Ascertaining the Francophone population in Ontario: validating the language variable in health data.

BACKGROUND: Language barriers can impact health care and outcomes. Valid and reliable language data is central to studying health inequalities in linguistic minorities. In Canada, language variables are available in administrative health databases; however, the validity of these variables has not been studied. This study assessed concordance between language variables from administrative health databases and language variables from the Canadian Community Health Survey (CCHS) to identify Francophones in Ontario. METHODS: An Ontario combined sample of CCHS cycles from 2000 to 2012 (from participants who consented to link their data) was individually linked to three administrative databases (home care, long-term care [LTC], and mental health admissions). In total, 27,111 respondents had at least one encounter in one of the three databases. Language spoken at home (LOSH) and first official language spoken (FOLS) from CCHS were used as reference standards to assess their concordance with the language variables in administrative health databases, using the Cohen kappa, sensitivity, specificity, positive predictive value (PPV), and negative predictive values (NPV). RESULTS: Language variables from home care and LTC databases had the highest agreement with LOSH (kappa = 0.76 [95%CI, 0.735-0.793] and 0.75 [95%CI, 0.70-0.80], respectively) and FOLS (kappa = 0.66 for both). Sensitivity was higher with LOSH as the reference standard (75.5% [95%CI, 71.6-79.0] and 74.2% [95%CI, 67.3-80.1] for home care and LTC, respectively). With FOLS as the reference standard, the language variables in both data sources had modest sensitivity (53.1% [95%CI, 49.8-56.4] and 54.1% [95%CI, 48.3-59.7] in home care and LTC, respectively) but very high specificity (99.8% [95%CI, 99.7-99.9] and 99.6% [95%CI, 99.4-99.8]) and predictive values. The language variable from mental health admissions had poor agreement with all language variables in the CCHS. CONCLUSIONS: Language variables in home care and LTC health databases were most consistent with the language often spoken at home. Studies using language variables from administrative data can use the sensitivity and specificity reported from this study to gauge the level of mis-ascertainment error and the resulting bias.

Validation of an algorithm for identifying incident cancer cases based on long-term illness and diagnosis related group program data from the French National Health Insurance Information System (SNDS).

PURPOSE: Three generic claims-based algorithms based on the Illness Classification of Diseases (10th revision- ICD-10) codes, French Long-Term Illness (LTI) data, and the Diagnosis Related Group program (DRG) were developed to identify retirees with cancer using data from the French national health insurance information system (Système national des données de santé or SNDS) which covers the entire French population. The present study aimed to calculate the algorithms' performances and to describe false positives and negatives in detail. METHODS: Between 2011 and 2016, data from 7544 participants of the French retired self-employed craftsperson cohort (ESPrI) were first matched to the SNDS data, and then toFrench population-based cancer registries data, used as the gold standard. Performance indicators, such as sensitivity and positive predictive values, were estimated for the three algorithms in a subcohort of ESPrI. RESULTS: The third algorithm, which combined the LTI and DRG program data, presented the best sensitivities (90.9%-100%) and positive predictive values (58.1%-95.2%) according to cancer sites. The majority of false positives were in fact nearby organ sites (e.g., stomach for esophagus) and carcinoma in situ. Most false negatives were probably due to under declaration of LTI. CONCLUSION: Validated algorithms using data from the SNDS can be used for passive epidemiological follow-up for some cancer sites in the ESPrI cohort.

Risk of severe infections after the introduction of biologic DMARDs in people with newly diagnosed rheumatoid arthritis: a population-based interrupted time-series analysis.

OBJECTIVES: To determine the impact of the introduction of biologic DMARDs (bDMARDs) on severe infections among people newly diagnosed with RA compared with non-RA individuals. METHODS: In this population-based retrospective cohort study using administrative data (from 1990-2015) for British Columbia, Canada, all incident RA patients diagnosed between 1995 and 2007 were identified. General population controls with no inflammatory arthritis were matched to RA patients based on age and gender, and were assigned the diagnosis date (i.e. index date) of the RA patients they were matched with. RA/controls were then divided into quarterly cohorts according to their index dates. The outcome of interest was all severe infections necessitating hospitalization or occurring during hospitalization after the index date. We calculated 8-year severe infection rates for each cohort and conducted interrupted time-series analyses to compare severe infection trends in RA/controls with index date during pre-bDMARDs (1995-2001) and post-bDMARDs (2003-2007) periods. RESULTS: A total of 60 226 and 588 499 incident RA/controls were identified. We identified 14 245 severe infections in RA, and 79 819 severe infections in controls. The 8-year severe infection rates decreased among RA/controls with increasing calendar year of index date in the pre-bDMARDs period, but increased over time only among RA, not controls, with index date in the post-bDMARDs period. The adjusted difference between the pre- and post-bDMARDs secular trends in 8-year severe infection rates was 1.85 (P = 0.001) in RA and 0.12 (P = 0.29) in non-RA. CONCLUSION: RA onset after bDMARDs introduction was associated with an elevated severe infection risk in RA patients compared with matched non-RA individuals.

