Ontario Health Insurance Plan Database Research Articles

Integrating unsupervised and supervised learning techniques to predict traumatic brain injury: A population-based study

This work aimed to identify pre-existing health conditions of patients with traumatic brain injury (TBI) and develop predictive models for the first TBI event and its external causes by employing a combination of unsupervised and supervised learning algorithms. We acquired up to five years of pre-injury diagnoses for 488,107 patients with TBI and 488,107 matched control patients who entered the emergency department or acute care hospitals between April 1st, 2002, and March 31st, 2020. Diagnoses were obtained from the Ontario Health Insurance Plan (OHIP) database which contains province-wide claims data by physicians in Ontario, Canada for inpatient and outpatient services. A screening process was conducted on the OHIP diagnostic codes to limit the subsequent analysis to codes that were predictive of TBI, which concluded that 314 codes were significantly associated with TBI. The Latent Dirichlet Allocation (LDA) model was applied to the diagnostic codes and generated an optimal number of 19 topics that concur with published literature but also suggest other unexplored areas. Estimated word-topic probabilities from the LDA model helped us detect pre-morbid conditions among patients with TBI by uncovering the underlying patterns of diagnoses, meanwhile estimated document-topic probabilities were utilized in variable creation as form of a dimension reduction. We created 19 topic scores for each patient in the cohort which were utilized along with socio-demographic factors for Random Forest binary classifier models. Test set performances evaluated using area under the receiver operating characteristic curve (AUC) were: TBI event (AUC = 0.85), external cause of injury: falls (AUC = 0.85), struck by/against (AUC = 0.83), cyclist collision (AUC = 0.76), motor vehicle collision (AUC = 0.83). Our analysis successfully demonstrated the feasibility of using machine learning to predict TBI due to various external causes and identified the most important factors that contribute to this prediction.

Stress Incontinence Surgery Does Not Cause Pelvic Malignancy: A Population-Based Cohort Study

We sought to determine if stress urinary incontinence surgery (mesh or nonmesh) is associated with the development of pelvic malignancies later in life. We performed a retrospective cohort study between January 1, 2002 and October 31, 2015 of all women in Ontario, Canada without a history of pelvic malignancy who underwent an index stress incontinence surgery. The primary outcome was a composite of any pelvic malignancy (including urological and gynecological cancers) following stress incontinence surgery. Secondarily, we considered each cancer individually. A survival analysis using a Cox proportional-hazards model with a 3-level categorical exposure (mesh surgery, nonmesh surgery, and control) was performed. Patients were followed until death, emigration or the study end (October 31, 2017). Of the women 74,968 underwent stress urinary incontinence surgery during the study period. There were 5,505,576 women in the control group. Over a median followup of 8.5 years (IQR, 5.5-11.9), 587 pelvic malignancies occurred in the surgery group. Women who underwent stress incontinence surgery had a reduced risk of pelvic malignancy independent of surgery type, compared to controls (Wald type 3 p <0.001; mesh HR, 0.68 [95% CI, 0.62-0.76]; p <0.0001; nonmesh HR, 0.37 [95% CI, 0.29-0.46]; p <0.0001). The individual pelvic cancers similarly demonstrated a reduced risk of malignancy following stress incontinence surgery. At a median followup of 8.5 years, women had no increased risk of pelvic malignancy following either mesh or nonmesh stress urinary incontinence surgery in a large population-based cohort.

Risk of infertility in female adolescents and young adults with cancer: a population-based cohort study.

