Effectiveness of advertising availability of prenatal ultrasound on uptake of antenatal care in rural Uganda: A cluster randomized trial

In rural Uganda pregnant women often lack access to health services, do not attend antenatal care, and tend to utilize traditional healers/birth attendants. We hypothesized that receiving a message advertising that “you will be able to see your baby by ultrasound” would motivate rural Ugandan women who otherwise might use a traditional birth attendant to attend antenatal care, and that those women would subsequently be more satisfied with care. A cluster randomized trial was conducted across eight rural sub-counties in southwestern Uganda. Subcounties were randomized to a control arm, with advertisement of antenatal care with no mention of portable obstetric ultrasound (four communities, n = 59), or an intervention arm, with advertisement of portable obstetric ultrasound. Advertisement of portable obstetric ultrasound was further divided into intervention A) word of mouth advertisement of portable obstetric ultrasound and antenatal care (one communitity, n = 16), B) radio advertisement of only antenatal care and word of mouth advertisement of antenatal care and portable obstetric ultrasound (one community, n = 7), or C) word of mouth + radio advertisement of both antenatal care and portable obstetric ultrasound (two communities, n = 75). The primary outcome was attendance to antenatal care. 159 women presented to antenatal care across eight subcounties. The rate of attendance was 65.1 (per 1000 pregnant women, 95% CI 38.3-110.4) where portable obstetric ultrasound was advertised by radio and word of mouth, as compared to a rate of 11.1 (95% CI 6.1-20.1) in control communities (rate ratio 5.9, 95% CI 2.6-13.0, p<0.0001). Attendance was also improved in women who had previously seen a traditional healer (13.0, 95% CI 5.4-31.2) compared to control (1.5, 95% CI 0.5-5.0, rate ratio 8.7, 95% CI 2.0-38.1, p = 0.004). By advertising antenatal care and portable obstetric ultrasound by radio attendance was significantly improved. This study suggests that women can be motivated to attend antenatal care when offered the concrete incentive of seeing their baby. Ethics approval was obtained for this study from Mbarara University of Science and Technology Ethics Board in Mbarara, Uganda as well as Lakeridge Health Ethics Board in Oshawa, Ontario (Canada). Approval was obtained on January 7th, 2014. All study participants were asked to provide written informed consent on a form developed in collaboration with the Mbarara University of Science and Technology (MUST) Ethics Board. Many study participants were illiterate or unable to read the consent form as it was written in English. A bilingual Ugandan midwife translated the consent form and verbally explained its contents for each study participant as they were checked into the clinic. It was clearly stated that any woman who chose to not engage in the study would still receive full ANC services. We conducted a non-blinded cluster-randomized controlled trial to evaluate the effectiveness of advertising prenatal pOBU on uptake of ANC services. This study design was selected to mitigate confounding variables and use the natural landscape of the region (described below). The trial was retrospectively registered at https://clinicaltrials.gov/ct2/show/{"type":"clinical-trial","attrs":{"text":"NCT02587091","term_id":"NCT02587091"}}NCT02587091 with registration number {"type":"clinical-trial","attrs":{"text":"NCT02587091","term_id":"NCT02587091"}}NCT02587091 and the original protocol is available at http://bridgetohealth.ca/study-protocol/. Retrospective registration was performed as our study was very low risk, and did not utilize any novel patient care. Rather, the standard of care was being provided to pregnant women in ANC clinics and the intervention was increasing access to care. There are no on-going trials and as such no further trial registries. The study was conducted from February 17th– 25th, 2014 in the Kabale District, a rural highland located in southwestern Uganda. It is roughly 560 km from the capital city of Kampala and has a population of approximately 522,000. It is divided into three regions—north, central and south, and is further subdivided into 22 sub-counties, with 11 in the north, four centrally and seven in the south (see S1 Fig). The population of this region predominantly engages in subsistence agriculture and are similar from a cultural, language and ethnic background, often times known as Abanya-Kigezi or “people of Kigezi”. There is very limited data describing the healthcare of this rural and isolated region, and government statistics are likely inaccurate. The Ugandan Ministry of Health describes that healthcare provision is a mix of public health, provided on a district health system with level one to four health centers (four being the most advanced) and private healthcare, delivered predominantly by faith-based organizations such as the Uganda Catholic Medical Bureau (UCMB).[25] Uptake of any ANC services is close to 66%, with only one third of these women returning for a fourth visit or delivering in a HCF.