reg_ord.html. Accessed 1 Jun 2017.

35. Waggoner TL, Nishimoto JH, Eng J. Eye injury from tarantula. J Am Optom Assoc 1997;68:188-90.

36. Belyea DA, Tuman DC, Ward TP, Babonis TR. The red eye revisited: ophthalmia nodosa due to tarantula hairs. South Med J 1998;91:565-7. Crossref

37. Blaikie AJ, Ellis J, Sanders R, MacEwen CJ. Eye disease associated with handling pet tarantulas: three case reports. BMJ 1997;314:1524-5. Crossref

With the ageing population in Hong Kong, the number of elderly people aged 65 years or above is projected to rise most rapidly in the next 20 years, with a projected increase from 15% in 2014 to 30% in 2034.1 With this surge in the elderly population, and as one of the most common injuries in the elderly, hip fracture is also projected to double its numbers in 20 years.2 This places a huge financial burden on health care resources. The sum of HK$310 million allocated to elderly patients with hip fracture in 2011 will rise in the next few years.2

Early surgical repair is a key element both for pain management and restoration of bone integrity after hip fracture.3 4 5 Systematic reviews show that surgery beyond 48 hours significantly increases 30-day and 1-year mortality and complication rates.6 7 8 9 Early surgical stabilisation and mobilisation has become the standard of care. As a result, and due to congested operating theatre schedules, non–life-threatening orthopaedic surgery may be performed at night. However, there is a perception that out-of-hours surgery may result in poorer outcomes due to insufficient technical support and surgeon fatigue or inexperience.

Studies that investigated the effect of out-of-hours surgery in different specialties have shown increased morbidity and mortality risk.10 11 12 Scant literature on the effect of time of the day of operation on hip surgery outcome shows controversial results. A German study in 200313 and a study by Chacko et al14 in 2011 showed no significant differences in mortality or complication rate 6 months after surgery when it was performed at night. Other studies, however, have shown that night-time surgeries for hip fracture may be associated with increased operating time and surgical complication rate.15 16

Owing to the controversial outcomes of these limited studies, this retrospective study aimed to evaluate the surgical outcome of elderly patients with hip fracture who underwent surgery in Hong Kong during the day or out-of-hours. It was hypothesised that surgical outcomes of out-of-hours surgery would not differ significantly to those of daytime surgery. It was hoped that findings of this study would help surgeons in making a decision about whether to operate out-of-hours or to delay surgery until the following day.

The Hospital Authority (HA) in Hong Kong manages all public hospitals serving more than 90% of the population. The Clinical Data Analysis and Reporting System (CDARS) includes in-patient data from all hospitals and forms a huge database. The Clinical Management System (CMS) is another computerised system that records all aspects of clinical management in the HA.

Using these two systems, a retrospective case series study was conducted to review individual records of patients in the Prince of Wales Hospital (PWH) in Hong Kong. This study was approved by the New Territories East Cluster Ethics Committee (reference number: 2015.665). Preliminary screening was performed using CDARS. All patients discharged in 2014 with a diagnosis of hip fracture (ICD-9 code: 820.00-820.03, 820.09, 820.20-820.23 820.8, 821.00 and 905.3) and who underwent surgical intervention (ICD-9 code: 79.15(0)-79.15(5), 79.15(7)-79.15(10)) were selected from CDARS. Records were also reviewed through the CMS for verification. Patients aged 65 years or older with an isolated hip fracture who underwent surgical intervention were included in the study. Those with high-energy trauma, periprosthetic fracture, bilateral hip fracture, or multiple lower limb fractures were excluded as well as those with a fracture as a result of primary or metastatic bone tumours.

Records of patients who fulfilled the criteria were divided into two groups based on the time of surgical incision. The daytime group included those with an operation between 08:00 and 16:59 (group 1). The out-of-hours group comprised patients of whom the procedure was commenced between 17:00 and 07:59. This group was further split into those having surgery before (group 2) or after midnight (group 3) to enable more detailed analysis.

Operation procedure was defined as either fixation or arthroplasty. Preoperative surgical risk was estimated by the American Society of Anesthesiologists (ASA) classification. Surgeon’s qualification was defined according to the list of specialist registration in Orthopaedics and Traumatology in the Medical Council of Hong Kong. Surgeons who qualified as a specialist in or before 2014 were considered a specialist in this study. Surgery performed by a non-specialist but in the presence of a specialist was classified as ‘non-specialist with supervision’.

Outcome measures were 30-day mortality and complications during hospital stay; 30-day mortality was chosen because a shorter period could include deaths directly related to the hip surgery. Surgical outcome was defined as complications related to surgical procedures only. General complications such as cardiovascular, respiratory, or cognitive complications were excluded.

Records were divided into groups based on the time of incision. The daytime group included patients operated on between 08:00 and 16:59 (group 1). The remaining patients were assigned to the out-of-hours group. More detailed comparison was performed with the out-of-hours group further split into those having surgery before (group 2: 17:00 to 23:59) or after midnight (group 3: 00:00 to 07:59).

For group comparisons, continuous variables were presented as means and standard deviations. Comparison between groups was performed by one-way analysis of variance with post-hoc Bonferroni test. Categorical data such as demographic data as well as mortality and complication rates were expressed as proportion and were compared by Pearson’s Chi squared test. Statistical analysis was performed using the SPSS (Windows version 20.0; IBM Corp, Armonk [NY], United States). The level of significance was set at P<0.05.

Using International Classification of Disease, 9th revision and identified from CDARS, there were 379 hip fracture patients operated on and discharged from PWH in 2014. Review of the related medical records in CMS led to elimination of 12 patients according to the inclusion and exclusion criteria. Of the remaining 367 patients, 242 patients were operated on between 08:00 and 16:59 (daytime group; group 1), and 125 patients were operated on during out-of-hours after 16:59 and before 08:00. Among these 125 patients, 104 were operated on before midnight (group 2: 17:00 to 23:59), and 21 were operated on after midnight (group 3: 00:00 to 07:59). Patient selection and grouping are shown in the Figure.

Demographic equivalency was assessed by comparing the daytime and out-of-hours group and revealed no difference in terms of age, sex, or type of fracture. Detailed comparison was performed with the out-of-hours group further divided into before and after midnight as shown in the Table. There remained no differences in terms of age, sex, or fracture type among the groups. The mean age of the three groups ranged from 83.2 to 84.3 years and there were more females than males in all groups, more intertrochanteric fractures in group 1 and group 2, and more femoral neck fractures in group 3.

Intra-operative variables were compared between the daytime and out-of-hours groups and revealed no significant differences in ASA class, type of surgery performed, or surgeon’s qualification. Again, a more complete comparison was made with the three groups.