Ascertaining asthma status in epidemiologic studies: a comparison between administrative health data and self-report.

BACKGROUND: Studies have suggested that agreement between administrative health data and self-report for asthma status ranges from fair to good, but few studies benefited from administrative health data over a long period. We aimed to (1) evaluate agreement between asthma status ascertained in administrative health data covering a period of 30 years and from self-report, and (2) identify determinants of agreement between the two sources. METHODS: We used administrative health data (1983-2012) from the Quebec Birth Cohort on Immunity and Health, which included 81,496 individuals born in the province of Quebec, Canada, in 1974. Additional information, including self-reported asthma, was collected by telephone interview with 1643 participants in 2012. By design, half of them had childhood asthma based on health services utilization. Results were weighted according to the inverse of the sampling probabilities. Five algorithms were applied to administrative health data (having ≥ 2 physician claims over a 1-, 2-, 3-, 5-, or 30-year interval or ≥ 1 hospitalization), to enable comparisons with previous studies. We estimated the proportion of overall agreement and Kappa, between asthma status derived from algorithms and self-reports. We used logistic regression to identify factors associated with agreement. RESULTS: Applying the five algorithms, the prevalence of asthma ranged from 49 to 55% among the 1643 participants. At interview (mean age = 37 years), 49% and 47% of participants respectively reported ever having asthma and asthma diagnosed by a physician. Proportions of agreement between administrative health data and self-report ranged from 88 to 91%, with Kappas ranging from 0.57 (95% CI: 0.52-0.63) to 0.67 (95% CI: 0.62-0.72); the highest values were obtained with the [≥ 2 physician claims over a 30-year interval or ≥ 1 hospitalization] algorithm. Having sought health services for allergic diseases other than asthma was related to lower agreement (Odds ratio = 0.41; 95% CI: 0.25-0.65 comparing ≥ 1 health services to none). CONCLUSIONS: These findings indicate good agreement between asthma status defined from administrative health data and self-report. Agreement was higher than previously observed, which may be due to the 30-year lookback window in administrative data. Our findings support using both administrative health data and self-report in population-based epidemiological studies.

A method for generating synthetic longitudinal health data.

Getting access to administrative health data for research purposes is a difficult and time-consuming process due to increasingly demanding privacy regulations. An alternative method for sharing administrative health data would be to share synthetic datasets where the records do not correspond to real individuals, but the patterns and relationships seen in the data are reproduced. This paper assesses the feasibility of generating synthetic administrative health data using a recurrent deep learning model. Our data comes from 120,000 individuals from Alberta Health's administrative health database. We assess how similar our synthetic data is to the real data using utility assessments that assess the structure and general patterns in the data as well as by recreating a specific analysis in the real data commonly applied to this type of administrative health data. We also assess the privacy risks associated with the use of this synthetic dataset. Generic utility assessments that used Hellinger distance to quantify the difference in distributions between real and synthetic datasets for event types (0.027), attributes (mean 0.0417), Markov transition matrices (order 1 mean absolute difference: 0.0896, sd: 0.159; order 2: mean Hellinger distance 0.2195, sd: 0.2724), the Hellinger distance between the joint distributions was 0.352, and the similarity of random cohorts generated from real and synthetic data had a mean Hellinger distance of 0.3 and mean Euclidean distance of 0.064, indicating small differences between the distributions in the real data and the synthetic data. By applying a realistic analysis to both real and synthetic datasets, Cox regression hazard ratios achieved a mean confidence interval overlap of 68% for adjusted hazard ratios among 5 key outcomes of interest, indicating synthetic data produces similar analytic results to real data. The privacy assessment concluded that the attribution disclosure risk associated with this synthetic dataset was substantially less than the typical 0.09 acceptable risk threshold. Based on these metrics our results show that our synthetic data is suitably similar to the real data and could be shared for research purposes thereby alleviating concerns associated with the sharing of real data in some circumstances.

