Fig. 1
The contingency tables used to assess the significance of the (conditional) prevalence of a diagnostic term in a cohort of interest. a The prevalence of a term d in cohort X as compared to the background cohort (BG).[Yes, Yes] indicates the number of records in cohort X that contain term d, [Yes, No] indicates the number of records that are not in cohort X but contain term d, [No, Yes] indicates the number of records in cohort X that do not contain term d, and [No, No] indicates the number of records that are not in cohort X and do not contain term d. These entries are computed using the following four statistics obtained from query results: The number of records in the background cohort (NBG), the number of records in cohort X (NX), the frequency of term in the background cohort (fBG(d)) and the frequency of term d in cohort X (fX(d)). b The conditional prevalence of a term d in the intimate partner violence (IPV) cohort (IPV ∩ X) as compared to cohort X.[Yes, Yes] indicates the number of records in the cohort of records that contain both X and IPV, and also contain term d, [Yes, No] indicates the number of records that are in cohort X and contain term d, but are not in the IPV cohort.[No, Yes] indicates the number of records that contain both X and IPV but do not contain term d, and [No, No] indicates the number of records that are in cohort X, do not contain term d, and are not in the IPV cohort. These entries are computed using the following four statistics obtained from query results: Number of records in cohort X (NX), the frequency of term d in cohort X (fX(d)), the number of records in cohort IPV ∩ X, i.e., those that contain both IPV and X (NIPV∩X), the frequency of term d in cohort IPV ∩ X (fIPV∩X(d))