Changes in care-seeking for common childhood illnesses in the context of Integrated Community Case Management (iCCM) program implementation in Benishangul Gumuz region of Ethiopia

Background Integrated Community Case Management (iCCM) is a strategy for promoting access of under-served populations to lifesaving treatments through extending case management of common childhood illnesses to trained frontline health workers. In Ethiopia iCCM is provided by health extension workers (HEWs) deployed at health posts. We evaluated the association between the implementation of iCCM program in Assosa Zuria zone, Benishangul Gumuz region and changes in care-seeking for common childhood illnesses. Methods We conducted a pre-post study without control arm to evaluate the association of interest. The iCCM program that incorporated training, mentoring and supportive supervision of HEWs with community-based demand creation activities was implemented for two years (2017–18). Baseline, midline and endline surveys were completed approximately one year apart. Across the surveys, children aged 2–59 months (n = 1,848) who recently had cough, fever or diarrhea were included. Data were analysed using mixed-effects logistic regression model. Results Over the two-year period, care-seeking from any health facility and from health posts significantly increased by 10.7 and 17.4 percentage points (PP) from baseline levels of 64.5 and 34.1%, respectively (p<0.001). Care sought from health centres (p = 0.420) and public hospitals (p = 0.129) did not meaningfully change while proportion of caregivers who approached private (p = 0.003) and informal providers (p<0.001) declined. Caregivers who visited health posts for the treatment of diarrhea (19.2 PP, p<0.001), fever (15.5 PP, p<0.001), cough (17.8 PP, p<0.001) and cough with respiratory difficulty (17.3 PP, p = 0.038) significantly increased. After accounting for extraneous variables, we observed that care-seeking from iCCM providers was almost doubled (adjusted odds ratio = 2.32: 95% confidence interval; 1.88–2.86) over the period. Conclusion iCCM implementation was associated with a meaningful shift in care-seeking to health posts. Pre-posttest design without control group was applied to evaluate changes in care-seeking for common childhood illnesses (diarrhea, fever and cough) secondary to an iCCM program implemented in Assosa zone over two-year period (2017–18). Baseline survey was conducted ahead of the program in January 2017; and midline and endline surveys completed in January 2018 and January 2019. The study was conducted in Assosa zone, one of the three zones of Benishangul Gumuz region of Ethiopia. Benishangul Gumuz is among the four emerging regions of Ethiopia having low socio-economic status, limited access to social services and weak human resources to implement development programs including health services. Assosa town, the regional and zonal capital, is located approximately 700 km northwest of the national capital Addis Ababa. In 2017 Assosa zone had 360,000 inhabitants, of whom 86% were rural dwellers [24]. Administratively, the area is divided into seven districts. The main sources of livelihood are subsistence mixed agriculture and artisan gold mining. Benishangul Gumuz region has the second highest poverty rates in Ethiopia and about a quarter of the population lives below the national poverty line [25]. Furthermore, road density and access to basic social services are extremely low. According to the Ethiopia’s three-tier healthcare system, primary care is provided by health posts (1 facility for 3,000 to 5,000 population), health centers (1 facility for 15,000 to 25,000 population) and primary hospitals (1 hospital for 60,000 to 100,000 population) [26]. At the time of the study, Assosa zone had 1 primary hospital, 7 health centers and 183 functional health posts. About 380 frontline health workers (mainly HEWs) were deployed at the health posts and provided preventive and basic curative services including the iCCM. In Benishangul Gumuz region including the study area, the HEP is relatively weak and health indicators are much lower than the national averages [6,17]. The implementation of iCCM in the emerging regions of Ethiopia including Benishangul Gumuz region was started in 2015. However, at the ground level the program brought limited changes due to weak HEP platform, limited iCCM coverage, shortage of human resources for health and turnover of trained health workers. In 2017, Emory University in collaboration with its partners revitalized the iCCM program in all the seven districts of Assosa zone. The Emory University’s program included system strengthening and stronger community mobilization components. The program was directly involved in health system strengthening through training of frontline health workers and district health office managers, consolidating referral linkage between health posts and health centers, and instating quality improvement framework at various levels of the system. The major programmatic activities the Emory University’s program implemented included training of 182 HEWs on iCCM and 26 health professionals on Integrated Management of Newborn and Childhood Illnesses (IMNCI). The six-day iCCM training that was aligned with the standard WHO/UNICEF iCCM training protocol [27] addressed topics including clinical practice on identification and classification of signs and symptoms of common childhood illnesses and providing appropriate treatment, including pre-referral treatment, at health post level. Supervisory skill training was also provided to 42 HEW-supervisors. Furthermore, the program provided quality improvement training to 46 health professionals and the same was cascaded to 182 HEWs. At community level, 885 health development army (HDA) members (network of community volunteers) and community leaders were trained on danger signs of childhood illness, community mobilization and prompt referral of sick children to health posts. Accordingly, the HDA members counseled caregivers and linked them with the primary health care system. Further, quarterly supportive supervision and monthly couching of HEWs were regularly implemented jointly by the RHB and Emory University’s team. Multiple rounds of community festivals for iCCM demand creation were conducted on quarterly basis. Further, the festivals were used as opportunity for motivating community-level iCCM actors. Emory University also facilitated timely supply of iCCM commodities including medications to the health posts and worked towards