Validation of a case definition for depression in administrative data against primary chart data as a reference standard

Abstract

BackgroundBecause the collection of mental health information through interviews is expensive and time consuming, interest in using population-based administrative health data to conduct research on depression has increased. However, there is concern that misclassification of disease diagnosis in the underlying data might bias the results. Our objective was to determine the validity of International Classification of Disease (ICD)-9 and ICD-10 administrative health data case definitions for depression using review of family physician (FP) charts as the reference standard.MethodsTrained chart reviewers reviewed 3362 randomly selected charts from years 2001 and 2004 at 64 FP clinics in Alberta (AB) and British Columbia (BC), Canada. Depression was defined as presence of either: 1) documentation of major depressive episode, or 2) documentation of specific antidepressant medication prescription plus recorded depressed mood. The charts were linked to administrative data (hospital discharge abstracts and physician claims data) using personal health numbers. Validity indices were estimated for six administrative data definitions of depression using three years of administrative data.ResultsDepression prevalence by chart review was 15.9–19.2% depending on year, region, and province. An ICD administrative data definition of ‘2 depression claims with depression ICD codes within a one-year window OR 1 discharge abstract data (DAD) depression diagnosis’ had the highest overall validity, with estimates being 61.4% for sensitivity, 94.3% for specificity, 69.7% for positive predictive value, and 92.0% for negative predictive value. Stratification of the validity parameters for this case definition showed that sensitivity was fairly consistent across groups, however the positive predictive value was significantly higher in 2004 data compared to 2001 data (78.8 and 59.6%, respectively), and in AB data compared to BC data (79.8 and 61.7%, respectively).ConclusionsSensitivity of the case definition is often moderate, and specificity is often high, possibly due to undercoding of depression. Limitations to this study include the use of FP charts data as the reference standard, given the potential for missed or incorrect depression diagnoses. These results suggest that that administrative data can be used as a source of information for both research and surveillance purposes, while remaining aware of these limitations.

Osteoporosis Epidemiology Among Adults With Cerebral Palsy: Findings From Private and Public Administrative Claims Data

“…Previous validation studies have shown that two or more claims for a medical condition tend to improve accurate identification of that medical condition . However, accurately identifying medical conditions depends on the number of years for the study period and the medical condition examined . A single claim‐based definition for identifying a pediatric‐onset disability and osteoporosis performs better compared with other medical conditions, with positive predictive values of approximately 80% and up to 92%, respectively.…”

“…(25,44) However, accurately identifying medical conditions depends on the number of years for the study period (26) and the medical condition examined. (25,26,45) A single claim-based definition for identifying a pediatric-onset disability and osteoporosis performs better compared with other medical conditions, with positive predictive values of approximately 80% (25) and up to 92%, (26) respectively. Third, we did not account for potential confounding factors, such as ethnicity, geographic region, or other socioeconomic status variables (eg, education level).…”

Unlocking the Potential of Electronic Health Records for Health Research

“…Previous research on data quality demonstrates that there are potential biases and other issues that need to be accounted for [19][20][21][22][23], and EHR data is no exception. Thus, a researcher must consider the following factors when attempting to design a study using EHR: Documentation in EHRs should be thorough and complete, as missing or incorrect information at this stage impacts the quality of downstream data.…”

Sociodemographic and Health-Related Characteristics of a Safety-Net Patient Population

“…Our SMR estimates may be sensitive to outcome misclassification because of errors in coding claims and varying coding practices across participating hospitals. For example, a prior study [27] reported that administrative claims data have 61% sensitivity and 94% specificity for classifying depression when electronic health records are the reference standard. If outcome misclassification is precisely nondifferential (i.e.…”