Study design and setting
A cross-sectional study was conducted during May–June in 2017 at the adult HIV clinic of Mbarara Regional Referral Hospital (MRRH) located in Mbarara city, 275 km southwest of the capital city, Kampala. The study was done as part of an oral health status assessment of PLWH in rural south western Uganda. The MRRH HIV clinic has an enrollment of over 10,000 patients, and initiates approximately 1000 new patients on antiretroviral therapy (ART) each year. MRRH serves a catchment population of over 8,000,000 people drawn from 17 districts of rural Southwestern Uganda.
All participants provided written informed consent to participate in the study. A separate consent to review participants’ clinical records was also obtained. For participants who were not formally educated and unable to read and write, informed consent was read to them verbatim in the local language of their preference. When they agreed to participate in the study, they were asked to append their thumb print to the informed consent form and the translator signed as a witness. The Mbarara University Research Ethics Committee (MUST-REC) reviewed and approved the proposal (Reference number: 17/02-17). Participants with dental treatment needs detected by this study were referred to the dental unit at MRRH for management.
Inclusion criteria and recruitment
We selectively recruited adult patients aged 18 years, and older receiving care at MRRH HIV clinic, and either taking or initiating ART. The selection was performed following a systematic sampling approach in which every fifth patient attending the clinic who met these criteria was included in the study. If a patient declined treatment or failed to meet the inclusion criteria, the subsequent patient was considered.
We determined the sample size of 194 participants using the Krejcie and Morgan table  basing on an average of 380 clients attending the adult HIV clinic per month assuming population proportion of 0.5 and confidence interval of 95%.
An interviewer administered questionnaire with structured questions to collect data was used. Socio-demographic and clinical data were collected including age, gender, ART use and history, smoking history, alcohol use, brushing practices and dental caries using a modified World Health Organization (WHO) oral health questionnaire for adults . The questionnaire was administered by trained research assistants.
Dental examinations were carried out by a trained and calibrated dental surgeon using disposable dental mirrors and probes under natural light. First, the number of missing teeth was noted, and surfaces of all teeth present were inspected for the presence or absence of dental caries, with or without fillings; and findings recorded on a dental chart. Caries experience were further classified using the decayed, missing, filled, teeth (DMFT) index . Teeth that were traumatized or malformed, missing naturally, or extracted after trauma, existing periodontal disease or surgical intervention involving the mouth were excluded from classification. Those teeth which are filled because of dental caries were defined as having caries, whereas those restored following trauma or for cosmetic purposes were not. The respondents with toothache or dental caries were referred to the dental clinic for management, while those with missing teeth were advised to have dentures if necessary according to standard care of the Uganda Ministry of Health. All guidelines by the Uganda National Council of Science and Technology for conducting human participants research were fully adhered to.
Treatment needs assessment
We computed treatment needs by adding participants who had decayed and missing teeth and dividing by the total number of participants who had decayed, missing and filled teeth as previously used by Aleksejūnienė and Vilma Brukienė .
We conducted descriptive analyses on socio-demographic, oral health, and HIV characteristics of the cohort, then stratified by the presence or absence of dental caries. Estimated crude prevalence of dental caries and summarized DMFT index were obtained. Using fitted logistic regression models we identified correlates of the presence of dental caries. Our predictor variables of interest included socio-demographic factors (age, sex, education and income, which was categorized based on an income greater than vs less than $1/day ), alcohol and tobacco use, dental health practices (number of times meals were eaten, frequency of brushing and dental checkups), dietary habits (eating snacks between meals) and HIV medical history (ART use, current CD4 count and CD4 at the time of initiation of ART). Variables with significance of p < 0.25 in univariable models were included in multivariable models. Analyses were conducted with STATA version 13 (Statacorp, College Station, TX).