Table 3:

Bias report for temporal validation dataset (year 2020)

	N	Substance misuse prevalence	False-discovery rate (95% CI)	False-positive rate (95% CI)	False-omission rate (95% CI)	False-negative rate (95% CI)
All encounters	16917	1023	0·25 (0·23–0·29)	0·02 (0·02–0·02)	0·01 (0·01–0·01)	0·18 (0·15–0·20)
Age group, years
18–44	5336	433	0·19 (0·16–0·23)	0·02 (0·01–0·02)	0·01 (0·01–0·02)	0·16 (0·13–0·20)
≥45	11 581	590	0·30 (0·26–0·33)	0·02 (0·02–0·02)	0·01 (0·01–0·01)	0·18 (0·15–0·21)
Sex
Female	9944	315	0·29 (0·24–0·34)	0·01 (0·01–0·01)	0·01 (0·01–0·01)	0·22 (0·17–0·27)
Male	6973	708	0·25 (0·22–0·28)	0·03 (0·03–0·04)	0·02 (0·02–0·02)	0·16 (0·13–0·19)
Race and ethnicity
Non-Hispanic Black	6190	445	0·29 (0·26–0·34)	0·03 (0·02–0·03)	0·03 (0·02–0·03)	0·18 (0·15–0·22)
Non-Hispanic White	6156	353	0·22 (0·18–0·26)	0·01 (0·01–0·02)	0·01 (0·01–0·02)	0·19 (0·15–0·23)
Hispanic	3329	159	0·20 (0·14–0·26)	0·01 (0·01–0·02)	0·01 (0·01–0·02)	0·15 (0·09–0·21)
*Mixed	1242	66	0·32 (0·22–0·43)	0·02 (0·02–0·03)	0·02 (0·02–0·03)	0·14 (0·06–0·24)

Substance misuse prevalence represents any occurrence of alcohol misuse, opioid misuse, or non-opioid drug misuse. The referent labels and predicted labels for any type of substance misuse were used to calculate the number of false positives (FP), true positives (TP), false negatives (FN), and true negatives (TN). Bias assessment metrics included: false-discovery rate (FP/[FP+TP]); false-positive rate (FP/[FP+TN]); false-omission rate (FN/[FN+TN]); and false-negative rate (FN/[FN+TP]).

^*Mixed=Asian, Native-American, or Pacific Islander, other, or refused to answer or answer unknown.