The ASA class was comparable among the groups, with almost two thirds of the patients categorised as ASA class 3. Fixation was more common in all the groups but the number of fixation and arthroplasty cases was not statistically significant. There was no difference in surgeon’s qualification among the groups, with most surgeries (>95%) performed in the presence of a specialist. Chi squared test revealed that significantly fewer surgeons were involved in the out-of-hours group, especially after midnight (P=0.02).

Regarding surgical outcome, the 30-day mortality rate and postoperative complication rate during hospital stay were obtained. There were eight deaths among 367 patients, accounting for 2.2% of the study population. The cause of death included chest infection and cardiac arrest. The mortality rates were 2.1% and 2.4% in the daytime and out-of-hours groups, respectively (P=0.84).

Surgical outcome was defined as complications related to surgical procedure only. The overall complication rate was 24.3% in the study population with a similar rate between daytime and out-of-hours groups. Comparable results were obtained when the out-of-hours group was further divided into two subgroups (P=0.53). A total of 89 patients among all groups had postoperative complications. Fall in haemoglobin level in 89 patients required blood transfusion in 96.7% of cases. Wound infection or implant infection occurred in only four patients. Because all patients with implant infection had revision surgery, rate of revision surgery was the same as implant infection. No patient had fixation failure, prosthetic dislocation, or peri-prosthetic fracture.

Comparison of surgical time revealed no significant difference in surgical outcome, or in surgeon’s qualification (P=0.21). For type of surgery performed, the fixation group showed a significantly higher surgical complication rate than the arthroplasty group (P=0.03), although mortality rate was similar.

Bone density insufficiency is the leading cause of major musculoskeletal trauma following a fall in the aged population.17 In 2000, the number of hip fractures worldwide was about 1.6 million. By 2050, the projected number will reach 4.5 million, and more than 50% of osteoporotic hip fractures will occur in Asia.18

Encouragement of early surgery after hip fracture will result in unavoidable out-of-hours surgery because of busy daytime operating room schedules. Safety of surgery performed outside routine daytime working hours, however, has long been a controversial issue. Surgery performed after-hours may be under less-ideal conditions with consequent poorer outcomes. This study was designed to assess if surgical outcomes for out-ofhours surgery significantly differ to those of daytime surgery.

In this study, patients were grouped according to the time of surgical incision. The normal shift in the operating theatre is 08:00 to 17:00. Surgeries performed after 17:00 and before 08:00 were considered out-of-hours. The time period correlates with the typical working hours and allows analysis based on a surgeon’s routine practice. Demographic characteristics were comparable among the groups.

Mortality and complication rates were comparable between the daytime and night-time groups. Even after midnight, when a surgeon is thought to be most affected by fatigue, there was no significant increase in complication rate or mortality. This was supported by a study in 2013 that showed no significant difference in postoperative complication rate or mortality rate after reviewing 220 dynamic hip screw surgeries in terms of their operating time.19 It concluded that out-of-hours surgery offers the benefit of early fixation and mobilisation, and hence may shorten the length of stay and reduce cost of treatment.19 Chacko et al14 also reported similar findings in 171 hip fracture patients with surgical intervention where mortality rate within 1 month and complication rate were comparable between the daytime and night-time groups. Switzer et al20 studied the relationship between surgical time of day and outcome after hip fracture fixation. They identified more than 1400 hip fracture patients with surgical intervention. Time of surgery was treated as a continuous variable and showed no association with complication rate at any time period. The authors concluded that there was no difference in 30-day mortality or complications based on the time of surgery and suggested that early operation after normal operating room hours was safe and reasonable.20

In addition, complex cases are generally scheduled for surgery during the daytime when more support can be obtained when needed. This may help explain the similar surgical outcomes among the groups. The comparable results for daytime and out-of-hours surgery shown in this study are supported by the literature suggesting that out-of-hours surgery is safe.

The overall 30-day mortality rate was 2.2% in this study, lower than the 3.5% to 10% reported in the UK,17 as well as the 4.96% in a 1997 local study.21 The lower mortality rate in this study may be attributed to advancements in surgical technique and design of prostheses. The introduction of an ortho-geriatrician in managing hip fracture patients has also been proven to decrease mortality and complication rates.22

Postoperative complications included chest infection and acute coronary syndrome. The effect of surgeon aspects on outcomes, however, was the main factor under investigation in this study. Thus, surgical outcome was defined as complications related to surgical procedure only. General complications were excluded. For surgical outcome, fall in haemoglobin level with the need for blood transfusion, wound infection, and implant infection were analysed.

The overall surgical complication rate was 24.2% in this study compared with previous reports of 5% to 32% in hip fracture fixation.15 19 20 23 24 Nonetheless, different analyses and definitions of complication rate were used in these studies. Some studies defined complications as medical complications or unplanned return to the operating room,15 24 whereas others reported only wound infection, urinary tract infection, and deep vein thrombosis.19 Thus direct comparison with these studies was not possible. Further comparison of blood transfusion rate with previous studies was performed, as it represented the most common complication. The blood transfusion rate was 23.4% in this study, similar to the results in previous studies where transfusion after hip fixation ranged from 19% to 69%.25 This may be due to incomplete reporting in the CMS as blood transfusion was not always noted in the discharge summary. Despite the difficulties in direct comparison of the complication rate with previous study, we suggest that the rate in this study was reasonable.

Significantly fewer surgeons were involved in out-of-hours surgery. This may be because training of junior staff commonly occurs during the daytime. Although fewer surgeons were involved in out-of-hours surgeries, this may be compensated by the experience of the surgeon since a larger proportion of out-of-hours surgeries was performed by a specialist. Nonetheless, the difference was not significant.

Furthermore, the qualification of the surgeon had no association with surgical outcomes in this study. This may be because cases were screened prior to allocation. Difficult and more complex cases would likely be operated on by a more experienced surgeon. Holt et al26 showed comparable results in their study of the Scottish Hip Fracture Audit Database published in 2008. They studied more than 18 000 patients and concluded that grade of surgeon did not significantly affect surgical outcome.26

This is the first local study based in a major hospital in Hong Kong to analyse the effect of operating time on surgical outcome. The out-of-hours group was split into before and after midnight so as to focus on surgeon fatigue. Analysis of surgeon expertise revealed that surgical outcome was not compromised by surgeon’s qualification.

