Validation of diagnosis of acute myocardial infarction and stroke in electronic medical records: a primary care cross-sectional study in Madrid, Spain (the e-MADVEVA Study)

Abstract

AbstractObjectivesTo validate the diagnoses of AMI and stroke recorded in EMR and to estimate the population prevalence of both diseases in people aged ≥ 18 yearsDesignCross-sectional validation study Setting: 45 primary care centresParticipantsSimple random sampling of diagnoses of AMI and stroke (ICPC-2 codes K75 and K90, respectively) registered by 55 physicians and random age- and sex-matched sampling of the records that included in primary care EMRs in Madrid (Spain).Primary and Secondary Outcomes MeasuresSensitivity, specificity, positive and negative predictive values, and overall agreement were calculated using the kappa statistic. Applied gold standards were electrocardiograms, brain imaging studies, hospital discharge reports, cardiology reports, and neurology reports. In the case of AMI, the ESC/ACCF/AHA/WHF Expert Consensus Document was also used. Secondary outcomes were the estimated prevalence of both diseases considering the sensitivity and specificity obtained (true prevalence).ResultsThe sensitivity of a diagnosis of AMI was 98.11% (95% CI, 96.29-99.03), and the specificity was 97.42% (95% CI, 95.44-98.55). The sensitivity of a diagnosis of stroke was 97.56% (95% CI, 95.56-98.68), and the specificity was 94.51% (95% CI, 91.96-96.28). No differences in the results were found after stratification by age and sex (both diseases). The prevalence of AMI and stroke was 1.38% and 1.27%, respectively.ConclusionThe validation results show that diagnoses of AMI and stroke in primary care EMRs constitute a helpful tool in epidemiological studies. The prevalence of AMI and stroke was lower than 2% in the population aged over 18 years.Strengths and limitations of this studyThe major strength of the e-MADVEVA study is the individual validation (manual validation) of electronic medical records by comparing each case and non-case, matched by age, with an accepted reference gold standard.The validation method allows for calculating PPV, NPV, Sensitivity, and Specificity, in contrast with other methods such as questionnaires for healthcare practitioners or patients and comparing rates in a comparable population.In-hospital mortality due to AMI and stroke may not have been recorded in primary care EMRs, with the result that the prevalence of patients who die in hospital due to a first AMI or stroke may be underestimated.The ICPC-2 codes studied for AMI (ICPC-2 K75) and stroke (ICPC-2 K90) do not allow differentiation between the different types of AMI and stroke.