Effect of Village Health Team home visits and mobile phone consultations on maternal and newborn care practices in Masindi and Kiryandongo, Uganda: A community-intervention trial

Introduction: The World Health Organisation recommends home visits conducted by Community Health Workers (in Uganda known as Village Health Teams-VHTs) in order to improve maternal and newborn health. This study measured the effect of home visits combined with mobile phone consultations on maternal and newborn care practices. Method: In a community intervention trial design 16 health centres in Masindi and Kiryandongo districts, Uganda were randomly and equally allocated to one of two arms: control and intervention arms. Eight control health centres received the usual maternal and newborn educational messages offered by professional health workers and eight intervention health centres that received an intervention package for maternal care and essential newborn care practices. In the intervention arm VHTs made two prenatal and one postnatal home visit to households. VHTs were provided with mobile phones to enable them make regular telephone consultations with health workers at the health centre serving the catchment area. The primary outcome was health facility delivery. Other outcomes included antenatal attendances, birth preparedness, cord and thermal care and breastfeeding practices. Analysis was by intention-to-treat. Results: A total of 1385 pregnant women were analysed: 758 and 627 in the control and intervention arms respectively. Significant post-intervention differences were: delivery place [adjusted Odds Ratio aOR: 17.94(95%CI: 6.26-51.37); p<0.001], cord care [aOR: 3.05(95%CI: 1.81-5.12); p<0.001] thermal care [aOR: 7.58(95%CI: 2.52-22.82); p<0.001], and timely careseeking for newborn illness [aOR: 4.93(95%CI: 1.59-15.31); p = 0.006]. Conclusion: VHTs can have an effect in promoting proper cord and thermal care for the newborn and improve timely care-seeking for health facility delivery and newborn illness, because they could answer questions and refer patients correctly. However, VHTs should be supported by professional health workers through the use of mobile phones. The study was conducted in Masindi and Kiryandongo districts of Uganda from May/June 2013 to October/December 2014. At the time of conceptualising this study in 2010/2011 these two districts were one administrative unit known as Masindi district, later split to form Kiryandongo and Masindi districts as separate administrative entities. Otherwise, the ethnicity, socio-demographic characteristics and access to health care in the two districts are comparable. The combined population for the two districts is estimated at 700,000 [18]. The district health system in Uganda is structured into four-tiers. Health centre (HC) level II, III, IV and general hospital. HC II is based in a parish with an estimated service population of 20,000 inhabitants and headed by a general nurse. HC level III is located at a sub county with estimated population of 50,000 inhabitants and headed by a clinical officer. HC level IV is located in a county (average population of 150,000 inhabitants) and headed by a senior medical officer while a general hospital is located at a district (estimated population of 300,000 inhabitants) and headed by a principal medical officer. HCs of levels II and III provide mainly ambulatory care while HC IV and hospital provide inpatient services and emergency surgery. The HCs relevant for our study were levels II and III. There are 44 HCs of level II and III in this region (26 in Masindi and 18 in Kiryandongo). Eighty per cent of the population lives within a five-kilometre of the nearest Health facility. About 97% of pregnant women make at least one antenatal care consultation, but less than 50% deliver in a health facility [19]. All villages have VHTs who have been trained on community mobilisation. The guidelines for selection of VHTs and their mandate were published elsewhere [20,21]. In Uganda VHT members are selected from among members of the communities by community members themselves. VHTs could be male or female adults and their levels of education ranges from 7–11 years of schooling whereby they can read and comprehend basic communications in the English language. District health systems train VHTs to capacitate them to provide health education and mobilise communities for preventive services such as vaccination campaigns, safe water programmes and medicines distribution for deworming and treatment for malaria, diarrhoea and pneumonia under the iCCM (Integrated Community Case Management for Malaria, Pneumonia and Diarrhoea) programme. All villages in Masindi and Kiryandongo have a fair coverage of mobile telecommunication network. Regional data suggests that up to 80% of the population can access mobile telephone networks [22]. This was a randomised community intervention trial. We used a pragmatic approach whereby the intervention was integrated within the routine healthcare delivery system as opposed to highly controlled experimental design [23–25]. However, standard operating procedures were developed for health workers and VHTs (see S1 File). Random allocation was at three levels-HCs, Villages then VHTs. Blinding was not necessary since randomization was at the level of HCs. HCs that offered antenatal and delivery care were eligible for randomisation. Random allocation was conducted by writing the names of HCs, VHTs or villages on small pieces of papers which were then folded to conceal the names. Two persons, each representing a study arm, but not associated to the study were asked to randomly pick folded papers from the pool of names. The process was repeated until allocation was completed for HCs, villages and VHTs. A proportionate number of 16 public HCs were selected from Masindi (10 HCs out of 26) and Kiryandongo (six HCs out of 18) then randomly assigned to control (eight HCs) or intervention (eight HCs) arm. Ten villages were randomly selected from the catchment areas of each HC. Normally, three VHTs per village are selected and trained. However, for purposes of this study, one VHT was assigned to work in two villages since the number of women to be recruited per village was estimated to be very low at 10 women per village over a period of 6–12 months. Therefore, five VHTs were randomly selected from the ten villages. Eight VHTs, two each for four of the intervention HCs, were added when it was realised that their catchment areas had wider geographical coverage, or selected villages were found to be far apart. For example, one catchment area was identified to be across a river bank that necessitated crossing by means of a ferry. Selected VHTs were given a monthly reimbursement of 10,000 Uganda shillings (equivalent to 5 USD in 2013) for their transport as suggested by the VHT policy guideline [21]. We used the formula for cluster randomised trials suggested by Hayes and Bennett [26] to estimate the sample size of 1,388 participants from 16 clusters. This was based on 80% power to detect a 10% difference between two comparison groups with a design effect of 2.0 and Coefficient of variation between clusters, km, 0.35 considering 10% loss to follow up. The current prevalence of institutional deliveries was considered to be 23.5% [27]. Pregnant women originating from selected villages were consecutively recruited during their first antenatal visits. Based on pre-intervention visit observations, the required sample size could be realised within seven months of initiation of the study if 80% of the women making first antenatal visits were recruited. The cut-off point of 28 weeks of gestation was assessed as suitable to achieve this end. Age of pregnancy was determined by palpation. During selection, the purpose of the study was explained and a written consent secured. Biodata including village, name of the spouse, previous and current obstetric histories were also recorded. No decline to participate were registered. Eligible pregnant women who were approached readily consented to participate in the study, partly in anticipation of material benefits. In this study we did not offer financial or material benefits to pregnant women. Consented women were issued with a counter-referral form to be delivered to the responsible VHT. The health worker further made a telephone call to notify the VHT about the newly consented woman. After receiving the phone call, VHTs located the pregnant woman and her family to make appointment for the first home visit. Three home visits were planned: The first one was conducted soon after enrolment and the second made four weeks later. The third visit was scheduled within three days after delivery of the baby. We expected the last woman recruited to deliver by August 2014 (15th month of the study), however, follow-up was extended to November 2014, by three extra months. In addition to routine educational messages offered in ANC clinics, the intervention arm received a package of two closely linked components: i) VHTs making home visits to provide educational messages for maternal and newborn care ii) each VHT was equipped with a mobile phone handset capable of making unlimited phone call consultation with professional health workers in case of further clarification or advice. Health workers involved in the study were also provided with mobile phone handsets. Voice communication was preferred over short messages systems because the former had the advantage of interactive consultation between VHTs and professional health workers. No special arrangements were made for evacuating women from the health centre to hospital. However, women from both intervention and control groups could access ambulance services through the routine means of evacuation to the hospital. Forty-eight VHTs were selected (29 from Masindi and 19 from Kiryandongo) for training. Two VHT training sessions in Masindi and one in Kiryandongo were conducted consecutively each lasting five days. Training sessions had between 14–17 participants each. Smaller groups were preferred for the training to facilitate better learning. Training was based on the VHT handbook locally designed by the Ministry of Health in Uganda as a job aid for VHTs [28]. The handbook elaborates promotional information for pregnant women and essential newborn