There are several limitations in this study. First, this was a retrospective study with no functional outcomes. Information on complications was retrieved from the CMS only which might not have recorded all complications. A fracture registry or prospective study with more representative complications including prosthetic dislocation, peri-prosthetic fracture, implant loosening, fixation failure, malreduction, malfixation, and implant malposition is suggested in future. Data collection was performed by the authors who were not blinded so this might have introduced bias. Blood transfusion, the most common complication reported, was believed to be related to the operative procedure. Fall in haemoglobin level due to other causes, however, could not be excluded simply from details in the CMS. Second, the overall population size and the relatively smaller number of cases in the after-midnight group might not have the statistical power to show any difference. Further study with a larger sample size is suggested. Finally, several potential confounders were not investigated, for example, fractures were not classified according to stability and time to surgery. These factors may be associated with poorer outcome.

This study demonstrates similar outcomes of elderly patients with hip fracture in terms of mortality and postoperative complications for daytime and out-of-hours surgery. Qualification and number of surgeons involved were not associated with outcome. To facilitate better outcome with early operation, out-of-hours surgery remains a safe option and the only means to overcome limited resources.

All authors have disclosed no conflicts of interest.

1. Hong Kong population projections 2015-2064. Census and Statistics Department, the Government of the Hong Kong Administrative Region. Available from: http://www.statistics.gov.hk/pub/B1120015062015XXXXB0100.pdf. Accessed Dec 2015.

2. Ngai WK. Fragility Fracture Registry in Hong Kong. Proceedings of the Hospital Authority Convention 2014; 2014 May 7-8; Hong Kong.

3. Mak JC, Cameron ID, March LM; National Health and Medical Research Council. Evidence-based guidelines for the management of hip fractures in older persons: an update. Med J Aust 2010;192:37-41.

4. Australian & New Zealand Hip Fracture Registry. Australian and New Zealand guideline for hip fracture care. September 2014.

5. Evidence update—Hip fracture. London: National Institute for Health and Clinical Excellence. March 2013.

6. Shiga T, Wajima Z, Ohe Y. Is operative delay associated with increased mortality of hip fracture patients? Systematic review, meta-analysis, and meta-regression. Can J Anaesth 2008;55:146-54. Crossref

7. Sircar P, Godkar D, Mahgerefteh S, Chambers K, Niranjan S, Cucco R. Morbidity and mortality among patients with hip fractures surgically repaired within and after 48 hours. Am J Ther 2007;14:508-13. Crossref

8. Siegmeth AW, Gurusamy K, Parker MJ. Delay to surgery prolongs hospital stay in patients with fractures of the proximal femur. J Bone Joint Surg Br 2005;87:1123-6.Crossref

9. Simunovic N, Devereaux PJ, Sprague S, et al. Effect of early surgery after hip fracture on mortality and complications: systematic review and meta-analysis. CMAJ 2010;182:1609-16.Crossref

10. Desai V, Gonda D, Ryan SL, et al. The effect of weekend and after-hours surgery on morbidity and mortality rates in pediatric neurosurgery patients. J Neurosurg Pediatr 2015;16:726-31. Crossref

11. Kelz RR, Freeman KM, Hosokawa PW, et al. Time of day is associated with postoperative morbidity: an analysis of the national surgical quality improvement program data. Ann Surg 2008;247:544-52. Crossref

12. Scott SW, Bowrey S, Clarke D, Choke E, Bown MJ, Thompson JP. Factors influencing short- and long-term mortality after lower limb amputation. Anaesthesia 2014;69:249-58. Crossref

13. Dorotka R, Schoechtner H, Buchinger W. Influence of nocturnal surgery on mortality and complications in patients with hip fractures [in German]. Unfallchirurg 2003;106:287-93. Crossref

14. Chacko AT, Ramirez MA, Ramappa AJ, Richardson LC, Appleton PT, Rodriguez EK. Does late night hip surgery affect outcome? J Trauma 2011;71:447-53; discussion 453. Crossref

15. Bhattacharyya T, Vrahas MS, Morrison SM, et al. The value of the dedicated orthopaedic trauma operating room. J Trauma 2006;60:1336-40. Crossref

16. Wixted JJ, Reed M, Eskander MS, et al. The effect of an orthopedic trauma room on after-hours surgery at a level one trauma center. J Orthop Trauma 2008;22:234-6. Crossref

17. Giannoulis D, Calori GM, Giannoudis PV. Thirty-day mortality after hip fractures: has anything changed? Eur J Orthop Surg Traumatol 2016;26:365-70. Crossref

18. Poh KS, Lingaraj K. Complications and their risk factors following hip fracture surgery. J Orthop Surg (Hong Kong) 2013;21:154-7. Crossref

19. Rashid RH, Zubairi AJ, Slote MU, Noordin S. Hip fracture surgery: does time of the day matter? A case-controlled study. Int J Surg 2013;11:923-5. Crossref

20. Switzer JA, Bennett RE, Wright DM, et al. Surgical time of day does not affect outcome following hip fracture fixation. Geriatr Orthop Surg Rehabil 2013;4:109-16. Crossref

21. Ho ST, Chau YS, Wong WC. Short-term outcome of operated geriatric hip fracture. Hong Kong J Orthop Surg 1997;1:7-12.

22. Vidán M, Serra JA, Moreno C, Riquelme G, Ortiz J. Efficacy of a comprehensive geriatric intervention in older patients hospitalized for hip fracture: a randomized, controlled trial. J Am Geriatr Soc 2005;53:1476-82. Crossref

23. Zuckerman JD, Skovron ML, Koval KJ, Aharonoff G, Frankel VH. Postoperative complications and mortality associated with operative delay in older patients who have a fracture of the hip. J Bone Joint Surg Am 1995;77:1551-6. Crossref

24. Lawrence VA, Hilsenbeck SG, Noveck H, Poses RM, Carson JL. Medical complications and outcomes after hip fracture repair. Arch Intern Med 2002;162:2053-7. Crossref

25. Liodakis E, Antoniou J, Zukor DJ, Huk OL, Epure LM, Bergeron SG. Major complications and transfusion rates after hemiarthroplasty and total hip arthroplasty for femoral neck fractures. J Arthroplasty 2016;31:2008-12. Crossref

26. Holt G, Smith R, Duncan K, Finlayson DF, Gregori A. Early mortality after surgical fixation of hip fractures in the elderly: an analysis of data from the scottish hip fracture audit. J Bone Joint Surg Br 2008;90:1357-63. Crossref

Nipple discharge is a relatively uncommon complaint in Hong Kong Chinese women. According to a study in 1997, nipple discharge constituted 1.5% of all presenting complaints for women who attended a breast clinic in Hong Kong.1 On the contrary, nipple discharge accounted for up to 4% to 7% of all presenting symptoms in other studies.2 3 This may be better explained by the unique Chinese culture and help-seeking pattern rather than a true disease pattern. With this understanding, any clinical survey will probably underestimate the prevalence of nipple discharge in Chinese women. When patients approach health care professionals because of nipple discharge, not only is it important to differentiate malignant from benign causes of nipple discharge, it is also a valuable opportunity to promote breast health awareness.

