The Ottawa SAH search algorithms: protocol for a multi- centre validation study of primary subarachnoid hemorrhage prediction models using health administrative data (the SAHepi prediction study protocol).

BACKGROUND: Conducting prospective epidemiological studies of hospitalized patients with rare diseases like primary subarachnoid hemorrhage (pSAH) are difficult due to time and budgetary constraints. Routinely collected administrative data could remove these barriers. We derived and validated 3 algorithms to identify hospitalized patients with a high probability of pSAH using administrative data. We aim to externally validate their performance in four hospitals across Canada. METHODS: Eligible patients include those ≥18 years of age admitted to these centres from January 1, 2012 to December 31, 2013. We will include patients whose discharge abstracts contain predictive variables identified in the models (ICD-10-CA diagnostic codes I60** (subarachnoid hemorrhage), I61** (intracranial hemorrhage), 162** (other nontrauma intracranial hemorrhage), I67** (other cerebrovascular disease), S06** (intracranial injury), G97 (other postprocedural nervous system disorder) and CCI procedural codes 1JW51 (occlusion of intracranial vessels), 1JE51 (carotid artery inclusion), 3JW10 (intracranial vessel imaging), 3FY20 (CT scan (soft tissue of neck)), and 3OT20 (CT scan (abdominal cavity)). The algorithms will be applied to each patient and the diagnosis confirmed via chart review. We will assess each model's sensitivity, specificity, negative and positive predictive value across the sites. DISCUSSION: Validating the Ottawa SAH Prediction Algorithms will provide a way to accurately identify large SAH cohorts, thereby furthering research and altering care.

Impact of immunosuppressive medication on the risk of renal allograft failure due to recurrent glomerulonephritis.

Recurrent glomerulonephritis is a major problem in kidney transplantation but the role of immunosuppression in preventing this complication is not known. We used data from the United States Renal Data System to examine the effect of immunosuppressive medication on allograft failure due to recurrent glomerulonephritis for 41,272 patients undergoing kidney transplantation from 1990 to 2003. Ten-year incidence of graft loss due to recurrent glomerulonephritis was 2.6% (95% confidence interval [CI]: 2.3-2.8%). After adjusting for important covariates, the use of cyclosporine, tacrolimus, azathioprine, mycophenolate mofetil, sirolimus or prednisone was not associated with graft failure due to recurrent glomerulonephritis. There was no difference between cyclosporine and tacrolimus or between azathioprine and mycophenolate mofetil in the risk of graft failure due to recurrent glomerulonephritis. However, any change in immunosuppression during follow-up was independently associated with graft loss due to recurrence (adjusted hazard ratio 1.30, 95% CI: 1.06-1.58, p = 0.01). In patients with a pretransplant diagnosis of glomerulonephritis, the risk of graft loss due to recurrence was not associated with any specific immunosuppressive medication. The selection of immunosuppression for kidney transplant recipients should not be made with the goal of reducing graft failure due to recurrent glomerulonephritis.

A population-based comparison of joint survival of hemiarthroplasty versus total shoulder arthroplasty in osteoarthritis and rheumatoid arthritis.

AIMS: Few studies have compared survivorship of total shoulder arthroplasty (TSA) with hemiarthroplasty (HA). This observational study compared survivorship of TSA with HA while controlling for important covariables and accounting for death as a competing risk. PATIENTS AND METHODS: All patients who underwent shoulder arthroplasty in Ontario, Canada between April 2002 and March 2012 were identified using population-based health administrative data. We used the Fine-Gray sub-distribution hazard model to measure the association of arthroplasty type with time to revision surgery (accounting for death as a competing risk) controlling for age, gender, Charlson Comorbidity Index, income quintile, diagnosis, and surgeon factors. RESULTS: During the study period, 5777 patients underwent shoulder arthroplasty (4079 TSA, 70.6%; 1698 HA, 29.4%), 321 (5.6%) underwent revision, and 1090 (18.9%) died. TSA patients were older (TSA mean age 68.4 years (sd 10.2) vs HA mean age 66.5 years (sd 12.7); p = 0.001). The proportion of female patients was slightly lower in the TSA group (58.0% vs 58.4%). The adjusted association between surgery type and time to shoulder revision interacted significantly with patient age. Compared with TSA patients, revision was more common in the HA group (adjusted-health ratio (HR) 1.214, 95% confidence interval (CI) 0.96 to 1.53) but this did not reach statistical significance. CONCLUSION: Although there was a trend towards higher revision risk in patients undergoing HA, we found no statistically significant difference in survivorship between patients undergoing TSA or HA. Cite this article: Bone Joint J 2019;101-B:454-460.

