Background: International Classification of Diseases codes for rheumatic heart disease (RHD) (I05-I08) include valvular heart disease of unspecified origin, limiting their usefulness for estimating RHD burden. An expert opinion-based algorithm was developed to improve the accuracy of these codes for epidemiological case ascertainment. The algorithm included unspecified diagnoses not defaulting to RHD plus selected codes pertaining to mitral valve involvement in people <60 years.
Aim: To determine the positive predictive value (PPV) of this algorithm
Methods: Chart reviews of RHD-coded admissions to three Western Australian adult tertiary hospitals (2009-2016) authenticated RHD status. We selected all cases with algorithm-positive codes from population groups at high risk of RHD and an age-stratified random sample from low-risk groups. RHD status was based on echocardiographic reports or clinical diagnosis in charts. PPVs were calculated and compared by population risk status (high/low risk), broad-age group, sex and principal/secondary diagnosis.
Results: High-risk cases (198/368=53.8%) had significantly higher PPV (83.8%) than low-risk cases (54.9%) (p=0.0012). The PPV of RHD as a principal diagnosis in the low-risk group was substantially higher than if it was an additional diagnosis (principal=84.5%; additional=44.4%) (p<0.0001) but it was not different in the high risk group (p=0.096). The PPV was highest (91.8%) for high-risk patients <35 years.
Conclusion: The PPVs of algorithm-defined cases were high for high-risk but not low-risk groups, suggesting that further research is needed before using the algorithm for high/low-risk population comparisons. The algorithm can be used for epidemiological monitoring in high-risk contexts or in people <35 years.