The study was conducted in Chamwino district which is one of the the seven districts of Dodoma Region. The district had 63 functioning health facilities, including one district hospital, five public rural health centres, 55 government dispensaries, and two faith based dispensaries. Chamwino district had an estimated 77,429 women of reproductive age (15–49 years) in 2013, 24,242 live births in theprevious two years and a crude birth rate (CBR) of 38.5 births per 1000 population [16, 17].
Study design and population
We conducted a cross-sectional in January 2014 that interviewed recently delivered women (RDW), defined as those women who delivered within two years prior to data collection regardless of newborn outcome.
Sample size and sampling technique
We calculated the sample size using the cluster sampling formula adopted from United Nations guideline , which provided a sample size of 410 women based on the following assumption: proportions of pregnant women who prepared for birth and its complications were estimated at 86 % from Mpwapwa district Tanzania , level of significance at 95 %, Marginal error of 5 %, Design effect of 2 and by considering 10 % non-response rate.
A multi-stage cluster sampling technique was used. At each stage a sampling frame was developed and simple random sampling technique was employed. First, Chamwino district constitutes 32 wards. To determine the representative sample of Chamwino district, 1/8th of the wards were selected. Based on above calculation, four wards were selected randomly from all the 32 wards which constitute the district. From each selected ward, two villages were randomly chosen. Next, one hamlet was randomly chosen from each selected village. All households in each selected hamlet with RDW were visited in sequence and those who were eligible were enrolled. Where more than one RDW found in the same household, only one was randomly selected by using the lottery method. RDW of permanent resident of the Chamwino district, willing to participate and respond to the questionnaire were included in the study. Women who were not permanent resident, not willing to participate in the study, mentally disabled and severely ill were excluded.
Measurements of variables
Socio demographic variables
Maternal age were grouped into “<20”, “20-29”, “30-39”, “>40”. Marital status were grouped into “single”, “ married/cohabiting” and “Widow/Separated/Divorced”. Later on grouped into “living with partner” and “not living with partner”. Education status grouped into “none”, “primary” and “secondary and above” leter on into “primary and above” and “no education”. Occupation was grouped into “Employed/Self-employed” and “Not employed”. Spouse’s education status was categorized into “none”, “primary” and “secondary and above” leter on into “primary and above” and “no education”. Spouse’s occupation status was categorized into “Employed/Self-employed” and “Not employed”. Monthly income of the mother in terms of cash were grouped into Tanzanian shillings “<50,000/=”, “50,000-100,000/=” and “>100,000/=”. Parity grouped into “1”, “2-4” and “≥5”.
ANC and reproductive health variables
Gestational age at booking categorized as “1st semester” and “other”. Number of ANC visit are grouped into “≥4 visits” and “<4 visits”. Receive counseling on birth preparedness (BP) was coded as “Yes” and “No”. Received counseling on complication readiness (CR) was coded as “Yes” and “No”.
Knowledge of the key obstetric danger signs during pregnancy, childbirth and postpartum: A woman was considered knowledgeable if she spontaneously mention a total of five danger sign in all three phases with at least one in each of phase. Phase 1: Danger signs during pregnancy (vaginal bleeding, swollen hands/face and blurred vision). Phase 2: Danger signs during labour/childbirth (severe vaginal bleeding, prolonged labour (>12hours), convulsion and retained placenta). Phase 3: Danger signs during postpartum (severe vaginal bleeding, foul-smelling vaginal discharge and high fever). This method of scoring has been previously used to assess women’s knowledge on obstetric danger signs .
The woman was assessed for the presence of the following five basic components of BPCR i) knew the expected date of delivery (EDD), ii) identified a skilled birth attendant or health facility for delivery/emergency, iii) identified mode of transport for delivery and obstetric emergency, iv) Saved money, v) Identified two compatible blood donors. A woman was considered as “prepared” for birth and its complication if she reported to follow at least three of five basic components of BPCR and the rest were considered as not “prepared”. This scoring has been previously used in studies which assessed women’s level of BPCR [5, 15].
Institution delivery: A woman was considered as having institution delivery if she was assisted by skilled birth attendant with midwifery skills (Physicians, Nurses, Midwives, and Health Officers) who can manage normal deliveries and diagnose, manage or refer obstetric complications.
Data collection, management and analysis
Interviewer administered questionnaires were used during data collection due to the fact that 40 % of the population of Chamwino district are illiterate . This questionnaire was adapted from a safe motherhood questionnaire developed by the Maternal Neonatal Program of JHPIEGO and modified to fit Tanzanian context. The expert translator translated it from English version to the local language (Swahili), and then another translator did a back translation to English to check for its original meaning. It was pretested in the neighboring district of Dodoma municipal. Five research assistants who have diploma in clinical medicine were trained for 3 days, participated in the pretesting and thereafter conducted the interviews under the supervision of principal investigator. All questionnaires were counter checked for completeness and consistency of the responses before leaving the field site. Those which were not filled properly, we returned back to the respective household and filled the information correctly.
Data was analyzed using STATA version 11.2. During descriptive analysis, continuous variables were summarized using mean and standard deviation while categorical variables were summarized using proportions and then presented in tables and graphs. Bivariate analysis was then done to test for associations between the dependent variable BPCR and other independent variables using Pearson’s chi square of Fischer’s exact test where appropriate. Then, all variables which showed association at bivariate analysis at (P value < 0.2) were fitted into the multiple logistic regression model by stepwise (forward selection) method to test for the association of each with the dependent variable at 95 % confidence level. Then final model obtained includes all variable which determine the BPCR. P-value and 95 % confidence interval (CI) for odds ratios (OR) were used to confirm the significance of the associations. P-value less than 0.05 was considered significantly different. In all our analyses we used the “svy” set command in Stata to adjust for clustering effect due to complex sampling.
Ethical clearance was approved by Muhimbili University of Health and Allied Sciences (MUHAS) Research Ethics Committee. Permission was also obtained from Chamwino district executive director, wards executive officers, and village executive officers. A written informed consent has been obtained from all the respondents after being explained about all the relevant aspects of the study, including its aim, interview procedures, anticipated benefits and potential hazards. Also their right to refuse not to participate in the study any time they want was assured.