Systematic radiologic detection of kidney stones in Canadian children: a new era of asymptomatic stones?

INTRODUCTION: Asymptomatic, or incidental, pediatric kidney stones detected on abdominal imaging pose a clinical challenge as their significance and expected outcome are not well described. OBJECTIVE: Our primary objective was to estimate the incidence of nephrolithiasis in pediatric patients undergoing abdominal ultrasound (US) or computerized tomography (CT), for all indications, in a pediatric tertiary care hospital. Our secondary objective was to determine the clinical outcome of these radiographically detected stones. METHODS: All abdominal US or CT radiology reports for patients younger than 18 years between January 1, 2011, and December 31, 2016, were retrieved. Reports were automatically flagged using textual analysis if they contained one of the 32 keywords potentially indicating nephrolithiasis. Flagged reports, as well as 10% of unflagged reports, were reviewed to confirm the radiologist impression of presence or absence of stones. Patient and stone-related clinical data were extracted. RESULTS: Two thousand four hundred forty-nine (5%) of 53 235 imaging reports cited at least one of the keywords. Manual review of flagged reports identified 498 studies having a radiologist impression of stones (244 unique patients). Stone incidence in children undergoing abdominal imaging more than doubled between 2011 and 2016. Medical record review found that 140 patients (57%) were symptomatic, while the other 104 patients (43%) were asymptomatic. Spontaneous resolution was greater (57% vs 34%) in asymptomatic patients compared to symptomatic patients. Asymptomatic patients were younger with a median age of 0.6 compared to 12.3 years. Asymptomatic and symptomatic stones were followed up for similar lengths of time (2.3 vs 2.0 years, p > 0.05) and had a similar number of follow-up USs (3.9 vs 4.6 studies, p > 0.05). CONCLUSIONS: The incidence of radiologically identified stones in children undergoing abdominal imaging at our center increased over time. Asymptomatic stones follow a benign course with minimal need for intervention; however, they were detected almost as frequently as symptomatic stones. Asymptomatic stones may represent increased healthcare resource utilization due to similar follow-up compared to symptomatic stones. Further research regarding the optimal imaging intervals and long-term follow-up for asymptomatic stones is warranted.

Real-time polymerase chain reaction detection of methicillin-resistant Staphylococcus aureus: impact on nosocomial transmission and costs.

OBJECTIVES: To assess the impact of real-time polymerase chain reaction (PCR) detection of methicillin-resistant Staphylococcus aureus (MRSA) on nosocomial transmission and costs. DESIGN: Monthly MRSA detection rates were measured from April 1, 2000, through December 31, 2005. Time series analysis was used to identify changes in MRSA detection rates, and decision analysis was used to compare the costs of detection by PCR and by culture.Setting. A 1,200-bed, tertiary care hospital in Canada. PATIENTS: Admitted patients at high risk for MRSA colonization. MRSA detection using culture-based screening was compared with a commercial PCR assay. RESULTS: The mean monthly incidence of nosocomial MRSA colonization or infection was 0.37 cases per 1,000 patient-days. The time-series model indicated an insignificant decrease of 0.14 cases per 1,000 patient-days per month (95% confidence interval, -0.18 to 0.46) after the introduction of PCR detection (P=.39). The mean interval from a reported positive result until contact precautions were initiated decreased from 3.8 to 1.6 days (P<.001). However, the cost of MRSA control increased from Can$605,034 to Can$771,609. Of 290 PCR-positive patients, 120 (41.4%) were placed under contact precautions unnecessarily because of low specificity of the PCR assay used in the study; these patients contributed 37% of the increased cost. The modeling study predicted that the cost per patient would be higher with detection by PCR (Can$96) than by culture (Can$67). CONCLUSION: Detection of MRSA by the PCR assay evaluated in this study was more costly than detection by culture for reducing MRSA transmission in our hospital. The cost benefit of screening by PCR varies according to incidences of MRSA colonization and infection, the predictive values of the assay used, and rates of compliance with infection control measures.

