Pathology test-ordering behaviour of Australian general practice trainees: a cross-sectional analysis†

International Journal for Quality in Health Care, 2015, 27(6), 528–535 doi: 10.1093/intqhc/mzv086 Advance Access Publication Date: 20 October 2015 Article Article Downloaded from https://academic.oup.com/intqhc/article-abstract/27/6/528/2357515 by guest on 22 May 2020 Pathology test-ordering behaviour of Australian general practice trainees: a cross-sectional analysis† SIMON MORGAN1, KIM M. HENDERSON1, AMANDA TAPLEY1, JOHN SCOTT1, MIEKE L. VAN DRIEL2, NEIL A. SPIKE3, LAWRIE A. MCARTHUR4, ANDREW R. DAVEY5, CHRIS OLDMEADOW6, JEAN BALL6, and PARKER J. MAGIN1,5 1 General Practice Training Valley to Coast, PO Box 573, HRMC, Mayfield, NSW 2310, Australia, 2Discipline of General Practice, School of Medicine, The University of Queensland, L8 Health Sciences Building, Royal Brisbane and Women’s Hospital, Brisbane ALD 4029, Australia, 3Victorian Metropolitan Alliance General Practice Training, 15 Cato Street, Hawthorn, VIC 3122, Australia, 4Adelaide to Outback General Practice Training, Lower Level, 183 Melbourne Street, North Adelaide, SA 5006, Australia, 5Discipline of General Practice, University of Newcastle, Newbolds Building, University Drive, Callaghan, NSW 2308, Australia, and 6Hunter Medical Research Unit, Locked Bag 1000, New Lambton, NSW 2305, Australia Address reprint requests to: Simon Morgan, General Practice Training Valley to Coast, PO Box 573, Hunter Regional Mail Centre, Mayfield, NSW 2310, Australia. Tel: +61-02-4968-6753; Fax: +61-02-4960-0417; E-mail: simon.morgan@gptvtc.com.au † The paper has not been previously presented at any academic meetings. Accepted 27 September 2015 Abstract Objective: In the context of increasing over-testing and the implications for patient safety, to establish the prevalence and nature of pathology test-ordering of GP trainees, and to describe the associations of this test-ordering. Design: A cross-sectional analysis of data from the Registrar Clinical Encounters in Training (ReCEnT) cohort study. Setting: Five of Australia’s 17 general practice regional training providers, encompassing urban-tovery remote practices. Participants: GP trainees. Main Outcome Measure(s): The number of pathology tests ordered per problem/diagnosis managed. Results: A total of 856 individual trainees (response rate 95.2%) contributed data from 1832 traineeterms, 108 759 encounters and 169 304 problems. Pathology test-ordering prevalence was 79.3 tests (95% CI: 78.8–79.8) per 100 encounters, 50.9 (95% CI: 50.6–51.3) per 100 problems, and at least 1 test was requested in 22.4% of consultations. Most commonly ordered was full blood count (6.1 per 100 problems). The commonest problem prompting test-ordering was ‘check-up’ (18.6%). Test-ordering was significantly associated, on multivariable analysis, with the trainee having worked at the practice previously; the patient being adult, male and new to both trainee and practice; the practice being urban; the problem/diagnosis being new; imaging being ordered; referral being made and followup being arranged. Trainees were significantly less likely to order tests for problems/diagnoses for which they had sought in-consultation information or advice. © The Author 2015. Published by Oxford University Press in association with the International Society for Quality in Health Care; all rights reserved 528 529 Test-ordering by trainees • Patient Safety Conclusions: Compared with the established GPs, trainees order more pathology tests per consultation and per problem managed, and in a higher proportion of consultations. Our findings will inform educational policy to enhance quality and safety in general practice training. Key words: appropriateness, under-use and over-use, appropriate health care, primary care/general practice, setting of care, training/education, human resources, laboratory test, pathology Introduction Methods Participants This was a cross-sectional analysis of data from the Registrar Clinical Encounters in Training (ReCEnT) cohort study. The study methodology has been described in detail elsewhere [20]. Briefly, ReCEnT is an ongoing cohort study of GP trainees’ in-practice clinical experiences undertaken in 5 of Australia’s 17 general practice regional training providers (RTPs), encompassing urban, rural, remote and very remote practices in 5 of Australia’s 6 states. General practice vocational training in Australia entails a minimum of three 6-month terms in the general practice setting. Procedures In ReCEnT, we document participating trainee characteristics and the characteristics of their training practice. Trainees record the details of 60 consecutive patient consultations, representing ∼1 week of consultations, each 6-month training term. Data collection is conducted around the mid-point of the term. The analyses in this study used data from nine collection periods for the period 2010–14. Outcome factors The primary outcome factor was the number of pathology tests ordered per problem/diagnosis encountered by the trainee. Trainees recorded all pathology tests ordered (up to 12 per consultation), and these were linked to the problem/diagnosis for which they were ordered. A single pathology test could be linked to more than one problem/diagnosis. Independent variables Other variables in this analysis related to the trainee, patient, practice and the consultation. Trainee factors were age, gender, training term, RTP enrolled with, place of medical