Identification of Doctors at Risk of Recurrent Complaints: A National Study of Healthcare Complaints in Australia PDF

ORIGINAL RESEARCH BMJ Qual Saf: first published as 10.1136/bmjqs-2012-001691 on 10 April 2013. Downloaded from http://qualitysafety.bmj.com/...

ORIGINAL RESEARCH BMJ Qual Saf: first published as 10.1136/bmjqs-2012-001691 on 10 April 2013. Downloaded from http://qualitysafety.bmj.com/ on August 31, 2024 by guest. Protected by copyright. Identification of doctors at risk of recurrent complaints: a national study of healthcare complaints in Australia Marie M Bismark,1 Matthew J Spittal,1 Lyle C Gurrin,1 Michael Ward,2 David M Studdert1,3 ▸ Additional material is ABSTRACT compared doctors who experienced mul- published online only. To view Objectives (1) To determine the distribution of tiple malpractice claims,1–5 complaints,6 7 please visit the journal online (http://dx.doi.org/10.1136/bmjqs- formal patient complaints across Australia’s and disciplinary actions8–10 with doctors 2012-001691). medical workforce and (2) to identify who experienced few or none, and identi- 1 characteristics of doctors at high risk of incurring fied differences in the sex, age and spe- Melbourne School of Population and Global Health, University of recurrent complaints. cialty profile of the two groups. Such Melbourne, Parkville, Victoria, Methods We assembled a national sample of all research helps to explain medico-legal risk Australia 18 907 formal patient complaints filed against retrospectively, but does not provide prac- 2 School of Medicine, University doctors with health service ombudsmen tical guidance for identifying risks pro- of Queensland, Brisbane, Queensland, Australia (‘Commissions’) in Australia over an 11-year spectively. Clinical leaders, risk managers, 3 Melbourne Law School, period. We analysed the distribution of liability insurers and regulators all lack University of Melbourne, complaints among practicing doctors. We then reliable methods for systematically deter- Parkville, Victoria, Australia used recurrent-event survival analysis to identify mining which doctors should be targeted Correspondence to characteristics of doctors at high risk of recurrent for assistance and preventive action before Dr David M Studdert, Melbourne complaints, and to estimate each individual they acquire troubling track records. School of Population and Global doctor’s risk of incurring future complaints. Consequently, the medico-legal enterprise Health, University of Melbourne, 207 Bouverie Street, Parkville, Results The distribution of complaints among remains reactive, dealing primarily with VIC 3052, Australia, doctors was highly skewed: 3% of Australia’s the aftermath of adverse events and beha- [email protected] medical workforce accounted for 49% of viours that lead to costly disputes. Received 15 November 2012 complaints and 1% accounted for a quarter of The conventional wisdom is that future Revised 23 January 2013 complaints. Short-term risks of recurrence varied medico-legal events cannot be predicted Accepted 24 January 2013 significantly among doctors: there was a strong at the doctor level with acceptable levels Published Online First dose-response relationship with number of of accuracy.11 12 Numerous studies have 11 April 2013 previous complaints and significant differences tried,13–23 most with limited success. This by doctor specialty and sex. At the practitioner body of research has two important short- level, risks varied widely, from doctors with comings. First, only a few studies15 17 21 80% risk. legal risk that is potentially replicable, and Conclusions A small group of doctors accounts these methods are statistically complex. Open Access for half of all patient complaints lodged with The practical consequence is that regula- Scan to access more free content Australian Commissions. It is feasible to predict tors and liability insurers today have no which doctors are at high risk of incurring more clear way of estimating risk at the practi- complaints in the near future. Widespread use of tioner level, and doing so is not a standard this approach to identify high-risk doctors and part of risk management practice. target quality improvement efforts coupled with Second, no study to date has found a ▸ http://dx.doi.org/10.1136/ effective interventions, could help reduce adverse way to deal well with temporal aspects bmjqs-2013-001880 events and patient dissatisfaction in health of risk, such as the evolving nature of ▸ http://dx.doi.org/10.1136/ bmjqs-2013-001902 systems. doctors’ medico-legal event histories, ▸ http://dx.doi.org/10.1136/ which can be crucial information in bmjqs-2013-002138 INTRODUCTION assembling a risk profile. Previous claims To cite: Bismark MM, To many doctors who are sued or com- and complaints have been identified as an Spittal MJ, Gurrin LC, et al. plained against, the event seems random. important predictor of future events, but BMJ Qual Saf 2013;22: At the