Influence of antibiotics and case exposure on hospital-acquired Clostridium difficile infection independent of illness severity.

BACKGROUND: Previous studies of the association between antibiotic exposure and risk of hospital-acquired Clostridium difficile-associated infection (CDI) have not fully accounted for patient severity of illness, and competing risks. AIM: To determine the potential effects of interventions on hospital-acquired CDI risk. METHODS: All adults admitted to a teaching hospital between 2004 and 2014 for more than two days were included. Exposures to all antibiotics and cases of CDI were determined. Patients were followed until discharge from hospital, death, or acquisition of hospital-acquired CDI (defined as positive toxin assay in unformed stool >2 days following admission). Multivariable proportional hazards competing-risks modelling with time-dependent covariates was used, accounting for patient severity of illness using the Escobar model. FINDINGS: In all, 208,104 patients were studied. Hospital-acquired CDI risk was 0.46 events per 1000 patient-days, decreasing significantly during the study period. Compared to the 5th percentile hospital death risk (0.02%), patients with a 50% risk of death in hospital had an adjusted hazard ratio (aHR) of hospital-acquired CDI of 5.5. Exposure to some antibiotics significantly increased hospital-acquired CDI risk, being highest for carbapenems (aHR: 1.47 after one week of continuous exposure) and intravenous vancomycin (aHR: 1.53). On the ward, sharing a room with other patients newly diagnosed with CDI significantly increased the risk of subsequent disease (aHR: 1.16 on CDI diagnosis day). CONCLUSION: The primary determinant of hospital-acquired CDI was patient severity of illness. Exposure to both antibiotics and other patients with CDI significantly increased the subsequent risk of hospital-acquired CDI but this risk was small relative to patient severity of illness.

Accuracy of coding for possible warfarin complications in hospital discharge abstracts.

BACKGROUND: Hospital discharge abstracts could be used to identify complications of warfarin if coding for bleeding and thromboembolic events are accurate. OBJECTIVES: To measure the accuracy of International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9CM) codes for bleeding and thromboembolic diagnoses. SETTING: University affiliated, tertiary care hospital in Ottawa, Canada. PATIENTS: A random sample of patients discharged between September 1999 and September 2000 with an ICD-9-CM code indicating a bleeding or thromboembolic diagnosis. METHODS: Gold-standard coding was determined by a trained chart abstractor using explicit standard diagnostic criteria for bleeding, major bleeding, and acute thromboembolism. The abstractor was blinded to the original coding. We calculated the sensitivity, specificity, positive, and negative predictive values of the original ICD-9CM codes for bleeding or thromboembolism diagnoses. RESULTS: We reviewed 616 medical records. 361 patients (59%) had a code indicating a bleeding diagnosis, 291 patients (47%) had a code indicating a thromboembolic diagnosis and 36 patients (6%) had a code indicating both. According to the gold standard criteria, 352 patients experienced bleeding, 333 experienced major bleeding, and 188 experienced an acute thromboembolism. For bleeding, the ICD-9CM codes had the following sensitivity, specificity, positive and negative predictive values [95% CI]: 93% [90-96], 88% [83-91], 91% [88-94], and 91% [87-94], respectively. For major bleeding, the ICD-9CM codes had the following sensitivity, specificity, positive and negative predictive values: 94% [91-96], 83% [78-87], 87% [83-90], and 92% [88-95], respectively. For thromboembolism, the ICD-9CM codes had the following sensitivity, specificity, positive and negative predictive values: 97% [94-99], 74% [70-79], 62% [57-68], and 98% [96-99], respectively. By selecting a sub-group of ICD-9CM codes for thromboembolism, the positive predictive value increased to 87%. CONCLUSION: In our centre, the discharge abstract could be used to identify and exclude patients hospitalized with a major bleed or thromboembolism. If coding quality for bleeding is similar in other hospitals, these ICD-9-CM diagnostic codes could be used to study population-based warfarin-associated hemorrhagic complications using administrative databases.

