Combining population-based administrative health records and electronic medical records for disease surveillance

Consistency Between Administrative Health Records and Self-Reported Health Status and Health Care Use Among Indigenous Wayuu Health Insurance Enrollees: La Guajira, Colombia

The objective of the study was to assess the consistency between self-reported demographic characteristics, health conditions, and healthcare use, and administrative healthcare records, in a sample of enrollees of an Indigenous health organization in Colombia. We conducted a phone survey of a random sample of 2113 enrollees September-2020/February-2021. Administrative health records were obtained for the sample. Using ICD-10 diagnostic codes, we identified individuals who had healthcare visits for diabetes, hypertension, and/or pregnancy. Using unique identifiers, we linked their survey data to the administrative dataset. Agreement percentages and Cohen's Kappa coefficients were calculated. Logistic regressions were performed for each health condition/state. Results showed high degree of agreement between data sources for sex and age, similar rates for diabetes and hypertension, 10% variation for pregnancy. Kappa statistics were in the moderate range. Age was significantly associated with agreement between data sources. Sex, language, and self-rated health were significant for diabetes. This is the first study with data from an Indigenous population assessing the consistency between self-reported data and administrative health records. Survey and administrative data produced similar results, suggesting that Anas Wauu can be confident in using their data for planning and research purposes, as part of the movement toward data sovereignty.

The validity of electronic health data for measuring smoking status: a systematic review and meta-analysis

BACKGROUND

Smoking is a risk factor for many chronic diseases. Multiple smoking status ascertainment algorithms have been developed for population-based electronic health databases such as administrative databases and electronic medical records (EMRs). Evidence syntheses of algorithm validation studies have often focused on chronic diseases rather than risk factors. We conducted a systematic review and meta-analysis of smoking status ascertainment algorithms to describe the characteristics and validity of these algorithms.

METHODS

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines were followed. We searched articles published from 1990 to 2022 in EMBASE, MEDLINE, Scopus, and Web of Science with key terms such as validity, administrative data, electronic health records, smoking, and tobacco use. The extracted information, including article characteristics, algorithm characteristics, and validity measures, was descriptively analyzed. Sources of heterogeneity in validity measures were estimated using a meta-regression model. Risk of bias (ROB) in the reviewed articles was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 tool.

RESULTS

The initial search yielded 2086 articles; 57 were selected for review and 116 algorithms were identified. Almost three-quarters (71.6%) of algorithms were based on EMR data. The algorithms were primarily constructed using diagnosis codes for smoking-related conditions, although prescription medication codes for smoking treatments were also adopted. About half of the algorithms were developed using machine-learning models. The pooled estimates of positive predictive value, sensitivity, and specificity were 0.843, 0.672, and 0.918 respectively. Algorithm sensitivity and specificity were highly variable and ranged from 3 to 100% and 36 to 100%, respectively. Model-based algorithms had significantly greater sensitivity (p = 0.006) than rule-based algorithms. Algorithms for EMR data had higher sensitivity than algorithms for administrative data (p = 0.001). The ROB was low in most of the articles (76.3%) that underwent the assessment.

CONCLUSIONS

Multiple algorithms using different data sources and methods have been proposed to ascertain smoking status in electronic health data. Many algorithms had low sensitivity and positive predictive value, but the data source influenced their validity. Algorithms based on machine-learning models for multiple linked data sources have improved validity.

Data quality assessment and subsampling strategies to correct distributional bias in prevalence studies

Background

Healthcare-associated infections (HAIs) represent a major Public Health issue. Hospital-based prevalence studies are a common tool of HAI surveillance, but data quality problems and non-representativeness can undermine their reliability.

Methods

This study proposes three algorithms that, given a convenience sample and variables relevant for the outcome of the study, select a subsample with specific distributional characteristics, boosting either representativeness (Probability and Distance procedures) or risk factors’ balance (Uniformity procedure). A “Quality Score” (QS) was also developed to grade sampled units according to data completeness and reliability.

The methodologies were evaluated through bootstrapping on a convenience sample of 135 hospitals collected during the 2016 Italian Point Prevalence Survey (PPS) on HAIs.

Results

The QS highlighted wide variations in data quality among hospitals (median QS 52.9 points, range 7.98–628, lower meaning better quality), with most problems ascribable to ward and hospital-related data reporting. Both Distance and Probability procedures produced subsamples with lower distributional bias (Log-likelihood score increased from 7.3 to 29 points). The Uniformity procedure increased the homogeneity of the sample characteristics (e.g., − 58.4% in geographical variability).

The procedures selected hospitals with higher data quality, especially the Probability procedure (lower QS in 100% of bootstrap simulations). The Distance procedure produced lower HAI prevalence estimates (6.98% compared to 7.44% in the convenience sample), more in line with the European median.

Conclusions

The QS and the subsampling procedures proposed in this study could represent effective tools to improve the quality of prevalence studies, decreasing the biases that can arise due to non-probabilistic sample collection.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-021-01277-y.

Is there an agreement between self-reported medical diagnosis in the CARTaGENE cohort and the Québec administrative health databases?

BACKGROUND

Population health studies often use existing databases that are not necessarily constituted for research purposes. The question arises as to whether different data sources such as in administrative health data (AHD) and self-report questionnaires are equivalent and lead to similar information.

OBJECTIVES

The main objective of this study was to assess the level of agreement between self-reported medical conditions and medical diagnosis captured in AHD. A secondary objective was to identify predictors of agreement among medical conditions between the two data sources. Therefore, the purposes of the study were to explore the extent to which these two methods of commonly used public health data collection provide concordant records and identify the main predictors of statistical variations.

METHODS

Data were extracted from CARTaGENE, a population-based cohort in Québec, Canada, which was linked to the provincial health insurance records of the same individuals, namely the MED-ÉCHO database from the Régie de l'assurance maladie du Québec (RAMQ) and the fee-for-service billing records provided by the physician, for the time period 1998-2012. Agreement statistics (kappa coefficient) along with sensitivity, specificity and predictive positive value were calculated for 19 chronic conditions and 12 types of cancers. Logistic regressions were used to identify predictors of concordance between self-report and AHD from significant covariates (sex, age groups, education, region, income, heavy utilization of health care system and Charlson comorbidity index).

RESULTS

Agreement between self-reported data and AHD across diseases ranged from kappa of 0.09 for chronic renal failure to 0.86 for type 2 diabetes. Sensitivity of self-reported data was higher than 50% for 14 out of the 31 medical conditions studied, especially for myocardial infarction (88.62%), breast cancer (86.28%), and diabetes (85.06%). Specificity was generally high with a minimum value of 89.70%. Lower concordance between data sources was observed for higher frequency of health care utilization and higher comorbidity scores.

CONCLUSION

Overall, there was moderate agreement between the two data sources but important variations were found depending on the type of disease. This suggests that CARTaGENE's participants were generally able to correctly identify the kind of diseases they suffer from, with some exceptions. These results may help researchers choose adequate data sources according to specific study objectives. These results also suggest that Québec's AHD seem to underestimate the prevalence of some chronic conditions, which might result in inaccurate estimates of morbidity with consequences for public health surveillance.