2.1 Study sites
The study was conducted at three health facilities (Masaka Regional Referral Hospital, Mityana Hospital and Luwero Hospital) located respectively in Masaka, Mityana, and Luwero districts of Central Uganda. These facilities were among the first to provide lifelong ART in October 2012 shortly after the Ugandan Ministry of Health rolled out the national guidelines for implementation of lifelong ART for the prevention of Mother-to-Child Transmission of HIV.
2.2 Study design
This study used data from a larger 3-year prospective cohort of 507 women who were enrolled on lifelong antiretroviral therapy during pregnancy and followed up to 18 months (between August 2013 and May 2015) to document retention in HIV care, adherence to treatment, uptake of ART, HIV positive status disclosure, partner HIV testing, sexual behaviours (including resumption of sexual intercourse), family planning use, childcare practices, and other PMTCT-related services. Detailed information on the study has been previously published [41, 43, 44].
2.3 Sample size
The larger study estimated a total enrolment of 500 HIV-positive pregnant women to address the primary objective of assessing retention in care and adherence to ART before and after delivery. The sample size was determined using the following assumptions: percent of HIV+ pregnant and lactating women enrolled for PMTCT was p=80%, a 5% level of precision, type-I error rate of 5%, a two-sided α= 0.05, and a design effect of 2 to account for the within facility clustering resulted in 492 women, which was rounded off to 500. The participants were drawn from three recruitment facilities where enrolment into the study was done concurrently, resulting in a total of 507 women. However, analysis for this study was restricted to 732 repeated observations contributed by 385 women for whom there was data on pregnancy outcomes for the first 90 days post-delivery.
2.4 Sampling procedure
On each clinic day, the attending nurses referred HIV-positive pregnant women to the study interviewer for eligibility screening and enrolment into the cohort. Study enrolment was done consecutively until the required sample size was obtained.
2.5 Data collection methods
Written informed consent was obtained from all eligible study participants who accepted to participate. A structured questionnaire with pre-coded responses was administered by a trained and experienced study interviewer who was stationed at the facility for the duration of the study period. During the six-month follow-up period, face-to-face interviews were conducted at different time intervals defined as number of days since delivery: 1-45, 46-90, 91-120, 121-150 and 151-183. These time intervals were selected based on the three phases of the postpartum period, which can be distinct but also continuous namely: the initial or acute phase which involves the first 6-12 hours after childbirth; sub-acute postpartum period which lasts 2–6 weeks, and the delayed postpartum period, which can last up to six months. For ethical reasons, interviews could not be conducted during the initial or acute phase (6-12 hours after childbirth). We selected the time interval of 1-45 days since delivery, to represent the sub-acute postpartum period within which sexual intercourse resumption is not recommended based on current Ugandan MoH guidelines. In addition, at 6 weeks (about 45 days after delivery), infants exposed to HIV are expected to obtain their first PCR test at the health facility for purposes of early infant diagnosis. The 46-90 days since delivery interval represents the delayed postpartum period which is a time of gradual restoration of muscle tone and connective tissue to the pre-pregnant state. All follow up interviews were, as much as possible, tied to the schedule of postnatal clinic visits for the women and their infants.
2.6 Data collection tools
The data collection tool used was a structured questionnaire which was translated into Luganda, the commonly spoken native language in the study area.
2.7 Study measures
At the initial visit, data were collected on women’s socio-demographic characteristics included: age, marital status, education level, religion and primary occupation. At subsequent visits, women were asked about the outcome of their latest pregnancy, HIV positive status disclosure to at least someone, currently having a sexual partner, partner HIV testing (for those who reported having sexual partners) and current contraceptive use which we defined as condom use combined with or without other contraceptive methods. To capture data on current contraceptive use we asked; are you or your spouse currently using any family planning method? Those who said, ‘yes’ to this question were additionally asked to select one or more options from a pre-coded list of contraceptive methods. We also asked about women’s desire for another child and resumption of sexual intercourse after childbirth. The primary outcome variable was the ‘time to resumption of sexual intercourse after childbirth’. To obtain data on this variable, women were asked to self-report how soon they resumed sexual intercourse after delivery: ‘Since you delivered your child, have you resumed sexual intercourse with your partner?’ Those who said, ‘yes’ to this question, were asked: how soon after delivery did you and your sexual partner resume sexual intercourse? The women had 3 options to choose from: 1) within 6 weeks; 2) after 6 weeks; 3) I don’t remember. Those who chose the third option were assumed to have resumed sexual intercourse within 6 weeks based on findings from a prior study where 68% of postnatal women who resumed sexual intercourse within 6 weeks did not attend their scheduled six-week postnatal clinic visit [10, 45].