population level, however, there only in analyses that specify this variable 532–540. are patterns. Previous studies have crudely—usually by ‘freezing’ a doctor’s 532 Bismark MM, et al. BMJ Qual Saf 2013;22:532–540. doi:10.1136/bmjqs-2012-001691  Original research BMJ Qual Saf: first published as 10.1136/bmjqs-2012-001691 on 10 April 2013. Downloaded from http://qualitysafety.bmj.com/ on August 31, 2024 by guest. Protected by copyright. track record at a specific point to estimate a Outside of the clinic or hospital in which care is ‘one-time’ effect.13 14 16 17 19 21 24 25 This approach received, Commissions are the primary avenue of redress is out of step with how claims and complaints are for patients dissatisfied with the quality of care they have managed. The frontline challenges are to determine received. Plaintiffs’ lawyers in Australia will rarely take how a practitioner’s risk profile changes over time as on cases unless they have first proceeded through new information (including new events) comes to Commission processes (although the vast majority of hand; when support or intervention measures to complaints do not become negligence claims). At least prevent further events are warranted; and how strong 10 other Organisation for Economic Co-operation and those measures should be. A risk prediction method Development (OECD) countries—including Austria, that helped to address these questions would have Finland, Israel, New Zealand and the UK—have similar considerable potential for boosting the contribution bodies.27 28 In the UK, the closest analogue is the of medico-legal institutions to quality improvement. Parliamentary and Health Service Ombudsman. We assembled a national sample of nearly 19 000 Commissions in all Australian states and territories formal healthcare complaints lodged against doctors except South Australia participated in the study. in Australia between 2000 and 2011. We then used a These seven jurisdictions have 21 million residents time-to-event method of analysis to determine charac- and 90% of the nation’s 88 000 registered doctors. teristics of doctors poised to incur recurrent com- The study was approved by the ethics committee at plaints, and to estimate each practitioner’s risk of the University of Melbourne. recurrence at specific time points. The study had two main goals: to identify predictors of complaint-prone Data doctors in Australia, and to develop a robust and Between May 2011 and February 2012 we collected useful method for forecasting medico-legal risk. data on-site at Commission offices in each participat- ing state and territory. Complaints against doctors were identified by querying the Commissions’ admin- METHODS istrative data systems. The filing period of interest Setting spanned 12 years and differed slightly by jurisdiction: Health service commissions (Commissions) are statu- 2000–2011 for the Australian Capital Territory, tory agencies established in each of Australia’s six states the Northern Territory, Queensland, Tasmania and and two territories. Commissions have responsibility Victoria; 2000–2010 for Western Australia; and for receiving and resolving patient complaints about 2006–2011 for New South Wales. the quality of healthcare services. Patients or their All Commissions record the names of persons and advocates must initiate complaints in writing, but the institutions that are the subject of complaints, as well process is free and legal representation is optional.26 as the filing date, the nature of the complaint, the Table 1 compares the jurisdiction and functions of type of health professional named and their practice Commissions to those of the two other agencies that location. Although all Commissions recorded doctors’ handle medico-legal matters in Australia—civil courts clinical specialty, the quality of this variable was and the Medical Board of Australia. mixed. Doctors’ age and sex were not routinely Table 1 Jurisdiction and functions of key agencies with responsibility for medico-legal matters in Australia Civil courts Health complaints commissions Medical Board of Australia Cases handled ▸ Negligence claims ▸ Patient complaints ▸ Conduct, competence, or health matters Jurisdictional ▸ Substandard care causing ▸ Low-quality care ▸ Professional misconduct focus patient harm ▸ Patient dissatisfaction with care ▸ Performance or competence falling below professional standards ▸ Ill-health, substance misuse, or impairment Procedures used ▸ Out-of-court negotiation ▸ Early resolution ▸ Review of doctor’s competence or health ▸ Alternative forms of dispute ▸ Conciliation status resolution (eg, mediation, ▸ Investigation ▸ Investigation arbitration) ▸ Disciplinary charges ▸ Trials before judges Remedies ▸ Monetary damages ▸ Communication (eg, facilitate apology or ▸ Correction (eg, requirement that practitioner explanation) undergo education, rehabilitation, ▸ Restoration (eg, facilitate provision of further monitoring etc) treatment, fee