CHAPTER 2 METHODOLOGY
2.2. RESEARCH QUESTION
2.2.2. DATA COLLECTION INSTRUMENT
A structured questionnaire was developed for use with the EPI-info statistical software package. Questions were grouped under six headings: general information, this pregnancy, previous pregnancies, delivery, post-delivery and economic information. The format of most questions was closed, offering a limited range of possible responses or requiring a numerical answer. The total length of the questionnaire aimed for an interview duration of 30 to 45 minutes. Interviews conducted during the pilot phase confirmed this to be a reasonable assumption.
The lead researcher, who is fluent in English and French, was responsible for the English original of the questionnaire and for its translation into French. The French version was
ix
The 2007 Demographic and Health Survey (EDS-DHS DRC) reported an average caesarean section rate of 4% nationwide, and a 2.7 % CS rate for the Oriental Province. (Reference 87)
27
translated to Swahili by volunteer local collaborators attached to the Institut Supérieur Technique Médical (ISTM). Pilot testing of the Swahili version took place in Bunia during the second half of May 2007.
Selection and training of research team
The choice of the research team was guided by ethical considerations. Preference was given to female interviewers with a nursing and/or midwifery background. Academic staff from the ISTM made a first selection among female candidates to be trained. Four selected nurses received three days of training before the start of the pilot phase. A fifth person (team leader) was charged with the overall field management of project activities. Pilot interviews of ten cases and five controls were conducted over the next week. The questions were reviewed and modified daily according to the feedback received until there were no more problems. The final version of the Swahili questionnaire was subsequently handed to a staff member of the Institut Supérieur Pan-Africain de Santé Communautaire (ISPASC) for back-translation into French and English and found to be accurate. The final, amended questionnaire was written in a slightly modified colloquial version of Swahili, as it is used in Bunia. Interviews were conducted from December 2007 until June 2008.
Selection of cases and controls
A list of cases was compiled twice weekly. Every Monday and Thursday, the team leader went to the participating hospitals to collect the names of all the women who had delivered by CS since the previous visit, and entered the names into a computerised randomisation programme. Cases were allocated to each of the interviewers by randomly selecting names from the “active” list. Only the latest (active) list of names was used to select new cases to be interviewed.
To identify a matching control, an interviewer went to the cases' place of residency. Information about deliveries in the area was obtained from staff at local health structures, from observation of the surroundings (e.g. baby clothes on washing line), and from members of the community. In each neighbourhood there was a government registered health centre. There were no formal networks of community workers or village health committees in Bunia. If there was no delivery in the area within the specified time period of two weeks before or after the delivery date of the case, the interviewers moved to the nearest adjacent neighbourhood to find a control.
28
As the study progressed, the team observed that the vast majority of CSs took place at the BM hospital. As a result, most case interviews were initially conducted in this structure. Because of the large number of surgical deliveries at BM, randomisation of cases at this hospital was necessary. At the time of the lead researcher’s first follow-up visit in January 2008, it was decided to limit randomisation to the CSs taking place in the predominant structure, and to exclude from randomisation all the women who had a CS in any of the three other hospitals. From then on, randomisation was used to select cases among women who had a CS at BM, while those who had a CS in any of the other selected structures were all recruited to be interviewed. The ethics committee of the Liverpool School of Tropical Medicine (LSTM) was informed in writing about this change in methodology. CS deliveries in private practices were counted, but patients were not considered for inclusion in the study. It was assumed that people using private healthcare were able to afford the cost and did not need humanitarian assistance.
The study was conducted from December 2007 until June 2008. This period was long enough to recruit the number of participants required. It also ensured that seasonal variations potentially affecting the number of deliveries, access to, and use of health facilities were taken into account. The Bunia area has two rainy seasons, one of which was covered by the study period. North of the Equator the rainy season is during the period November to April; to the south, it is from June to September. (79) In Bunia, it rains most of the year, although December-January is considered to be relatively dry. (80) Malaria is endemic.