STUDY QUESTIONDo female adolescents and young adults (AYAs) with cancer have a higher risk of subsequent infertility diagnosis than AYAs without cancer?SUMMARY ANSWERFemale AYAs with breast, hematological, thyroid and melanoma cancer have a higher risk of subsequent infertility diagnosis.WHAT IS KNOWN ALREADYCancer therapies have improved substantially, leading to dramatic increases in survival. As survival improves, there is an increasing emphasis on optimizing the quality of life among cancer survivors. Many cancer therapies increase the risk of infertility, but we lack population-based studies that quantify the risk of subsequent infertility diagnosis in female AYAs with non-gynecological cancers. The literature is limited to population-based studies comparing pregnancy or birth rates after cancer against unexposed women, or smaller studies using markers of the ovarian reserve as a proxy of infertility among female survivors of cancer.STUDY DESIGN, SIZE, DURATIONWe conducted a population-based cohort study using universal health care databases in the province of Ontario, Canada. Using data from the Ontario Cancer Registry, we identified all women 15–39 years of age diagnosed with the most common cancers in AYAs (brain, breast, colorectal, leukemia, Hodgkin lymphoma, non-Hodgkin lymphoma, thyroid and melanoma) from 1992 to 2011 who lived at least 5 years recurrence-free (Exposed, n = 14,316). Women with a tubal ligation, bilateral oophorectomy or hysterectomy previous to their cancer diagnosis were excluded. We matched each exposed woman by age, census subdivision, and parity to five randomly selected unexposed women (n = 60,975) and followed subjects until 31 December 2016.PARTICIPANTS/MATERIALS, SETTING, METHODSInfertility diagnosis after 1 year of cancer was identified using information on physician billing codes through the Ontario Health Insurance Plan database (ICD-9 628). Modified Poisson regression models were used to assess the risk of infertility diagnosis (relative risk, RR) adjusted for income quintile and further stratified by parity at the time of cancer diagnosis (nulliparous and parous).MAIN RESULTS AND THE ROLE OF CHANCEMean age at cancer diagnosis was 31.4 years. Overall, the proportion of infertility diagnosis was higher in cancer survivors compared to unexposed women. Mean age of infertility diagnosis was similar among cancer survivors and unexposed women (34.8 years and 34.9 years, respectively). The overall risk of infertility diagnosis was higher in cancer survivors (RR 1.30; 95% CI 1.23–1.37). Differences in infertility risk varied by type of cancer. Survivors of breast cancer (RR 1.46; 95% CI 1.30–1.65), leukemia (RR 1.56; 95% CI 1.09–2.22), Hodgkin lymphoma (RR 1.49; 95% CI 1.28–1.74), non-Hodgkin lymphoma (RR 1.42; 95% CI 1.14, 1.76), thyroid cancer (RR 1.20; 95% CI 1.10–1.30) and melanoma (RR 1.17; 95% CI 1.01, 1.35) had a higher risk of infertility diagnosis compared to women without cancer. After stratification by parity, the association remained in nulliparous women survivors of breast cancer, leukemia, lymphoma and melanoma, whereas it was attenuated in parous women. In survivors of thyroid cancer, the association remained statistically significant in both nulliparous and parous women. In survivors of brain or colorectal cancer, the association was not significant, overall or after stratification by parity.LIMITATIONS, REASONS FOR CAUTIONNon-biological factors that may influence the likelihood of seeking a fertility assessment may not be captured in administrative databases. The effects of additional risk factors, including cancer treatment, which may modify the associations, need to be assessed in future studies.WIDER IMPLICATIONS OF THE FINDINGSReproductive health surveillance in female AYAs with cancer is a priority, especially those with breast cancer, leukemia and lymphoma. Our finding of a potential effects of thyroid cancer (subject to over-diagnosis) and, to a lesser extent, melanoma need to be further studied, and, if an effect is confirmed, possible mechanisms need to be elucidated.STUDY FUNDING/COMPETING INTEREST(S)Funding was provided by the Faculty of Health Sciences and Department of Obstetrics and Gynecology, Queen’s University. There are no competing interests to declare.TRIAL REGISTRATION NUMBERN/A

Job Strain, Overweight, and Diabetes: A 13-Year Prospective Study Among 12,896 Men and Women in Ontario.

The American Diabetes Association recently called for research on social and environmental determinants of diabetes to intensify primary prevention. Recent epidemiological evidence suggests that frequent and modifiable psychosocial stressors at work might contribute to the development of diabetes, but more prospective studies are needed. We evaluated the relationship between job strain and diabetes incidence in 12,896 workers followed up over a 13-year period in Ontario, Canada. We also examined the modifying effect of body mass index in this relationship. Data from Ontario respondents (35-74 years of age) to the 2000-2001, 2002, and 2003 cycles of the Canadian Community Health Survey were prospectively linked to the Ontario Health Insurance Plan database for physician services and the Canadian Institute for Health Information Discharge Abstract Database for hospital admissions. The sample consisted of actively employed participants with no previous diagnosis for diabetes. Cox proportional hazard regression models were performed to evaluate the relationship between job strain, obesity, and the incidence of diabetes. Overall, job strain was not associated with the incidence of diabetes (hazard ratio [HR] = 1.05; 95% confidence interval [CI] = 0.83-1.34). Among women, job strain was associated with an elevated risk of diabetes, although this finding did not reach statistical significance (HR = 1.36; 95% CI = 0.94-1.96). Among men, no association was observed (HR = 0.89; 95% CI = 0.65-1.22). Also, job strain increased the risk of diabetes among women with obesity (HR = 1.88; 95% CI = 1.14-3.08), whereas these stressors reduced the risk among men with obesity (HR = 0.58; 95% CI = 0.36-0.95). The current study suggests that lowering job strain might be an effective strategy for preventing diabetes among women, especially the high-risk group comprising women with obesity.