Numerous studies have demonstrated the relationship between breast cancer and nipple discharge, with malignancy reported in up to 9.3% to 21% of all patients who present with nipple discharge.4 5 The most challenging role of breast surgeons is to accurately identify these patients. Notwithstanding, controversy persists about the value and accuracy of individual investigative tools for nipple discharge.6

There are no recent data on nipple discharge and its association with malignancy in Chinese women in Hong Kong. The primary aim of this study was to report our recent experience in the management of patients with nipple discharge in a single surgical centre. The secondary aim was to report our experience of individual investigative tools in an attempt to determine their role in the management of nipple discharge.

We retrospectively reviewed the clinical data of patients who attended our Breast Clinic at the Queen Mary Hospital, a university-affiliated hospital in Hong Kong, for nipple discharge from January 2007 to December 2011. Potential subjects were identified when diagnosis coding 611.79 (other signs and symptoms in breast) was entered into our Clinical Management System, which is a territory-wide computer-based medical record system designed for use in public hospitals, and also from the prospective database of the Division of Breast Surgery, The University of Hong Kong.

Data extraction and coding were performed by the first author (WM Kan) and included duration of follow-up until December 2011, age at presentation, history of breast condition, and laterality and duration of nipple discharge before first consultation. Clinical variables included colour of nipple discharge, single- or multiple-duct discharge, associated symptoms, mammographic and ultrasonographic imaging results, as well as ductogram and cytology results. Pathology results were recorded for patients who underwent surgery or biopsy.

In order to make a meaningful comparison, we divided patients into malignant and benign subgroups. The malignant subgroup was defined by malignant pathology on a surgically resected specimen. The benign subgroup was defined by benign pathology of a surgically resected or biopsy specimen, or clinical non-progression after more than 2 years of follow-up. Patients who did not undergo surgery or biopsy and who were followed up for less than 2 years were excluded (Fig).

In the first part of our study, we compared the background clinical variables and investigative results between the two subgroups. In the second part of our analysis, we reported the sensitivity, specificity, positive predictive value, and negative predictive value of individual investigative tools.

For the purpose of this analysis, we also classified the results of clinical examination, mammography, ultrasonography, and cytology as ‘test positive’ or ‘test negative’ for underlying malignancy. Presence of a palpable breast mass (regardless of mobility) was considered a positive result and no palpable breast mass a negative result. For mammographic findings, microcalcifications were considered a positive result. For ultrasonography, a detectable mass was ‘test positive’ for underlying malignancy; non-solitary dilated ducts, cysts, and normal ultrasonogram were regarded as ‘test negative’. For ductogram results, dilated ducts, irregularity, and the presence of ductal filling defects were considered positive. For cytology, atypical, suspicious, and malignant were considered ‘test positive’, and benign as ‘test negative’. This study was done in accordance with the principles outlined in the Declaration of Helsinki.

R version 3.0.2 (the R Foundation) and the SPSS (Windows version 14.0; SPSS Inc, Chicago [IL], United States) were used for data analysis. To determine the differences between subgroups, Wilcoxon rank sum test and Fisher’s exact test were used for numerical data and categorical data, respectively. Multiple logistic regression was performed to examine the odds ratios of the factors. Backward selection through likelihood ratio test with removal of P value of 0.1 was conducted for model selection. Variables in univariate analysis with a P value of <0.1 were included in the full model. A P value of <0.05 was considered statistically significant.

Table 1 summarises the first part of our analysis. We identified 102 patients who presented to our Breast Clinic during the study period. They had either a tissue diagnosis or had been followed up for longer than 2 years without tissue diagnosis. There were 31 and 71 patients in the malignant and benign subgroups, respectively.

The median age at presentation of the benign subgroup was significantly younger than that of the malignant subgroup (48 vs 59 years; P=0.003). The median interval between onset of nipple discharge and first presentation was significantly longer in the benign subgroup than in the malignant subgroup (13 vs 4 weeks; P=0.002).

Comparing the two subgroups, a larger proportion of patients in the malignant subgroup presented with blood-stained discharge (87.1% vs 47.9%; P=0.002) and had a breast mass at presentation (46.7% vs 7.0%; P<0.001). For the individual investigative modalities, with the exception of ultrasonography, neither mammography, ductography nor cytology showed any statistically significant difference between the malignant and benign subgroups.

Table 2 summarises the second part of the study. We calculated the sensitivity, specificity, and positive and negative predictive values of mammographic, ultrasonographic, cytological, and ductographic findings. There were 83, 95, 27, and 46 patients who underwent mammography, ultrasonography, cytology, and ductography, respectively. The positive and negative predictive values of cytology were 41.2% and 80.0%, respectively. Ductography had a sensitivity of 100%, specificity of 7.5%, positive predictive value of 14.0%, and negative predictive value of 100%.

Multiple logistic regression analysis with backward selection was performed. Covariates with a P value of <0.1 were included in the full model (Table 1). By likelihood ratio test and removal of variables with a P value of >0.1, duration of nipple discharge, colour of nipple discharge, mastalgia, and associated mass remained in the final model (Table 3).

Compared with serous, milky and brownish discharge, patients with blood-stained discharge had a significantly higher risk for malignancy (odds ratio=13.368; 95% confidence interval, 1.926-92.809). In addition, compared with patients having no symptoms, those with a breast mass had a significantly higher risk for malignancy (odds ratio=14.648; 95% confidence interval, 3.155-68.000) [Table 3].

A methodologically ideal study of nipple discharge would require every patient to undergo the same investigations and also surgery for final pathology. This, however, would be unethical. For patients who opted for non-operative management of nipple discharge, our retrospective study considered 2-year clinical non-progression a reasonable surrogate for benign breast pathology.

Women in the malignant subgroup were significantly older at presentation than their benign counterparts. This was in agreement with the fact that physiological nipple discharge is more common in younger premenopausal women. Caution should be exercised in postmenopausal women who present with nipple discharge and the possibility of malignancy investigated before concluding a benign pathology.

With respect to the colour of nipple discharge, underlying benign and malignant causes had a different pattern. Benign pathology was more likely to be associated with non–blood-stained discharge (n=37, 52.1%), whereas malignant pathology was more likely to be associated with blood-stained discharge (n=27, 87.1%). This is not pathognomonic but did reach statistical significance.

The differentiation between multiple-duct and single-duct discharge showed no association with underlying pathology.