The impact of frailty on outcomes and healthcare resource usage after total joint arthroplasty: a population-based cohort study.

AIMS: Total joint arthroplasty (TJA) is commonly performed in elderly patients. Frailty, an aggregate expression of vulnerability, becomes increasingly common with advanced age, and independently predicts adverse outcomes and the use of resources after a variety of non-cardiac surgical procedures. Our aim was to assess the impact of frailty on outcomes after TJA. PATIENTS AND METHODS: We analysed the impact of pre-operative frailty on death and the use of resources after elective TJA in a population-based cohort study using linked administrative data from Ontario, Canada. RESULTS: Of 125 163 patients aged > 65 years having elective TJA, 3023 (2.4%) were frail according to the Johns Hopkins ACG frailty-defining diagnoses indicator. One year follow-up was complete for all patients. Frail patients had a higher adjusted one year risk of mortality (hazard ratio 3.03, 95% confidence interval (CI) 2.62 to 3.51), a higher rate of admission to intensive care (odds ratio (OR) 2.52, 95% CI 2.21 to 2.89), increased length of stay (incidence rate ratio 1.62, 95% CI 1.59 to 1.65), a higher rate of discharge to institutional care (OR 2.09, 95% CI 1.93 to 2.25), a higher rate of re-admission (OR 1.33, 95% CI 1.07 to 1.66) and increased costs at 30, 90 and 365 days post-operatively. Frailty affected outcomes after total hip arthroplasty more than after total knee arthroplasty. TAKE HOME MESSAGE: Frailty is an important risk factor for death after elective TJA, and increases post-operative resource utilisation across many metrics. Processes to optimise the outcomes and efficiency of TJA in frail patients are needed. Cite this article: Bone Joint J 2016;98-B:799-805.

Derivation of a clinical decision rule for the discontinuation of in-hospital cardiac arrest resuscitations.

BACKGROUND: Most patients undergoing in-hospital cardiac resuscitation will not survive to hospital discharge. OBJECTIVE: To derive a decision rule permitting the discontinuation of futile resuscitation attempts by identifying patients with no chance of surviving to hospital discharge. PATIENTS AND METHODS: Patient, arrest, and outcome data for 1077 adult patients undergoing in-hospital cardiac resuscitation was retrieved from 2 randomized clinical trials involving 5 teaching hospitals at 2 university centers. Recursive partitioning was used to identify a decision rule using variables significantly associated with death in hospital. RESULTS: One hundred three patients (9.6%) survived to hospital discharge. Death in hospital was significantly more likely if patients were older than 75 years (P<.001), the arrest was unwitnessed (P = .003), the resuscitation lasted longer than 10 minutes (P<.001), and the initial cardiac rhythm was not ventricular tachycardia or fibrillation (P<.001). All patients died if there was no pulse 10 minutes after the start of cardiopulmonary resuscitation, the initial cardiac rhythm was not ventricular tachycardia or fibrillation, and the arrest was not witnessed. As a resuscitation rule, these parameters identified all patients who survived to hospital discharge (sensitivity, 100%; 95% confidence interval, 97.1%-100%). Resuscitation could have been discontinued for 119 (12.1%) of 974 patients who did not survive, thereby avoiding 47 days of postresuscitative care. CONCLUSIONS: A practical and highly sensitive decision rule has been derived that identifies patients with no chance of surviving in-hospital cardiac arrest. Prospective validation of the rule is necessary before it can be used clinically.