A cohort study evaluating the association between concurrent mental disorders, mortality, morbidity, and continuous treatment retention for patients in opioid agonist treatment (OAT) across Ontario, Canada, using administrative health data

BackgroundDue to the high prevalence of mental disorders among people with opioid use disorder, the objective of this study was to determine the association between concurrent mental disorders, mortality, morbidity, and continuous treatment retention for patients in opioid agonist treatment in Ontario, Canada.MethodsWe conducted a retrospective cohort study of patients enrolled in opioid agonist treatment between January 1, 2011, and December 31, 2015. Patients were stratified into two groups: those diagnosed with concurrent mental disorders and opioid use disorder and those with opioid use disorder only, using data from the Ontario Health Insurance Plan Database, Ontario Drug Benefit Plan Database. The primary outcome studied was all-cause mortality using data from the Registered Persons Database. Emergency department visits from the National Ambulatory Care Database, hospitalizations Discharge Database, and continuous retention in treatment, defined as 1 year of uninterrupted opioid agonist treatment using data from the Ontario Drug Benefit Plan Database were measured as secondary outcomes. Encrypted patient identifiers were used to link information across databases.ResultsWe identified 55,924 individuals enrolled in opioid agonist treatment, and 87% had a concurrent mental disorder diagnosis during this period. We observed that having a mental disorder was associated with an increased likelihood of all-cause mortality (odds ratio (OR) 1.4; 95% confidence interval (CI) 1.2–1.5). For patients diagnosed with mental disorders, the estimated rate of ED visits per year was 2.25 times higher and estimated rate of hospitalization per year was 1.67 times higher than for patients with no mental disorders. However, there was no association between having a diagnosis of a mental disorder and 1-year treatment retention in OAT-adjusted hazard ratio (HR) = 1.0; 95% CI 0.9 to 1.1.ConclusionOur findings highlight the consequences of the high prevalence of mental disorders for individuals with opioid use disorder in Ontario, Canada.

A retrospective cohort study evaluating correlates of deep tissue infections among patients enrolled in opioid agonist treatment using administrative data in Ontario, Canada.

The objective of this study was to evaluate the relationship between individual characteristics and deep tissue infections in patients enrolled in opioid agonist treatment in Ontario, Canada. A retrospective cohort study was conducted on patients in opioid agonist treatment between January 1, 2011, and December 31, 2015 in Ontario, Canada. Patients were identified using data from the Ontario Health Insurance Plan Database, and the Ontario Drug Benefit Plan Database. We identified other study variables including all-cause mortality using data from the Registered Persons Database. Encrypted patient identifiers were used to link across databases. Logistic regression models were used to measure potential correlates of deep tissue infections. An increase in the incidence of deep tissue infections was observed between 2011 and 2016 for patients on opioid agonist treatment. Additionally, age, sex, positive HIV diagnosis, and all-cause mortality was correlated with deep tissue infection in our study population. The study indicates factors that are associated with deep tissue infections in the opioid use disorder population and can be used to identify opportunities to reduce the incidence of new infections.

