Menu
Menu
Menu
Menu
Background: Deficiencies in the provision of evidence-based obstetric care are common in low-income countries, including Mozambique. Constraints relate to lack of human and financial resources and weak health systems, however limited resources alone do not explain the variance. Understanding the healthcare context ahead of implementing new interventions can inform the choice of strategies to achieve a successful implementation. The Context Assessment for Community Health (COACH) tool was developed to assess modifiable aspects of the healthcare context that theoretically influence the implementation of evidence. Objectives: To investigate the comprehensibility and the internal reliability of COACH and its use to describe the healthcare context as perceived by health providers involved in maternal care in Mozambique. Methods: A response process evaluation was completed with six purposively selected health providers to uncover difficulties in understanding the tool. Internal reliability was tested using Cronbach’s α. Subsequently, a cross-sectional survey using COACH, which contains 49 items assessing eight dimensions, was administered to 175 health providers in 38 health facilities within six districts in Mozambique. Results: The content of COACH was clear and most items were understood. All dimensions were near to or exceeded the commonly accepted standard for satisfactory internal reliability (0.70). Analysis of the survey data indicated that items on all dimensions were rated highly, revealing positive perception of context. Significant differences between districts were found for the Work culture, Leadership, and Informal payment dimensions. Responses to many items had low variance and were left-skewed. Conclusions: COACH was comprehensible and demonstrated good reliability, although biases may have influenced participants’ responses. The study suggests that COACH has the potential to evaluate the healthcare context to identify shortcomings and enable the tailoring of strategies ahead of implementation. Supplementing the tool with qualitative approaches will provide an in-depth understanding of the healthcare context.
This was a cross-sectional survey in which the COACH tool was administered to health providers involved in maternal and neonatal care in 38 health facilities of six districts in southern Mozambique (Figure 1): Bilene-Macia, Chibuto, Chokwe and Xai-Xai districts (in Gaza province), and Magude and Manhiça districts (in Maputo province). Study setting displaying included districts and health facilities, in Maputo and Gaza provinces, Mozambique. Of the 38 facilities, 32 are primary health centres providing essential preventive and curative services, including antenatal and intrapartum care for uncomplicated deliveries. The remaining six facilities are hospitals (four rural, one district and one provincial) to which complicated cases are referred, and routine surgical interventions such as caesarean sections or obstetric hysterectomies, are performed. The tool has 49 items that measure eight dimensions of context (some dimensions have sub-dimensions) and is available in English, Bangla, Vietnamese, Lusoga, isiXhosa, and Spanish [21]. Items for seven of the eight dimensions (see Table 1 for the definitions) measure agreement with statements that theoretically reflect a context supportive of change (hereinafter referred to as the context’s readiness to change). Items on these dimensions were measured on a five-point Likert scale (ranging from ‘strongly disagree’ to ‘strongly agree’). Definitions of COACH dimensions. *Unit refers to the department or primary health care centre where the respondent is working. For the Sources of knowledge dimension, respondents indicate for each of five knowledge sources whether the source is available, and, where available, the frequency of its use (never, rarely, occasionally, frequently and almost always) [14]. In addition to the 49 original COACH items, the version used in this study contained seven demographic questions (age, gender, professional qualification, year professional qualification obtained, health facility, department (if applicable) and years working at the current facility). The translation of the COACH tool from English to Portuguese followed Brislin’s model as summarized by Yu et al. (2008) [22]. The translation was conducted in four phases: (1) Forward translation (English to Portuguese) by a bilingual professional translator with knowledge of the tool in order to assure appropriate language use; (2) Review of the translated tool by a monolingual reviewer with no familiarity of or access to the English version; (3) Backward translation (Portuguese to English) by a different bilingual professional translator from the one engaged in step (1); and (4) Comparison of the original version and the backward-translated version focusing on conceptual clarity and aimed to ensure an appropriate Portuguese translation of the tool. Comprehensibility, in this study, refers to the extent to which a statement is easy to understand by the reader. To uncover difficulties in understanding the instructions for completing COACH or items in the tool, the Portuguese version was administered by structured interview to six purposively selected health providers (two physicians, two midwives and two auxiliary nurses) representing the provider categories the main survey would target. In each interview, the first author introduced COACH before the participants were asked to read and state their level of agreement with each of the items in the tool and reflect upon whether they had any difficulty understanding its content. Attention was paid to the participants’ level of understanding and whether they had any challenges in rating their level of agreement with the items. Identified problems were translated into English and categorized in two ways: (a) by the magnitude of their effect on the collected data (prominent vs. minor) [14]; and (b) by Conrad and Blair’s taxonomy [23] (see Table 2). All identified problems were also discussed in relation to the underlying cause of the problem, i.e. relating to the content of the item or the Portuguese translation of the item. Based on the findings from the response process, we produced the final Portuguese version of the COACH tool for data collection (http://www.kbh.uu.se/imch/coach). Analysis framework for the COACH tool response process in Mozambique. The original COACH tool, designed to be a self-administered questionnaire [16], was amended for administration via an individual structured interview to maximize response and item response rates [24]. An interview guide was designed to ensure that the data collection was standardized and that clear, complete and unambiguous responses to the statements were obtained from the respondents. A member of the research team (R.C.) carried out the interviews, which were undertaken in secluded rooms in the health facilities. Eligible respondents were health providers (doctors, medical assistants, nurses, midwives and auxiliary nurses) who had worked in the targeted facilities for at least 12 months before the study (n = 273). Data were collected between April and June 2016. We were able to interview 175 health providers from the identified 273 eligible respondents (64% response rate). From the 98 who did not participate (46% were nurses and 37% auxiliaries), 55 were absent (vacation, illness leave or not on shift), 42 were not able to answer (busy with patients), and one refused to be interviewed. The non-response rate was higher in hospitals, 41% (47 out of 114) compared to 32% (51 out of 159) in primary facilities. The 175 questionnaires were checked for completeness of responses, with no missing responses detected. Data were double-entered in OpenClinica software, version 3.1 [25] and imported into SPSS v. 24 [26] and R software (version 3.3.1) [27] for further analyses. For the demographic variables age, gender, professional category, healthcare level, district and years working in the current facility, mean and standard deviation or median and interquartile range as appropriate were calculated for continuous variables and proportion (%) for categorical variables. Items 42 to 47 described elements of context obstructive to the implementation of interventions and EBPs and scores were therefore reverse scored to be consistent with the connotation of the other items. Items from the Sources of Knowledge dimension were recoded into 0 (not available, never and rarely), 0.5 (occasionally), and 1 (frequently and always). The internal consistency reliability of each dimension was tested using Cronbach’s α analyses with item trial removal where indicated. Once satisfactory reliability was demonstrated, items within dimensions were summed, and descriptive analysis (minimum and maximum scores, means and standard deviations) of dimensions was performed. Subsequently, individual-level data were aggregated within districts and one-way analysis of variance (ANOVA) with the post hoc Tukey HSD test was performed for each dimension using the district as the group variable. Level of significance was set at p < .05.
N/A