improving referral linkage between health posts and health centers. Mothers/caregivers from the seven districts of the of the zone having children 2–59 months of age with at least one of the illnesses in the past two weeks were eligible for the survey. Infants younger than two months of age and children from urban areas were excluded because they were not targeted by the iCCM program. The sample size for the survey was determined using the Cochran's single population proportion formula [28] assuming 39% expected prevalence of care-seeking from formal providers [29], 95% confidence level, 5% margin of error and design effect (DEFF) of 1.5. The DEFF is determined using the standard formula (DEEF = 1 + (cluster size-1)*intra-cluster correlation) specifying average cluster size of 23 and an intra-cluster correlation of 2% [30]. Based on this, 616 children were needed in each survey-round. Furthermore, using double population proportion formula, we assured that the sample size is adequate to detect 5 percentage points change in care-seeking between any two of the three surveys at 95% confidence level and 80% power. We used multistage cluster sampling approach for selecting the study participants. From each of the seven districts, 4 rural kebeles–the smallest administrative units in Ethiopia with approximately 1000 households, were randomly drawn. Then from each of the selected kebeles, 1 village (got) was chosen using simple random sampling (SRS) technique. In each selected village, a rapid listing of eligible children was completed and sampling frames were developed. Ultimately from every village 23 eligible children got selected using SRS technique. With the intention of maximizing the sample size, study subjects who were not willing to take part in the study were replaced by randomly chosen eligible subjects from the same clusters. During each survey round, the same kebele and village were studied; however, study subjects were selected independently. The two primary outcomes of the study were care-seeking (yes/no) from any health facility and care-seeking (yes/no) from health posts. Care-seeking was also determined for any of the individual conditions the child had in the past two weeks. Secondary outcomes were care-seeking from informal sources (e.g. traditional and religious healers) and caregivers' knowledge of childhood illness danger signs. The practice of consulting friends, family members or neighbors about the illness of the child was also considered as a secondary outcome. The factor of interest was implementation of the iCCM program (baseline, midline, endline). Control variables included distance of the household from the nearest health facility, household wealth index, maternal educational status, age and occupation, number of under-five children in the household and type of the primary caregiver (mothers vs other caregivers). The control variables were selected based on review of relevant literature [6,10,11]. Data were gathered by trained and experienced enumerators and supervisors from the primary caregivers of the children using pretested and interviewer administered questionnaire. The questionnaire had not been validated before but was used in a similar survey conducted in Ethiopia before. The questionnaire used in the survey is provided as a supporting file (S1 Table). The questionnaire was finalized in English, translated to Amharic language and administered to the respondents in their local languages. On top of socio-economic features, the questionnaire characterized the illness the child recently had and assessed the practice of care-seeking from different sources. Care-seeking practice was assessed based on the reports of the caregiver without reviewing any formal medical records. Occurrence of diarrhea was assessed by asking the caregiver whether the child had three or more loose stools per day in the reference period. For all children who reportedly had cough, the presence of concomitant difficulty of breathing and whether that was due to nasal congestion or chest problem, were explored. Formal care providers were classified as public (health post, health center, public hospitals) or private (private clinic, hospital, charity clinics and drug vendors). Further, informal care-seeking was defined as care sought from informal practitioners (e.g. traditional or religious healers) or attempting traditional treatments at home. The knowledge of the caregivers on thirteen danger signs of childhood illness [31] were assessed and a summated score (minimum and maximum possible scores of 0 and 13) was developed. Household wealth status was measured based on ownership of livestock, durable household assets, land size, materials used for house construction and access to electricity and improved drinking water source. We used STATA version 14 for data analysis. Descriptive data analysis was made using frequency distributions and measures of central tendency and dispersion. Weighted analysis was made using sampling weights and post-stratifications weights developed based on the population sizes of the districts. Changes in care-seeking and knowledge of danger signs of childhood illness were compared across the surveys using chi-square for trend test. Household wealth index was developed using principal component analysis (PCA) as commonly done in national demographic and health surveys and classified into three tertiles: lower, middle and upper third. The association between the iCCM implementation (baseline, midline and endline) and formal care-seeking was evaluated using mixed effects multivariable logistic regression model. Random intercepts were set at district and kebele levels. Control variables were selected for adjustment based on statistical criteria. Initially, the comparability of the three surveys in selected basic socio-demographic characteristics was assessed using Pearson’s chi-square test and significantly or marginally unbalanced variables (p<0.1) were statistically adjusted. The analyzed dataset is provided as a supporting table (S2 Table). The study was implemented in conformation with international ethical standards including the Helsinki Declaration. The work was approved by the institutional review board (IRB) of the Benishangul Gumuz Regional Health Bureau. The data were collected after taking informed verbal consent form the study subjects. Verbal, rather that written consent was used because significant proportion of the population in the area had no formal education. The same was approved by the ethics committee that cleared the protocol.