Risk of intracerebral hemorrhage in patients with arterial versus cardiac origin of cerebral ischemia on aspirin or placebo: analysis of individual patient data from 9 trials.

BACKGROUND AND PURPOSE: Patients who are anticoagulated after cerebral ischemia have a 19-fold-higher risk of intracerebral hemorrhage (ICH) if they had an arterial rather than a cardiac source. To determine whether this excess risk of ICH was due to the underlying disease (cerebral ischemia of arterial versus cardiac origin) or whether it depended on the antithrombotic regimen, we studied the risk of ICH in arterial versus cardiac origin of cerebral ischemia in patients who received aspirin or no antithrombotic drugs. METHODS: Individual patient data of patients who received aspirin or placebo after cerebral ischemia were obtained from 9 clinical trials. Presence of atrial fibrillation was considered evidence of a cardiac source. Otherwise, events were considered of arterial origin. Cox proportional-hazards modeling was used for univariate and multivariate analyses. RESULTS: Fifty-four ICHs occurred in 16 625 patient-years in the aspirin-treated patients, and 7 ICHs occurred in 4317 patient-years in those on placebo. After multivariate adjustment for age, sex, current smoking, history of hypertension and diabetes, and aspirin dose (aspirin-treated patients only), the hazard ratio for ICH in patients with an arterial versus a cardiac source was 0.74 (95% confidence interval, 0.30 to 1.82) for aspirin-treated patients and 4.34 (95% confidence interval, 0.35 to 54) for placebo-randomized patients. CONCLUSIONS: Our findings do not confirm the previous finding of an excess risk of ICH in patients with cerebral ischemia of arterial origin. Therefore, it seems that having cerebral ischemia of arterial origin by itself is not associated with an increased risk of ICH, but only in combination with high-intensity anticoagulation.

Surveying physicians to determine the minimal important difference: implications for sample-size calculation.

The minimal important difference (MID) is the smallest benefit of treatment that would result in clinicians recommending it to their patients. The MID is necessary to calculate sample size for randomized clinical trials, but its chosen value is often arbitrary. This study set out to determine the practicability of surveying physicians to elicit the MID for clinical trial sample-size calculation. Using a mail survey, we elicited the MID of different physician specialties (family medicine, internal medicine, vascular surgery) for using propranolol to slow abdominal aortic aneurysm (AAA) growth assuming that propranolol was efficacious in this condition. We used different outcome measures (growth rate or proportion of patients requiring surgery) and different methods of data presentation for the proportion of patients requiring surgery (absolute risk reduction or number needed to treat). The MID varied significantly by physician specialty, experience with AAA and propranolol, and the method used to elicit the MID. Consequently, sample-size calculations using these various MIDs varied from 116 to 3015. Future attempts to elicit the MID need to consider carefully who is surveyed, how data are presented, and how opinions are elicited.

Surveillance mammography after treatment of primary breast cancer: a systematic review.

As the prevalence of diagnosed breast cancer increases, it is important to define how best to provide long-term follow-up. Whereas many aspects of follow-up remain controversial, guidelines recommend surveillance mammograms as the only investigation to be performed routinely. We conducted a systematic review of the literature to elucidate the effect of routine surveillance mammograms on detecting ipsilateral recurrence (IR) and contralateral breast cancers (CBC). The systematic review yielded 15 articles. All were observational studies and ranked as level II-2 or III evidence. There were no randomized controlled trials identified. Most of the ten studies on detection of IR did not report on outcomes after detection. When reported, most studies found that the method of detection of IR did not influence overall survival or disease-free survival. Two of the nine studies on detection of CBC found that the CBC was detected at an earlier stage than the initial breast cancer, but did not report on long-term outcomes. This systematic review highlights the need for further research to help better define the optimum surveillance mammography regimen.

False-positive coding for acute myocardial infarction on hospital discharge records: chart audit results from a tertiary centre.