Individualized Prospective Prediction of Opioid Use Disorder.

OBJECTIVE: Opioid use disorder (OUD) is a chronic relapsing disorder with a problematic pattern of opioid use, affecting nearly 27 million people worldwide. Machine learning (ML)-based prediction of OUD may lead to early detection and intervention. However, most ML prediction studies were not based on representative data sources and prospective validations, limiting their potential to predict future new cases. In the current study, we aimed to develop and prospectively validate an ML model that could predict individual OUD cases based on representative large-scale health data. METHOD: We present an ensemble machine-learning model trained on a cross-linked Canadian administrative health data set from 2014 to 2018 (n = 699,164), with validation of model-predicted OUD cases on a hold-out sample from 2014 to 2018 (n = 174,791) and prospective prediction of OUD cases on a non-overlapping sample from 2019 (n = 316,039). We used administrative records of OUD diagnosis for each subject based on International Classification of Diseases (ICD) codes. RESULTS: With 6409 OUD cases in 2019 (mean [SD], 45.34 [14.28], 3400 males), our model prospectively predicted OUD cases at a high accuracy (balanced accuracy, 86%, sensitivity, 93%; specificity 79%). In accord with prior findings, the top risk factors for OUD in this model were opioid use indicators and a history of other substance use disorders. CONCLUSION: Our study presents an individualized prospective prediction of OUD cases by applying ML to large administrative health datasets. Such prospective predictions based on ML would be essential for potential future clinical applications in the early detection of OUD.

Real word challenges in integrating electronic medical record and administrative health data for regional quality improvement in diabetes: a retrospective cross-sectional analysis.

BACKGROUND: Linked electronic medical records and administrative data have the potential to support a learning health system and data-driven quality improvement. However, data completeness and accuracy must first be assessed before their application. We evaluated the processes, feasibility, and limitations of linking electronic medical records and administrative data for the purpose of quality improvement within five specialist diabetes clinics in Edmonton, Alberta, a province known for its robust health data infrastructure. METHODS: We conducted a retrospective cross-sectional analysis using electronic medical record and administrative data for individuals ≥ 18 years attending the clinics between March 2017 and December 2018. Descriptive statistics were produced for demographics, service use, diabetes type, and standard diabetes benchmarks. The systematic and iterative process of obtaining results is described. RESULTS: The process of integrating electronic medical record with administrative data for quality improvement was found to be non-linear and iterative and involved four phases: project planning, information generating, limitations analysis, and action. After limitations analysis, questions were grouped into those that were answerable with confidence, answerable with limitations, and not answerable with available data. Factors contributing to data limitations included inaccurate data entry, coding, collation, migration and synthesis, changes in laboratory reporting, and information not captured in existing databases. CONCLUSION: Electronic medical records and administrative databases can be powerful tools to establish clinical practice patterns, inform data-driven quality improvement at a regional level, and support a learning health system. However, there are substantial data limitations that must be addressed before these sources can be reliably leveraged.

Has the incidence of total joint arthroplasty in rheumatoid arthritis decreased in the era of biologics use? A population-based cohort study.

OBJECTIVES: To determine whether the introduction of biological DMARDs (bDMARDs) was associated with reduced incidences of total hip and knee arthroplasty (THA/TKA) among patients with RA compared with OA. METHODS: Using a population-based cohort in British Columbia, Canada, RA and OA patients diagnosed between 1995 and 2007 were divided into semi-annual cohorts according to diagnosis date. For each cohort, we calculated 8-year incidence rates of THA and TKA. We compared levels and trends of THA/TKA incidence in RA/OA patients diagnosed during pre-bDMARDs (1995-2001) and post-bDMARDs (2003-2007) periods using interrupted time-series analysis, adjusting for baseline characteristics. Adjusted 8-year total joint arthroplasty incidence estimated for RA/OA cohorts diagnosed five years after bDMARDs introduction were compared with expected rates assuming no bDMARDs introduction, based on extrapolation of pre-bDMARDs trends. RESULTS: We identified 60 227 RA and 288 260 OA incident cases. For cohorts diagnosed pre-bDMARDs, 8-year THA/TKA incidence rates increased over time in both RA and OA. For cohorts diagnosed post-bDMARDs, these rates decreased over time in RA but continued to increase for OA. For RA, differences between the post- and pre-bDMARDs secular trends in incidence rates were -0.49 (P = 0.002) for THA and -0.36 (P = 0.003) for TKA, compared with +0.40 (P = 0.006) and +0.54 (P < 0.001), respectively, for OA. For RA cohorts diagnosed five years after bDMARDs introduction, 8-year incidences were 26.9% and 12.6% lower for THA and TKA, respectively, than expected rates. In contrast, corresponding rates in OA were higher by 11.7% and 16.6%, respectively. CONCLUSION: Arthritis onset after bDMARDs introduction is associated with a significant reduction in THA/TKA incidence in RA, but not in OA. The reduction reflects a significant improvement in RA treatment during the biological era.

