Search

Abstract Background Health administrative data is increasingly being used for chronic disease surveillance. This study explored agreement between administrative and survey data for ascertainment of seven key chronic diseases, using individually linked data from a large population of individuals in Ontario, Canada. Methods All adults who completed any one of three cycles of the Canadian Community Health Survey (2001, 2003 or 2005) and agreed to have their responses linked to provincial health administrative data were included. The sample population included 85,549 persons. Previously validated case definitions for myocardial infarction, asthma, diabetes, chronic lung disease, stroke, hypertension and congestive heart failure based on hospital and physician billing codes were used to identify cases in health administrative data and these were compared with self-report of each disease from the survey. Concordance was measured using the Kappa statistic, percent positive and negative agreement and prevalence estimates. Results Agreement using the Kappa statistic was good or very good (kappa range: 0.66-0.80) for diabetes and hypertension, moderate for myocardial infarction and asthma and poor or fair (kappa range: 0.29-0.36) for stroke, congestive heart failure and COPD. Prevalence was higher in health administrative data for all diseases except stroke and myocardial infarction. Health Utilities Index scores were higher for cases identified by health administrative data compared with self-reported data for some chronic diseases (acute myocardial infarction, stroke, heart failure), suggesting that administrative data may pick up less severe cases. Conclusions In the general population, discordance between self-report and administrative data was large for many chronic diseases, particularly disease with low prevalence, and differences were not easily explained by individual and disease characteristics.

Abstract Background Computer simulation models are used increasingly to support public health research and policy, but questions about their quality persist. The purpose of this article is to review the principles and methods for validation of population-based disease simulation models. Methods We developed a comprehensive framework for validating population-based chronic disease simulation models and used this framework in a review of published model validation guidelines. Based on the review, we formulated a set of recommendations for gathering evidence of model credibility. Results Evidence of model credibility derives from examining: 1) the process of model development, 2) the performance of a model, and 3) the quality of decisions based on the model. Many important issues in model validation are insufficiently addressed by current guidelines. These issues include a detailed evaluation of different data sources, graphical representation of models, computer programming, model calibration, between-model comparisons, sensitivity analysis, and predictive validity. The role of external data in model validation depends on the purpose of the model (e.g., decision analysis versus prediction). More research is needed on the methods of comparing the quality of decisions based on different models. Conclusion As the role of simulation modeling in population health is increasing and models are becoming more complex, there is a need for further improvements in model validation methodology and common standards for evaluating model credibility.

Abstract Background Little is known about the efficiency and effectiveness of the current level of diabetes screening activity in Ontario where there is universal access to health services. Our study aims were to: (i) determine how often Ontarians are screened for diabetes; (ii) estimate screening efficiency based on the number needed to screen (NNS) to diagnosis one diabetes case; (iii) examine the population effectiveness of screening as estimated by the number of undiagnosed diabetes cases. Methods Ontario respondents of the Canadian Community Health Survey who agreed to have their responses linked to health care data (n = 37,400) provided the cohort. The five-year probabilities of glucose testing and diabetes diagnoses were estimated using a Cox Proportional Hazards Model. We defined NNS as the ratio of diabetes tests to number of diabetes diagnoses over the study period. We estimated the number of undiagnosed diabetes by dividing the number not tested at the end of study period by the NNS. Results 80% of women and 66% of men had a blood glucose test within 5 years. The efficiency of screening was estimated by a NNS of 14 among men and 22 among women. 127,100 cases of undiagnosed diabetes were estimated, representing 1.4% of the Ontario adult population. Increasing age, hypertension, immigrant and non-white ethnicity, and number of general practitioner visits were associated with an increased likelihood of having a glucose test (LR χ2 p < 0.001). Low income men were less likely to be tested. Conclusions Diabetes screening was high in this population-based cohort of Ontarians. Screening efficiency varied considerably in the population. Undiagnosed diabetes continues to be prevalent and remains concentrated in the highest risk groups for diabetes, especially among men.

