Study design bias | Transmission bias | Social desirability/recall bias | Selection bias | Barrier effects | Popularity bias | |
---|---|---|---|---|---|---|
Cote D'Ivoire | Assessment status: Assessed, low risk of bias Mode of adjustment: No adjustments made | Assessment status: Assessed, high risk of bias Mode of adjustment: They included confidante abortions that respondents were less certain about if abortion methods were reported. They also imputed confidantes with similar sociodemographic characteristics for confidants who reported zero confidantes and imputed the probability they had obtained abortions in previous years. Thereafter they constructed post-stratification weights to ensure confidante characteristics matched respondents | Assessment status: Assessed, low risk of bias Mode of adjustment: Not adjusted for | Assessment status: Assessed, high risk of bias Mode of adjustment: They included the characteristics of respondents who reported zero confidantes in the confidante one sample and those who reported zero or one confidante in the confidante two sample and applied post-stratification weights to each sample | Assessment status: Assessed, high risk of bias Mode of adjustment: The adjusted for “missing” confidantes by assuming that each respondent who reported zero confidantes has one confidante who shared their characteristics (i.e., assuming perfect homophily). Next, the relationships between reported confidante abortions and other respondent and confidante characteristics are used to predict the probability that each “missing” confidante had an abortion | Not assessed |
Ethiopia | Assessment status: Assessed, low risk of bias Mode of adjustment: No adjustments made | Assessment status: Assessed, high risk of bias Mode of adjustment: They calculated a transmission bias adjustment factor that is the inverse of the proportion of respondents who told their confidantes about induced abortion. Then they then apply the confidante-specific adjustment factors to inflate the reported abortions in the past year separately among confidantes. To get the one-year transmission bias-adjusted abortion incidence estimates among all confidantes, they summed the inflated number of abortions by each confidante, divided by the total number of confidantes, and multiplied this estimate by 1000 | Assessment status: Assessed, moderate risk of bias Mode of adjustment: Not adjusted for | Assessment status: Assessed, high risk of bias Mode of adjustment: They used multivariable logistic regression to create post-stratification weights so that the confidante samples were nationally representative based on available characteristics | Assessment status: Assessed, high risk of bias Mode of adjustment: Not adjusted for | Not assessed |
Ghana | Assessment status: Assessed, low risk of bias Mode of adjustment: No adjustment made | Assessment status: Assessed, high risk of bias Mode of adjustment: Based on an assumption of reciprocity, between respondents and confidantes, they asked respondents who reported an abortion in the past 3 years whether they had disclosed it to each of their confidantes. They calculated the proportion of respondents who disclosed to each confidante, took the inverse of this proportion as the transmission bias factor, computed an average of all three factors (weighted on the number of confidantes) and applied it to the incidence rate adjusted for missing confidantes | Assessment status: Assessed, high risk of bias Mode of adjustment: Given considerable telescoping to the past year and some attrition over time in both approaches, they present direct report and confidante estimates as annualized rates over the past 3 years (mid-2015 to mid-2018) | Assessment status: Assessed, unclear risk of bias Mode of adjustment: They predicted the likelihood of recent abortion for the missing confidante of each confidante-less respondent, using Poisson regression with the respondent’s sociodemographic characteristics as covariates. They then calculated the confidante abortion rates including the missing confidantes’ data | Assessment status: Assessed, high risk of bias Mode of adjustment: They predicted the likelihood of recent abortion for the missing confidante of each confidante-less respondent, using Poisson regression with the respondent’s sociodemographic characteristics as covariates. They then calculated the confidante abortion rates including the missing confidantes’ data | Assessment status: Assessed, Uncertain risk of bias Mode of adjustment: Not adjusted for in analysis |