Hospitalizations During the COVID-19 Pandemic Among Recently Homeless Individuals: a Retrospective Population-Based Matched Cohort Study.

BACKGROUND: Hospitalizations fell precipitously among the general population during the COVID-19 pandemic. It remains unclear whether individuals experiencing homelessness experienced similar reductions. OBJECTIVE: To examine how overall and cause-specific hospitalizations changed among individuals with a recent history of homelessness (IRHH) and their housed counterparts during the first wave of the COVID-19 pandemic, using corresponding weeks in 2019 as a historical control. DESIGN: Population-based cohort study conducted in Ontario, Canada, between September 30, 2018, and September 26, 2020. PARTICIPANTS: In total, 38,617 IRHH, 15,022,368 housed individuals, and 186,858 low-income housed individuals matched on age, sex, rurality, and comorbidity burden. MAIN MEASURES: Primary outcomes included medical-surgical, non-elective (overall and cause-specific), elective surgical, and psychiatric hospital admissions. KEY RESULTS: Average rates of medical-surgical (rate ratio: 3.8, 95% CI: 3.7-3.8), non-elective (10.3, 95% CI: 10.1-10.4), and psychiatric admissions (128.1, 95% CI: 126.1-130.1) between January and September 2020 were substantially higher among IRHH compared to housed individuals. During the peak period (March 17 to June 16, 2020), rates of medical-surgical (0.47, 95% CI: 0.47-0.47), non-elective (0.80, 95% CI: 0.79-0.80), and psychiatric admissions (0.86, 95% CI: 0.84-0.88) were significantly lower among housed individuals relative to equivalent weeks in 2019. No significant changes were observed among IRHH. During the re-opening period (June 17-September 26, 2020), rates of non-elective hospitalizations for liver disease (1.41, 95% CI: 1.23-1.69), kidney disease (1.29, 95% CI: 1.14-1.47), and trauma (1.19, 95% CI: 1.07-1.32) increased substantially among IRHH but not housed individuals. Distinct hospitalization patterns were observed among IRHH even in comparison with more medically and socially vulnerable matched housed individuals. CONCLUSIONS: Persistence in overall hospital admissions and increases in non-elective hospitalizations for liver disease, kidney disease, and trauma indicate that the COVID-19 pandemic presented unique challenges for recently homeless individuals. Health systems must better address the needs of this population during public health crises.

Association between Bacillus Calmette-Guerin vaccination and type 1 diabetes in adolescence: A population-based birth cohort study in Quebec, Canada.

The Bacillus Calmette-Guerin (BCG) vaccine could reduce the incidence of type 1 diabetes through non-specific immunomodulation. Previous epidemiological studies, presenting some limitations, report no association. We examined this association of early life BCG vaccination and age at vaccination with type 1 diabetes incidence in adolescence in a large representative cohort in Quebec. The cohort included 387,704 individuals born in Quebec between 1970 and 1974 whose BCG vaccination status was determined from a provincial registry. Individuals were followed up from 1985 to their 19th birthday (maximum to 1993) for their use of physician services. Individuals were defined as type 1 diabetes cases if they had ≥4 related physician claims over a 2-year period, with at least 30 days between two claims. Cox proportional hazards regression was used to estimate the association of BCG vaccination and age at vaccination with type 1 diabetes. Covariates were selected based on a directed acyclic graph. Interaction according to sex was evaluated. A total of 178,133 (45.9%) individuals were vaccinated and 442 (0.11%) incident cases of type 1 diabetes were identified. The risk of type 1 diabetes was similar in vaccinated compared with unvaccinated individuals (adjusted hazard ratio = 1.06 [95% CI: 0.88-1.29]). There was no association with age at vaccination, and results did not differ by sex (Interaction, p = 0.60). Our results suggest that BCG vaccination does not prevent type 1 diabetes in adolescence.