qualification (Australia or international) and fulltime/part-time status. Patient factors were age, gender, Aboriginal and Torres Strait Islander status, non-English-speaking background (NESB), new patient to the practice and new patient to the trainee. Downloaded from https://academic.oup.com/intqhc/article-abstract/27/6/528/2357515 by guest on 22 May 2020 Pathology tests play an important role in the diagnosis, monitoring and screening for disease in medical practice. The number of available tests has risen rapidly in recent decades, with the Royal College of Pathologists of Australasia manual now listing over 750 individual tests [1]. The use of laboratory tests is increasing in many countries [2]. In Australia, the number of Medicare-funded (governmentfunded) pathology tests increased by 54% from 2000–01 to 2007– 08 [3]. Over this period, pathology costs increased from A$ 1.2 billion to almost A$ 1.9 billion. General practitioners (GPs) are responsible for initiating 70% of these tests [3]. While much of this increase in testing may be appropriate, reflecting advances in technology and clinical knowledge, a growing body of evidence suggests that over-testing is a significant problem [4]. Australian data suggest that pathology testing is often not congruent with evidence-based consensus guidelines, with 25–75% of tests not supported by evidence or expert opinion [3]. Concerns have been raised about the inappropriate use of many common pathology tests, including full blood count (FBC) [5], liver function tests (LFTs) [6], vitamin B12 and folate [7], thyroid function tests (TFTs) [8], vitamin D [9] and prostate specific antigen [10]. Inappropriate test-ordering directly increases health care expenditure, as well as resulting in opportunity costs (of appropriate evidence-based health care). Additionally, unexpected abnormal results can be problematic for the clinician to interpret and manage. But, most importantly, over-testing can lead to patient harm. Overtesting is especially problematic in general practice, a setting where the pre-test probability of serious disease is generally low. This means that false-positive results are common, even in tests with reasonable specificity [11]. Both false-positive results and incidental findings can lead to a cascade of further tests [12], so-called investigation momentum [13]. This, in turn, leads to a greater risk of complications and patient harm, as well as the potential for significant patient anxiety. Lastly, over-testing may lead to over-diagnosis, the circumstance where people without symptoms are diagnosed with a disease that ultimately will not cause them to experience symptoms or early death [14]. This can lead to unnecessary treatment, adding to the risk of harm. In seeking to address the problem of over-testing, the epidemiology of test-ordering and other influences upon the test-ordering behaviours of doctors must be appreciated. A range of factors including doctor (demographics, knowledge, prior experience, personality, fear of litigation), patient (trust, anxiety), practice (billing practices) and systems (development of new tests) have been associated with test-ordering behaviours [2, 11, 15, 16]. It is particularly important to understand the factors influencing test-ordering practice in early-career doctors, as clinical behaviours established in training and early practice tend to persist [17]. Given that the majority of testing occurs in general practice, the test-ordering patterns of GP trainees are of particular interest. Consulting with patients is the core learning activity of general practice training in Australia. Trainees (registrars) learn by the ‘apprenticeship model’. While they have recourse to assistance and advice from experienced GP supervisors (trainers), they operate as independent practitioners (including for the purposes of ordering tests). Critical use of investigations is one of the core skills of the Royal Australian College of General Practitioners Common Training Outcomes [18]. However, there is evidence for a relative lack of training for Australian GP trainees in quality use of pathology [19]. The test-ordering behaviour of GP trainees has not previously been described, in Australia or internationally. We aimed to describe the rate of pathology test-ordering of GP trainees, the type of tests ordered and for which problems they were ordered. We also aimed to establish trainee, patient, practice and consultation associations of test-ordering behaviour. 530 who undertook their primary medical degree in Australia comprised 78.5% (95% CI: 75.6–81.1) of the total. The 856 trainees contributed data from 1832 trainee-terms, 108 759 encounters and 169 304 problems. Characteristics of participating trainees, practices and traineeterms are displayed in Table 1. Pathology ordering Trainees ordered pathology at a rate of 79.3 (95% CI: 78.8–79.8) tests per 100 encounters, and at least 1 pathology test was requested in 22.4% of all encounters. This equates to 50.9 (95% CI: 50.6–51.3) tests per 100 problems, with at least 1 pathology test requested in 17.2% of all problems managed. When the decision to order was made, the rate of test-ordering was 296.8 (95% CI: 294.8–298.8) tests per 100 problems, or approximately 3 tests per problem. Nature of tests ordered The top twelve most common tests ordered are listed in Table 2. The most common tests ordered were FBC (6.1 tests per 100 problems), electrolytes, urea and creatinine (4.6 tests per 100 problems) and LFTs (4.4 tests per 100 problems). Data analysis Descriptive analysis with consultation as the unit of analysis The mean (with SD) and median (with IQR) number of pathology tests per consultation were calculated. The proportion of consultations in which any pathology tests were ordered was calculated with 95% confidence intervals (CIs). Analyses with problem/diagnosis as the unit of analysis We used the individual problem/diagnosis as the unit of analysis for our other analyses as much of our relevant data are linked to the problem/diagnosis (including tests ordered). The mean (with SD) and median (with IQR) number of pathology tests per problem/diagnosis were calculated. The proportion of problems/diagnoses for which any pathology tests were ordered was calculated with 95% CIs. Univariate and multivariable analyses The outcome variable (number of tests ordered per problem/diagnosis) had 13 response levels (0–12 tests ordered), and 83% of these responses were zero. A zero-inflated negative binomial (ZINB) regression model was used to account for this heavily skewed distribution. Likelihood ratio chi-square tests are presented for each independent variable to assess the overall contribution to the outcome, as well as Wald Z-tests (and 95% confidence intervals) assessing the contribution of the levels within an independent variable on the outcome. Parameter estimation was within the generalized estimating equations (GEEs) framework to account for the repeated measures on trainees. All variables with a P-value of <0.20 and a relevant effect-size in the univariate (ZINB regression) analysis were included in the multivariable regression model. Analyses were programmed using STATA 13.1 and SAS V9.4. Ethics approval The ReCEnT project has approval from the University of Newcastle Human Research Ethics Committee, Reference H-2009-0323. Results Demographics of trainees, patients and practices A total of 856 individual trainees (response rate 95.2%) contributed data to the analysis. Overall, 65.7% [95% CI: 62.4–68.8] of the trainees were female, with a mean age of 32.5 years [SD 6.3]. Trainees Problems for which tests were ordered The top ten problems for which pathology was most frequently ordered are listed in Table 3, with the most common being for ‘check-up’ (18.6%) and urinary tract infection (5.9%). Associations of test-ordering The associations of pathology test-ordering for a problem/diagnosis are presented in Table 4. The multivariable analysis is presented in Table 5. Test-ordering was significantly associated in the adjusted model with the trainee having worked at the practice previously, and the patient being adult, male and new to both the trainee and to the practice. The only statistically significant practice-level association was rurality, with trainees in major city practices more likely to order tests. Consultation-related factors were the problem/diagnosis being new, imaging being ordered, referral made and follow-up being arranged. Trainees were significantly less likely to order tests for problems/diagnoses for which they had sought in-consultation information or advice but were more likely to generate learning goals for problem/diagnoses for which tests were ordered. Discussion We have found that trainees order pathology tests in about one in five encounters, with ‘check-up’ being the most common problem for which tests are ordered. As well, we identified a number of significant associations of test-ordering, relating to the trainee, practice and encounter. This is the first time the test-ordering behaviour of GP trainees has been described and one of very few studies describing test-ordering in general practice. Comparison with other literature and interpretation of findings Test-ordering prevalence and type of test Compared with established Australian GPs, trainees order pathology tests in more encounters (22.4% compared with 19.1%), in more problems managed (17.2% compared with 13.9%) and at a higher prevalence (50.9 compared with 31.0 per 100 problems managed) [25]. As well, once the decision to order is made, trainees order more pathology Downloaded from https://academic.oup.com/intqhc/article-abstract/27/6/528/2357515 by guest on 22 May 2020 Practice factors included rurality, socioeconomic status of the practice location, practice size (number of GPs) and if the practice routinely bulk-bills (that is, there is no financial cost to the patient for the consultation). Practice postcode was used to define the Australian Standard Geographical Classification-Remoteness Area (ASGC-RA) classification [21] (the degree of rurality) and the Socioeconomic Index for Areas (SEIFA) Index of Disadvantage [22] of the practice location. Consultation factors were duration of consultation, number of problems managed, whether the problem was new, whether imaging was ordered, whether a specialist referral was made and whether follow-up was ordered. Educational consultation factors included whether the trainee sought clinical information or assistance during the consultation (from their trainer, from a specialist or from electronic or hard-copy resources) or generated any learning goals. We coded problems managed/diagnoses according to the International Classification of Primary Care, second edition classification system (ICPC-2PLUS) [23]. Individual diseases/problems are categorized in ICPC-2PLUS to 17 system-based chapters (cardiovascular, psychological, etc.). We coded problems/diagnoses as chronic diseases via an existing