Implications of a Coverage Survey in Ethiopia
By Simon Kiarie Karanja
Simon Karanja is currently the regional nutrition advisor with GOAL in East Africa. Previously he worked as the CTC Coordinator for GOAL Ethiopia and as the nutritionist on a Merlin International CTC programme in Wajir, Kenya.
The author gratefully acknowledges the assistance of GOAL Ethiopia Rapid Response programme staff for working hard and ensuring that the data were collected and recorded in the best way possible. Particular thanks to Jessica Barney for her help in organising the survey and special thanks to Angela Davis and Hatty Newhouse for their technical advice and for reviewing and editing the survey report.
This article presents the results of a survey that took place in December 2006 to assess the coverage of a community-based programme and discusses the implications of the findings.
Screening using MUAC in the GOAL
community based programme
Fedis woreda is located in East Harerghe Zone of Oromiya regional state. The woreda consists of 25 rural and one urban kebeles1. The total population is estimated at 246,437, with children under five years estimated at 28,057 (Fedis woreda Council, 2005). The population is predominantly of the Oromo ethnic group, are Muslim, and Afaan Oromo is widely spoken.
The woreda lacks many basic essential services and traditionally suffers from high prevalence of malnutrition and high morbidity rates. Fedis, which is vulnerable to recurring droughts and shocks, is one of the most food insecure woredas in the East Harerghe zone and has received aid relief for over twenty years.
The livelihood of people in the area centres on agriculture, complemented by livestock holdings. The predominant cash crop is chat. All kebeles members are recipients of various relief activities including the Extended Outreach Strategy/Targeted Supplementary Feeding Programme and Safety Net Programming (general food distribution, Cash for Work and Food for Work).
There are nine functional health facilities within the woreda including one health centre at Fechatu, three clinics and five health posts. There are a total of 27 professional health workers in these facilities including 14 nurses, seven health assistants, two environmental health technicians, two health extension workers, and two frontline health workers. The overall health service coverage is 32% of the total population. The facilities are understaffed and lack basic facilities and equipment. The facility utilisation rate is low in the woreda. The nearest referral hospital for patients in the woreda is Hiwot Fana in Harar City, which is located approximately 35 kilometres away from Fedis.
A team measuring
length during
coverage survey
Community based approach
GOAL Ethiopia has been managing severely malnourished children in Fedis woreda using the community based therapeutic approach since mid 2005, following an initial joint assessment by Care and GOAL Ethiopia at the end of July 2005 that revealed an extensive nutritional crisis, confirming the findings of an earlier survey2. The GOAL intervention was designed to promote sustainable capacity within the woreda to address malnutrition. Goal's community-based approach targets all 26 kebeles in the woreda, through five outpatient therapeutic programme (OTP) sites, one stabilisation centre (SC) and eleven targeted supplementary feeding programme (TSFP) sites. The catchment area for the community-based programme is defined by Gandas (villages) within kebeles in Fedis woreda, which the outreach workers visit to find new cases and do follow-ups. The catchment area includes all kebeles currently within what was one but now comprises two defined woredas: Fedis and Midegha (Midegha woreda was initially part of the larger Fedis woreda but was recently created as a separate woreda). As of November 30, 2006, a total of 2352 cases had been admitted to the community-based programme and 442 outreach workers trained. Table 1 presents GOALs OTP and SC admission criteria.
| Table 1 OTP/SC admission criteria |
| Outpatient Therapeutic Programme (OTP) |
Stabilisation Centre (SC) |
| Weight for Height (WFH) < 70% OR |
WFH< 80% AND severe medical complications OR |
| Bilateral pitting oedema Grade + or ++ OR |
MUAC <11cm AND No appetite or severe medical condition OR |
| MUAC < 11.0cm (age >1year or height >70cm) |
Oedema Grade + or ++ AND No appetite or severe medical condition OR |
| |
Marasmic kwashiorkor (alone) OR |
| |
Bilateral pitting oedema Grade +++ (alone) |
| Table 2 Area surveyed |
| Woreda area2: |
1010 sq km |
| Number of quadrats: |
35* |
| Area of each quadrat: |
5km x 5km = 25 sq km |
| Area covered by the survey: |
35 x 25 = 875 sq km4 |
| Period of survey: |
7-16 December 2006 |
*36 quadrats were planned but one was not surveyed due to problems relating to mapping.