Personalized prediction of incident hospitalization for cardiovascular disease in patients with hypertension using machine learning.

BACKGROUND: Prognostic information for patients with hypertension is largely based on population averages. The purpose of this study was to compare the performance of four machine learning approaches for personalized prediction of incident hospitalization for cardiovascular disease among newly diagnosed hypertensive patients. METHODS: Using province-wide linked administrative health data in Alberta, we analyzed a cohort of 259,873 newly-diagnosed hypertensive patients from 2009 to 2015 who collectively had 11,863 incident hospitalizations for heart failure, myocardial infarction, and stroke. Linear multi-task logistic regression, neural multi-task logistic regression, random survival forest and Cox proportional hazard models were used to determine the number of event-free survivors at each time-point and to construct individual event-free survival probability curves. The predictive performance was evaluated by root mean squared error, mean absolute error, concordance index, and the Brier score. RESULTS: The random survival forest model has the lowest root mean squared error value at 33.94 and lowest mean absolute error value at 28.37. Machine learning methods provide similar discrimination and calibration in the personalized survival prediction of hospitalizations for cardiovascular events in patients with hypertension. Neural multi-task logistic regression model has the highest concordance index at 0.8149 and lowest Brier score at 0.0242 for the personalized survival prediction. CONCLUSIONS: This is the first personalized survival prediction for cardiovascular diseases among hypertensive patients using administrative data. The four models tested in this analysis exhibited a similar discrimination and calibration ability in predicting personalized survival prediction of hypertension patients.

The impact of lookback windows on the prevalence and incidence of chronic diseases among people living with HIV: an exploration in administrative health data in Canada.

BACKGROUND: We described the impact of different lengths of lookback window (LW), a retrospective time period to observe diagnoses in administrative data, on the prevalence and incidence of eight chronic diseases. METHODS: Our study populations included people living with HIV (N = 5151) and 1:5 age-sex-matched HIV-negative individuals (N = 25,755) in British Columbia, Canada, with complete follow-up between 1996 and 2012. We measured period prevalence and incidence of diseases in 2012 using LWs ranging from 1 to 16 years. Cases were deemed prevalent if identified in 2012 or within a defined LW, and incident if newly identified in 2012 with no previous cases detected within a defined LW. Chronic disease cases were ascertained using published case-finding algorithms applied to population-based provincial administrative health datasets. RESULTS: Overall, using cases identified by the full 16-year LW as the reference, LWs ≥8 years and ≥ 4 years reduced the proportion of misclassified prevalent and incidence cases of most diseases to < 20%, respectively. The impact of LWs varied across diseases and populations. CONCLUSIONS: This study underscored the importance of carefully choosing LWs and demonstrated data-driven approaches that may inform these choices. To improve comparability of prevalence and incidence estimates across different settings, we recommend transparent reporting of the rationale and limitations of chosen LWs.

The effect of disease co-occurrence measurement on multimorbidity networks: a population-based study.

BACKGROUND: Network analysis, a technique for describing relationships, can provide insights into patterns of co-occurring chronic health conditions. The effect that co-occurrence measurement has on disease network structure and resulting inferences has not been well studied. The purpose of the study was to compare structural differences among multimorbidity networks constructed using different co-occurrence measures. METHODS: A retrospective cohort study was conducted using four fiscal years of administrative health data (2015/16 - 2018/19) from the province of Manitoba, Canada (population 1.5 million). Chronic conditions were identified using diagnosis codes from electronic records of physician visits, surgeries, and inpatient hospitalizations, and grouped into categories using the Johns Hopkins Adjusted Clinical Group (ACG) System. Pairwise disease networks were separately constructed using each of seven co-occurrence measures: lift, relative risk, phi, Jaccard, cosine, Kulczynski, and joint prevalence. Centrality analysis was limited to the top 20 central nodes, with degree centrality used to identify potentially influential chronic conditions. Community detection was used to identify disease clusters. Similarities in community structure between networks was measured using the adjusted Rand index (ARI). Network edges were described using disease prevalence categorized as low (< 1%), moderate (1 to < 7%), and high (≥7%). Network complexity was measured using network density and frequencies of nodes and edges. RESULTS: Relative risk and lift highlighted co-occurrences between pairs of low prevalence health conditions. Kulczynski emphasized relationships between high and low prevalence conditions. Joint prevalence focused on highly-prevalent conditions. Phi, Jaccard, and cosine emphasized associations involving moderately prevalent conditions. Co-occurrence measurement differences significantly affected the number and structure of identified disease clusters. When limiting the number of edges to produce visually interpretable graphs, networks had significant dissimilarity in the percentage of co-occurrence relationships in common, and in their selection of the highest-degree nodes. CONCLUSIONS: Multimorbidity network analyses are sensitive to disease co-occurrence measurement. Co-occurrence measures should be selected considering their intrinsic properties, research objectives, and the health condition prevalence relationships of greatest interest. Researchers should consider conducting sensitivity analyses using different co-occurrence measures.