classification system derived from ICPC-2 PLUS [24]. Morgan et al. 531 Test-ordering by trainees • Patient Safety Table 1 Participating trainee, trainee-term and practice characteristics Variable Trainee variables (n = 856) Trainee gender Qualified as a doctor in Australia Trainee age (years) Trainee-term or practice-term variables (n = 1832) Trainee training term Trainee works full-time Does the practice routinely bulk-billa Number of GPs working at the practice Rurality of practice SEIFAb Index (decile) of practice [22] n (%) [95% CIs] or mean (SD) Male Female No Yes Mean (SD) 294 (34.4) 562 (65.7) 182 (21.5) 664 (78.5) 32.5 (6.3) Term 1 Term 2 Term 3 Term 4 No Yes No Yes No Yes 1–5 6–10+ Major city Inner regional Outer regional, remote or very remote Mean (SD) 765 (42.8) [39.5–44.0] 538 (29.4) [27.3–31.5] 454 (24.8) [22.9–26.8] 75 (4.1) [3.3–5.1] 1321 (73.1) [71.0–75.1] 486 (26.9) [24.9–29.0] 399 (22.2) [20.4–24.2] 1395 (77.8) [75.8–79.6] 1502 (82.6) [80.8–84.2] 317 (17.4) [15.8–19.2] 604 (33.7) [31.6–35.9] 1187 (66.3) [64.1–68.4] 1060 (57.9) [55.6–60.1] 521 (28.4) [26.4–30.6] 251 (13.7) [12.2–15.4] 5.4 (2.9) [31.2–37.6] [62.4–68.8] [18.9–24.4] [75.6–81.1] a No financial cost to patient. SEIFA Relative Index of Disadvantage [22]. b Table 2 Top twelve pathology tests ordered by GP trainees Table 3 Top ten problems for which pathology was ordered Pathology Percentage of all pathology tests ordered Rate per 100 problems (95% CI) Problem/diagnosis FBC Electrolytes LFTs TFTs Lipids profile Glucose Urine microscopy and culture Ferritin/iron studies Cervical cytology CRP Vitamin D Vitamin B12 12.1 (11.8–12.3) 11.2 (11.0–11.4) 9.4 (9.2–9.6) 5.9 (5.8–6.1) 5.7 (5.5–5.8) 5.1 (4.9–5.2) 4.7 (4.5–4.8) 3.5 (3.4–3.7) 3.1 (3.0–3.2) 2.7 (2.6–2.8) 2.3 (2.2–2.4) 2.0 (1.9–2.1) 6.2 (6.0–6.3) 5.7 (5.6–5.8) 4.8 (4.7–4.9) 3.0 (2.9–3.1) 2.9 (2.8–3.0) 2.6 (2.5–2.7) 2.4 (2.3–2.5) 1.8 (1.7–1.9) 1.6 (1.5–1.6) 1.4 (1.3–1.4) 1.2 (1.1–1.2) 1.0 (1.0–1.1) Check-up Urinary tract infection Follow-up of abnormal test results Hypertension Tiredness Diabetes STI screen Hypercholesterolaemia Hypothyroidism Abdominal pain tests compared with the established GPs (296.1 compared with 232.2 per 100 problems). There are a number of possible explanations for the higher test-ordering prevalence in GP trainees. Trainees usually enter general practice after exclusive hospital-based experience, a setting with a much greater focus on investigation and diagnostic certainty. As well, GP trainees may be less tolerant of uncertainty, due to their relative inexperience and unfamiliarity with managing undifferentiated illness. A low tolerance to uncertainty has been described as a causative factor in higher rates of testing [26]. This may be further compounded by other known factors in driving test-ordering: the ‘need to reassure the patient’ and patient pressure to order tests [26, 27], both of which the trainee may be less-well-equipped to deal with than the experienced clinician. Furthermore, trainees may encounter a more acutely Percentage of all pathology tests ordered 18.6 (18–2–19.1) 5.9 (5.6–6.2) 3.4 (3.2–3.6) 2.8 (2.6–3.0) 2.7 (2.5–2.8) 2.6 (2.5–2.8) 2.4 (2.2–2.6) 2.4 (2.2–2.6) 2.3 (2.1–2.5) 2.3 (2.1–2.4) Percentage of these problems where pathology is ordered 40.1 (39.3–40.9) 71.9 (70.1–73.7) 54.7 (52.1–57.3) 14.3 (13.4–15.2) 96.5 (95.0–97.7) 37.5 (35.4–39.7) 80.2 (77.4–82.8) 30.4 (28.6–32.4) 90.2 (87.8–92.3) 45.0 (42.4–47.6) unwell patient population than established GPs due to structural issues in appointment allocation within practices. While test-ordering by trainees was at a higher prevalence than the established GPs overall, the difference in the rates of specific tests ordered varied considerably. Trainees ordered a similar proportion of lipid tests (2.9 compared with 2.6 per 100 problems), but close to twice as many urine cultures (2.4 compared with 1.3) and C-reactive protein (CRP) tests (1.4 compared with 0.7), and three times as many LFTs (4.8 compared with 1.5) [25].This is consistent with trainees seeing a more acutely unwell population with more undifferentiated presentations, and less chronic disease, than established GPs [28]. The problems for which tests were ordered were similar to that of established GPs [25]. 3.4% of all pathology tests were ordered as follow-up of previously abnormal tests, the third most common problem recorded. While some of this further testing may well be Downloaded from https://academic.oup.com/intqhc/article-abstract/27/6/528/2357515 by guest on 22 May 2020 Trainee worked at the practice previously Class 532 Morgan et al. Table 4 Characteristics associated with the number of pathology tests ordered (n = 131 423) Variable Patient age group Patient gender Patient/practice status Trainee gender Trainee FT or PTb Training term/post Worked at practice previously Qualified as doctor in Australia Practice sizec Practice routinely bulk-bills Rurality RTP New problem Chronic disease Sought help any source Imaging ordered Learning goals generated Referral ordered Follow-up ordered Trainee age SEIFA Indexd Consultation durationd Number of problemsd 0–14 15–34 35–64 65+ Male Female No Yes No Yes Existing patient of registrar New to trainee New to practice Male Female Part-time Full-time Term 1 Term 2 Term 3 Term 4 No Yes No Yes Small Large No Yes Major city Inner regional Outer regional remote 1 2 3 4 5 No Yes No Yes No Yes No Yes No Yes No Yes No Yes Mean (SD) Mean (SD) Mean (SD) Mean (SD) Pa Number of pathology tests ordered 0 path (n = 140 