Prevalence estimation by joint use of big data and health survey: a demonstration study using electronic health records in New York city

Table 2 Simulation studies: square root of MSE of four methods

True Population Prevalence		Squared Root of MSE
Prevalence (p₁) based on outcome used in health survey (Y₁)	Prevalence (p₂) based on outcome used in EHR (Y₂)	Health Survey (n₁ = 500)	Post-stratified EHR	Mosteller estimator	Subject-level imputation model
0.3	0.30	0.021	0.002	0.015	0.019
0.3	0.31	0.021	0.009	0.017	0.019
0.3	0.32	0.022	0.019	0.018	0.021
0.3	0.33	0.021	0.029	0.020	0.021
0.3	0.35	0.021	0.049	0.023	0.021

Square root of MSE for estimating p₁ is shown. The size of health survey (n₁) and the size of subjects linked between two sources (n₁₂) are both 500. For each row, the best performing method in each row is highlighted in bold

ISSN: 1471-2288