Effectiveness and safety of gonadotropins used in female infertility: a population-based study in the Lazio region, Italy.

PURPOSE: Infertility is a topic of growing interest, and female infertility is often treated with gonadotropins. Evidence regarding comparative safety and efficacy of different gonadotropin formulations is available from clinical studies, while real-world data are missing. The present study aims to investigate effectiveness and safety of treatment with different gonadotropin formulations in women undergoing medically assisted procreation treatments in Latium, a region in central Italy, through a real-world data approach. METHODS: A retrospective population-based cohort study in women between the ages of 18 and 45 years who were prescribed with at least one gonadotropin between 2007 and 2019 was conducted. Women were enrolled from the regional drug dispense registry, and data on their clinical history, exposure to therapeutic cycles (based on recombinant "REC" or extractives "EXT" gonadotropin, or combined protocol "CMD" (REC + EXT)), and maternal/infantile outcomes were linked from the regional healthcare administrative databases. Multivariate logistic regression models were applied to estimate the association between exposure and outcomes. RESULTS: Overall, 90,292 therapeutic cycles prescribed to 35,899 women were linked to pregnancies. Overall, 15.8% of cycles successfully led to pregnancy. Compared to extractives, recombinant and combined treatments showed a stronger association with conception rate (RRREC adj = 1.06, 95% CI: 1.01-1.12; RRCBD adj = 1.17, 95% CI: 1.11-1.24). Maternal outcomes occurred in less than 5% of deliveries, and no significant differences between treatments were observed (REC vs EXT, pre-eclampsia: RR adj = 1.24, 95% CI: 0.86-1.79, ovarian hyperstimulation syndrome: RR adj = 1.25, 95% CI: 0.59-2.65, gestational diabetes: RR adj = 1.06, 95% CI: 0.84-1.35). Regarding infantile outcomes, similar results were obtained for different gonadotropin formulations (REC vs EXT: low birth weight: RR adj = 0.98, 95% CI: 0.83-1.26, multiple births: RR adj = 1.06, 95% CI: 0.92-1.23, preterm birth: RR adj = 1.03, 95% CI: 0.92-1.26). CONCLUSIONS: Efficacy and safety profiles of REC proved to be similar to those of EXT. Regarding the efficacy in terms of conception rate and birth rate, protocols using the combined approach performed slightly better. Outcomes related to maternal and infantile safety were generally very rare, and safety features were overlapping between gonadotropin formulations.

Medical Care Among Individuals with a Concussion in Ontario: A Population-based Study.

BACKGROUND: Concussion affects 1.2% of the population annually; rural regions and children have higher rates of concussion. METHODS: Using administrative health care linked databases, all residents of Ontario with a physician diagnosed concussion were identified using ICD-9 code 850 or ICD-10 code S06. Cases were tracked for 2 years for concussion-related health care utilization with relevant specialist physicians (i.e., neurology, otolaryngology, physiatry, psychiatry, ophthalmology). Billing codes, specialist codes, and time from index to visit were analyzed. Factors associated with increased specialist visits were also examined. RESULTS: In total, 1,022,588 cases were identified between 2008 and 2014 with 2 years of post-concussion health care utilization available. Follow-up by physician within 3 days of injury occurred in only 14% of cases. Mean time between ED diagnosis and follow-up by a physician was 83.9 days, whereas for rural regions it was >100 days. About half of adults (51.9%) and children (50.3%) had at least 1 specialist visit following concussion. Mean time between injury and first specialist visit was 203.8 (SD 192.9) days for adults, 213.5 (SD 201.0) days for rural adults, and 276.0 (SD 202.6) days for children. There were 67,420 neurology visits, 70,404 psychiatry visits, 13,571 neurosurgery visits, 19,780 physiatry visits, 101,788 ENT visits, and 103,417 ophthalmology visits in the 2 years tracking period. Factors associated with more specialist use included age > 18 years, urban residence, and pre-injury psychiatric history. CONCLUSIONS: There are discrepancies in post-concussion health care utilization based on age group and rural/urban residence. Addressing these risk factors could improve concussion care access.