242) 1 test (n = 13 783) 2–3 tests (n = 5621) 4–5 tests (n = 4592) 6–12 tests (n = 5069) 21 025 (91.9%) 33 982 (79.7%) 55 176 (81.1%) 27 996 (84.5%) 53 108 (86.3%) 83 534 (80.9%) 130 987 (82.8%) 1768 (83.4%) 125 038 (82.9%) 8707 (81.5%) 61 231 (84.2%) 1294 (5.7%) 3857 (9.0%) 5814 (8.6%) 2564 (7.7%) 3541 (5.8%) 9840 (9.5%) 12 905 (8.2%) 145 (6.8%) 12 231 (8.1%) 898 (8.4%) 5911 (8.1%) 286 (1.3%) 1855 (4.4%) 2398 (3.5%) 988 (3.0%) 1706 (2.8%) 3754 (3.6%) 5243 (3.3%) 84 (4.0%) 4974 (3.3%) 393 (3.7%) 2399 (3.3%) 129 (0.6%) 1266 (3.0%) 2222 (3.3%) 886 (2.7%) 1677 (2.7%) 2781 (2.7%) 4286 (2.7%) 57 (2.7%) 4075 (2.7%) 307 (2.9%) 1639 (2.3%) 136 (0.6%) 1691 (4.0%) 2429 (3.6%) 710 (2.1%) 1537 (2.5%) 3383 (3.3%) 4764 (3.0%) 66 (3.1%) 4486 (3.0%) 376 (3.5%) 1546 (2.1%) 66 538 (82.2%) 8682 (78.2%) 49 596 (85.7%) 90 646 (81.3%) 30 525 (81.7%) 106 858 (83.2%) 59 686 (82.5%) 40 595 (83.1%) 34 432 (83.1%) 5529 (82.8%) 101 061 (82.7%) 37 346 (83.3%) 31 341 (83.3%) 107 091 (82.7%) 47 126 (82.9%) 90 053 (82.8%) 114 948 (82.8%) 24 423 (83.2%) 81 384 (83.0%) 39 395 (82.5%) 19 463 (82.8%) 46 267 (82.4%) 17 778 (81.9%) 15 978 (83.4%) 58 814 (83.4%) 1405 (79.7%) 60 263 (86.3%) 68 176 (79.8%) 108 403 (82.4%) 31 456 (84.2%) 12 0831 (83.4%) 19 411 (79.5%) 130 775 (83.5%) 9467 (74.3%) 120 787 (83.8%) 19 455 (77.1%) 122 142 (82.1%) 18 100 (88.5%) 84 324 (89.5%) 55 918 (74.4%) 32.9 (6.6) 5.37 (2.85) 18.1 (9.7) 1.95 (0.96) 6629 (8.2%) 875 (7.9%) 3760 (6.5%) 10 023 (9.0%) 3213 (8.6%) 10 262 (8.0%) 5939 (8.2%) 3943 (8.1%) 3393 (8.2%) 508 (7.6%) 9965 (8.2%) 3621 (8.1%) 3047 (8.1%) 10 567 (8.2%) 4530 (8.0%) 8917 (8.2%) 11 397 (8.2%) 2289 (7.8%) 7764 (7.9%) 4076 (8.5%) 1943 (8.3%) 4656 (8.3%) 1909 (8.8%) 1594 (8.3%) 5474 (7.8%) 150 (8.5%) 4483 (6.4%) 8175 (9.6%) 11 551 (8.8%) 2218 (5.9%) 11 543 (8.0%) 2240 (9.2%) 13 005 (8.3%) 778 (6.1%) 11 584 (8.0%) 2199 (8.7%) 13 003 (8.7%) 780 (3.8%) 5408 (5.7%) 8375 (11.2%) 32.7 (6.4) 5.35 (2.82) 20.4 (9.7) 2.02 (1.0) 2628 (3.3%) 458 (4.1%) 1508 (2.6%) 4113 (3.7%) 1375 (3.7%) 4141 (3.2%) 2575 (3.6%) 1584 (3.2%) 1229 (3.0%) 233 (3.5%) 4153 (3.4%) 1392 (3.1%) 1193 (3.2%) 4378 (3.4%) 1900 (3.3%) 3596 (3.3%) 4628 (3.3%) 965 (3.3%) 3225 (3.3%) 1604 (3.4%) 792 (3.4%) 1930 (3.4%) 745 (3.4%) 618 (3.2%) 2261 (3.2%) 67 (3.8%) 2154 (3.1%) 3080 (3.6%) 4192 (3.2%) 1426 (3.8%) 4517 (3.1%) 1104 (4.5%) 4837 (3.1%) 784 (6.2%) 4299 (3%) 1322 (5.2%) 5091 (3.4%) 530 (2.6%) 1800 (1.9%) 3821 (5.1%) 32.7 (6.5) 5.42 (2.83) 21.7 (10.0) 2.16 (1.03) 2393 (3.0%) 434 (3.9%) 1495 (2.6%) 3097 (2.8%) 1036 (2.8%) 3448 (2.7%) 2048 (2.8%) 1252 (2.6%) 1094 (2.6%) 198 (3.0%) 3334 (2.7%) 1191 (2.7%) 1071 (2.9%) 3470 (2.7%) 1580 (2.8%) 2918 (2.7%) 3823 (2.8%) 737 (2.5%) 2488 (2.5%) 1362 (2.9%) 742 (3.2%) 1612 (2.9%) 704 (3.2%) 515 (2.7%) 1691 (2.4%) 70 (4.0%) 1519 (2.2%) 2678 (3.1%) 3342 (2.5%) 1247 (3.3%) 3775 (2.6%) 817 (3.3%) 3820 (2.4%) 772 (6.1%) 3487 (2.4%) 1105 (4.4%) 4127 (2.8%) 465 (2.3%) 1336 (1.4%) 3256 (4.3%) 32.9 (6.6) 5.29 (2.77) 21.4 (9.8) 2.12 (0.99) 2762 (3.4%) 653 (5.9%) 1513 (2.6%) 3556 (3.2%) 1235 (3.3%) 3770 (2.9%) 2104 (2.9%) 1462 (3.0%) 1296 (3.1%) 207 (3.1%) 3736 (3.1%) 1261 (2.8%) 961 (2.6%) 4040 (3.1%) 1694 (3.0%) 3289 (3.0%) 4091 (3.0%) 940 (3.2%) 3175 (3.2%) 1326 (2.8%) 568 (2.4%) 1693 (3.0%) 578 (2.7%) 458 (2.4%) 2270 (3.2%) 70 (4.0%) 1393 (2.0%) 3321 (3.9%) 4043 (3.1%) 1017 (2.7%) 4212 (2.9%) 857 (3.5%) 4135 (2.6%) 934 (7.3%) 3927 (2.7%) 1142 (4.5%) 4492 (3.0%) 577 (2.8%) 1316 (1.4%) 3753 (5.0%) 32.6 (6.5) 5.56 (2.8) 22.8 (10.3) 2.06 (0.98) <0.0001 0.0059 0.57 0.029 <0.0001 0.34 0.40 0.35 0.085 0.0029 0.73 0.16 <0.0001 <0.0001 <0.0001 0.75 0.11 <0.0001 <0.0001 <0.0001 <0.0001 0.030 0.002 <0.0001 0.71 Care should be used when interpreting frequencies in this table. This analysis uses problems, not encounters, as the population unit. Reported frequencies at the problem level may not reflect the observed frequencies at the consultation level. a The P-values are results from a likelihood ratio chi-square test for the overall contribution of each variable to the uncategorized counts, estimated from the ZINB regression GEE model. b Trainee part-time status is defined as less than eight sessions per week. c Practices defined as small if less than six GPs were working in the practice. d SEIFA, age, duration and number of tests at consultation level. Downloaded from https://academic.oup.com/intqhc/article-abstract/27/6/528/2357515 by guest on 22 May 2020 Aboriginal and Torres Strait Islander NESB Class 533 Test-ordering by trainees • Patient Safety Table 5 Univariable and multivariable adjusted ZINB regression: associations of the number of pathology tests ordered per problem Class Univariable IRRa (95% CI) P Patient age group (referent: 0–14) 15–34 35–64 65+ Female New to Practice New to Registrar Yes 2.5 (2.2, 2.8) 2.3 (2.0, 2.6) 1.9 (1.6, 2.1) 0.94 (0.89, 0.98) 1.4 (1.3, 1.4) 1.8 (1.7, 1.9) 1.1 (1.04, 1.2) 0.995 (0.99, 0.999) 0.95 (0.89, 1.01) 0.84 (0.79, 0.9) 0.79 (0.72, 0.86) 1.015 (1.006, 1.025) 0.94 (0.85, 1.04) 0.87 (0.8, 0.95) 1.1 (1.04, 1.2) 1.1 (0.8, 1.5) 1.01 (1.007, 1.01) 1.2 (1.2, 1.3) 0.95 (0.9, 1.01) 1.7 (1.6, 1.8) 1.2 (1.1, 1.2) 1.3 (1.2, 1.4) 1.5 (1.4, 1.5) <0.0001 <0.0001 <0.0001 0.006 <0.0001 <0.0001 0.003 0.030 0.085 <0.0001 <0.0001 0.002 0.23 0.003 0.002 0.54 <0.0001 <0.0001 0.11 <0.0001 <0.0001 <0.0001 <0.0001 Patient gender Patient/practice status Referent: existing patient Qualified as doctor in Australia Trainee age Worked at practice previously Rurality Referent: major city SEIFA Index RTP (referent: 1) Consultation