As shown in Table 2, mammography had a higher specificity of 98.4% and positive predictive value of 66.7% but a disappointingly low sensitivity of 9.5%. Therefore, a normal mammogram did not confidently exclude malignancy. On the other hand, breast ultrasonography had a specificity and negative predictive value of 87.0% and 75.0%, respectively. Mammography was routinely offered to patients who presented with nipple discharge. Complementary breast ultrasonography was also arranged, especially for younger Asian women with denser breasts on mammography.7 In our experience, complementary ultrasonography increases the overall sensitivity and negative predictive value compared with mammography alone.

Opinion is divided on the value of cytological examination. While some studies report a complementary diagnostic value and recommend its routine use,8 9 others report it has little such value and advise against its routine use.10

Of the 102 patients, 36 had demonstrable nipple discharge at consultation with a sample collected for examination. Of these 36 specimens, only 27 showed a sufficient number of cells to make a cytological diagnosis. Nonetheless, we attempted to analyse its accuracy. The sensitivity and specificity of cytological examination were 77.8% and 44.4%, respectively. Its positive predictive value was disappointingly low at 41.2% and its negative predictive value was 80.0%. The diagnostic value of this investigation was limited as not every patient had demonstrable nipple discharge and not every specimen contained adequate cells for testing. Nonetheless, this investigation is minimally invasive so was always performed if there was demonstrable nipple discharge, although it rarely affected the clinical decision or plan of management.

The value of ductography is debatable. While some studies have validated the diagnostic value of preoperative ductography in differentiating benign and malignant pathology,11 12 others doubt its value.13 Rather than differentiating benign and malignant pathology, we used preoperative ductogram to aid in the localisation of non-palpable lesions.14 15 The sensitivity was 100% whereas the specificity was low at 7.5%, with a positive predictive value of 14.0% and a negative predictive value of 100%.

Magnetic resonance imaging was not included in our routine evaluation of patients with nipple discharge although we acknowledge its value in the detection of carcinoma in these patients. It has an exceptionally high sensitivity for both invasive and in-situ carcinoma.16 Its routine use in patients with a breast lesion is nonetheless limited by its relatively low specificity of 72% (95% confidence interval, 67%-77%).17 The role of magnetic resonance imaging in patients with nipple discharge has been extensively validated,18 19 20 21 suggesting that it may detect or exclude the presence of carcinoma with a high degree of certainty. Magnetic resonance imaging may be considered when all other available strategies are inconclusive.

Emerging evidence suggests that neither clinical variables nor preoperative investigations reliably distinguish benign and malignant pathology so duct excision should be offered to every patient with nipple discharge.22 23 24 25 26 We offered microdochectomy to patients with no palpable breast lesion based on two indications: clinical or radiological suspicion, or a patient’s wish to stop nipple discharge by surgery. It is likely that offering microdochectomy to all patients with nipple discharge would result in overtreatment as the final pathology was benign in most cases. In patients with negative clinical examination and negative imaging findings, a period of watchful waiting with regular follow-up is a reasonable alternative to surgical intervention.

The association of blood-stained discharge with malignancy is controversial. Morrogh et al24 reported that haemorrhagic discharge did not indicate malignancy or high risk, and non-haemorrhagic discharge did not exclude malignancy. In our study, we showed that blood-stained discharge was associated with malignancy but was not pathognomonic.

On the other hand, presence of an associated breast mass was a significant finding. This may be because it is the most common presenting symptom of breast cancer, and its incidence rises with age.

Our study had several limitations. First, as data collection was retrospective, there might have been inconsistent or incomplete recording of clinical findings. Study subjects might not be representative and some data for importable variables might have been missing. No blinding during information extraction or coding could be achieved as it was performed by the first author. Second, the small sample size limited the power of our study although this could in part be due to the relatively conservative culture and help-seeking pattern of Hong Kong Chinese women. The unequal arm size also limited the interpretation of statistical significance of comparisons. Third, our assumption of 2-year clinical non-progression as benign pathology might have underestimated the true incidence of malignancy in our group of patients. Lastly, the small number of adequate cytology specimens limited meaningful analysis of this investigation. As the sample taken for cytology is usually small, it will affect the sensitivity.

Clinical variables including age at presentation, duration and colour of discharge, presence of an associated breast mass, and abnormal sonographic findings were important in suggesting the underlying pathology of nipple discharge. Only blood-stained nipple discharge and an associated breast mass remained in the multiple logistic regression model and were statistically significant. In patients with non–blood-stained nipple discharge, as well as a negative clinical breast examination and imaging, we may infer an underlying benign pathology. Further prospective studies with a larger sample size are advocated.

All authors have disclosed no conflicts of interest.

The authors would like to thank Mr Wing-pan Luk and Mr Ling-hiu Fung, Medical Physics & Research Department, Hong Kong Sanatorium & Hospital, Hong Kong for their statistical contribution to this paper.

1. Cheung KL, Alagaratnam TT. A review of nipple discharge in Chinese women. J R Coll Surg Edinb 1997;42:179-81.

2. Murphy IG, Dillon MF, Doherty AO, et al. Analysis of patients with false negative mammography and symptomatic breast carcinoma. J Surg Oncol 2007;96:457-63. Crossref

3. Vargas HI, Vargas MP, Eldrageely K, Gonzalez KD, Khalkhali I. Outcomes of clinical and surgical assessment of women with pathological nipple discharge. Am Surg 2006;72:124-8.

4. Murad TM, Contesso G, Mouriesse H. Nipple discharge from the breast. Ann Surg 1982;195:259-64. Crossref

5. King TA, Carter KM, Bolton JS, Fuhrman GM. A simple approach to nipple discharge. Am Surg 2000;66:960-6.

6. Jain A, Crawford S, Larkin A, Quinlan R, Rahman RL. Management of nipple discharge: technology chasing application. Breast J 2010;16:451-2. Crossref

7. Kwong A, Cheung PS, Wong AY, et al. The acceptance and feasibility of breast cancer screening in the East. Breast 2008;17:42-50. Crossref

8. Pritt B, Pang Y, Kellogg M, St. John T, Elhosseiny A. Diagnostic value of nipple cytology: study of 466 cases. Cancer 2004;102:233-8. Crossref

9. Kalu ON, Chow C, Wheeler A, Kong C, Wapnir I. The diagnostic value of nipple discharge cytology: breast imaging complements predictive value of nipple discharge cytology. J Surg Oncol 2012;106:381-5. Crossref

10. Kooistra BW, Wauters C, van de Ven S, Strobbe L. The diagnostic value of nipple discharge cytology in 618 consecutive patients. Eur J Surg Oncol 2009;35:573-7. Crossref

11. Hou MF, Huang TJ, Liu GC. The diagnostic value of galactography in patients with nipple discharge. Clin Imaging 2001;25:75-81. Crossref