Control charts for chronic disease surveillance: testing algorithm sensitivity to changes in data coding.

BACKGROUND: Algorithms used to identify disease cases in administrative health data may be sensitive to changes in the data over time. Control charts can be used to assess how variations in administrative health data impact the stability of estimated trends in incidence and prevalence for administrative data algorithms. We compared the stability of incidence and prevalence trends for multiple juvenile diabetes algorithms using observed-expected control charts. METHODS: Eighteen validated algorithms for juvenile diabetes were applied to administrative health data from Manitoba, Canada between 1975 and 2018. Trends in disease incidence and prevalence for each algorithm were modelled using negative binomial regression and generalized estimating equations; model-predicted case counts were plotted against observed counts. Control limits were set as predicted case count ±0.8*standard deviation. Differences in the frequency of out-of-control observations for each algorithm were assessed using McNemar's test with Holm-Bonferroni adjustment. RESULTS: The proportion of out-of-control observations for incidence and prevalence ranged from 0.57 to 0.76 and 0.45 to 0.83, respectively. McNemar's test revealed no difference in the frequency of out-of-control observations across algorithms. A sensitivity analysis with relaxed control limits (2*standard deviation) detected fewer out-of-control years (incidence 0.19 to 0.33; prevalence 0.07 to 0.52), but differences in stability across some algorithms for prevalence. CONCLUSIONS: Our study using control charts to compare stability of trends in incidence and prevalence for juvenile diabetes algorithms found no differences for disease incidence. Differences were observed between select algorithms for disease prevalence when using wider control limits.

Detection of adverse drug events in e-prescribing and administrative health data: a validation study.

BACKGROUND: Administrative health data are increasingly used to detect adverse drug events (ADEs). However, the few studies evaluating diagnostic codes for ADE detection demonstrated low sensitivity, likely due to narrow code sets, physician under-recognition of ADEs, and underreporting in administrative data. The objective of this study was to determine if combining an expanded ICD code set in administrative data with e-prescribing data improves ADE detection. METHODS: We conducted a prospective cohort study among patients newly prescribed antidepressant or antihypertensive medication in primary care and followed for 2 months. Gold standard ADEs were defined as patient-reported symptoms adjudicated as medication-related by a clinical expert. Potential ADEs in administrative data were defined as physician, ED, or hospital visits during follow-up for known adverse effects of the study medication, as identified by ICD codes. Potential ADEs in e-prescribing data were defined as study drug discontinuations or dose changes made during follow-up for safety or effectiveness reasons. RESULTS: Of 688 study participants, 445 (64.7%) were female and mean age was 64.2 (SD 13.9). The study drug for 386 (56.1%) patients was an antihypertensive, and for 302 (43.9%) an antidepressant. Using the gold standard definition, 114 (16.6%) patients experienced an ADE, with 40 (10.4%) among antihypertensive users and 74 (24.5%) among antidepressant users. The sensitivity of the expanded ICD code set was 7.0%, of e-prescribing data 9.7%, and of the two combined 14.0%. Specificities were high (86.0-95.0%). The sensitivity of the combined approach increased to 25.8% when analysis was restricted to the 27% of patients who indicated having reported symptoms to a physician. CONCLUSION: Combining an expanded diagnostic code set with e-prescribing data improves ADE detection. As few patients report symptoms to their physician, higher detection rates may be achieved by collecting patient-reported outcomes via emerging digital technologies such as patient portals and mHealth applications.

Incidence of acute myocardial infarction in the workforce: Findings from the Occupational Disease Surveillance System.