duration New problem Sought help any source Imaging ordered Learning goals generated Referral ordered Follow-up ordered Yes Inner Regional Outer Regional Remote 2 3 4 5 Yes Yes Yes Yes Yes Yes Adjusted IRR (95% CI) 2.4 (2.1, 2.7) 2.3 (2.1, 2.6) 2.03 (1.8, 2.3) 0.87 (0.83, 0.91) 1.4 (1.3, 1.4) 1.7 (1.5, 1.8) 1.05 (0.98, 1.1) 0.998 (0.99, 1.002) 1.07 (1.01, 1.1) 0.93 (0.86, 0.99) 0.89 (0.8, 0.98) 1.01 (0.999, 1.02) 0.93 (0.85, 1.02) 0.94 (0.86, 1.04) 1.04 (0.97, 1.1) 1.1 (0.81, 1.6) 1.002 (0.999, 1.005) 1.1 (1.06, 1.2) 0.83 (0.78, 0.88) 1.6 (1.5, 1.6) 1.2 (1.2, 1.3) 1.4 (1.3, 1.5) 1.4 (1.4, 1.5) P <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.160 0.35 0.024 0.033 0.02 0.094 0.14 0.21 0.28 0.44 0.18 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 a IRRs are interpreted as the multiplicative increase/decrease in rates for a unit change in predictor variable. appropriate, this also probably illustrates ‘investigation momentum’ that can occur as a result of an initial abnormal or equivocal test [13]. Associations of test-ordering A number of ‘physician-related factors’ that affect test-ordering behaviour have been described [15], including age and gender. The only significant trainee-associated factor in our study was the trainee having worked in the practice previously (and this association was of a modest effect-size). The association of increased test-ordering in adult patients compared with children reflects the increased screening, investigation and monitoring of disease in an adult population. We found the rate of test-ordering to be significantly increased when the patient was new to the trainee and/or the practice. There is evidence that test-ordering can be driven by the imperative to ‘strike while the iron’s hot’, thoroughly screening the patient while they are present [11]. This may also explain our finding of increased test-ordering in male patients, who are known to present less frequently than female patients [25]. The rate of test-ordering was significantly increased in a major city compared with inner or outer regional location. This may reflect access to pathology providers being greater in the major cities than those in inner regional areas (which include rural and semi-rural areas), or issues related to health care access in rural areas [29]. Consultation-related factors associated with more test-ordering were the problem being new, imaging being ordered, referral made and follow-up arranged. This is not unexpected and is consistent with tests being ordered at the first presentation of a complaint and for more complex problems. The greater complexity of the problem is also suggested by the positive association with the generation of learning goals, meaning that trainees find such problems challenging. However, test-ordering was negatively associated with in-consultation information-seeking. This is a very important finding and may suggest that trainees use test-ordering as a diagnostic strategy for complex problems in preference to seeking information from sources like clinical guidelines. An alternative (but not mutually exclusive) explanation is that when trainees consult their trainer (with their greater experience, higher tolerance of uncertainty and greater knowledge) or other evidence-based information sources, this leads to less test-ordering. Strengths and limitations Our study has a number of strengths. The trainee participants had similar demographics (age, gender and IMG status) to the national GP trainee cohort [30]. As well, we conducted this study in five regional training providers across five Australian states, making the findings broadly generalizable to Australian general practice training. Our participant response rate was 95.2%, which is singularly high for a study recruiting GPs [31]. We used a paper-based collection system. Though electronic formats have previously been used for tracking patient encounters, there is little evidence that this format improves accuracy or completeness of data collection [32]. Due to the large and diverse variety of software packages in Australian general practices, efficient extraction of routinely collected electronic data is currently impractical [33]. Furthermore, routinely recorded data in Australian general practice are likely to be of relatively poor quality compared with deliberately collected records, especially for problems managed. We coded our data using ICPC2-plus, the international standard for classifying primary care data. The validity of this system has previously been demonstrated [34]. We compared our data with that of established Australian GPs from the same time period (2013–14), incorporating a similar methodology to ours [25]. Limitations of this study include this being a ‘broad brush’ analysis with the appropriateness of individual test-ordering decisions being beyond the resolution of our data. Downloaded from https://academic.oup.com/intqhc/article-abstract/27/6/528/2357515 by guest on 22 May 2020 Variable 534 A further limitation is that, from the patient point of view, our data contain a ‘snapshot’ of a single consultation with problems/diagnoses (and any tests ordered for them) recorded only for those problems/ diagnoses addressed in the consultation. We have no data on other diagnoses not addressed. Thus, we have no measure of comorbidities or overall patient health status, which may potentially be associated with trainees’ test-ordering behaviour. Furthermore, though our study is a longitudinal study of registrars, this