2.7.1 Study variables by specific objective
This study used three interviewer-administered questionnaires: 1) the baseline questionnaire administered at the time of study enrolment; 2) the PNC-1st interview questionnaire administered during the first postnatal visit following childbirth and; 3) the PNC-subsequent questionnaire administered at all subsequent postnatal interviews (Table 1). The PNC-subsequent questionnaire did not include some questions from the PNC-1st postnatal interview as they did not need to be repeated, such as questions on pregnancy outcome. The English versions of these questionnaires are presented as additional files: 1, 2 and 3.
Table 1: Summary of study variables by specific objective
Specific objective
|
Questionnaire type
|
Variable name
|
1. To describe characteristics of postnatal women enrolled on lifelong antiretroviral therapy in Uganda.
|
Baseline questionnaire
|
- Enrolment health facility
- Age, marital status
- Education level
- Occupation, religion
|
PNC (1) interview
|
- HIV positive disclosure status
- Latest pregnancy outcome
- Currently having sexual partner
- Current contraceptive use
- Desire for another child
- Timing for wanting another child
|
2. To establish time to sexual intercourse resumption after child birth among women enrolled on lifelong antiretroviral therapy in Uganda.
|
PNC (1) interview
|
- Sexual intercourse resumption
- Timing of sexual resumption
Within 6 weeks…….1
After 6 weeks…..… 2
I do not remember....3
|
3. To establish factors associated with time to sexual intercourse resumption after child birth among women enrolled on lifelong antiretroviral therapy in Uganda.
|
PNC (Subsequent) questionnaire
|
- Sex intercourse resumption (dependent variable)
- All variables highlighted in specific objective 1 above.
|
2.8 Data management
All completed questionnaires were reviewed and edited onsite with support from the study coordinator. Each questionnaire was independently entered twice by two data entrants who conducted all the data capture into the computer using CSPro version 5. The data manager compared the two data entries to identify and edit any inconsistencies.
2.9 Data analysis
Analysis for this study included 732 repeated observations contributed by the 385 women for whom data were collected on pregnancy outcomes for the first 90 days since delivery. We used descriptive statistics, survival analysis and Cox proportional regression models to analyse the data. Descriptive analyses were conducted to describe women’s characteristics using frequencies and proportions, means, and standard deviation. Survival analysis was used because of the nature of our primary outcome variable, ‘time to resumption of sexual intercourse within six months after childbirth’, which is a time variable. The probability of sexual intercourse resumption was assessed by fixed covariates such as age, marital status, education at enrolment and by time-varying covariates such as HIV partner testing and disclosure of HIV positive status. Repeated measures were used at different time intervals because the nature of our data did not provide a common starting point, as women included in the analysis did not deliver on the same date. Due to the discrete nature of the timing of follow-up, data from visits were ascribed to the time of follow-up from date of delivery to self-report of first sexual intercourse. For the continuous predictors, we used a univariate Cox proportional model. We also used a life-table approach to obtain hazards of sexual intercourse resumption by categories of time from delivery to resumption. The cumulative probability of sexual intercourse resumption was assessed using Kaplan Meier curves. In order to determine independent factors associated with time-to-sexual intercourse resumption, we conducted semi-parametric Cox proportional hazards regression analyses. We obtained adjusted hazard ratios (HRs) as a measure of association with corresponding 95% confidence intervals (CIs) at 5% level of statistical significance. All data analyses were performed using STATA version 14.0 (Stata-Corp, College Station, Texas, USA).