forgiveness, monetary ▸ Sanction (eg, suspension or revocation of settlement) practice licence*) ▸ Correction (eg, recommend system change) *Typically, such sanctions are imposed by external administrative tribunals in proceedings initiated by the Medical Board of Australia. Bismark MM, et al. BMJ Qual Saf 2013;22:532–540. doi:10.1136/bmjqs-2012-001691 533  Original research BMJ Qual Saf: first published as 10.1136/bmjqs-2012-001691 on 10 April 2013. Downloaded from http://qualitysafety.bmj.com/ on August 31, 2024 by guest. Protected by copyright. collected. We therefore supplemented the jurisdiction, and the doctor’s specialty, age, sex and Commissions’ administrative data with data from principal practice location. another source. The number of prior complaints was specified as a AMPCo Direct, a subsidiary of the Australian time-varying covariate. Age was also time-varying in Medical Association, maintains a comprehensive list of the sense that we allowed doctors to move into higher doctors in Australia, including information on their age categories, commensurate with their age at the sex, date of birth, specialty and subspeciality, and prac- time of the complaint. We fit cluster-adjusted robust tice location. We purchased the AMPCo Direct data- SEs to account for doctors who experienced repeated base and matched doctors listed in it to doctors named complaints over time. in the complaints databases. The matching method is Details of model selection and specification are described in an online supplementary appendix. described in the online supplementary appendix. All statistical analyses were conducted using Stata 12.1. Variables Risk predictions We coded specialty into 13 categories, based on those To estimate doctors’ risks of experiencing complaints promulgated by the Medical Board of Australia.29 over time, we plotted adjusted failure curves.33 34 Details Doctors’ principal practice address was classified as of the statistical techniques used to create these curves urban or rural, based on the location of its postcode are provided in the online supplementary appendix. We within a standard geographic classification system.30 also plotted failure curves showing the predicted risk of The nature of concerns raised in complaints was recurrent complaints for several individual doctors. sorted into 20 broad ‘issue’ categories. Commissions Values for all failure curves were computed using coeffi- run dispute resolution processes; they generally do cients from the main multivariable model, and hence, not rule on the merit of complaints, nor make find- derived from the survivor function, S(t). ings for or against parties, so it was not possible to Sensitivity analysis include a variable indicating how meritorious com- We tested the robustness of estimates from the main plaints were. multivariable analysis by rerunning the analysis on a subsample of complaints (n=10 010) with issue codes Statistical analysis suggestive of relatively serious concerns (namely, poor Distributional analysis clinical care, breach of conditions, rough or painful We plotted the cumulative distribution of complaints treatment and sexual contact or relationship). among two populations of doctors: (1) all unique doctors named in complaints and (2) all practicing RESULTS doctors in the seven jurisdictions under study (ie, Characteristics of complained-against doctors regardless of whether they had been named in com- and complaints plaints). The size of this second population was based The study sample consisted of 18 907 complaints against on the number of doctors in employment in 2006,31 11 148 doctors. Sixty-one percent of the complaints the median study year. Because certain classes of com- addressed clinical aspects of care, most commonly con- plaints do not name doctors individually (eg. com- cerns with treatment (41%), diagnosis (16%) and medi- plaints arising in public hospitals in several of the cations (8%) (table 2). Nearly one quarter of complaints study jurisdictions), we adjusted the proportions in addressed communication issues, including concerns the distributional calculations to ensure the numera- with the attitude or manner of doctors (15%), and the tors (number of complaints) matched the denomina- quality or amount of information provided (6%). tors (size of the ‘exposed’ segment of the medical Seventy-nine percent of the doctors named in com- workforce). Details are provided in the online supple- plaints were male, 47% were general practitioners and mentary appendix. 