12. Hou MF, Huang CJ, Huang YS, et al. Evaluation of galactography for nipple discharge. Clin Imaging 1998;22:89-94. Crossref

13. Dawes LG, Bowen C, Venta LA, Morrow M. Ductography for nipple discharge: no replacement for ductal excision. Surgery 1998;124:685-91. Crossref

14. Peters J, Thalhammer A, Jacobi V, Vogl TJ. Galactography: an important and highly effective procedure. Eur Radiol 2003;13:1744-7. Crossref

15. Lamont JP, Dultz RP, Kuhn JA, Grant MD, Jones RC. Galactography in patients with nipple discharge. Proc (Bayl Univ Med Cent) 2000;13:214-6. Crossref

16. Heywang-Koebrunner SH. Diagnosis of breast cancer with MR—review after 1250 patients. Electromedica 1993;61:43-52.

17. Peters NH, Borel Rinkes IH, Zuithoff NP, Mali WP, Moons KG, Peeters PH. Meta-analysis of MR imaging in the diagnosis of breast lesions. Radiology 2008;246:116-24. Crossref

18. Orel SG, Dougherty CS, Reynolds C, Czerniecki BJ, Siegelman ES, Schnall MD. MR imaging in patients with nipple discharge: initial experience. Radiology 2000;216:248-54. Crossref

19. Nakahara H, Namba K, Watanabe R, et al. A comparison of MR imaging, galactography and ultrasonography in patients with nipple discharge. Breast Cancer 2003;10:320-9. Crossref

20. Hirose M, Otsuki N, Hayano D, et al. Multi-volume fusion imaging of MR ductography and MR mammography for patients with nipple discharge. Magn Reson Med Sci 2006;5:105-12. Crossref

21. Ballesio L, Maggi C, Savelli S, et al. Role of breast magnetic resonance imaging (MRI) in patients with unilateral nipple discharge: preliminary study [in English, Italian]. Radiol Med 2008;113:249-64. Crossref

22. Adepoju LJ, Chun J, El-Tamer M, Ditkoff BA, Schnabel F, Joseph KA. The value of clinical characteristics and breast-imaging studies in predicting a histopathologic diagnosis of cancer or high-risk lesion in patients with spontaneous nipple discharge. Am J Surg 2005;190:644-6. Crossref

23. Lanitis S, Filippakis G, Thomas J, Christofides T, Al Mufti R, Hadjiminas DJ. Microdochectomy for single-duct pathologic nipple discharge and normal or benign imaging and cytology. Breast 2008;17:309-13. Crossref

24. Morrogh M, Park A, Elkin EB, King TA. Lessons learned from 416 cases of nipple discharge of the breast. Am J Surg 2010;200:73-80. Crossref

25. Alcock C, Layer GT. Predicting occult malignancy in nipple discharge. ANZ J Surg 2010;80:646-9. Crossref

26. Foulkes RE, Heard G, Boyce T, Skyrme R, Holland PA, Gateley CA. Duct excision is still necessary to rule out breast cancer in patients presenting with spontaneous bloodstained nipple discharge. Int J Breast Cancer 2011;2011:495315. Crossref

Prostate-specific antigen (PSA) plays a significant role in the early detection of prostate cancer in current practice.1 2 It is, however, a double-edged sword that leads to overdiagnosis, especially for clinically insignificant prostate cancer.3 4 Curative treatments for low-risk prostate cancer include radical prostatectomy and radiotherapy, both of which are associated with significant morbidities.5 6 7 In recent years, the concept of active surveillance (AS) has been adopted with the aim of monitoring clinically insignificant prostate cancer until disease progression, at which point radical prostatectomy or radiotherapy is considered. The ultimate objective is to delay or avoid the morbidities associated with radical treatments without compromising survival.8 9 10

Although AS is an established management option for low-risk prostate cancer, different AS protocols have been adopted.11 12 13 14 15 16 17 The most commonly used include those from the University of Toronto,11 Royal Marsden,12 John Hopkins,13 14 University of California San Francisco (UCSF),15 Memorial Sloan Kettering Cancer Center (MSKCC),16 and Prostate Cancer Research International: Active Surveillance (PRIAS).17 Most AS protocols select prostate cancer with a Gleason score of ≤6, PSA level of ≤10 ng/mL, and clinical stage of ≤T2. Other parameters that are considered by some protocols include PSA density, number of positive biopsy cores, and percentage of core involvement (Table 111 12 13 14 15 16 17).

Currently, there is no consensus regarding which AS protocol we should adopt for our patients. In addition, direct comparisons between different AS protocols are few. Before deciding to follow any particular AS protocol, urologists and oncologists should be aware of their individual strengths and limitations. Our study aimed to provide some insight into this issue by performing a head-to-head comparison of six AS protocols.

Patients who underwent radical prostatectomy for prostate cancer from January 1998 to December 2012 at a university teaching hospital in Hong Kong were reviewed. Indication for radical prostatectomy was localised prostate cancer in patients with a life expectancy exceeding 10 years. All patients underwent clinical assessment including clinical T staging by digital rectal examination, serum PSA level, and transrectal ultrasound-guided prostate biopsy. Sextant biopsies were performed from 1998 to 2002, but changed to 10-core biopsies from 2002 to 2011 and subsequently 12-core biopsies thereafter. Preoperative magnetic resonance imaging of the prostate was routinely performed from 2007. From 1998 to 2007, open or laparoscopic radical prostatectomies were performed. After November 2007, all prostatectomies at our institution were performed with the da Vinci robotic surgery system. Pathological assessment of transrectal ultrasound-guided biopsy and radical prostatectomy specimens was performed by a specialist pathologist in our institution. All patients attended a follow-up visit with physical examination 2 weeks after operation, and physical examination with serum PSA level checked every 3 months for the first year, every 6 months for the second year, and then annually thereafter. Data on patient demographics, clinical T stage, serum PSA level, transrectal ultrasound-guided biopsy results, and final pathology of radical prostatectomy specimen were retrospectively retrieved by an independent third party. Pathological assessment of the radical prostatectomy specimen was performed by independent specialist pathologists.

In our current study, we compared six different AS protocols, specifically from the University of Toronto,11 Royal Marsden,12 John Hopkins,13 14 UCSF,15 MSKCC,16 and PRIAS17 (Table 1). The six protocols were retrospectively applied to our cohort and patients were stratified accordingly based on clinical T stage, serum PSA level, PSA density, Gleason score on biopsy, number of positive biopsy cores, and percentage of positive core involvement. Data from the pathological assessment of radical prostatectomy specimens including extracapsular extension, seminal vesicle invasion, upgrading to pathological T3 disease, and upgrading of Gleason score were analysed. The clinical data used in the AS protocols were those available on diagnosis of prostate cancer and operations were performed within 12 weeks of diagnosis.

