Variables with P<0.20 in the univariate analyses were candidates for inclusion in final multivariable logistic regression models for unintended
pregnancies. When multiple covariates measured similar phenomena, the variable representing each construct with the most statistical significance was chosen. We carried out an additional analysis of interest to determine the level of happiness with the participants’ last pregnancy analysed by whether the pregnancy was intended or unintended, including all pregnancies with an a priori hypothesis that HIV status at the time of the pregnancy and ethnicity may be predictors of happiness with an unintended pregnancy. The question used to represent the level of happiness with the participants’ last pregnancy asked ‘How happy were you learn more with being pregnant the LAST time you were pregnant?’ A five-point Likert scale was used for the answer from ‘not happy at all’ to ‘very happy’ and ‘neither happy nor unhappy’ in the middle. The Cochran–Armitage test for
trend was used for the comparison of the degree of happiness with the participants’ last pregnancy based on whether it was intended or unintended. Levels of happiness according to whether or not the last pregnancy was unintended were compared among ethnic groups (African, Caribbean, European-British or French-Canadian, Aboriginal and Other). Also, univariate MDV3100 order and multivariable logistic regression models were fitted to predict happiness with the last unintended pregnancy. Only women who indicated that their last pregnancy was unintended were included in this analysis. HIV diagnosis at the time of the pregnancy and ethnicity were included as covariates of interest to assess whether they influenced happiness with unintended pregnancy.
Statistical analyses were performed using sas version 9.2 (SAS Institute, Cary, NC, USA). A total of 504 HIV-positive next women living in Ontario, Canada were recruited. Four participants did not meet the inclusion criteria (two were over the age of 52 years, and two were not living in Ontario). Fifty-nine women had never been pregnant, 13 did not answer and 12 answered ‘I don’t know’ to the question used to represent unintended pregnancy. Therefore, 416 surveys were included in the final analysis. There was a small amount of missing data for a number of survey questions, resulting in different denominators for percentages and Ns used to calculate medians. The final study sample had a median age of 38 years (IQR 33–44 years; range 18–52 years) at the time of the survey. The respondents’ last pregnancy had been a median of 8 years (IQR 3–14 years) prior to the completion of the survey (n=283 for those with data). Of the 416 women included in the study, 60% (246/411) were born outside Canada, 51% (211/416) were living in Toronto, 47% (187/400) defined themselves as being of African ethnicity and 74% (303/408) were currently on ART.