Background
Methods
Literature search
Study population and design
HPV DNA detection and genotyping
Statistical analysis
Results
Literature search
Type of study | Definition of persistence | Duration of persistent infection | VE analysis method | Calculation of infection rates |
---|---|---|---|---|
- 2 consecutive measurements: positive preceded by a negative measurement [14] | - Conditional exact for comparing two independent Poisson rates using a binomial distribution (14,15,24–28 [39, 41–43] ,16, 35–44,17–23) - Prentice Williams Peterson total time approach [14] |
Conditional exact method for comparing two independent Poisson rates using a binomial distribution | Cox proportional hazard | GEE Poisson | Prentice Williams Peterson-Total time | |
---|---|---|---|---|
Outcome | Rate ratio | Hazard ratio | Rate ratio | Hazard ratio |
Assumption(s) | * Rate of events constant over time * Groups are considered to be equally exposed [52] | * Proportional hazard assumption (hazard ratio over time should be constant) * Independence assumption (estimate only for 1st event) [52] | * Rate of events constant over time [52] * Measurements are independent across subjects * Measurements may be correlated within subjects | * Event specific baseline hazard (baseline hazard for kth event allowed to be different) [51] |
Check assumptions in HAVANA | Assumption for constant rate over time violated among unvaccinated | Proportional hazard assumption not violated | Assumption for constant rate over time violated among unvaccinated | Assumption for event-specific hazard not violated |
Assumptions
Yrs. Since vaccination | Vaccination status | N | Vaccine types (HPV16/18) | Cross-protective types (HPV31/33/45) | ||||
---|---|---|---|---|---|---|---|---|
# infections | PR per 100 PY (95%CI) | PRR (95%CI) | # infections | PR per 100 PY (95%CI) | PRR (95%CI) | |||
2 | Unvaccinated | 551 | 2 | 0.18 (0.05–0.73) | 2 | 0.18 (0.05–0.73) | Ref | |
Vaccinated | 626 | 0 | 0.00 (0.00–0.59) | 1 | 0.08 (0.01–0.57) | 0.44 (0.04–4.85) | ||
3 | Unvaccinated | 513 | 2 | 0.19 (0.05–0.78) | 2 | 0.19 (0.05–0.78) | Ref | |
Vaccinated | 567 | 0 | 0.00 (0.00–0.65) | 2 | 0.18 (0.04–0.71) | 0.90 (0.13–6.42) | ||
4 | Unvaccinated | 472 | 7 | 0.74 (0.35–1.56) | 4 | 0.42 (0.16–1.13) | Ref | |
Vaccinated | 515 | 0 | 0.00 (0.00–0.72) | 3 | 0.29 (0.09–0.90) | 0.69 (0.15–3.074) | ||
5 | Unvaccinated | 455 | 10 | 1.10 (0.59–2.04) | 3 | 0.33 (0.11–1.02) | Ref | |
Vaccinated | 472 | 0 | 0.00 (0.00–0.78) | 1 | 0.11 (0.01–0.75) | 0.32 (0.03–3.09) | ||
6 | Unvaccinated | 447 | 18 | 2.01 (1.27–3.20) | Ref | 12 | 1.34 (0.76–2.36) | Ref |
Vaccinated | 438 | 1 | 0.11 (0.02–0.81) | 0.06 (0.01–0.42) | 2 | 0.23 (0.06–0.91) | 0.17 (0.04–0.76) | |
7 | Unvaccinated | 433 | 11 | 1.27 (0.70–2.99) | Ref | 5 | 0.58 (0.24–1.39) | |
Vaccinated | 448 | 2 | 0.22 (0.06–0.89) | 0.18 (0.04–0.79) | 0 | 0.00 (0.00–0.82) | ||
8 | Unvaccinated | 414 | 6 | 0.72 (0.33–1.61) | 9 | 1.09 (0.57–2.09) | Ref | |
Vaccinated | 429 | 0 | 0.00 (0.00–0.86) | 5 | 0.58 (0.24–1.40) | 0.54 (0.18–0.60) |
Vaccination Status | Event | Cases | Person time | PR per 100 PY | PRR per 100 PY |
---|---|---|---|---|---|
Vaccine types (HPV16/18) | |||||
Unvaccinated | 1 | 51 | 3792 | 1.35 (1.02–1.77) | Ref |
2 | 5 | 95 | 5.26 (2.19–2.64) | 3.91 (1.56–9.80) | |
3 | 0 | 17 | |||
Vaccinated | 1 | 3 | 4100 | 0.07 (0.02–0.23) | |
2 | 0 | 4 | |||
3 | 0 | 0 | |||
Cross-protective types (HPV31/33/45) | |||||
Unvaccinated | 1 | 34 | 3813 | 0.89 (0.64–1.25) | Ref |
2 | 3 | 60 | 5.00 (1.61–15.50) | 5.61 (1.72–18.26) | |
3 | 0 | 6 | |||
Vaccinated | 1 | 14 | 4082 | 0.34 (0.20–0.58) | |
2 | 0 | 60 | |||
3 | 0 | 0 |
Vaccine effectiveness
Analysis Method | Case definition | Person-time definition |
---|---|---|
Conditional exact method for comparing two independent Poisson rates using a binomial distribution | Two consecutive measurements positive for the same HPV type. The participant is counted as a case if one or more persistent infections occur. | Data for two consecutive rounds counts as 1 person-year, each additional consecutive round adds another person-year. After a missing data point counting continues. Counting stops after event or at the end of follow-up. |
Cox PH | Two consecutive measurements positive for the same HPV type. The participant is counted as a case if one or more persistent infections occur. | Data for two consecutive rounds counts as 1 person-year, each additional consecutive round adds another person-year. Person time is censored at event, loss to follow-up or end of follow-up; half-time censoring was applied. |
GEE Poisson | Two consecutive measurements positive for the same HPV type. Multiple events can occur within one participant. In our study to be counted as next infection after at least one negative round was observed. The number of infections is counted. | Data for two consecutive rounds counts as 1 person-year, each additional consecutive round adds another person-year. After a missing data point counting continues. Counting stops at the end of follow-up. |
PWP-TT | Two consecutive measurements positive for the same HPV type. Multiple events can occur within one participant, in our study to be counted as next infection; at least one negative round should be observed. The number of infections is counted. Analyses are stratified for sequential events. | Data for two consecutive rounds counts as 1 person-year. After a missing data point counting continues. Counting stops at the end of follow-up. |