Statistical analyses to compare the rate of not diagnosing clinically significant prostate cancer—defined as extracapsular extension, seminal vesicle invasion, upgrading to T3 disease, and upgrading of Gleason score in the final prostatectomy specimens—were performed. The sensitivity and specificity of each protocol in selecting localised prostate cancer (defined as pathological stage <T3) and histological low-risk disease (defined as no upgrading of Gleason score on final pathology) were compared.

Statistical analysis was performed using the SPSS (Windows version 20.0; IBM Corp, Armonk [NY], US). Independent sample t test and Pearson Chi-squared test were used for continuous and categorical variables, respectively. A P value of <0.05 was considered statistically significant. This study was done in accordance with the principles outlined in the Declaration of Helsinki.

A total of 287 patients were included in the cohort. The mean age was 66 years, mean serum PSA level was 10 ng/mL, mean number of positive cores during biopsy was 3, and mean Gleason sum at biopsy was 6. In the current cohort, 266 (93%) patients had clinical T1c or T2a prostate cancer—198 (69%) had clinical T1c disease and 68 (24%) had clinical T2a disease. Table 2 summarises the basic demographics of all patients.

When the six AS protocols were applied to the cohort, 30 to 152 patients were identified as low-risk; their mean serum PSA level ranged from 5.3 ng/mL to 7.7 ng/mL, and mean PSA density ranged from 0.12 ng/mL/mL to 0.25 ng/mL/mL. All six protocols had a mean biopsy Gleason sum of 6. Table 3 summarises the clinical characteristics of patients stratified according to different AS protocols.

In the analyses of final pathological outcomes in patients stratified into different AS protocols, extracapsular extension rate varied from 3.3% to 17.1%. The incidence of seminal vesicle invasion was low in all six protocols, ranging from 0% to 3.3%. The rate of pathological T3 disease was lowest according to the John Hopkins criteria (3.3%), while the University of Toronto criteria had the highest incidence (19.1%). Regarding the upgrading of Gleason score in the radical prostatectomy specimens, all six protocols had a relatively high rate ranging from 20.6% to 34.5%. Table 4 summarises the pathological outcomes among the six AS protocols.

Comparative analyses of individual AS protocols against each other were also performed (Table 5). The University of Toronto protocol had a significantly higher rate of extracapsular extension at 17.1% and pathological T3 disease at 19.1% when compared with the more stringent protocol from John Hopkins (3.3% extracapsular extension, P=0.05 and 3.3% pathological T3 disease, P=0.03) and PRIAS (8.0% pathological T3 disease, P=0.04). In addition, the Royal Marsden protocol had a significantly higher rate of upgrading of Gleason score at 34.5% when compared with the more stringent protocol of PRIAS at 20.6% (P=0.04). There was no significant difference in the incidence of seminal vesicle invasion between the six protocols.

In terms of the ability of each protocol to identify pathological localised disease (defined as pathological stage <T3) and histologically low-risk cancer (defined as no upgrading of Gleason score), sensitivity varied from 13%-61% and 14-71%, respectively. The John Hopkins criteria demonstrated highest specificity in identifying pathological localised disease (98%) and histological low-risk cancer (94%). Table 6 illustrates the sensitivity and specificity of identifying localised and histological low-risk disease for the six AS protocols.

Prostate cancer screening has always been a controversial issue and evidence of improved survival is awaited.1 2 Nonetheless, PSA screening has undoubtedly led to overdiagnosis of insignificant prostate cancer.3 4 Active surveillance, with the purpose to delay or even avoid radical treatments and their associated morbidities, plays an important role in managing these patients. Unfortunately there are different AS protocols with various inclusion criteria, and urologists and oncologists may have difficulty deciding which protocol to adopt. The gold standard to answer this question will be a prospective randomised trial to compare overall survival following the application of different AS protocols. This, however, will require decades to observe low-risk prostate cancer patients before survival endpoints are reached. Our study provides data on pathological outcomes when different AS protocols were compared.

In our cohort, the proportion of patients eligible for active surveillance varied widely from approximately 11% to 58% according to different selection criteria (Table 3). Two recent series showed similar findings of a large discrepancy in the proportion of patients eligible for different AS protocols, varying from 16% to 63% and 28% to 69%.18 19 We demonstrated that although all AS protocols aim to select low-risk prostate cancer, the heterogeneity between them can be quite large. Clinicians need to be vigilant before adopting any of the AS protocols for their patients when further data from comparative analyses among different protocols are unavailable. The proportion of patients who were eligible for AS protocols in our study was lower than that in previous series.18 19 This may be because some patients with localised prostate cancer were treated with radiotherapy. The proportion of patients who can be selected in different AS protocols will be affected by the proportion of patients who undergo radiotherapy instead of surgery. In our centre, it is also possible that low-risk patients were selected to undergo a non-operative approach.

When the six protocols were compared after stratifying patients according to different AS criteria, the University of Toronto protocol had a significantly higher rate of extracapsular extension at 17.1% and pathological T3 disease at 19.1% than the John Hopkins protocol (3.3% extracapsular extension, P=0.05 and 3.3% pathological T3 disease, P=0.03) and PRIAS criteria (8.0% pathological T3 disease, P=0.04) [Table 5]. This observation can be explained by the difference in stringency of the two protocols. The University of Toronto criteria selected patients by two factors only: PSA of <10 ng/mL and Gleason score of ≤6; PSA density, number of positive biopsy cores, and percentage of core involvement were not considered. On the contrary, the John Hopkins criteria applied very strict criteria: a PSA density of 0.15 ng/mL/mL. In addition, only patients with T1 disease with at most two positive cores during biopsy and no more than 50% involvement of each core were selected (Table 1). Contrary to our findings, El Hajj et al19 found no significant difference in the rate of extracapsular extension, upgrading of Gleason score, or unfavourable disease when they compared the University of Toronto protocol with the John Hopkins protocol. The difference can be explained by the high rate of extracapsular extension (15%) and unfavourable disease (46%) within the John Hopkins criteria in their series, compared with 3% extracapsular extension and 3% pathological T3 disease in our cohort. This also implies that disease heterogeneity among different populations may influence the choice and results of different AS protocols.