analysis of test-ordering is cross-sectional and so we have established associations of test-ordering rather than causal relationships. Implications for policy Implications for further research Particular aspects of trainee test-ordering demand further analysis, including qualitative exploration for reasons for greater test-ordering, the appropriateness of specific tests ordered for specific problems and the effect of educational interventions on rational test-ordering. The ReCEnT study methodology, as a cohort study, will also allow examination of changes in trainee test-ordering over the course of training. Funding This work was supported by the Department of Health, Commonweath of Australia [grant number D14/17024]. References 1. Royal College of Pathologists Australasia (RCPA). RCPA Manual. Sydney: RCPA, 2009. http://www.rcpamanual.edu.au/ (7 September 2013, date last accessed). 2. Verstappen WH, ter Riet G, Dubois WI et al. Variation in test ordering behaviour of GPs: professional or context-related factors? Fam Pract 2004;21:387–95. 3. Bayram C, Britt H, Miller G et al. Evidence-practice gap in GP pathology test ordering: a comparison of BEACH pathology data and recommended testing. Final report. Sydney: University of Sydney, Family Medicine Research Centre, 2009. 4. McGregor MJ, Martin D. Testing 1,2,3. Is overtesting undermining patient and system health? Can Fam Physician 2012;58:1191–3. 5. Bubner T, Laurence C. Quality use of pathology program. Enhancing the quality use of pathology for GP registrars and international medical graduates – assessing the need. Final Report. Adelaide, Australia: University of Adelaide, Discipline of General Practice, 2009. 6. Lilford RJ, Bentham LM, Armstrong MA et al. What is the best strategy for investigating abnormal liver function tests in primary care? Implications from a prospective study. BMJ Open 2013;3:e003099. 7. Willis CD, Metz MP, Hiller JE et al. Vitamin B12 and folate tests: the ongoing need to determine appropriate use and public funding. Med J Aust 2013;198:586–8. 8. Roti E, Gardini E, Magotti MG et al. Are thyroid function tests too frequently and inappropriately requested? J Endocrinol Invest 1999;22:184–90. 9. Bilinski KL, Boyages SC. The rising cost of vitamin D testing in Australia: time to establish guidelines for testing. Med J Aust 2012;197:90. 10. Del Mar CB, Glasziou PP, Hirst GH et al. Should we screen for prostate cancer? A re-examination of the evidence. Med J Aust 2013;198:525–7. 11. Axt-Adam P, van der Wouden JC, van der Does E. Influencing behaviour of physicians ordering laboratory test: a literature study. Med Care 1993;31:784–94. 12. Deyo RA. Cascade effects of medical technology. Annu Rev Public Health 2002;23:23–44. Epub 25 October 2001. 13. Sah S, Elias P, Ariely D. Investigation momentum: the relentless pursuit to resolve uncertainty. JAMA Intern Med 2013;173:932–3. 14. Moynihan R, Doust J, Henry D. Preventing overdiagnosis: how to stop harming the healthy. BMJ 2012;344:e3502. 15. Sood R, Sood A, Ghosh AK. Non-evidence-based variables affecting physicians’ test-ordering tendencies: a systematic review. Neth J Med 2007;65:167–77. 16. Van Bokhoven MA, Pleunis-van Empel MC, Koch H et al. Why do patients want to have their blood tested? A qualitative study of patient expectations in general practice. BMC Fam Pract 2006;7:75. 17. Bjornsdottir I, Kristinsson KG, Hansen EH. Diagnosing infections: a qualitative view on prescription decisions in general practice over time. Pharm World Sci 2010;32:805–14. 18. Royal Australian College of General Practitioners (RACGP). Common Training Outcomes. Melbourne: RACGP, 2011. http://curriculum.racgp. org.au/statements/common-training-outcomes.pdf (2 February 2015, date last accessed). 19. Bubner T, Laurence C, Tirimacco R. Assessing pathology training needs; results from a survey of general practice registrars. Aust Fam Physician 2012;41:721–4. 20. Morgan S, Magin PJ, Henderson KM et al. Study protocol: the registrar clinical encounters in training (ReCEnT) study. BMC Fam Pract 2012;13:50. 21. Australian Bureau of Statistics. ASGC remoteness classification: purpose and use. Census paper No. 03/01. Canberra: Commonwealth of Australia, 2003. http://www.abs.gov.au/Websitedbs/D3110122.NSF/ 4a255eef008309e44a255eef00061e57/f9c96fb635cce780ca256d420005 Downloaded from https://academic.oup.com/intqhc/article-abstract/27/6/528/2357515 by guest on 22 May 2020 Vocational training is arguably the most critical period in the development of future patterns of clinical practice for the GP. This includes the development of test-ordering behaviour, with the context being that over-testing is linked to increased health care expenditure and to patient harm [4]. Our findings are therefore of considerable importance in providing evidence on which to inform educational policy to enhance quality and safety in general practice training. The greater prevalence of test-ordering that we have demonstrated compared with established GPs (in whom it is regarded that over-testing is also problematic) suggests a focus on teaching of rational test-ordering strategies, management of uncertainty and evidence-based practice is indicated. The associations of test-ordering that we have established will inform specific educational approaches within these areas of focus. A number of general and specific strategies for teaching and learning rational test-ordering have been described [35]. These include explicitly practising patient-centred care [36] and shared decisionmaking [37], and undertaking a thorough clinical assessment [38]. This is in the context of evidence that requesting diagnostic tests for patients with a low risk of serious illness generally does little to reassure patients or reduce anxiety [39]. Dealing with uncertainty is an essential skill for GPs. One of the most important drivers for ‘superfluous’ test-ordering in the context of an unexplained complaint is diagnostic uncertainty [26]. A number of practical and teachable strategies have been described for this purpose, including access to and use of guidelines and watchful waiting [40, 41]. Test-ordering should be guided by evidence-based clinical guidelines where they exist, and trainees should be directed to these guidelines and educated in their use, e.g. investigation of fatigue [42], and preventive health and screening [18]. We found that in-consultation use of information sources was associated with less test-ordering. This supports specific training in guideline use and trainer rolemodelling [43]. Our findings also reinforce the need for review of general practice policy and procedures for test-ordering. There is evidence that changing the format of laboratory test order forms can change the test-ordering behaviour of primary care physicians [44], and practice test-ordering software can be modified to avoid the option of batch testing (ordering of multiple tests for a clinical syndrome, for example ‘antenatal screen’, ‘diabetes check’). Morgan et al. Test-ordering by trainees • Patient Safety 22. 23. 24. 25. 26. 28. 29. 30. 31. 32. Denton GD, DeMott C, Pangaro LN et al. Narrative review: use of studentgenerated logbooks in undergraduate medical education. Teach Learn Med 2006;18:153–64. 33. Britt HC, Miller GC. The Bettering the evaluation and care of health (BEACH) program: where to from here? Med J Aust 2013;198:125–6. 34. Britt H, Meza RA, Del Mar C. Methodology of morbidity and treatment data collection in general practice in Australia: a comparison of two methods. Fam Pract 1996;13:462–7. 35. Morgan S, Coleman J. We live in testing times—teaching rational test ordering in general practice. Aust Fam Physician 2014;43:273–6. 36. Epstein RM, Franks P, Shields GC et al. Patient-centered communication and diagnostic testing. Ann Fam Med 2005;20:415–21. 37. Charles C, Gafni A, Whelan T. Shared decision-making in the medical encounter: what does it mean? (or it takes at least two to tango). Soc Sci Med 1997;44:681–92. 38. Palfrey S. Daring to practice low cost medicine in a high tech era. N Engl J Med 2011;364:e21. 39. Rolfe A, Burton C. Reassurance after diagnostic testing with a low pretest probability of serious disease: systematic review and meta-analysis. JAMA Intern Med 2013;173:407–16. 40. O’Riordan M, Dahinden A, Akturk Z et al. Dealing with uncertainty in general practice: an essential skill for the general practitioner. Qual Prim Care 2011;19:175–81. 41. Heneghan C, Glasziou P, Thompson M et al. Diagnostic strategies used in primary care. BMJ 2009;338:1003–6. 42. Moulds R EB, van Driel M, Greenberg P et al. Fatigue: diagnostic approach in primary care. Revised Oct 2011. In: eTG complete [Internet]. Therapeutic Guidelines Limited, 2011. 43. Passi V, Johnson S, Peile E et al. Doctor role modelling in medical education. BEME Guide No 27. Medical Teacher 2013;35:e1422–36. 44. Shalev V, Chodick G, Heymann AD. Format change of a laboratory test order form affects physician behavior. Int J Med Inform 2009;78: 639–44. Epub 2009 Jun 24. Downloaded from https://academic.oup.com/intqhc/article-abstract/27/6/528/2357515 by guest on 22 May 2020 27. dc02/$FILE/Remoteness_Paper_text_final.pdf (2 February 2015, date last accessed). Australian Bureau of Statistics. 2039.0 - Information paper: an introduction to Socio-Economic Indexes of Areas (SEIFA), 2006. Canberra: Commonwealth of Australia. http://www.abs.gov.au/ausstats/abs@.nsf/mf/2039.0/ (17 July 2013, date last accessed). Lamberts H, Woods M. International Classification of Primary Care. Oxford: Oxford University Press, 1987. O’Halloran J, Miller GC, Britt H. Defining chronic conditions for primary care with ICPC-2. Fam Pract 2004;21:381–6. Britt H, Miller GC, Henderson J et al. General Practice Activity in Australia 2013–14. General Practice series no. 36. Sydney: Sydney University Press, 2014. van der Weijden T, van Velsen M, Dinant GJ et al. Unexplained complaints in general practice: prevalence, patients expectations and professionals test ordering behaviour. Med Decis Making 2003;23:226–31. Ring A, Dowrick C, Humphris G et al. Do patients with unexplained physical symptoms pressurise general practitioners for somatic treatment? A qualitative study. BMJ 2004;328:1057. Magin P, Morgan S, Henderson K et al. Family medicine trainees’ clinical experience of chronic disease during training: a cross-sectional analysis from the registrars’ clinical encounters training study. BMC Med Educ 2014;14:260. Goins RT, Williams KA, Carter MW et al. Perceived barriers to health care access among rural older adults: a qualitative study. J Rural Health 2005;21:206–13. General Practice Education and Training. General practice education and training limited annual report to 30 June 2012. General Practice Education and Training Limited, 2012. http://www.gpet.com.au/About-Us/AnnualReport (5 March 2013, date last accessed). Bonevski B, Magin P, Horton G et al. Response rates in GP surveystrialling two recruitment strategies. Aust Fam Physician 2011;40: 427–30. 535