[26] All women who were currently aware of being pregnant and presented to ANC were eligible for inclusion into the study. There were no exclusion criteria for women in this study. Advertisement was carried out through a sequentially increasing intensity of word of mouth and/or radio messaging revolving around the central hypothesis. Twenty-two sub-counties were assessed for inclusion. Four were intentionally excluded from the study design because they are primarily composed of urban and sub-urban communities and effectively divide the northern and southern regions. From the remaining 18 sub-counties four were randomly selected from the southern wing as the control arm (n = 4) and four from the northern wing as the intervention arm (n = 4) (Fig 1). Each sub-county represented one cluster. Selection of northern versus southern sub-counties for intervention or control arms, as well as selection of which sub-counties within each region and in which order they would be visited was all selected at random. Random selection was conducted by the statisticians at MUST using a simple random number program in MS-Excel. The control sub-counties were chosen to be visited en bloc in the first four days, as opposed to alternating control then intervention on each day. This was done specifically to avoid contamination with messaging of pOBU, and was important in mitigating contamination with radio messaging in particular, which was felt to possibly cross geographic boundaries. Similarly the messages regarding pOBU were only delivered by radio on the last two days of the study in order to prevent contamination. Sunday was intentionally avoided as a study day as most study participants in the population ascribe to the Catholic faith and would not attend any clinics if held on that particular day. When consent was obtained in the control arm there was intentional masking of pOBU. Following completion of initial consent and an entrance survey all women were provided with a debriefing form explaining the presence of ultrasound, including communities where it had not been advertised, and again consent was obtained to participate in the study. Any women who did not consent were provided with complete ANC services, including pOBU. The intervention itself was advertising the availability pOBU at the cluster level (sub-county). The initial study design had only two arms as described above in randomization. However, the protocol was amended during the study due to interim analysis results demonstrating word of mouth (WOM) advertising of pOBU was not successful in increasing uptake of ANC. WOM advertising was carried out in the control arm by local community leaders and church pastors who announced the presence of the free ANC at community gatherings. In interventions A, B and C, ANC was advertised in the same way by WOM. In addition, a message regarding free pOBU was advertised by WOM. The message deliverd by WOM was not scripted for community leaders. They were given an introduction to the study and asked to tell their communities that pOBU would allow them to see a picture of their baby and that a free maternal health clinic would be offered along with the pOBU. It was hypothesized that many in the population—including community leaders—did not clearly understand what pOBU was, and therefore could not relay the message effectively. The intervention arm was subsequently modified to include radio advertising so that the message could be delivered more clearly and directly. Three nearly identical radio messages were delivered by the same Ugandan medical doctor through the same radio station over a four day period of time. The first radio message only discussed a free maternal health clinic, while the final two messages included a description of pOBU and the statement that “you will be able to see a picture of your baby”. This resulted in the following three distinct intervention arms: WOM advertisement by community leaders of both pOBU and ANC (Intervention A), WOM advertisement of pOBU and WOM + radio advertisement of ANC with no radio mention of pOBU (Intervention B), and WOM + radio advertisement of both pOBU and ANC (Intervention C). See Fig 2 for an outline of the intervention and control arms. The radio message in Intervention C included the words "You will be able to see a picture of your baby" delivered in both English and Ruchiga, the local dialect. WOM = word of mouth. rANC = radio advertisement of ANC, the radio message was, “there will be a free clinic for pregnant women at X community on X day, all tests and treatment will be provided free of charge”. rOBU = radio advertisement of portable obstetric ultrasound, the radio message was, “there will be a free clinic for pregnant women at X community on X day, all tests and treatment will be provided free of charge. Free ultrasound will also be offered at the clinic, if you come “you will be able to see a picture of your baby” and was delivered in both English and Ruchiga (the local dialect). All women attending clinic were offered pOBU delivered by a certified Ugandan ultrasound technician. Study participants rotated through an ANC clinic that was based on the WHO guidelines for high quality ANC[5] as well as the WHO four-pronged approach of elimination of mother to child transmission of HIV (EMTCT).