Identifying diabetes cases from administrative data: a population-based validation study

Lipscombe, Lorraine L.; Hwee, Jeremiah; Webster, Lauren; Shah, Baiju R.; Booth, Gillian L.; Tu, Karen

doi:10.1186/s12913-018-3148-0

Cited by 183 publications

(173 citation statements)

References 25 publications

(24 reference statements)

Supporting

Mentioning

165

Contrasting

Order By: Relevance

“…CCDSS methods for using administrative data to capture disease incidence have been validated for several chronic diseases, including asthma, COPD, IHD, hypertension, and diabetes. [37][38][39][40][41] The first date of disease diagnosis is used in administrative data as a proxy for disease onset. Incidence trends captured by the CCDSS can provide valuable insights into changes in Canadian population health; however, they may be influenced by changes in administrative data quality, such as modifications in data collection methods, coding and classification systems, or billing practices.…”

Section: Discussionmentioning

confidence: 99%

Trends in chronic disease incidence rates from the Canadian Chronic Disease Surveillance System

Hamm

Pelletier

Ellison

et al. 2019

Health Promot Chronic Dis Prev Can

View full text Add to dashboard Cite

Introduction The Public Health Agency of Canada’s Canadian Chronic Disease Surveillance System (CCDSS) produces population-based estimates of chronic disease prevalence and incidence using administrative health data. Our aim was to assess trends in incidence rates over time, trends are essential to understand changes in population risk and to inform policy development. Methods Incident cases of diagnosed asthma, chronic obstructive pulmonary disease (COPD), diabetes, hypertension, ischemic heart disease (IHD), and stroke were obtained from the CCDSS online infobase for 1999 to 2012. Trends in national and regional incidence estimates were tested using a negative binomial regression model with year as a linear predictor. Subsequently, models with year as a restricted cubic spline were used to test for departures from linearity using the likelihood ratio test. Age and sex were covariates in all models. Results Based on the models with year as a linear predictor, national incidence rates were estimated to have decreased over time for all diseases, except diabetes; regional incidence rates for most diseases and regions were also estimated to have decreased. However, likelihood ratio tests revealed statistically significant departures from a linear year effect for many diseases and regions, particularly for hypertension. Conclusion Chronic disease incidence estimates based on CCDSS data are decreasing over time, but not at a constant rate. Further investigations are needed to assess if this decrease is associated with changes in health status, data quality, or physician practices. As well, population characteristics that may influence changing incidence trends also require exploration.

show abstract

Section: Discussionmentioning

confidence: 99%

Trends in chronic disease incidence rates from the Canadian Chronic Disease Surveillance System

Hamm

Pelletier

Ellison

et al. 2019

Health Promot Chronic Dis Prev Can

View full text Add to dashboard Cite

show abstract

“…Because of the single‐payer universal nature of the healthcare system, our data included records of virtually all hospital and physician services for Ontario residents. We identified individuals with diabetes using the Ontario Diabetes Database, a validated registry of physician‐diagnosed diabetes derived using these claims databases . To maximize the likelihood that individuals included in the study actually had diabetes, we used a highly specific definition of diabetes based on physician claims and hospital discharge records, which has 99.2% specificity, 77.2% sensitivity and a 97.3% negative predictive value .…”

Section: Methodsmentioning

confidence: 99%

“…We identified individuals with diabetes using the Ontario Diabetes Database, a validated registry of physician‐diagnosed diabetes derived using these claims databases . To maximize the likelihood that individuals included in the study actually had diabetes, we used a highly specific definition of diabetes based on physician claims and hospital discharge records, which has 99.2% specificity, 77.2% sensitivity and a 97.3% negative predictive value . HbA 1c results came from the Ontario Laboratory Information Service, which includes information on laboratory test orders and results from community, hospital and public health laboratories across Ontario.…”