care practices. To maintain enthusiasm among trainees each training-day lasted between 9.00 am to 2.00 pm. The first day was dedicated to the problem of newborn care and its importance. On Day two, VHTs were taught how to conduct an educational session with the families. On Day three and Day four role-plays were conducted and pre-test in a real life situation using a field-based practicum was conducted. The practicum sessions were critiqued by fellow VHTs and improvements suggested. During the fifth day, VHTs were initiated into the actual study protocol with the aid of a standard operating procedure. Two professional health workers (one midwife and a nurse) from each of the intervention health centres were included in the training. Home visit sessions adopted a group-discussion approach whereby VHTs became facilitators of a discussion rather than the provider of knowledge per se [16]. In the first home visit VHTs held discussions on two topics: general care (importance of: Intermittent Presumptive Treatment for malaria (IPT), regular folic acid supplements, consistent use of bed nets, maternal diet, antenatal consultations and institutionalized delivery); danger signs in pregnancy (Vaginal bleeding, Convulsions, fever, Water loss [drainage of amniotic fluid], abdominal pains, severe headaches, blurred vision, swelling of limbs, absent or diminished foetal movement). The second visit required VHTs to discuss about birth preparation (Identifying place of delivery, identify a skilled birth attendant, organizing transport, setting aside some money, Planning for emergency evacuation, Planning with a family member) and the items needed for delivery (Clean plastic sheet for delivery, Clean dry towel for mother and baby, New razorblade, Clean threads, Pairs of gloves). VHTs explained how to care for the newborn baby by hygienic cord care, initiation of breast feeding within the first hour after birth, avoiding pre-lacteal feeds and delayed bathing for at least three days. These topics were adapted from the WHO recommendations for home visits [9] and based on formative studies [16,19]. During home visits, VHTs could make mobile phone consultations with professional health workers on issues that they considered challenging [20]. The home visit session was concluded with a recap by the VHT highlighting important areas for the family’s attention. Each session lasted 60–90 minutes. In the control group the same recruitment process as in the intervention arm was applied. Participants’ bio-data were recorded in personal files. The major difference between the two groups was that the control group did not receive follow-up visits by VHTs and VHTs in the control group were not provided with mobile phones. However, pregnant women in the control group continued to receive group education routinely offered in the health centres. Details of how antenatal care is provided in routine care has been described elsewhere [16]. Women in the control group were next visited in their homes during quantitative data collection. Data collection was limited to two periods February-March and October-December 2014 in order to maximise efficiencies of resources. Also, these are the seasons when the pastoralist study population is settled in one place. During each session of data collection ten research assistants and three supervisors were engaged. Supervisors ensured appropriate deployment of research assistants and checked to ensure all questionnaires were correctly filled. Prior to data collection, lists of recruited women were obtained from HCs and provided to the VHTs for guidance to women’s homes. Where the VHT was not sure especially in the control arm, local councils were asked for direction. At the household, a verbal consent was obtained to permit second data collection. All women approached consented to the second interview. Structured questionnaires were used to collect data on maternal practices such as antenatal attendance, birth preparation and place of delivery; newborn care practices such as tying and cutting of the cord, wrapping and bathing of the newborn, initiation of exclusive breastfeeding and care-seeking in case of newborn illness. The longest interval between delivery of the woman and data collection was 4–6 months. Women who were not found at home on first visit were followed up on the second day and two weeks later. Women were declared lost to follow-up if not found on third attempt. Health facility delivery was the primary outcome for this study. Pregnant women who delivered at home, with TBAs or along the way were grouped as home delivery and those who reported HC or hospital delivery were grouped as health facility delivery = 1. Other maternal outcomes of interest comprised of antenatal consultations and birth preparation, while newborn care practices comprised of cord care, thermal care, initiation of exclusive breastfeeding and care-seeking for newborn illness. All variables were dichotomised by building composite variables from discrete variables. Women who reported having three or more antenatal consultations were categorised as adequate and the rest were grouped as inadequate. Women were considered