Table 2. Risk of bias within studies.
First Author | Selection of participants1 | Time of exposure ascertainment2 | Loss to follow- up3 |
Outcome ascertainment4 |
Confounding5,6 | Selective outcome reporting7 |
---|---|---|---|---|---|---|
Auvert[29] | All trial participants who provided at least one HPV result (47% of trial population). Those excluded (53%) had similar baseline characteristics Low risk |
First available HPV status was used (median follow-up 2.4y/2.2y in HPV uninfected/infected). Consistent results were seen using recent HPV results. Low risk |
12% loss to follow- up at 48 weeks. Unclear risk |
Median follow-up 2.5 years. Low risk |
Adjusted for sexual behavior at baseline, no adjustment for HSV2 High risk |
No a-priori plan. No response to request for additional analyses High risk |
Averbach | 93% of cases included, exclusions | Repeated HPV testing, median 80 days | 12% loss to follow- | Median follow-up | Adjusted for prospective | Any HPV infection was the primary |
[24] | due to missing information. Controls selected from within same cohort study Low risk |
between exposure and outcome Low risk | up at completion. Low risk |
21.9 months Low risk |
measurement of sexual behavior and HSV2 Low risk |
exposure of interest. Responded to request for additional analyses Low risk |
Chin-Hong [23] |
No description of how subset (30% of EXPLORE study) were selected from participants in main study Unclear risk |
HPV assessed at baseline only, median time to outcome between 6 and 36 months8 Unclear risk |
Retention rate not documented. Unclear risk |
Median follow-up 3 years Low risk |
Adjusted for prospective sexual behavior, no adjustment for HSV2 High risk |
Primary analysis assessed the association of HPV infection with HIV acquisition No additional analysis provided Unclear risk |
Low[25] | No description of how subset (40%)of participants were selected from participants in main study Unclear risk |
HPV assessed at baseline only, median time to outcome between 4 and 24 months Unclear risk |
Retention rate not documented. Unclear risk |
Median follow-up 1.7 years Low risk |
Adjusted for HSV2, no adjustment for sexual behavior High risk |
HPV analysis was not the primary aim of the publication. Additional analysis provided Low risk |
Myer[27] | Cohort comprised of all HPV positives in main study and HPV negatives recruited over 1 year Unclear risk |
HPV assessed at baseline only, median time to outcome between 6 and 24 months Unclear risk |
25% loss to follow- up at 12months and 68% at 24 months. High risk |
Median follow-up 14.3 months Low risk |
Adjusted for baseline sexual behavior, no adjustment for HSV2 High risk |
HPV analysis was not the primary aim of the publication. Additional analysis provided Unclear risk |
Smith[30] | Cohort comprised of trial participants consenting to HPV testing (80%) Low risk |
HPV assessed at baseline only, median time to outcome between 3 and 42 months Unclear risk |
Retention rate not documented. Unclear risk |
Median follow-up 42 months for 71% Low risk |
Adjusted for HSV2, no adjustment for sexual behaviour High risk |
No a-priori analysis plan. Additional analysis provided Low risk |
Smith- McCune[26] |
Cohort comprised of trial participants consenting to HPV sub-study (98%) Low risk |
HPV assessed every 3 months. Recent infection was within 6 months excluding concurrent visit Low risk |
6% loss to follow-up. Low risk | Median follow-up 21 months Low risk |
Adjusted for prospective sexual behavior and HSV2 Low risk |
A-priori analysis plan. Additional analysis provided Low risk |
Veldhuijzen [28] |
Cohort comprised of those with HPV results from main study (92%) Low risk |
HPV assessed at baseline only, median time to outcome between 3 and 18 months Unclear risk |
12% loss to follow- up. Low risk |
Median follow-up 16.6 months Low risk |
No adjustment for confounders High risk |
No a-priori plan. No additional analysis provided Unclear risk |
In case control studies here was a high risk of bias if >90% of cases were not included (and/or exclusion was related to exposure or outcome) or controls were not selected from within the cohort study. Unclear risk when this information was not available. In cohort studies, there was low risk of bias if the cohort were representative of the average individual in the population of interest, and both the exposed and unexposed were drawn from the same population. Unclear risk if this was not the case or not documented.
Low risk if HPV was tested prospectively and/or the time between exposure and outcome was <1year. High risk if this time was ≥ 1 year and unclear risk if insufficient information was available.
Low risk if retention was ≥80% at the end of the study. Unclear risk if information not available.
High risk of bias if follow-up was not long enough for HIV to be acquired (less than 1 year). Low risk if median follow-up was at least 1 year and unclear risk if information not available.
See table 1 for full list of confounders adjusted for.
Low risk of bias if adjustment for prospective measurements of sexual behavior AND HSV2. Unclear risk if adjustment for sexual behavior OR HSV2 serology at baseline only. High risk if HSV2 OR sexual behavior not adjusted for.
If (i)authors stated there was an a priori plan or stated which outcomes were of primary interest or (ii) authors responded to a request for further information and analysis then there was a low risk of bias. Of one of these was met this was unclear risk, if none were met this was high risk
Where median time was not available in studies which tested baseline HPV only it is assumed to be between minimum follow-up HIV test frequency and total study follow-up.