Abstract Background Health administrative data are increasingly used to examine disease occurrence. However, health administrative data are typically available for a limited number of years – posing challenges for estimating disease prevalence and incidence. The objective of this study is to estimate the prevalence of people previously hospitalized with an acute myocardial infarction (AMI) using 17 years of hospital data and to create a registry of people with myocardial infarction. Methods Myocardial infarction prevalence in Ontario 2004 was estimated using four methods: 1) observed hospital admissions from 1988 to 2004; 2) observed (1988 to 2004) and extrapolated unobserved events (prior to 1988) using a "back tracing" method using Poisson models; 3) DisMod incidence-prevalence-mortality model; 4) self-reported heart disease from the population-based Canadian Community Health Survey (CCHS) in 2000/2001. Individual respondents of the CCHS were individually linked to hospital discharge records to examine the agreement between self-report and hospital AMI admission. Results 170,061 Ontario residents who were alive on March 31, 2004, and over age 20 years survived an AMI hospital admission between 1988 to 2004 (cumulative incidence 1.8%). This estimate increased to 2.03% (95% CI 2.01 to 2.05) after adding extrapolated cases that likely occurred before 1988. The estimated prevalence appeared stable with 5 to 10 years of historic hospital data. All 17 years of data were needed to create a reasonably complete registry (90% of estimated prevalent cases). The estimated prevalence using both DisMod and self-reported "heart attack" was higher (2.5% and 2.7% respectively). There was poor agreement between self-reported "heart attack" and the likelihood of having an observed AMI admission (sensitivity = 63.5%, positive predictive value = 54.3%). Conclusion Estimating myocardial infarction prevalence using a limited number of years of hospital data is feasible, and validity increases when unobserved events are added to observed events. The "back tracing" method is simple, reliable, and produces a myocardial infarction registry with high estimated "completeness" for jurisdictions with linked hospital data.

Abstract Background While research to date has consistently demonstrated that socioeconomic status (SES) is inversely associated with injury mortality in both children and adults, findings have been less consistent for non-fatal injuries. The literature addressing SES and injury morbidity among adolescents has been particularly inconclusive. To explore potential explanations for these discrepant research findings, this study uniquely compared the relationship across different measures of SES and different causes of injury (recreation versus non-recreation injuries) within a sample of Canadian adolescents. Methods The sample included adolescent participants (aged 12 to 19 years) in the Canadian 1996–1997 cross-sectional National Population Health Survey (n = 6967). Five SES measures (household income, two neighbourhood-level proxy measures, two parental indicators) were examined in relation to three injury outcomes (total, recreation, and non-recreation injuries) using multivariable logistic regression. Results Among males, a clear relationship with injury was observed only for a parental SES index, which was positively associated with total and recreation injuries (odds ratios for the highest versus lowest SES category of 1.9 for total and 2.5 for recreation injuries). Among females, there was some evidence of a positive relationship between SES and injuries, particularly for a neighbourhood-level education measure with total and recreation injuries (odds ratios of 1.7 for total and 2.0 for recreation injuries). Conclusion The results suggest that differences related to the measures of SES chosen and the causes of injury under study may both contribute to discrepancies in past research on SES and non-fatal injuries among adolescents. To clarify the potential SES-injury relationship among youth, the findings emphasize a need for a greater understanding of the meaning and relevance of different SES measures for adolescents, and for an exploration of the pathways through which SES may be related to injury risk.

Background: Computer simulation models are used increasingly to support public health research and policy, but questions about their quality persist. The purpose of this article is to review the principles and methods for validation of population-based disease simulation models. Methods: We developed a comprehensive framework for validating population-based chronic disease simulation models and used this framework in a review of published model validation guidelines. Based on the review, we formulated a set of recommendations for gathering evidence of model credibility. Results: Evidence of model credibility derives from examining: 1) the process of model development, 2) the performance of a model, and 3) the quality of decisions based on the model. Many important issues in model validation are insufficiently addressed by current guidelines. These issues include a detailed evaluation of different data sources, graphical representation of models, computer programming, model calibration, between-model comparisons, sensitivity analysis, and predictive validity. The role of external data in model validation depends on the purpose of the model (e.g., decision analysis versus prediction). More research is needed on the methods of comparing the quality of decisions based on different models. Conclusion: As the role of simulation modeling in population health is increasing and models are becoming more complex, there is a need for further improvements in model validation methodology and common standards for evaluating model credibility. [ABSTRACT FROM AUTHOR]