Java, Indonesia | Assessment status: Assessed, moderate risk of bias Mode of adjustment: Not adjusted for | Assessment status: Assessed, high risk of bias Mode of adjustment: They calculated a visibility factor that is the inverse of the proportion of respondents who reported disclosing their own abortion(s) to each confidante. They then applied this visibility factor to the confidante-specific abortion rates to adjust for women’s imperfect knowledge of their confidantes’ abortions | Assessment status: Assessed, moderate risk of bias Mode of adjustment: Not adjusted for | Assessment status: Assessed, high risk of bias Mode of adjustment: They recalculated the community-based survey sample weights so that their respondent sample would more closely match that of the Indonesian Demographic and Health Survey (IDHS). As with the respondents, we calculated adjusted sample weights for the pooled confidantes to match the IDHS sample | Assessment status: Assessed, high risk of bias Mode of adjustment: Not adjusted for | Not assessed |
Nigeria | Assessment status: Assessed, low risk of bias Mode of adjustment: No adjustments made | Assessment status: Assessed, high risk of bias Mode of adjustment: They included confidante abortions that respondents were less certain about if abortion methods were reported. They also imputed confidantes with similar sociodemographic characteristics for confidants who reported zero confidantes and imputed the probability they had obtained abortions in previous years. Thereafter they constructed post-stratification weights to ensure confidante characteristics matched respondents | Assessment status: Assessed, low risk of bias Mode of adjustment: Not adjusted for | Assessment status: Assessed, high risk of bias Mode of adjustment: They included the characteristics of respondents who reported zero confidantes in the confidante one sample and those who reported zero or one confidante in the confidante two sample and applied post-stratification weights to each sample | Assessment status: Assessed, high risk of bias Mode of adjustment: The adjusted for “missing” confidantes by assuming that each respondent who reported zero confidantes has one confidante who shared their characteristics (i.e., assuming perfect homophily). Next, the relationships between reported confidante abortions and other respondent and confidante characteristics are used to predict the probability that each “missing” confidante had an abortion | Not assessed |
Rajasthan, India | Assessment status: Assessed, low risk of bias Mode of adjustment: No adjustments made | Assessment status: Assessed, high risk of bias Mode of adjustment: They included confidante abortions that respondents were less certain about if abortion methods were reported. They also imputed confidantes with similar sociodemographic characteristics for confidants who reported zero confidantes and imputed the probability they had obtained abortions in previous years. Thereafter they constructed post-stratification weights to ensure confidante characteristics matched respondents | Assessment status: Assessed, moderate risk of bias Mode of adjustment: Not adjusted for | Assessment status: Assessed, high risk of bias Mode of adjustment: They included the characteristics of respondents who reported zero confidantes in the confidante one sample and those who reported zero or one confidante in the confidante two sample and applied post-stratification weights to each sample | Assessment status: Assessed, high risk of bias Mode of adjustment: The adjusted for “missing” confidantes by assuming that each respondent who reported zero confidantes has one confidante who shared their characteristics (i.e., assuming perfect homophily). Next, the relationships between reported confidante abortions and other respondent and confidante characteristics are used to predict the probability that each “missing” confidante had an abortion | Not assessed |
Uganda | Assessment status: Assessed, low risk of bias Mode of adjustment: No adjustments made | Assessment status: Assessed, high risk of bias Mode of adjustment: They calculated a transmission bias adjustment factor that is the inverse of the proportion of respondents who told their confidantes about induced abortion. Then they then apply the confidante-specific adjustment factors to inflate the reported abortions in the past year separately among confidantes. To get the one-year transmission bias-adjusted abortion incidence estimates among all confidantes, they summed the inflated number of abortions by each confidante, divided by the total number of confidantes, and multiplied this estimate by 1000 | Assessment status: Assessed, low risk of bias Mode of adjustment: Not adjusted for | Assessment status: Assessed, high risk of bias Mode of adjustment: They used multivariable logistic regression to create post-stratification weights so that the confidante samples were nationally representative based on available characteristics | Assessment status: Assessed, high risk of bias Mode of adjustment: Not adjusted for | Not assessed |