[27] S2 Fig depicts study participant flow in detail through the clinic. The ANC clinic was held alongside a mobile medical and dental health program which served as the site of the trial on each unique clinic day, with each day representing a clinic serving a unique sub-county. The clinics were coordinated and implemented by Bridge to Health Medical and Dental, a registered Canadian charity and humanitarian organization, and Kigezi Healthcare Foundation, a Ugandan not for profit community development organization. Women received pre- and post-test counseling and rapid testing for HIV. Those who were positive were enrolled into the National HIV program and started on antiretroviral therapy as per national guidelines. Women were also offered intermittent presumptive therapy for malaria, rapid treponemal testing and treatment for syphilis, hepatitis B screening and syndromic management of sexually transmitted infections. Iron and folate supplementation was provided to all women. Any woman in active labour or with a critically low hemoglobin was transported to hospital by ambulance. Any woman with abnormal findings on ultrasound was referred for obstetric specialty follow up. All women were encouraged to deliver at a local HCF. All care was provided free of charge, with costs covered by Bridge to Health Medical and Dental. Women were interviewed on arrival and when leaving clinic using a standardized electronic entrance and exit survey (S1 File). This interview was performed with the assistance of a bilingual Ugandan midwife who translated the pre-developed survey into the local dialect, then translated responses into English. The entrance survey took approximately thirty minutes to complete and the exit survey approximately 10 minutes. Sub-counties were chosen as the unit of allocation and number of women accessing ANC in each sub-county as the unit of analysis. The primary outcome was attendance to ANC between the intervention arm and control arm. Secondary outcomes included A) the attendance by women who had previously used a TBA, B) the attendance by women who had not yet attended ANC, C) the attendance between the three interventions and D) the number of women stating they came for pOBU. The sample size calculation performed a priori was not utilized for this study as the design changed during implementation for reasons specified previously. We performed a post-hoc power calculation. We assumed that the distribution of pregnant women attending ANC was Poisson distributed. We assumed that the number of pregnant women in each sub-county was equal. We assumed that over-dispersion is not present when conducting the power calculation. Our goal was to test whether the rate of women attending ANC was equal in the control and intervention arms. Assuming a rate ratio of 2, we required 25 women attending ANC in the control arm and 50 women attending ANC in the intervention arm to achieve 80% power to reject the null hypothesis of equal rates of attendance. Demographic data was summarized using simple descriptive statistics. We computed rates of ANC uptake, first ANC use and ANC uptake due to pOBU, respectively, and all at the cluster level. The count of the number of women attending ANC, for each specific outcome, formed the numerator of the rate. We used the number of women attending first ANC in 2013–2014 through government run clinics as our denominator. The data used in our denominator calculation was provided by the District Health Office through the District HMIS database. DHS and recent 2014 national census data only present bulk district data, and do not differentiate between subcounties. First ANC visit was chosen as the denominator as it represents the total number of recorded pregnancies in each sub-county over a specific and equal period of time. The ideal denominator would have been the total number of pregnant women in each community during the time period of each clinic day. This information was not available as sub-county lines were changed following the most recent census, and sub-county level data were not reported in the 2014 census. We expressed all rates per 1,000 women. Using Poisson regression, and Pearson’s correction factor for over-dispersion, we tested whether the ratio of rates in the control and intervention arms were equal. Following our initial study design, we computed the results of the control arm versus all intervention arms combined together. We then computed all possible pair-wise rate ratios between the individual trial arms and the control arm. We chose to reject the null hypothesis of equal rates of ANC uptake at a 5% alpha level. When considering pair-wise comparison we apply the Bonferroni correction procedure, adjusting the statistical significance threshold by the number of pair-wise comparison conducted within a given model (six in our design). In this study design, this results in a Bonferroni corrected alpha level of: 0.05/6 = 0.0083. All statistical analyses were conducted using SAS version 9.4 (Cary, North Carolina). Since the outcomes reported were at the health facility (cluster) level, no further special analyses to account for clustering were conducted as per CONSORT statement guidelines. Qualitative answers recorded in the exit survey were reviewed by a single analyst. Specific words pertinent to the context of the study (i.e., the idea of the ultrasound itself) were identified. Descriptors considered to be positive and negative were recorded.