COMPARISON OF READMISSION AND DEATH AMONG CARDIAC PATIENTS IN NORTHERN VS. SOUTHERN ONTARIO

Northern Ontarians (residents of the Northeast and Northwest Local Health Integration Networks) have higher population age-standardized mortality rates than the rest of the province, and the majority of deaths in Northern Ontario were attributed to circulatory diseases. It is unknown if differences in hospital-based acute cardiovascular care contribute to the differential outcomes in Northern vs. Southern Ontario. To examine whether mortality and readmission rates differ in Northern and Southern Ontario among patients admitted with acute cardiovascular disease conditions. Using the Canadian Institute for Health Information Discharge Abstract Database (CIHI-DAD), we identified index admissions for acute myocardial infarction (MI), heart failure (HF), atrial fibrillation (AF), or stroke, from April 2005 to March 2015. We determined the risk of readmission (using CIHI-DAD) or death (using the Registered Persons Database) for Northern compared to Southern Ontarians, adjusting for validated comorbidity risk models, Johns Hopkins ADG, socioeconomic status, year and season of admission, and distance from home to hospital, using multivariable Cox regression analysis. Physician care was determined using the Ontario Health Insurance Plan (OHIP) database. We identified 523,029 patients who were admitted with acute MI (n=183,402), HF (n=131,374), AF (n=72,799), and stroke (n=128,673). Median ages of patients (Northern, Southern) were: 69, 69 years for acute MI; 78, 80 years for HF; 74, 75 years for AF; and 74, 76 years for stroke. Cumulative incidence of mortality over 1 year did not differ in the North vs. the South for acute MI and HF, but there was significantly higher incidence of readmission in the North (Figure). Similar findings were observed for AF and stroke. After multivariable adjustment, the hazard ratio [HR (95%CI)] for mortality in the North was 1.06 (1.03-1.10) for acute MI (p=0.001), but HRs were 1.02 (0.99-1.06) for HF, 0.95 for AF (0.89-1.02), and 1.00 for stroke (0.97-1.04) compared to the South. However, readmission rate was significantly increased in the North for all conditions (all p < .001): acute MI (HR 1.28, 95%CI; 1.25-1.32), HF (HR 1.20, 95%CI; 1.16-1.23), AF (HR 1.27, 95%CI; 1.21-1.32), and stroke (HR 1.31, 95%CI; 1.26-1.37). In all conditions, early specialist care (cardiologist or neurologist) was significantly less likely in the north with multivariable-adjusted HRs for 30-day specialist consults/visits ranging from 0.51 to 0.70 (p < .001). Readmissions for cardiovascular conditions were substantially higher in Northern compared to Southern Ontario, while differences in mortality were not different or less pronounced.

Internal and External Data Linkage of Complex Relational Database: Results from CorHealth Ontario

IntroductionCorHealth Ontario, formerly Cardiac Care Network (CCN), maintains a registry of patients undergoing select cardiac procedures/surgeries in Ontario, Canada. This population-based database contains over 35 datasets with complex structure, linked by unique primary key or multiple keys.&#x0D; Objectives and ApproachWe aimed to simplify the complex CorHealth database so that research analysts could create study cohorts more efficiently and effectively, and to enrich the study cohort by getting more clinical information through database linkage. Through internal linkage, we could combine clinical fields from multiple CorHealth datasets. While the CorHealth dataset may not have all the clinical information needed for a given study, we may link the CorHealth study cohort externally to other administration databases to obtain additional fields via the probability matching (i.e., identical patient ID, hospital ID and procedure/surgery date).&#x0D; ResultsAfter identifying the primary keys on the relational database flowchart, we designed new data structures by combining similar topic datasets. The total number of datasets was reduced from 35 to 13. This simplified CorHealth dataset includes one main CorHealth dataset (including demographic information, referral data, comorbidities) plus 12 other linkable specific datasets (including stent, vessel, TAVI, STEMI). Through internal linkage, we can get the stent numbers, lengths and types of Percutaneous Coronary Interventions from the Stent dataset. Linking to Discharge Abstract Database (DAD), we can get the hospital length of stay and the episode of care of hospital transfer for each procedure; linking to The Ontario Health Insurance Plan database (OHIP), we can find the graft numbers and vessel types of Coronary Artery Bypass Graft.&#x0D; Conclusion/ImplicationsTo improve the research capacity and increase the value of the CorHealth database, analysts could create enhanced cardiovascular study cohorts derived from the simplified CorHealth database, plus internal linkage from other CorHealth datasets, and external data linkage from population-based administrative sources. We have accomplished three reports (PCI/CABG/TAVI) accordingly in 2017/18.

Complications and Interventions in Patients with an Artificial Urinary Sphincter: Long-Term Results

