Improving pregnant women’s knowledge on danger signs and birth preparedness practices using an interactive mobile messaging alert system in Dodoma region, Tanzania: a controlled quasi experimental study

Background: Unacceptably high maternal and perinatal mortality remain a major challenge in many low income countries. Early detection and management of danger signs through improved access to maternal services is highly needed for better maternal and infant outcomes. The aim of this study was to test the effectiveness of an interactive mobile messaging alert system on improving knowledge on danger signs, birth preparedness and complication readiness practices among pregnant women in Dodoma region, Tanzania. Methods: A controlled quasi experimental study of 450 randomly selected pregnant women attending antenatal care was carried in Dodoma municipal. Participants were recruited at less than 20 weeks of gestation during the first visit where 150 were assigned to the intervention and 300 to the control group. The intervention groups was enrolled in an interactive mobile messaging system and received health education messages and were also able to send and receive individualized responses on a need basis. The control group continued receiving usual antenatal care services offered at the ANC centers. Pregnant women were followed from their initial visit to the point of delivery. Level of knowledge on danger signs and birth preparedness were assessed at baseline and a post test was again given after delivery for both groups. Analyses of covariance, linear regression were employed to test the effectiveness of the intervention. Results: The mean age of participants was 25.6 years ranging from 16 to 48 years. There was significant mean scores differences for both knowleadge and birth preparedness between the intervention and the control group after the intervention (p <.001). The mean knowleadge score was (M = 9.531,SD = 2.666 in the intervention compared to M = 6.518,SD = 4.304 in the control, equivalent to an effect size of 85% of the intervention. Meanwhile, the mean score for IBPACR was M = 4.165,SD = 1.365 for the intervention compared to M = 2.631,SD = 1.775 in the control group with an effect size of 90% A multivariate linear regression showed a positive association between the intervention (p < 0.001) and level of knowledge (B = 2.910,95%CI = 2.199-3.621) and birth preparediness (B = 1.463,95%CI = 1.185-1.740). Conclusion: The Interactive mobile messaging alert system demonstrated to be effective in increasing women's knowledge on danger signs and improving their birth preparedness practices. This study was carried out at Dodoma Municipal for both the intervention and control group from January to November, 2018. Dodoma Municipality is found in Dodoma Urban District. Dodoma Region is one of Tanzania’s 30 administrative regions and the location of the capital city of the country. It lies centrally in the eastern-central part of the country; it is about 300 miles (480 km) in the coast. Dodoma Urban District is one of the seven districts of Dodoma region. According to the 2012 Tanzania national census, the population of Dodoma Urban District was 410,956 and the area of 2576 km2 [11] . Dodoma was one of the regions with the highest maternal mortality rates in Tanzania as in 2012, Dodoma ranked the ninth high burdened region with a maternal mortality rate of 512/100,000 live births [12] . Within the municipal, there are two major public health facilities; the Makole Health Center which serves as the main antenatal care facility and the Dodoma Regional Referral Hospital which is the highest level referral hospital in the region. In this study, an interactive messaging alert system was developed and pregnant women attending antenatal care services at Makole Health Centre and Chamwino Dispensary were followed and received health education messages regarding their pregnancy. The control groups were taken from other facilities offering ANC and delivery services in Dodoma Municipal which were not selected for the intervention. This was a quasi-experimental study design with control (pre and post-test with a control group). The intervention group was enrolled in the interactive messaging alert system and received health education messages on antenatal care as per WHO guidelines, where as the control group received the normal health education being offered in the local ANC clinics. Both the intervention group and control group consisted of pregnant women who began Antenatal care at less than 20 weeks gestation. Controls were matched to the intervention group by age group, education level and gravidity (number of previous pregnancies) . The sample size for the Intervention and Control group was obtained by using the formula for comparing two independent samples (Intervention group versus control group), The values for π0 and π1 was taken from a quasi- experiment that was done in Uganda on Maternal health service utilization and newborn care; In which the proportion of women attending four visit or more at baseline was 51% and after intervention it was 63% [13]. Standard normal deviation of 1.96 at 95% confidence interval (CI) with 5% attrition rate. So, the minimum sample size was 142 plus 5% Attrition =150.The ratio of the Intervention group to control was 1:2, so controls were 300. So, the total sample size in this study was 450 pregnant women. A purposive sampling method was used to select healthcare facilities offering ANC and delivery care services in Dodoma Municipal. Purposeful sampling was employed to ensure that data came from all levels of the maternal services referral system in the region. That way, the regional hospital and the Makole Health Center were included in the study. One out of 8 dispensaries was randomly selected. Systematic random sampling method was used to select participants for both the intervention and control groups. Every third pregnant woman among those who met the criteria in a given day and agreed to participate in the study was selected until the required number of sample was obtained. In each setting, the criteria of less than 20 weeks gestation was utilized and attending ANC visit for the first time. All pregnant women who started their ANC first visit below 20 weeks, attending ANC and planning to deliver in Dodoma Municipal and who owned phones and consented to participate were considered for the study. Pregnant women who were receiving pregnancy-related health text SMS from other sources through their mobile phones and those who meet the inclusion criteria and refused to participate were excluded from the study. In this study all pregnant women who reported at the centres had mobile phone,this might be due to the nature of the study site, Dodoma urban where people are more educated and wealthier compared to rural. However the findings of this study should not be generalized in rural areas. More study need to be done in rural areas to come up with precise conclusion. Generally in Tanzania percentage of Phone ownership grows exponentially at the rate of 118%, it is therefore not uncommon to find most women especially in urban areas owning cellphones. According to Tanzania Communications Regulatory Authority, in June 2018 mobile users exceeded 41 million in Tanzania. Knowledge on obstetric and newborn danger signs was measured with 25 items. Participants were asked to mention spontaneously the key danger signs in the four phases (During pregnancy, labor/delivery, after