BACKGROUND: Increased risks of acute myocardial infarction (AMI) may be attributable to the workplace, however, associations are not well-established. Using the Occupational Disease Surveillance System (ODSS), we sought to estimate associations between occupation and industry of employment and AMI risk among workers in Ontario, Canada. METHODS: The study population was derived by linking provincial accepted lost-time workers' compensation claims data, to inpatient hospitalization records. Workers aged 15-65 years with an accepted non-AMI compensation claim were followed for an AMI event between 2007 and 2016. Adjusted Cox proportional hazard models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for each industry and occupation group, compared to all other workers in the cohort. Sex-stratified analyses were also performed. RESULTS: In all, 24,514 incident cases of AMI were identified among 1,502,072 Ontario workers. Increased incidence rates of AMI were found across forestry (HR 1.37, 95% CI 1.19-1.58) and wood processing (HR 1.50, 1.27-1.77) job-titles. Elevated rates were also detected within industries and occupations both broadly related to mining and quarrying (HR 1.52, 1.17-1.97), trucking (HR 1.32, 1.27-1.38), construction (HR 1.32, 1.14-1.54), and the manufacturing and processing of metal (HR 1.41, 1.19-1.68), textile (HR 1.41, 1.07-1.88), non-metallic mineral (HR 1.30, 0.82-2.07), and rubber and plastic (HR 1.42, 1.27-1.60) products. Female food service workers also had elevated AMI rates (HR 1.36, 1.23-1.51). CONCLUSION: This study found occupational variation in AMI incidence. Future studies should examine work-related hazards possibly contributing to such excess risks, like noise, vibration, occupational physical activity, shift work, and chemical and particulate exposures.

Value of the Rare Disease Registry of the Italian Region Friuli Venezia Giulia.

BACKGROUND: The lack of epidemiological and clinical data is a major obstacle in health service planning for rare diseases. Patient registries are examples of real-world data that may fill the information gap. OBJECTIVE: We describe the Rare Disease Registry of the Friuli Venezia Giulia region of Italy and its potential for research and health planning. METHODS: The Rare Disease Registry data were linked with information on mortality, hospital discharges, ambulatory care, and drug prescriptions contained in administrative databases. All information is anonymous, and data linkage was based on a stochastic key univocal for each patient. Average annual costs owing to hospitalizations, outpatient care, and medications were estimated. RESULTS: Implementation of the Registry started in 2010, and 4250 participants were registered up to 2017. A total of 2696 patients were living in the region as of January 1, 2017. The overall raw prevalence of rare diseases was 22 per 10,000 inhabitants, with higher prevalence in the pediatric population. The most common disease groups were congenital malformations, chromosomal and genetic syndromes, and circulatory and nervous diseases. In 2017, 30 patients died, 648 were hospitalized, and 2355 received some type of ambulatory care. The total annual estimated cost was approximately €6.5 million, with great variability in the average patient cost across diseases. CONCLUSIONS: The possibility of following the detailed real-world care experience of patients with each specific rare disease and assessing the costs related to each step in their care path represents a unique opportunity to identify inefficiencies, optimize care, and reduce waste of resources.

Measuring colorectal cancer incidence: the performance of an algorithm using administrative health data.

BACKGROUND: Certain cancer case ascertainment methods used in Quebec and elsewhere are known to underestimate the burden of cancer, particularly for some subgroups. Algorithms using claims data are a low-cost option to improve the quality of cancer surveillance, but have not frequently been implemented at the population-level. Our objectives were to 1) develop a colorectal cancer (CRC) case ascertainment algorithm using population-level hospitalization and physician billing data, 2) validate the algorithm, and 3) describe the characteristics of cases. METHODS: We linked physician billing, hospitalization, and tumor registry data for 2,013,430 Montreal residents age 20+ (2000-2010). We compared the performance of three algorithms based on diagnosis and treatment codes from different data sources. We described identified cases according to age, sex, socioeconomic status, treatment patterns, site distribution, and time trends. All statistical tests were two-sided. RESULTS: Our algorithm based on diagnosis and treatment codes identified 11,476 of the 12,933 incident CRC cases contained in the tumor registry as well as 2317 newly-captured cases. Our cases share similar overall time trends and site distributions to existing data, which increases our confidence in the algorithm. Our algorithm captured proportionally 35% more individuals age 50 and younger among CRC cases: 8.2% vs. 5.3%. The newly captured cases were also more likely to be living in socioeconomically advantaged areas. CONCLUSIONS: Our algorithm provides a more complete picture of population-wide CRC incidence than existing case ascertainment methods. It could be used to estimate long-term incidence trends, aid in timely surveillance, and to inform interventions, in both Quebec and other jurisdictions.