Table 1 Input parameters used in sensitivity analyses

All sensitivity analyses

Meta-analysis pooled risk ratios of CASM with HIV risk behaviors

Tables 2 and 3 (Figure S53 for secondary analysis)

Figures S27-S52

RR_XU

1.54 (ATOS¹ and RP²)

1.02 (DAP³ and RP)

[42, 43]

RR_UY

1.40 (RP and medication non-adherence)

1.71 (RP and unprotected sex, transactional sex, multiple sexual partners)

[44, 45]

Proportion of Meaningfully Strong Effects

Estimated heterogeneity (τ²)

τ² from meta-analysis statistical output

Table S1

Mean bias factor across studies

Tables 2 and 3 (single bias factor per meta-analysis assuming generalizable)

Corresponding RR_XU and RR_UY values

Proportion of heterogeneity (τ²) due to variation in confounding bias

0.80

(assumed high heterogeneity of bias across studies)

[53]

Threshold (q) for scientifically meaningfully strong effect size

1.10; 0.90

[46, 47]

Minimum proportion of constituent studies with true effects above q deemed to indicate moderate to strong evidence of an effect (r)

0.20 if 10 ≤ k ≤ 15

0.10 if k > 15

[47]

Tail

Above for q = 1.10

Below for q = 0.90

RR > 1 indicates causative effect, RR < 1 indicates protective effect