to be adequately prepared = 1 if they secured warm clothing for the baby, saved some money, decided on place of delivery, identified a caretaker and a means of transport to the health facility in case of labour. Care for the cord was considered to be clean = 1 if the mother reported using new strings for tying the cord, sterile instrument or new razorblade for cutting and did not apply any substance on the cord. Regarding thermal care, women were categorised as appropriate = 1 if they reported drying the baby before or immediately after expulsion of the placenta, wrapping the baby before or soon after expulsion of placenta and delayed bathing for 24 hours or more. Breastfeeding was considered appropriate = 1, if the mother reported initiating breastfeeding within six hours of delivery and did not give any pre-lacteal feeds. Care-seeking for newborn illness was considered to be timely when a mother reported seeking for care within 24 hours on recognition of symptoms of newborn illness. A newborn was considered to have complete = 1 vaccination status if they had received both Polio 0 and BCG vaccines at the time of data collection. Data were entered in EpiData version 3.1 ("The EpiData Association" Odense, Denmark) and exported to STATA version 12 (StataCorp, College Station, Texas 77845 USA) for analysis. Analysis was by intention-to-treat whereby participants were kept in the groups in which they were randomised and final analysis excluded those with ‘missing of data’ about outcomes of interest (lost to follow up, had abortions) [29]. The dataset was declared a survey data by using svy-commands in STATA. Pre-intervention household and individual characteristics were compared between control and intervention arms and their differences estimated using the chi-square test statistic. Pre-intervention differences were considered to have occurred by chance, however covariates that showed significant differences were included in multi-variable analysis. Since individuals within a cluster are likely to be correlated we used cluster level analysis to cater for intra-cluster correlation. Odds Ratios were calculated using random effects, employing the xtmelogit command in STATA because our outcomes were of binary nature. Unadjusted models were fitted with the primary outcome and treatment arms only. Covariates that showed statistical significance (p-value <0.05) were fitted into the backward regression model. The generic STATA command was: xtmelogit [primary outcome] [trial arm] X1 X2 X3…|| hc level:, covariance (unstructured) OR where, X = explanatory variables; hc level = Health Centre level; OR = Odds Ratio Only covariates that showed p-values equal ≤0.05 were left, then the Odds Ratio, confidence interval and p-value of the treatment arm were reported as adjusted values. The study was approved by the Higher Degrees and Ethics Committee of the School of Public Health, Makerere University in March 2012 and in March 2014 registered with clinicalTrials.gov: {"type":"clinical-trial","attrs":{"text":"NCT02084680","term_id":"NCT02084680"}}NCT02084680. This registration was done when recruitment of study participants had already started because the first author (RMA) became aware about the need for registration after reviewing other trials studies in the literature. Participants provided written consent at the time of enrolment and verbal consent during post-intervention evaluation thereafter. Personal files for recruited women were kept in locked cupboards which were accessible only to the research team members. The dataset for this study is also available as: S2 File. In addition to role-plays conducted during the training, three researchers with social science background periodically participated in some of the home visit sessions to observe how VHTs steered the discussions. In order to minimize the effects of an observer whereby women may be shy to participate [30], researchers were allowed to also participate in the discussions. Pregnant women were randomly selected and interviewed to ensure that the sessions were conducted as required. Where disparities were noticed, refresher trainings were organized for all VHTs and the health workers. Consequently, two refresher trainings were conducted for both VHTs and health workers in the months of October 2013 and January 2014 by the first author assisted by the two district trainers. The two-days refresher trainings focused on key maternal and newborn care practises. The same training group as the initial training were used. Health worker attrition was an important setback because it delayed the recruitment of study participants. For example, two health workers got six months’ study leave, three health workers were transferred to non-participating health centres within the district, and one got employment in another district local government. Some female staff got 60 days’ maternity leave and combined with 30 days’ annual leave making 90 working days out of duty station. Again midwives frequently attended workshops causing further delays in the recruitment process. Interestingly, we did not register attrition among VHTs among VHTs during the study period [20].