In our study, analyses of final pathology revealed that the Royal Marsden protocol had a significantly higher rate of upgrading of Gleason score at 34.5% compared with the PRIAS criteria at 20.6% (P=0.04; Table 5). This result can be explained by the less-stringent selection criteria of the Royal Marsden protocol. First, it is the only protocol that allowed a Gleason score of 3+4 to be selected. Second, PSA level up to 15 ng/mL was permitted. These factors will invariably result in the inclusion of a proportion of patients with higher-risk disease. In the study by El Hajj et al,19 the Royal Marsden protocol were compared with the John Hopkins protocol and significantly more unfavourable disease was observed in the Royal Marsden group. Klotz et al11 also demonstrated that inclusion of Gleason score of 4 on biopsy into AS was a risk factor in predicting definitive treatment during active surveillance. These findings illustrate that active surveillance in patients with Gleason score of 3+4 is likely to miss higher-risk disease. It should be used cautiously and preferably not in young patients who are otherwise fit for radical treatments.

We have shown that less pathological T3 disease and Gleason score upgrading were present in the more-stringent John Hopkins and PRIAS protocols compared with the less stringent University of Toronto and Royal Marsden criteria. Nonetheless their sensitivity in identifying low-risk disease may be compromised by the more stringent selection criteria. More low-risk disease may therefore be excluded from surveillance by these stringent criteria. We addressed this issue in the last part of our analyses. The sensitivity and specificity in identifying localised disease (pathological stage <T3) and low-risk histology (no upgrading of Gleason score) among different AS protocols were compared (Table 6). The most stringent protocols of the John Hopkins and PRIAS had the highest specificity when selecting localised disease (94%-98%) and low-risk histology (91%-94%). However, inclusion of less pathological T3 disease and Gleason score upgrading by the more stringent protocols of John Hopkins and PRIAS should be cautious because it will, inevitably, be at the expense of low-risk patients who is excluded from AS and may receive unnecessary aggressive treatments. A recent study by Iremashvili et al18 showed that the PRIAS criteria had a better balance of sensitivity and specificity compared with the UCSF and MSKCC criteria. From our point of view, we tend to place more emphasis on high specificity since low specificity will include patients with high-risk tumours into active surveillance and thus patient survival may be jeopardised.

The present study had several limitations. First, the number of biopsy cores was not consistent throughout the study period. A proportion of patients had six-core biopsies in the early period of the cohort versus the current more recent standard of 10-12–core biopsies. Second, the sample size was relatively small due to the low incidence of prostate cancer in our population. Third, the tumour volume in prostatectomy specimens that might predict low-risk prostate cancer was not assessed. Lastly, the final prostatectomy pathology in this study was from patients who were operated on soon after diagnosis and not after a period of post-diagnosis surveillance. As a note of caution, it would be expected that the final pathology would show even worse pathological features if the patients were put on AS and operated on later. This should be noted when interpreting the results of the current study and counselling patients.

In conclusion, there is a wide range of variation in the selection criteria of different AS protocols. Active surveillance protocols based on PSA and Gleason score alone or including Gleason score of 3+4 may miss higher-risk disease and should be applied cautiously. The more stringent criteria of John Hopkins protocol and the PRIAS protocol were highly specific in identifying localised disease and low-risk histology.

All authors have disclosed no conflicts of interest

1. Schröder FH, Hugosson J, Roobol MJ, et al. Screening and prostate-cancer mortality in a randomized European study. N Engl J Med 2009;360:1320-8. Crossref

2. Andriole GL, Crawford ED, Grubb RL 3rd, et al. Mortality results from a randomized prostate-cancer screening trial. N Engl J Med 2009;360:1310-9. Crossref

3. Ploussard G, Epstein JI, Montironi R, et al. The contemporary concept of significant versus insignificant prostate cancer. Eur Urol 2011;60:291-303. Crossref

4. Etzioni R, Penson DF, Legler JM, et al. Overdiagnosis due to prostate-specific antigen screening: lessons from U.S. prostate cancer incidence trends. J Natl Cancer Inst 2002;94:981-90. Crossref

5. 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

6. Boorjian SA, Eastham JA, Graefen M, et al. A critical analysis of the long-term impact of radical prostatectomy on cancer control and function outcomes. Eur Urol 2012;61:664-75. Crossref

7. Zaorsky NG, Harrison AS, Trabulsi EJ, et al. Evolution of advanced technologies in prostate cancer radiotherapy. Nat Rev Urol 2013;10:565-79. Crossref

8. Bastian PJ, Carter BH, Bjartell A, et al. Insignificant prostate cancer and active surveillance: from definition to clinical implications. Eur Urol 2009;55:1321-30. Crossref

9. Cooperberg MR, Carroll PR, Klotz L. Active surveillance for prostate cancer: progress and promise. J Clin Oncol 2011;29:3669-76. Crossref

10. Dall’Era MA, Albertsen PC, Bangma C, et al. Active surveillance for prostate cancer: a systematic review of the literature. Eur Urol 2012;62:976-83. Crossref

11. Klotz L, Zhang L, Lam A, Nam R, Mamedov A, Loblaw A. Clinical results of long-term follow-up of a large, active surveillance cohort with localized prostate cancer. J Clin Oncol 2010;28:126-31. Crossref

12. van As NJ, Norman AR, Thomas K, et al. Predicting the probability of deferred radical treatment for localised prostate cancer managed by active surveillance. Eur Urol 2008;54:1297-305. Crossref

13. Carter HB, Kettermann A, Warlick C, et al. Expectant management of prostate cancer with curative intent: an update of the Johns Hopkins experience. J Urol 2007;178:2359-64. Crossref

14. Tosoian JJ, Trock BJ, Landis P, et al. Active surveillance program for prostate cancer: an update of the Johns Hopkins experience. J Clin Oncol 2011;29:2185-90. Crossref

15. Dall’Era MA, Konety BR, Cowan JE, et al. Active surveillance for the management of prostate cancer in a contemporary cohort. Cancer 2008;112:2664-70. Crossref

16. Berglund RK, Masterson TA, Vora KC, Eggener SE, Eastham JA, Guillonneau BD. Pathological upgrading and up staging with immediate repeat biopsy in patients eligible for active surveillance. J Urol 2008;180:1964-7. Crossref

17. van den Bergh RC, Roemeling S, Roobol MJ, Roobol W, Schröder FH, Bangma CH. Prospective validation of active surveillance in prostate cancer: the PRIAS study. Eur Urol 2007;52:1560-3. Crossref

18. Iremashvili V, Pelaez L, Manoharan M, Jorda M, Rosenberg DL, Soloway MS. Pathologic prostate cancer characteristics in patients eligible for active surveillance: a head-to-head comparison of contemporary protocols. Eur Urol 2012;62:462-8. Crossref

19. El Hajj A, Ploussard G, de la Taille A, et al. Patient selection and pathological outcomes using currently available active surveillance criteria. BJU Int 2013;112:471-7. Crossref