Menu
Menu
Menu
Menu
Background: Although the association between the presence of maternity waiting homes (MWHs) and the personal and environmental factors that affect the use of MWHs has been explained in qualitative terms, it has never been tested in quantitative terms. The aim of this study was to test the association between the presence of MWHs and personal and environmental factors that affect the use of MWHs. Methods: A cross-sectional study was conducted using an interviewer-administered questionnaire from 1st July to 31st August, 2014 among 340 women of reproductive age in 15 rural health centres in Kalomo district, Zambia. Tests of association (chi square, logistic regression analysis, odds ratio) were conducted to determine the strength of the association between the presence of MWHs and personal and environmental factors. Differences between respondents who used MWHs and those who did not were also tested. Results: Compared to respondents from health centres without MWHs, those from centres with MWHs had higher odds of expressing willingness to use MWHs (adjusted odds ratio [aOR] = 4.58; 95% confidence interval [CI]:1.39-15.17), perceived more benefits from using a MWH (aOR = 8.63; 95% CI: 3.13-23.79), perceived more social pressure from important others to use MWH (aOR = 27.09; 95% CI: 12.23-60.03) and higher personal risk from pregnancy and childbirth related complications (aOR = 11.63; 95% CI: 2.52-53.62). Furthermore, these respondents had higher odds of staying at a health centre before delivery (aOR = 1.78; 95% CI: 1.05-3.02), giving birth at a health facility (aOR = 3.36; 95% CI: 1.85-6.12) and receiving care from a skilled birth attendant (aOR = 3.24; 95% CI: 1.80-5.84). In contrast, these respondents had lower odds of perceiving barriers regarding the use of MWHs (aOR = 0.27; 95% CI: 0.16-0.47). Factors positively associated with the use of MWHs included longer distances to the nearest health centre (p = 0.004), higher number of antenatal care (ANC) visits (p = 0.001), higher proportions of complications during ANC (p = 0.09) and women’s perception of benefits gained from staying in a MWH while waiting for delivery at the health centre (p = 0.001). Conclusion: These findings suggest a need for health interventions that focus on promoting ANC use, raising awareness about the risk and severity of pregnancy complications, promoting family and community support, and mitigating logistical barriers.
The study used cross-sectional design, and data were collected from 15 health centre catchment areas of Kalomo district, Zambia [27–29]. For details on Kalomo district profile see Sialubanje et al., [4, 5, 8, 9]. The study participants were sampled from women of childbearing age (mean age 25.60 years, SD = 6.85). Of these, 203 women (mean age = 24.69 years, SD = 6.61) were recruited from health facilities with a MWH, and 137 women (mean age = 26.94, SD = 7.01) from health facilities without a MWH. To be eligible to participate in the interview, women must have delivered in the past 12 months prior to the survey and resided in the area for more than six months. The study utilised a multi-stage convenience sampling method. All ten health centres with a MWH in the district were identified and included in the study, after which, five out of a total 25 health centres without a MWH were purposefully selected and included in the study. Fourteen villages from the fourteen rural health centres and one compound from the semi-urban health centre were randomly sampled. Since there was more than one village in each health centre catchment area, one village was purposively selected based on accessibility and advice from community health workers and headmen. The number of respondents surveyed from each village was evenly distributed. However, due to a lack of information and the unstructured nature of housing units in the area, it was not possible to select respondents using systematic sampling methods. The Tropical Disease Research Centre Ethics Review Committee and the Ministry of Health Research and Ethics Committee in Zambia provided the ethical approval for the study (study number TDRC/ERC/2OO5/29/12). Before starting the survey, research assistants read out the aims of the study to the participants. They also explained that the respondents’ names would not be written on the questionnaire or on the informed consent form. Moreover, respondents were informed that survey participation was voluntary, that they would not receive any direct benefits from the study, and that they were free to discontinue the survey at any point if they felt uncomfortable. Participants were informed that the purpose of the survey was to collect information on what they thought affected their use of MWHs, and that the information would be used to inform and guide future government policies on MWHs. Written informed consent was obtained by having the participants either sign the consent form or mark with an ‘X’. Respondents who were able to write were made to sign on the consent form, whereas those who could not write were made to mark with an ‘X’. Two trained research assistants who were supervised by the principal investigator collected the data. The research assistants were recruited from within Kalomo district and were both female, aged 22 and 25 respectively, and had a full grade 12 certificate. The research