The artificial urinary sphincter is a common treatment of male urinary incontinence. We sought to characterize long-term rates of artificial urinary sphincter revision/removal and reimplantation, and associated risk factors. We performed a population based, retrospective study using the Ontario Health Insurance Plan database of all male patients who underwent artificial urinary sphincter implantation from 1994 to 2013. Hospital, diagnostic and billing codes were used to identify patients. The Kaplan-Meier method and multivariable Cox proportional hazards models were applied to examine the cumulative incidence of artificial urinary sphincter reimplantation and revision/removal, and identify risk factors, respectively. A total of 1,632 male patients underwent initial implantation of an artificial urinary sphincter. The 10-year revision/removal and reimplantation rates were 34% and 27%, respectively. There was no difference between high and low volume hospitals or between community and academic hospitals in terms of revision/removal. A high level of comorbidity was associated with an increasing risk of removal/revision. The reimplantation rate was significantly lower in patients who underwent insertion at hospitals with the highest volume of surgery (HR 0.55, 95% CI 0.37-0.82, p <0.01). A high level of comorbidity was not associated with an increasing risk of reimplantation. Preimplantation radiotherapy was not significantly associated with the risk of reimplantation (p = 0.17) or revision/removal (p = 0.95). Other factors were not significantly associated with reimplantation or revision/removal. Most men who undergo artificial urinary sphincter placement still have a device without repeat surgery 10 years following insertion. Radiotherapy does not increase the risk of repeat surgery. A high level of comorbidity was associated with an increasing risk of removal/revision.

Adverse effect of long work hours on incident diabetes in 7065 Ontario workers followed for 12 years

ObjectiveAccording to the International Diabetes Federation, the most important challenge for prevention is now to identify social and environmental modifiable risk factors of diabetes. In this regard, long work hours...

Variation in Access to Kidney Transplantation Across Renal Programs in Ontario, Canada.

In the United States, kidney transplant rates vary significantly across end-stage renal disease (ESRD) networks. We conducted a population-based cohort study to determine whether there was variability in kidney transplant rates across renal programs in a health care system distinct from the United States. We included incident chronic dialysis patients in Ontario, Canada, from 2003 to 2013 and determined the 1-, 5-, and 10-year cumulative incidence of kidney transplantation in 27 regional renal programs (similar to U.S. ESRD networks). We also assessed the cumulative incidence of kidney transplant for "healthy" dialysis patients (aged 18-50 years without diabetes, coronary disease, or malignancy). We calculated standardized transplant ratios (STRs) using a Cox proportional hazards model, adjusting for patient characteristics (maximum possible follow-up of 11 years). Among 23 022 chronic dialysis patients, the 10-year cumulative incidence of kidney transplantation ranged from 7.4% (95% confidence interval [CI] 4.8-10.7%) to 31.4% (95% CI 16.5-47.5%) across renal programs. Similar variability was observed in our healthy cohort. STRs ranged from 0.3 (95% CI 0.2-0.5) to 1.5 (95% CI 1.4-1.7) across renal programs. There was significant variation in kidney transplant rates across Ontario renal programs despite patients having access to the same publicly funded health care system.

Using physician billing claims from the Ontario Health Insurance Plan to determine individual influenza vaccination status: an updated validation study.

Owing to the absence of a vaccination registry in Ontario, administrative data are currently the best available source to determine population-based individual-level influenza vaccination status. Our objective was to validate physician billing claims for influenza vaccination in the Ontario Health Insurance Plan database against the Canadian Community Health Survey. We used self-reported seasonal influenza vaccination status of Ontario residents surveyed between 2007 and 2009 as the reference standard. The survey responses were linked to physician claims database records to validate billing codes for influenza vaccination. We calculated sensitivity, specificity, positive predictive value and negative predictive value with 95% confidence intervals (CIs). We stratified the data by several covariates and comorbidities to determine stratum-specific performance characteristics. We used these estimates to adjust an estimate of influenza vaccine effectiveness for the 2010/11 influenza season. For the 47 301 patients included in the analysis, the sensitivity for the billing codes was 49.8% (95% CI 49.0%-50.5%), specificity was 95.7% (95% CI 95.5%-96.0%), positive predictive value was 88.4% (95% CI 87.8%-89.0%) and negative predictive value was 74.5% (95% CI 74.0%-74.9%). Performance measures were optimized in patients aged 65 years and older, particularly those with comorbidities. Although administrative data have limitations for determining influenza vaccination status, owing to the high positive predictive value, they are well suited for self-controlled study designs that are often used to assess vaccine safety. For studies of coverage and effectiveness, restricting the cohort to patients aged 65 years and older will minimize misclassification bias. Performance characteristics from this study can be used to mitigate misclassification bias.

