This cross-sectional study uses data from the follow-up of women with HIV-I involved in the efficacy trial of clinical HIV-1 peri-exposure prophylaxis with ritonavir boosted lopinavir- (LPV/r), the ANRS12174 PROMISE- PEP trial (ClinicalTrials.gov, number NCT00640263). The ANRS12174 PROMISE- PEP trial is described in detail in previous papers . The trial conducted between 2009 and 2013, was a multi-center randomized trial including pregnant women with HIV-1, recruited at gestational age of 28–40 weeks at antenatal clinics in four African sites; Ouagadougou, Burkina Faso; East London, South Africa; Mbale, Eastern Uganda; and Lusaka, Zambia. HIV-1 infected pregnant women were referred for further assessment of inclusion criteria and again with their HIV uninfected children for enrolment at day 7 post-partum. Infants were eligible for inclusion if they were: a singleton; breastfed at day seven by their mothers; had a negative HIV-1 DNA PCR blood test and had received any Prevention of-Mother-to-Child-Transmission, PMTCT. Inclusion criteria for mother was age 18 years or older, intention to continue breastfeeding, being HIV-1 infected, and not being eligible for ART (either clinically or because CD4 count > 350 cells/µL at that time). All eligible mothers and infants followed the routine PMTCT with antepartum zidovudine (ZDV), intrapartum nevirapine (NVP), zidovudine-lamivudine (ZDV/3TC) for mothers and NVP for infants 7 days postpartum.
In Uganda, 278, seven-day old uninfected children born to HIV-1 infected women were randomized to receive infant prophylaxis (either 3TC or LPV/r daily) throughout the breastfeeding period from day 7 to 50 weeks. The primary outcome was mother to child HIV-1 transmission, diagnosed at day 7 and 6, 14, 26, 38 and 50 weeks with HIV-1 DNA PCR between 7 days and 50 weeks post-delivery. Findings from the PROMISE- PEP trial showed that infection rates, and clinical and biological severe adverse events did not differ between the two drug regimens suggesting that infant HIV-1 prophylaxis with either drug was not superior as both led to very low rates of HIV-1 postnatal transmission during 50 weeks of breastfeeding .
In 2017, 244 out of 278 mothers with HIV-1 infection and their uninfected children were eligible for re-enrollment in the follow-up study: the PROMISE-PEP Mechanism Safety study (PROMISE-PEP M&S ANRS12341). Of the 166 HIV exposed uninfected children (HEU) re-enrolled, 2 were excluded because they had contracted HIV during the follow up period. Thus, 68% of the eligible cohort of HEU children (164/244) and their HIV-1 infected mothers were followed up with 32% (n = 112) missed due to attrition. A comparison group of 199 HIV unexposed uninfected (HUU) children matched on age and sex, as well as their HIV uninfected mothers were recruited from communities located in Mbale, Eastern Uganda, which was the site for the ANRS12174 PROMISE- PEP trial. Of the 199 HUU control children, 19 were excluded due to a positive HIV-1 test result, leaving 181 HUU children and their uninfected mothers enrolled. The present cross-sectional study is based on information from interviews and clinical oral examinations of 164 HIV-1 infected- and 181 HIV-1 uninfected mothers participating in the follow-up study, the PROMISE-PEP Mechanism Safety study (PROMISE-PEP M&S ANRS12341) in 2017.
Interviews with HIV-1 infected mothers and HIV − 1 negative controls
Trained interviewers performed face-to-face interviews with mothers using semi-structured interviews in one of the local Ugandan languages, Lumasaba. The interview was constructed in English and translated into Lumasaba for use in the field. The schedule had been reviewed previously by project staff for semantic, experiental and conceptual equivalence of the source version. Sensitivity to culture and selection of appropriate words were considered [22, 23]. Mothers responded to questions about themselves and their children. Information was documented on case record forms (CRFs) and electronically with Capture software System (Clinsight) and Epidata program www.epidata.dk for the clinical oral examinations.
Socio-demographic characteristics of caretakers were assessed in terms of level of education, type of income and marital status. Level of education was categorized into ‘did not finish primary school (1), end of primary school (2) higher education’ (3). Marital status was categorized: ‘divorced (1), cohabiting/married (2), single (3), widowed’ (4) and recoded into single/divorced/widow (0) and married/cohabiting (1). Mother’s behavioral characteristics were assessed in terms of tooth brushing frequency and frequency intake of sugared snacks. Type of income was categorized into (1) no regular income (2) regular income.
Clinical oral examination
Two experienced and calibrated dental surgeons (NB and MM) performed the oral assessments among the study participants and duplicate full-mouth oral clinical examination among HIV-1 positive and HIV negative mothers not included in the main study. Dental caries was assessed on surface and tooth level (5 surfaces per tooth) in terms of decayed (D), missing (M), and filled (F) surface/teeth (DMFS/DMFT) in accordance with the World Health Organization (WHO) guidelines for field conditions . Each surface was recoded 0 for sound and 1 for caries experience and documented as decayed if it was visually cavitated with the aid of a dental mirror and periodontal probe. A surface was recorded filled when treated and a tooth was recorded missing when extracted due to caries, as confirmed by the participant. To assess gingival bleeding of the individual, the modified community periodontal index (CPI) was used. Each tooth was scored according to the presence or absence of gingival bleeding, using a periodontal probe across the gingival margins of the teeth. An individual score of presence of gingival bleeding was given if a tooth had bleeding on probing.
Maternal HIV-1 status of HIV-1 uninfected controls
Mothers in the comparison group were tested for their HIV-1 status using serial and parallel HIV rapid testing with Determine, Stat-Pak and Uni-Gold, three test algorithms as recommended by the Ugandan Ministry of Health .
STATA SE 16 (College Station, Texas 77845 USA) was used for data analysis. Cross-tabulation and Chi-square tests for categorical variables were used to assess the crude associations of covariates and outcome variables according to HIV-1 status. Site specific dental caries data has a clustered 3-level hierarchical structure with surfaces (level 1) clustered within tooth (level 2) and teeth clustered within individuals (level 3). Using site specific measures of caries as independent observations and ignoring that observations are clustered and correlated within individual will lead to an underestimation of the standard errors, too narrow confidence intervals and higher type 1 error rates. In this study we first applied ordinary logistic regression on the individual level with aggregated measures of Decayed Missed and Filled Teeth (DMFT) as dichotomized outcome variable (DMFT > 0). Socio-behavioral and clinical covariates statistically significantly associated both with HIV-1 status and DMFT were included in the ordinary logistic, and potential confounding variables. Secondly, we applied caries on all observed sites within individual. This was done using a random intercept model (RIM) (random intercepts for teeth and individuals) using mixed effect logistic regression with the melogit package within the Stata program (version SE 16) fitted for caries experience. Due to high correlation of caries on surface level within tooth, a three-level model including random intercepts for both individuals and teeth could not be fitted. Thus, we fitted a two-level model with sites within individuals. Relationships between outcome and covariates were assumed with all the cluster regression lines having a fixed slope and different intercepts. The effect of dependency of caries on sites within individual was assessed by calculating intra class correlation coefficients (ICC). The ICC expresses variations between individuals as a proportion of the total variance. ICC varies from 0, which implies that caries is independent within individuals to 1 indicating no variation of caries within individual. We applied a likelihood-ratio test to test if ICC equals 0. A statistical significance implies that the multilevel model is preferable. P-values less than 0.05 were considered statistically significant.