Section: Methodsmentioning

confidence: 99%

Universal drug coverage and income‐related disparities in glycaemic control

2019

View full text Add to dashboard Cite

Aims To examine whether income-related disparities in glycaemic control decline after the age of 65 years, when publicly funded universal drug insurance is acquired in Ontario, Canada. MethodsWe conducted a population-based cross-sectional study using linked administrative healthcare databases. Adults with diabetes, aged 40-89 years, with available HbA 1c data were included (N = 716 297). Income was based on median neighbourhood household income. Multiple linear regression was used to test for effect modification of age ≥65 years on the relationship between income and HbA 1c .Results There was a significant inverse association between income and HbA 1c level. After adjusting for baseline factors, the effect of income on HbA 1c level was significantly greater for individuals aged <65 years (mean difference HbA 1c for lowest vs highest income group +2.5 mmol/mol, 95% CI +2.3 to +2.7 [+0.23%, 95% CI 0.21 to 0.24]) than for those aged ≥65 years (+1.2 mmol/mol, 95% CI +1.0 to +1.3 [+0.11%, 95% CI 0.10 to 0.12]; P < 0.0001 for interaction).Conclusions Despite universal access to healthcare, people with diabetes with lower incomes had significantly worse glycaemic control compared with their counterparts on higher incomes. However, income gradients in glycaemic control were markedly reduced after the age of 65 years, possibly as a result of access to prescription drug coverage.

show abstract

“…Considering the study's goal and the validation test results, the optimal Chronic Pain Algorithm was determined to be: 1) a single encounter date recording a chronic pain-related provincial MCP procedure code in the MCP Claims File; OR 2) five or more encounter dates recording a pain-related diagnostic code in a five-year period with more than 183 days separating at least two pain-related encounter dates in the MCP Claims File. This algorithm identified 42 area under the Receiver Operating Characteristic curve (or adequate indicator of selection accuracy) [83], and 0.298(0.280-0.316 95% CI) Kappa agreement (or fair) [84].…”

Section: Administrative Data Algorithm Development and Preliminary Sementioning

confidence: 99%

“…The challenge of using health administrative datasets is that its record level data is not collected for research purposes and is often plagued with poor quality and significant variation [25,38]. This is exacerbated by chronic pain often being considered a symptom of another trauma or disease process with no objective diagnostic "gold standard" to use for validation [1,4,11,39,40], unlike other chronic diseases with standard objective diagnostic tests such as Diabetes [41,42], Multiple Sclerosis [38], and Rheumatoid Arthritis [43].…”

Section: Introductionmentioning

confidence: 99%

Identifying cases of chronic pain using health administrative data: A validation study

Foley

Knight

Ploughman

et al. 2019

Preprint

View full text Add to dashboard Cite

Background: Chronic pain is a pervasive and challenging public health issue. Prevalence estimates vary widely, both globally (2-54%) and in Canada (6.5-44%). Most estimates are derived from surveys, which are expensive and labour intensive. Health administrative data is a potential easily-accessed, low-cost method to obtain epidemiological and cost estimates of chronic pain, but its validity as a source to ascertain cases of this complex and multi-faceted condition is unknown. The purpose of this study was to derive and validate an algorithm to identify cases of chronic pain using health administrative data.Methods: A reference standard was developed and applied to the electronic medical records data of a general population sample participating in the Canadian Primary Care Sentinel Surveillance Network. Preliminary chronic pain algorithms were then created and refined through health administrative data analysis of four populations with known chronic pain diagnoses. Classification performance of the chronic pain administrative data algorithms was compared to that of the reference standard, and statistical tests of selection accuracy (sensitivity, specificity, positive predictive value, negative predictive value, likelihood ratio positive, likelihood ratio negative, diagnostic odds ratio, Kappa agreement, and area under the Receiver Operating Characteristic curve) were calculated. Results: The optimal algorithm to ascertain cases of chronic pain from health administrative data was a combination of one chronic pain clinic procedure code OR five physician claims with a pain-related diagnostic code in five years with more than 183 days separating at least two claims. The sensitivity was 0.703 (0.684-0.721 95% confidence interval), the specificity was 0.668 (0.657-0.679 95% confidence interval), and the positive predictive value was 0.408(0.398-0.418 95% confidence interval). The Chronic Pain Algorithm identified 37.6% of a Newfoundland and Labrador provincial cohort. Conclusions: Health administrative data is a valid source for information on chronic pain in Canada. The optimal chronic pain algorithm can be used to determine epidemiology and health care utilization statistics on people seeking physician care for chronic pain, and to evaluate cost effectiveness of any change to policy or health service delivery.

show abstract

Identifying diabetes cases from administrative data: a population-based validation study

Cited by 183 publications

References 25 publications

Trends in chronic disease incidence rates from the Canadian Chronic Disease Surveillance System

Trends in chronic disease incidence rates from the Canadian Chronic Disease Surveillance System

Universal drug coverage and income‐related disparities in glycaemic control

Identifying cases of chronic pain using health administrative data: A validation study

Contact Info

Product

Resources

About