Survival and cardiovascular events in men treated with testosterone replacement therapy: an intention-to-treat observational cohort study

Conflicting evidence exists for the association between testosterone replacement therapy and mortality and cardiovascular events. The US Food and Drug Administration recently cautioned that testosterone replacement therapy might increase risk of heart attack and stroke, based on evidence from studies with short treatment duration and follow-up. No previous study has assessed the effect of duration of testosterone treatment on these outcomes. We aimed to assess the association between long-term use of testosterone replacement therapy and mortality, cardiovascular events, and prostate cancer diagnoses, using a time-varying exposure analysis. We did a population-based matched cohort study of men aged 66 years or older newly treated with testosterone replacement therapy and controls matched for age, region of residence, comorbidity, diabetes status, and index year from 2007-12 in Ontario, Canada, using data from the Ontario Drug Benefit database, the Canadian Institute for Health Information (CIHI) Discharge Abstract Database, the CIHI National Ambulatory Care Reporting System, the Ontario Health Insurance Plan database, the Ontario Myocardial Infarction Database, the Ontario Diabetes Database, the Ontario Cancer Registry, and the Registered Persons database. We assessed the association between cumulative testosterone replacement therapy exposure and mortality, cardiovascular events, and prostate cancer using marginal models with a time-varying testosterone exposure. We included 10 311 men treated with testosterone replacement therapy and 28 029 controls between Jan 1, 2007, and June 30, 2012. Over a median follow-up of 5·3 years (IQR 3·6-7·5) in the testosterone replacement therapy group and 5·1 years (3·4-7·4) in the control group, patients treated with testosterone replacement therapy had lower mortality than did controls (hazard ratio [HR] 0·88, 95% CI 0·84-0·93). Patients in the lowest tertile of testosterone exposure had increased risk of mortality (HR 1·11, 95% CI 1·03-1·20) and cardiovascular events (HR 1·26, 95% CI 1·09-1·46) compared with controls. By contrast, those in the highest tertile of testosterone exposure had decreased risk of mortality (HR 0·67, 95% CI 0·62-0·73) and cardiovascular events (HR 0·84, 95% CI 0·72-0·98), with a significant trend across tertiles (p<0·0001). Risk of prostate cancer diagnosis was decreased for those with the highest tertile of exposure (HR 0·60, 95% CI 0·45-0·80) compared with controls, but not for those with the shortest exposure. Long-term exposure to testosterone replacement therapy was associated with reduced risks of mortality, cardiovascular events, and prostate cancer. However, testosterone replacement therapy increased the risk of mortality and cardiovascular events with short durations of therapy. In view of the limitations of observational data and the potential for selection bias, these results warrant confirmation in a randomised trial. Physicians' Services Incorporated Foundation and Ajmera Family Chair in Urologic Oncology.

Repeat colonoscopy after a colonoscopy with a negative result in Ontario: a population-based cohort study.

Data suggest the overuse of repeat colonoscopies, especially in patients at low risk for colorectal cancer. Our objective was to evaluate the time to repeat colonoscopies in low-risk patients aged 50-79 years old and the associated patient- and endoscopist-related factors. All patients aged 50-79 years of age who underwent a complete outpatient colonoscopy with a negative result between 2000 and 2007 were identified from the Ontario Health Insurance Plan database. A colonoscopy performed within 5.5 years of follow-up after the index colonoscopy was considered an early repeat colonoscopy. Patient, endoscopist and endoscopy setting characteristics were recorded and their association with an early repeat colonoscopy was determined using an extended Cox proportional hazards regression model. The cohort consisted of 546 467 patients: 55.4% of the patients were female with a mean age of 61.1 years (95% confidence interval [CI] 61.1-61.2). The cumulative percentage of early repeat colonoscopy after 5.5 years was 33.7%. The rate decreased significantly between 2000 and 2007 (hazard ratio [HR] 0.35, 95% CI 0.34-0.36). General surgeons were associated with a higher risk of early repeat colonoscopy than gastroenterologists (HR 1.27, 95% CI 1.25-1.28). Endoscopists practising in a nonhospital setting were more likely to perform an early repeat colonoscopy (HR 1.26, 95% CI 1.22-1.30) than endoscopists at a hospital. This study showed that there was overuse of early repeat colonoscopy in more than 30% of patients who were at low risk for colorectal cancer. The risk decreased significantly between 2000 and 2007 but was still greater than 20% in 2007. Our findings can be used to develop targeted educational interventions among subgroups of endoscopists with a higher rate of early repeat colonoscopy.