delivery and danger sign in newborn. One point was given for a correct mentioned item with the maximum score being 25 and lowest score 0 for those who failed to mention any of the key danger sign. The key danger signs in the four phases included: Phase 1: Danger signs during pregnancy (vaginal bleeding, swollen hands/face, severe headache, blurred vision, lower abdominal pain). Phase 2: Danger signs during labor/childbirth (severe vaginal bleeding, prolonged labor (> 12 h), convulsions, difficulty in breathing and retained placenta). Phase 3: Danger signs during postpartum (severe vaginal bleeding, foul-smelling vaginal discharge, and fever). Phase 4: Danger signs in the newborn; pitched cry, difficult feeding (unable to suckle), fits (convulsions), loss of consciousness, hot to touch (hyperthermia), difficult breathing, jaundice, failure to pass urine /stool in the first 24 h [14]. ii) Individual birth preparedness and complications readiness: This was measured with nine items. The woman was asked to tell important things/supplies that she prepared for birth and for emergencies, and see whether she knew the basic steps of IBPACR i) Knowing expected date of delivery (EDD) which was confirmed in her RCH 4 card, ii) Identified a skilled birth attendant iii) Identified the mode of transport for delivery and/or for obstetric emergency, iv) Saved money v) Identified two blood donors vi) Prepare supplies for birth and emergencies vii) Prepared a person to escort her during labor or in case of emergency viii) Prepared a person to take care of the family in her absence, and ix) Identified health facility for delivery/ or for an emergency. A score of one was awarded for a correct mentioned response and a maximum score was 9 for those who mentioned all the nine basic components of IBPACR and a score of 0 for those who mentioned none. Semi-structured questionnaire (with both closed and open-ended questions) was developed to be interviewer-administered. This ensured that those unable to read and write could fully participate and also to ensure optimal capturing of all the needed information. The questionnaire included questions on socio-demographic characteristics, knowledge of key danger signs during pregnancy, childbirth, postpartum and danger signs in newborn and individual birth preparedness and complication readiness. The questionnaire was first developed in English and then translated later to Kiswahili which is the national language of Tanzania and the language used by the study population. The questionnaire was adopted and modified from Jhpiego and modified to fit the Tanzanian context [15] also from Tanzania Demographic and Health Survey 2015/2016 and from Nepal Demographic and Health Survey [16]. An interactive messaging alert system was developed and integrated into the computerized database. The application was developed in January, 2018 and moved to a server and connected to a mobile gateway with enhanced capability to handle multiple and simultaneous SMS problems from the system. Specially designed software automatically generated and sent text messages. The information required for the interactive messaging software such as gestational age and mobile phone number were gathered in the first visit and entered into system by the registering nurses with the help of the system administrator. The aim of the SMS components was to provide simple health education information on obstetric and newborn danger signs and information on individual birth preparedness and complication readiness. Text messages were sent to both expecting parents (mother and father). The content of the messages was developed by an inter-professional team of nurse midwives and obstetricians from the College of Health Sciences at the University of Dodoma. The intervention group was pretested with a baseline questionnaire and then started receiving messages. After the intervention they were again given the same questionnaire to ensure consistency as a post-test. The control group was also pre-tested and received the current standard ANC service. They also completed a post-test after delivery. The communication was two-way communication whereby participants were able to send and receive health education messages (see Fig. ​Fig.1).1). With the interactive message, a pregnant woman could send text SMS through the system and reach a doctor or nurse. The health provider could respond the message through the system and it was directed back to the pregnant women. The health education SMS messages sent to pregnant women were free of charge for the recipients. The pregnant women were only charged if they called or texted the system in which case, their cost was the standard charge by their individual local network provider. Women were allowed to ask as many questions as they wanted and all the communication were recorded to the system database to identify the patterns of messages and the frequently asked questions. The majority of the participants in the intervention group accessed services and at the end there were more compliments than complaints about the system and appreciations from mothers and their partner. The logical flow diagram which show how this SMS module for two SMS worked Message content was checked for standard and provided as simple SMS in the local language of Swahili. The constructed SMS did not exceed 480 words equivalent to three SMS, sent at different interval of time depending on the gestational age. In the first trimester one message was sent per week, in the second trimester two messages were sent per week and in the third trimester three messages were sent per week. It was done this way to avoid frequently repetition and irritations and to monitor the flow and frequencies of message. The majority of women responded at least once, with 96.5% of pregnant mother texting back for appreciation or to ask questions. In this study, data was nalyzed by using the Statistical Product for Service Solutions (SPSS) software program version 21. Before conducting the analysis the questionnaires were checked for completeness followed by error checking (data cleaning) by using frequency distribution tables to see if all the data were entered correctly. Both descriptive and inferential analyses were carried out as per the objectives of the study. Descriptive analysis was used to analyze participant’s characteristics to determine the frequencies and a percentage of their distributions between the two groups. Analysis of covariance (ANCOVA) was used to compare the mean scores between intervention and control group. Simple and multiple linear regression analysis were employed to test the predictors of change in knowledge level and birth preparedness between the intervention and control group adjusting for possible confounders. A confidence interval of 95% with the margin of error 5% (0.05) were used as statistical measure of significance (< 0.05 was regarded as significant while > 0.05 not significant). Sample characteristics and the type of analysis to be done in this study were determined by running the normality test. Findings showed that parametric methods were the recommended type for data analysis. The visual inspection of their histograms, normal P-P plots showed that the variables were approximately normally distributed. Therefore, the mean, median and range were used as measures of central tendency. The data were also tested for kurtosis and skewness and the data were within the normal range.