Hospital medical records staff enter diagnostic codes on charts using the International Classification of Diseases (Clinically Modified), Ninth Revision (ICD-9-CM). In a downtown Toronto tertiary hospital, 209 consecutive charts coded for acute myocardial infarction as the primary diagnosis in 1987-88 were reviewed. Criteria for documentation of acute myocardial infarction included symptomatic, electrocardiographic and enzymatic elements. Forty-three (21%) false-positives, ie, charts coded acute myocardial infarction where criteria were not fulfilled, were found (95% confidence interval 15 to 26%). Physician diagnosis of acute myocardial infarction appeared on the face sheet of 30 of the false-positive cases. Common reasons for false-positive face sheet entries and chart coding were acute myocardial infarction within the previous eight weeks with transfer or readmission for coronary angiography and other procedures; and presumed acute myocardial infarction on admission subsequently unproven or disproved. The false-positive proportion was similar to a Canadian study drawing on charts from hospitals of various sizes in 1977, lower than in recent reports from various American tertiary teaching hospitals (P less than 0.0001), and higher than in five Boston area community hospitals (P = 0.0005) where procedure-related transfers or readmissions of previous acute myocardial infarction patients were less likely. This audit lends credence to arguments that changes are needed in ICD-9-CM codes for acute myocardial infarction and in the assignation of reasons for hospitalization.

An introduction to multilevel regression models.

Data in health research are frequently structured hierarchically. For example, data may consist of patients nested within physicians, who in turn may be nested in hospitals or geographic regions. Fitting regression models that ignore the hierarchical structure of the data can lead to false inferences being drawn from the data. Implementing a statistical analysis that takes into account the hierarchical structure of the data requires special methodologies. In this paper, we introduce the concept of hierarchically structured data, and present an introduction to hierarchical regression models. We then compare the performance of a traditional regression model with that of a hierarchical regression model on a dataset relating test utilization at the annual health exam with patient and physician characteristics. In comparing the resultant models, we see that false inferences can be drawn by ignoring the structure of the data.

What is necessary for high-quality discharge summaries?

The objective of this study was to determine what physicians perceive to be necessary for high-quality discharge summaries. One-on-one surveys of 100 hospital-based physicians-in-training and community family physicians were conducted. Participants indicated the amount that 56 items contributed to discharge summary quality on a 15-category ordinal scale. Results were transformed to a continuous scale, extending from -6.6 ("item makes summary useless") through 0 ("item has no effect on discharge summary quality") to 10 ("item is so essential that summary is useless without it"). Quality decreased significantly when summary length exceeded 2 pages and when the delay from patient discharge to summary delivery increased. Summary content that increased quality most included admission diagnosis (mean 8.2; 95% confidence interval [7.7, 8.6]), pertinent physical examination findings (6.6 [6.0, 7.2]) and laboratory results (6.8 [6.3, 7.4]), procedures (7.1 [6.7, 7.6]) and complications in hospital (7.1 [6.6, 7.5]), discharge diagnosis (8.8 [8.4, 9.1]), discharge medications (7.9 [7.4, 8.4]), active medical problems at discharge (7.8 [7.4, 8.2]), and follow up (6.6 [6.0, 7.1]). With minor exceptions, hospital and family physicians agreed on contributors to summary quality. For this sample of physicians, summaries were of high quality when they were short, delivered quickly, and contained pertinent data that concentrated upon discharge information.

Coding diagnoses and procedures using a high-quality clinical database instead of a medical record review.