Validation of infant immunization billing codes in administrative data

Ontario has a single payer provincial health insurance program. Administrative data may provide a potentially robust source of information for post-marketing vaccine studies. Vaccine-specific immunization billing codes were introduced in 2011. Our objective was to validate Ontario's universal health care administrative datasets to assess infant immunization status. Electronic medical record data from the Electronic Medical Record Administrative data Linked Database (EMRALD) was used as the reference standard to calculate performance characteristics of the Ontario Health Insurance Plan (OHIP) database vaccine-specific and general immunization codes for 4 primary infant immunizations: diphtheria, tetanus, acellular pertussis, inactivated polio, Haemophilus influenzae type B (DTaP-IPV-Hib) combination vaccine, pneumococcal conjugate vaccine, measles, mumps, rubella (MMR) vaccine, and meningococcal conjugate serogroup C vaccine. OHIP billing claims had specificity ranging from 81% to 92%, sensitivity 70% to 83%, positive predictive value (PPV) 97% to 99%, and negative predictive value (NPV) 13% to 46% for identifying the various specific vaccines in administrative data. For cohorts vaccinated in the new code introduction phase, using both the vaccine-specific and general codes had higher sensitivity than the vaccine-specific codes alone. In conclusion, immunization billing claims from administrative data in Ontario had high specificity and PPV, moderate sensitivity, and low NPV. This study identifies some of the applications of utilizing administrative data for post-marketing vaccine studies. However, limitations of these data decrease their utility for measuring vaccine coverage and effectiveness. Therefore, the establishment of a comprehensive and linkable immunization registry should be a provincial priority.

Validation of kidney transplantation using administrative data.

BackgroundAdministrative data are increasingly being used to assess outcomes in kidney transplant recipients.ObjectiveTo assess the validity of transplant data in healthcare administrative databases compared to the reference standard of information collected directly from transplant centres.DesignRetrospective cohort study.SettingOne of three major transplant centres in Ontario (Toronto General Hospital, University Hospital – London, and Ottawa Hospital).PatientsRecipients who received a kidney-only transplant between 2008 and 2011.MeasurementsFor each data source, we identified kidney transplants performed. We calculated the sensitivity and positive predictive value (PPV) of the administrative data for the reference standard data.MethodsThe data collected from transplant centres were compared with data from the Canadian Organ Replacement Register (CORR) database, a hospital procedural code from the Canadian Institute for Health Information Discharge Database (CIHI-DAD), and provincial physician billing claims from the Ontario Health Insurance Plan (OHIP) database.ResultsDuring the study period, the three centres reported a total of 1112 kidney transplants performed. The probability of identifying kidney transplant recipients in CORR, CIHI, and OHIP, given they were identified by the transplant centres (sensitivity), was 96%, 98%, and 98% respectively. The probability that the database code correctly identified a transplant recipient (positive predictive value) in CORR, CIHI, and OHIP was 98%, 98%, and 96% respectively.LimitationsWe validated the information from 2008 to 2011 and cannot attest to the reliability of the data beyond the study period. Specifically, we would not regard this as evidence that applies to the earlier years, shortly after the inception of the databases. Secondly, we were unable to distinguish between first and repeat transplantation.ConclusionsCodes in CORR, CIHI, and OHIP each operate well in the detection of kidney transplant recipients. These data sources can be used to efficiently identify and follow kidney transplant recipients for post-transplant outcomes.

Current use of domperidone and co-prescribing of medications that increase its arrhythmogenic potential among older adults: a population-based cohort study in Ontario, Canada.

Domperidone is commonly used to treat nausea and gastrointestinal disorders. Recent data suggests that it may increase the risk of sudden cardiac death, particularly in older people. Little is known about how it is used in contemporary practice. This study sought to characterize the population of older adults newly dispensed domperidone, describe dosages of domperidone used, and determine the frequency of co-prescribing domperidone with medications that may increase the arrhythmogenic potential of domperidone. This is a retrospective cohort study using administrative health database information from Ontario, Canada. Prescription medication records were obtained from the Ontario Drug Benefit Claims Database. Diagnostic codes were obtained from the Ontario Health Insurance Plan Database, the Canadian Institute for Health Information Discharge Abstract Database, and the same-day surgery database. Patients who received a new prescription for domperidone between April 1, 2003 and March 31, 2010 were included. A total of 122,233 patients met inclusion criteria; 85 % were between 66 and 84 years old and 63 % were female. The mean estimated daily domperidone dose was 35 mg, and the estimated daily dose was <40 mg for 62 % of users. Strong or moderately strong cytochrome P-450 (CYP) 3A4 inhibitors were co-prescribed for 4.3 and 10.7 % of users, while medications with a known risk or possible risk for torsades de pointes (TdP) were co-prescribed to 18.3 and 18.8 % of users. Older domperidone users were commonly co-prescribed drugs with the potential to increase the risk for TdP. These combinations should be avoided, as iatrogenic QT prolongation is a modifiable risk factor for TdP.