assistants received a one day face-to-face training on the study and the questionnaire. Female research assistants were preferred to male for cultural reasons and in order to ensure optimal interaction with the mothers. Moreover, to minimise information concealment from the respondents during the survey, research assistants spoke both English (the official language) and Tonga, the local language. A week before the survey, women were informed about the survey date by the village headmen and neighbourhood health committee (NHC) members. On the agreed day, the principal investigator and the research assistants travelled to the respective health centres from which the research assistants were directed into the households by the NHC members and community volunteers. Because of high illiteracy levels in the area, the questionnaire was translated into Tonga. Women who were able to read were allowed to go through the questionnaire by themselves; the interviewer merely confirmed whether the questionnaire was correctly and completely answered. All the interviews took place in the participant’s home or at a nearby convenient place-normally a quiet place under a tree, a few meters from the participant’s house. Each survey lasted between thirty to forty minutes. The questionnaire was developed by the research team based on variables described by social cognitive theories of human behaviour, including the theory of Reasoned Action Approach [24] and the Health Belief Model [25] as well as findings from our previous studies in the area [4, 8, 9]. The research instrument was first developed in English, translated to Tonga by an independent bi-lingual expert, and then back-translated to English. The final version of the questionnaire was both in Tonga and English (see supplementary file for the English version of the questionnaire). All items were answered on a 5-point Likert scale ranging from 1 = fully disagree to 5 = fully agree, or similar labels. We used factor analysis to check which items, based on theory, should measure a particular psychosocial construct combined into one factor or not. Items that showed strong internal consistency (Cronbach’s alpha > 0.6 or r > 0.40) were combined and averaged into one index. If items did not combine into one index, factor analyses were conducted using principal axis factoring as an extraction method, and an oblimin rotation. After inspection of the scree plot (that is, a plot which displays the eigenvalues associated with a component or factor in descending order versus the number of the component or factor), sum measures were created with Eigenvalue score of 1.0 or higher and included those items that had factor loadings of 0.4 or higher. See Table 1 for the different items used in the present study and how they were clustered to measure underlying psychosocial constructs. Factor analysis Intention was measured using one item: “If I am pregnant again and due for labour, I will make efforts to go and stay at the maternity waiting home as I wait for labour at the clinic”. In total, 25 items were constructed to measure attitude (table 1). Factor and reliability analyses revealed two underlying variables: cognitive attitude towards MWHs (17 items, α = 0.75). The other attitude variable was affective attitude toward staying in a MWH (7 items, α = 0.72). Similarly, factor and reliability analyses were performed on the 22 items measuring perceived social norms, which resulted in two variables: one of these was descriptive social norms towards MWH use (13 items, α = 0.60), and injunctive social norms towards MWHs (9 items, α = 0.82). Seventeen items were constructed to measure perceived behavioural control (PBC), and factor and reliability analyses resulted in one variable (15 items, α = 0.60). The five items measuring risk perception were also averaged into one variable (with five items, α = 0.83). Finally, perceived barriers towards using MWHs were measured using seventeen items. Factor analysis revealed one variable (14 items, α = 0.70). Descriptive statistics were used to compute percentages of respondents’ demographic and past maternal health seeking behaviour. After inspection of the data and descriptive analysis, we noticed that the data were severely negatively skewed and the assumption of normality was violated. We performed a median split procedure on the psychosocial measures-such that scores including the median and below were dummy-coded as 0 (representing low to moderate scores); and scores above the median were dummy-coded as 1 (representing high scores). To investigate the univariate association between psychosocial measures and intention to use a MWH, and to compare scores on psychosocial measures, sociodemographic variables and past behaviour between those with and those without access to MWHs, Chi-square tests and logistic regression analyses were used. Crude odds ratios (ORs) with 95% confidence intervals (CI) were computed to estimate the effect size. Furthermore, independent t-tests and Cohen’s d [30] were used to investigate whether the respondents from the two groups differed with regard to sociodemographic and economic factors (age, number of children, and distance to the nearest health centre). Finally, adjusted odds ratios (aOR) were calculated to control for confounding due to age, parity, and distance to the nearest health centre (p < 0.05).
N/A