A discharge abstract must be completed for each hospitalization. The most time-consuming component of this task is a complete review of the doctors' progress notes to identify and code all diagnoses and procedures. We have developed a clinical database that creates hospital discharge summaries. To compare diagnostic and procedural coding from a clinical database vs. the standard chart review by health records analysts (HRA). All patients admitted and discharged from general medical and surgical services at a teaching hospital in Ontario, Canada. Diagnostic and procedural codes were identified by reviewing discharge summaries generated from a clinical database. Independently, codes were identified by hospital health records analysts using chart review alone. Codes were compared with a gold standard case review conducted by a health records analyst and a doctor. Coding accuracy (percentage of codes in gold standard review) and completeness (percentage of gold standard codes identified). The study included 124 patients (mean length of stay 5.5 days; 66.4% medical patients). The accuracy of the most responsible diagnosis was 68.5% and 62.9% for the database (D) and chart review (C), respectively (P = 0.18). Overall, the database significantly improved the accuracy (D = 78.9% vs. C = 74.5%; P = 0.02) and completeness (D = 63.9% vs. C = 36.7%; P < 0.0001) of diagnostic coding. Although completeness of procedural coding was similar (D = 5.4% vs. C = 64.2%; P = NS), accuracy decreased with the database (D = 70.3% vs. C = 92.2%; P < 0.0001). Mean resource intensity weightings calculated from the codes (D = 1.3 vs. C = 1.4; P = NS) were similar. Coding from a clinical database may circumvent the need for HRAs to review doctors' progress notes, while maintaining the quality of coding in the discharge abstract.

Why are investigations not recommended by practice guidelines ordered at the periodic health examination?

Evidence-based guidelines recommend few routine investigations for healthy adults at the periodic health examination (PHE). However, small studies indicate that laboratory tests are commonly ordered at the PHE. This study examined PHE laboratory testing that is not recommended by recognized guidelines ('discretionary'). Using administrative data from the universal health care system in Ontario, Canada, we studied 792,844 adults having a PHE in 1996 and the 3,727 physicians who administered them. We measured the number of discretionary laboratory tests per PHE along with the patient and physician factors potentially influencing laboratory testing. A multilevel, multivariate model was used to examine the association between the number of discretionary laboratory tests at the PHE with patient and physician characteristics. A mean of 7.1 discretionary tests (SD 7.1) was ordered per PHE. Renal, haematological, glucose and lipid tests each were conducted in more than a third of PHEs. Testing varied extensively between physicians and was more common in healthy patients. With the exception of age, patient factors had little effect on discretionary testing. However, each physician factor we examined was independently associated with the number of discretionary tests. Physician specialty, practice volume and previous testing patterns had the strongest influence on discretionary testing. Discretionary investigations are common at the PHE. Testing varies extensively between physicians and seems to be driven more by physician than by patient factors. Interventions to modify discretionary test utilization at the PHE should consider these physician factors.

Inhibition of serotonin reuptake by antidepressants and upper gastrointestinal bleeding in elderly patients: retrospective cohort study.

OBJECTIVES: To determine the association between inhibition of serotonin reuptake by antidepressants and upper gastrointestinal bleeding. DESIGN: Retrospective cohort study from population based databases. SETTING: Ontario, Canada. PARTICIPANTS: 317 824 elderly people observed for more than 130 000 person years. The patients started taking an antidepressant between 1992 and 1998 and were grouped by how much the drug inhibited serotonin reuptake. Patients were observed until they stopped the drug, had an upper gastrointestinal bleed, or died or the study ended. MAIN OUTCOME MEASURE: Admission to hospital for acute upper gastrointestinal bleeding. RESULTS: Overall, 974 bleeds were observed, with an overall bleeding rate of 7.3 per 1000 person years. After controlling for age or previous gastrointestinal bleeding, the risk of bleeding significantly increased by 10.7% and 9.8%, respectively, with increasing inhibition of serotonin reuptake. Absolute differences in bleeding between antidepressant groups were greatest for octogenarians (low inhibition of serotonin reuptake, 10.6 bleeds/1000 person years v high inhibition of serotonin reuptake, 14.7 bleeds/1000 person years; number needed to harm 244) and those with previous upper gastrointestinal bleeding (low, 28.6 bleeds/1000 person years v high, 40.3 bleeds/1000 person years; number needed to harm 85). CONCLUSIONS: After age or previous upper gastrointestinal bleeding were controlled for, antidepressants with high inhibition of serotonin reuptake increased the risk of upper gastrointestinal bleeding. These increases are clinically important for elderly patients and those with previous gastrointestinal bleeding.