Clinic-Based Glaucoma Care in the Era of Surgical Subspecialization

To evaluate the impact of surgeon practice profile on clinic-based glaucoma care. Population-based study of glaucoma care patterns in Ontario, Canada from 2000-2010. Using comprehensive physician services data from the Ontario Health Insurance Plan database, ophthalmologists were divided into 5 surgical practice subgroups. The role of each subgroup in the provision of glaucoma care was evaluated. Consultations and office visits were used to assess nonsurgical care, while laser trabeculoplasty procedures were used to assess clinic-based procedural care. Between 2000 and 2010, the population rate of glaucoma consultations and follow-up visits provided by ophthalmologists who do not perform incisional glaucoma surgery increased at average annual rates of 1.6% (P < .0002) and 3.3% (P < .0001), respectively. In contrast, no significant growth in the rate of glaucoma consultations or follow-up visits provided by glaucoma surgeons was observed (0.8%/year [P = .2] for consultations; 0.2%/year [P = .6] for follow-up visits). Between 2000 and 2010, the rate of laser trabeculoplasty procedures provided by ophthalmologists who do not perform incisional glaucoma surgery increased 19.3% annually (P < .0001), while growth among glaucoma surgeons was more modest (annual growth of 9.2% [P = .0002]). While subspecialization is a growing reality in most areas of medicine, we found that the provision of clinic-based glaucoma care remains dependent on ophthalmologists who do not perform incisional glaucoma surgery. With increasing focus on integrated care, these findings will have important implications for residency education programs and their accrediting bodies and will inform decisions of health care policymakers, hospitals, and academic departments.

Quality of Diabetes and Hyperlipidemia Screening Before a First Myocardial Infarction

BackgroundThere has been significant attention to the quality of care for acute myocardial infarction (MI). However, little is known about the quality of preventive care before a patient's first MI. MethodsWe conducted a retrospective, cohort analysis of 5688 patients admitted with their first MI to 96 acute care hospitals in Ontario, Canada, from April 2004 to March 2005 using the Enhanced Feedback For Effective Cardiac Treatment clinical study database. We calculated rates of screening for diabetes and hyperlipidemia according to guidelines using linkages to the Ontario Health Insurance Plan database. Screening rates were stratified by age, sex, socioeconomic status, and number of primary care visits in the past 5 years. ResultsAmong the 5688 eligible patients, 27.1% did not receive serum cholesterol screening in the 5 years preceding their MI and 27.5% of patients did not receive a fasting blood glucose or glucose tolerance test in the 3 years before their MI. Women were more likely to be screened than men. Screening rates generally increased with age and were similar across socioeconomic categories. There was a positive association between the number of primary care visits and the likelihood of being screened. ConclusionsA significant number of patients admitted with their first MI were not screened for important modifiable risk factors. Opportunities for the prevention of coronary disease are being missed. More emphasis is needed on identifying risk factors before the development of acute coronary disease.

The psychosocial work environment and incident diabetes in Ontario, Canada

Relatively few longitudinal studies have explored the relationship between psychosocial work conditions and diabetes incidence. Given the increasing global burden of diabetes this is an important area for public health research. To examine the relationships between dimensions of the psychosocial work environment on the subsequent incidence of diabetes among men and women in Ontario, Canada over a 9 year period. We used data from Ontario respondents (35 to 60 years of age) to the 2000-01 Canadian Community Health Survey linked to the Ontario Health Insurance Plan database for physician services and the Canadian Institute for Health Information Discharge Abstract Database for hospital admissions. Our sample of actively employed labour market participants with no previous diagnoses for diabetes was followed for a 9 year period to ascertain incident diabetes. There were 7443 participants. Low levels of job control were associated with an increased risk of diabetes among women, but not among men. Counter to our hypotheses high levels of social support were also associated with increased diabetes risk among women, but not among men. No relationship was found between any psychosocial work measure and risk of diabetes among men. Given the increasing prevalence of diabetes worldwide, job control could potentially be an import ant modifiable risk factor to reduce the incidence of diabetes among female, but not among male, workers. More research is needed to understand the pathways through which low social support may protect against the development of diabetes.