Menu
Menu
Menu
Menu
Introduction: Two million women worldwide are living with genital fistula with an annual incidence of 50,000-100,000 women. Risk factors for obstetric fistula are context bound. Studies from other countries show variation in the risk factors for obstetric fistula. This study was conducted to identify risk factors for obstetric fistula in western Ugandan context.
The study was conducted at Kagadi and Kyenjojo general hospitals and Hoima regional referral hospital in western Uganda. The three sites are established sites for fistula outreach treatment. Data was collected from November 2011 to May 2012. This area is a rural, predominantly farming area with poor reproductive health indicators, and have the highest prevalence of fistula (4%) in the country [10]. Maternal health care utilization is low with only 56 per cent of women giving birth assisted by skilled birth attendants [11]. A case control study design was used in which 140 cases and 280 controls were included using face-to-face interviews. The interviewers were four research assistants, the first author, who is a gynaecologist trained in fistula surgery and three trained midwives from Mulago and the respective hospitals. Cases were patients confirmed by a doctor to have obstetric fistula irrespective of type and duration. The controls were other women without fistula who had ever given birth and were seeking treatment or attending to patients in the study units. Since these are community units, it was assumed that both cases and controls had a similar environmental exposure and were representative of the population in the area. The sample size was calculated apriori using OpenEpi, based on the formula described by Kelsey et al [30], with 95% two sided confidence level, 80% power and two controls per case. The exposure factor was the proportion of women delivering with no skilled birth attendant. We assumed that women with fistula were likely to have had no skilled labour monitoring and delivery. We also assumed that the controls were like any other Ugandan women reported in the 2006 Uganda demographic and health survey, where 58 per cent of the women were delivering with no skilled attendance [10]. With the resources at our disposal we aimed to have an effect difference of 14 per cent and hypothetically assumed that the proportion of those with fistula delivering without skilled attendance was 72.4 per cent. From the OpenEpi calculator, our sample size was hence fixed at 140 cases and 280 controls (one case to two controls). Cases were then recruited consecutively and for each case two controls were identified and interviewed. Women from the study area, who presented for treatment, were screened, confirmed to have obstetric fistula and then enrolled as cases. The cases were confirmed to have obstetric fistula through history and pelvic examination. Controls were women, who had delivered before, with similar or higher parity corresponding to the pregnancy that resulted into fistula in the corresponding case and were frequency age-matched within a range of 5 years. Women with fistula not following labour process or its management like those with carcinoma, trauma, infections and others were excluded. Also women who were not from the study geographical location were excluded as either cases or controls. The study variables included socio-demographic, physical and obstetric factors highlighted in the literature to predispose women to fistula. The socio-demographic factors were; age at interview, age at marriage, age at first pregnancy, marital status, religion, respondent’s education, spouse’s education, occupation of spouse, occupation of respondent, and distance to the nearest health facility providing emergency obstetric care including caesarean section. The physical characteristics were height of the respondent and the baby’s weight. The obstetric factors were: parity, antenatal care attendance, number of antenatal visits, being accompanied by husband, having a delivery plan, use of herbs in pregnancy and labour, attending antenatal health education classes, and being told the babies presentation. Other obstetric factors included mode of delivery, delivery attendant, and whether there was delay at facility (time spent at the health facility before delivery) or delay in making a decision to seek care, and the duration of labour. The data were collected using an interviewer-administered questionnaire by the first author assisted by trained research assistants who were midwives. A similar questionnaire was administered to both women who were the cases and those without fistula in the control group. Women who fulfilled the inclusion criteria were interviewed from a quiet and private room identified from outpatient department of the respective hospitals. The interviewers were knowledgeable in the local language and would translate the information and fill the data directly in English. The first author checked that data were filled in before respondents left the study site. All the data were double entered in a computer and cleaned using Epidata version 3.1. Prior to data entry, the Epidata computer screen had been fitted with range and consistency checks. The data were exported to STATA version 12 [31] for further cleaning and then analysed by the first author assisted by the second author. All variables were tested for significance at bivariate level using chi-square and the student’s t-test for categorical variables and continuous variables respectively. Covariates that were significant at bivariate level with a P-value of less than 0.1 were entered in a multivariate stepwise (backwards and forwards) logistic regression model and the covariates included were tested for interaction and confounding. Odds ratios and 95% confidence interval were computed. The backward likelihood ratio method was used to select the best fitting model. Duration of labour in hours was the most significant variable in the model and was hence taken as the main predictor for obstetric fistula. Interaction terms for duration of labour and other variables were added in the models. We used the log ratio test where the fitness of the model with all the interaction terms included was compared with a fit of the model with none of the interaction terms. During the log ratio tests, the negative two-log likelihood (-2LL) of the full model and the reduced models were compared. Interaction was considered present when the difference between the -2LL were significant at P≤0.5 with a chi square test. Confounding was considered present if the difference between crude and adjusted odds ratios was greater than or equal to ten per cent. Depending on contribution to the goodness of fit of the model, variables left out were brought back into the model. Hosmer and Lemeshow’s goodness of fit test was applied to check on quality of the model [32]. Respondents were given detailed information about the study: that participation was voluntary, no one would be denied access to services because of refusal to participate in the study, and that information obtained was confidential and would be used only for the purpose of the study. The study received ethical approval from institutional review boards in Uganda and Sweden (requirement for the Makerere University and Karolinska Institutet collaboration). In Uganda, the study received ethical approval from Makerere University, School of Medicine Research and Ethics Committee (#REC REF 2011-104). We also received ethical clearance and approval from the Uganda National Council for Science and Technology (UNCST) registration number HS 1337. We got verbal permission from the respective medical directors/superintendents of Hoima, Kagadi and Kyenjojo hospitals to conduct the study in the respective hospitals. From Sweden, the study protocol was presented and we received approval from the Regional Ethics Committee in Stockholm, (Protocol 2012/2∶4). Informed written consent was obtained from respondents before inclusion in the study. The three participants who were under 18 years assented and also their accompanying parents/guardians gave a written consent. The study conformed to the principles in the Helsinki declaration. All the data were kept confidential and participants were compensated for their time spent during the interviews with 5000 Uganda shillings (USD 2). Those cases not yet operated had their fistulas closed in the week following the interviews by the first author who is trained and skilled in fistula surgery.
N/A