14% were surgeons (table 3). Examples of several complaints are included in the online supplementary Multivariable survival analysis appendix. We used multivariable survival analysis to identify pre- dictors of doctors’ risks of recurrent complaints. Incidence and distribution of complaints Specifically, we used an Anderson–Gill model32 in Doctors in the sample were complained against an which the time-scale ran from time from first event average of 1.98 times (SD 2.31). The distribution was (ie, a doctor’s earliest complaint) and allowed each highly skewed, with a small subgroup of doctors doctor in the sample to accrue multiple complaints accounting for a disproportionate share of complaints. over the period of observation. The outcome variable Figure 1 plots the cumulative distribution of com- was the occurrence of a complaint against a doctor, plaints among doctors in six jurisdictions over a conditional on the doctor having been named in an decade. (New South Wales data was not included in earlier complaint. The covariates were the number of these plots because the complaints window there prior complaints a doctor had experienced, spanned only 5 years.) The curve on the left side of 534 Bismark MM, et al. BMJ Qual Saf 2013;22:532–540. doi:10.1136/bmjqs-2012-001691  Original research BMJ Qual Saf: first published as 10.1136/bmjqs-2012-001691 on 10 April 2013. Downloaded from http://qualitysafety.bmj.com/ on August 31, 2024 by guest. Protected by copyright. Table 2 Issues in a national sample of 18 907 complaints filed Table 3 Characteristics of 11 148 doctors named in complaints by patients n % n %* Gender Clinical care 11579 61 Male 8818 79 Treatment 7746 41 Female 2255 20 Diagnosis 3080 16 Missing 75 1 Medication 1572 8 Speciality Hygiene/infection control 190 1 General practice 5289 47 Discharge/transfer 113 0.6 Surgery 1540 14 Other clinical care 127 0.7 Orthopaedic 432 4 Communication 4279 23 General 398 4 Attitude or manner 2823 15 Plastic 177 2 Information 1132 6 Other surgical 533 5 Consent 582 3 Internal medicine 1243 11 Other communication 32 0.2 Obstetrics and gynaecology 541 5 Costs or billing 1309 7 Psychiatry 672 6 Medical records, certificates, or reports 1304 7 Anaesthesia 404 4 Access and timeliness 1257 7 Ophthalmology 243 2 Sexual contact or relationship 625 3 Dermatology 157 1 Rough or painful treatment 477 3 Radiology 200 2 Confidentiality or information privacy 392 2 Other 501 4 Breach of conditions 332 2 Missing 358 3 Grievance handling 213 1 Age Discrimination 103 0.5 22–35 years 757 7 Other 126 0.7 36–45 years 2624 24 *Complaint issues sum to more than 100% because some complaints 46–55 years 3354 30 involved multiple issues. 56–65 years 2184 20 66+ years 691 6 Missing 1583 14 the figure shows the distribution of complaints among Practice location doctors who experienced one or more complaints in Urban 8241 74 the decade. Fifteen percent of doctors named in com- Rural 2775 25 plaints accounted for 49% of all complaints, and 4% Missing 132 1 accounted for a quarter of all complaints. The curve on the right side of the figure shows the distribution of complaints across the full population of practicing surgeons had twice the risk (HR 2.04; 95% CI 1.75 doctors, not just those who experienced complaints. to 2.38), and risks were approximately 50% higher Three percent of all doctors accounted for 49% of among dermatologists (HR 1.56; 95% CI 1.30 to all complaints, and 1% accounted for a quarter of all 1.88) and obstetrician-gynecologists (HR 1.50; 95% complaints. CI 1.29 to 1.76). Anaesthetists had significantly lower risks of recurrence (HR 0.65; 95% CI 0.54 to 0.79). Multivariable predictors of recurrent complaints Male doctors had a 40% higher risk of recurrence In multivariable analyses, the number of prior com- than their female colleagues (HR 1.36; 95% CI 1.23 plaints doctors had experienced was a strong pre- to 1.50). Location of practice (urban vs rural) was not dictor of subsequent complaints, and a dose-response significantly associated with recurrence. Compared relationship was evident (table 4). Compared with with doctors 35 years of age or younger, older doctors with one prior complaint, doctors with doctors had 30–40% higher risks of recurrence; this two complaints had nearly double the risk of recur- level of heightened risk was similar through the rence (HR 1.93; 95% CI 1.79 to 2.09), and doctors middle-aged and older-aged groups. with five prior complaints had six times the risk of recurrence (HR 6.16; 95% CI 5.09 to 7.46). Risks of recurrence over time Doctors with 10 or more prior complaints had Doctors named in a third complaint had a 38% 30 times the risk of recurrence (HR 29.56; 95% CI chance of being the subject of a further complaint 19.24 to 45.41). within a year, and a 57% probability of being com- Risk of recurrence also varied significantly by spe- plained against again within 2 years (figure 2A). cialty. Compared with general practitioners, plastic Doctors named in a fifth complaint had a 59% 1-year Bismark MM, et al. BMJ Qual Saf 2013;22:532–540. doi:10.1136/bmjqs-2012-001691 535  Original research BMJ Qual Saf: first published as 10.1136/bmjqs-2012-001691 on 10 April 2013. Downloaded from http://qualitysafety.bmj.com/ on August 31, 2024 by guest. Protected by copyright. Table 4 Multivariable regression analysis estimating risk of recurrent complaints* HR (95% CI) p Value Number of prior complaints