Box 1 Sampling procedure5
First, each team had to visit one quadrat per day. The four Gandas
(villages) closest to the centre of the quadrat were identified for
sampling. As a general rule, each team had to complete the screening of
at least four Gandas in one day.
Secondly, an active case-finding strategy was used to identify severely
malnourished children. A key informant such as a community health
worker (CHW), community leader, traditional birth attendant (TBA), and
in many cases, a carer, helped to identify severely malnourished children
in each Ganda.
The teams were asking for 'types' of children who were likely to be
severely malnourished, e.g. thin, sick, oedematous, twins, orphans and
children registered in the OTP programme. These terms were defined
by the survey team, based on their local knowledge of how people in
the woreda would describe malnourished children. All four teams had
two or more members fluent in Afaan Oromo, the local language spoken
throughout the woreda.
A number of steps were taken to ensure that all cases in a study village
were identified through the case-finding method6:
- Village heads and mothers were notified in writing or by other means
regarding the active case finding and house screening, which was to
take place on a specific day.
- No sampling of villages took place on OTP distribution days.
- Villages surrounding a market town were not visited on the market day.
- Every carer, who was visited at their homestead, was probed about
their knowledge of other sick, oedematous, or thin children in the
community.
- If the sick, oedematous and thin child was absent from home, the
team tried to find him/her or come back later.
- If the household had a child in the hospital or the SC, the team noted
his name and the child's anthropometric data were obtained from the
hospital/SC.
- All children considered sick, thin or oedematous were seen until carers
referred back to those children already assessed (sampling to
redundancy).
- Outlying houses set away from the main sub-Gandas were include
| Formula 1 Period Cover |
| Number of respondents in the feeding programme |
X 100 |
| Number of cases NOT in the feeding programme +
number of respondents in the feeding programme |
| Formula 2 Point cover10 |
| Number of cases in the feeding programme |
X 100 |
| Total number of cases |
Objectives of the Study
The main objective of the GOAL study was to assess the community-based programme coverage for children 6-59 months of age. Specific objectives were:
- To assess the OTP/SC coverage.
- To map out both point and period coverage for the whole woreda and for specific areas within it.
- To determine factors that affect uptake of the programme services.
Methodology
Study Design
The study involved a cross-sectional survey using centric systematic area sampling (CSAS) and adaptive (or purposive) sampling (two stages). The programme area was stratified geographically using the CSAS method3. This method, based on an exhaustive geographical sample, involves dividing the survey area into non-overlapping squares of equal size. The total area surveyed (see table 2) excludes the pastoral area that covers 50% of the total woreda area approximately. In most instances when less than 50% of the quadrat fell within the woreda boundaries, the area was not surveyed - however, some quadrats with less than 50% of the quadrat in the woreda were included due to their high population density. Areas classified as pastoralist areas were also excluded due to difficult access, low population density and because seasonal migration is practised.
Active case finding was used in the second stage of the study. Approval for the coverage survey was obtained from both the Zonal Health bureau and the woreda Health Office. The sampling method was based on a spatial sampling method and an active case-finding strategy (see box 1).
Implementation of the Survey
Training/Data Quality Control
Four teams, each made up of two enumerators and one team leader, carried out the survey. To ensure standard data collection, all team members underwent two days training on anthropometric measurement, admission and discharge criteria, and signs of malnutrition. A local events calendar was developed to establish the age of children without vaccination card/exact age.
Validation Sub-study
A one day capture-recapture study was carried out (using active case finding and house-to-house methods)7 to test the sensitivity of the active case-finding method used during this coverage survey8. This confirmed 100% accuracy of the method in locating children enrolled in OTP/SC in Fedis woreda. In addition, team members gained practical experience while the study helped identify those without sufficient experience and skill in anthropometric measurement9.
Data Processing and Analysis
Data were analysed in Microsoft Excel and SPSS v10. Point coverage and period coverage were calculated as below. Period coverage is the conventional approach used by agencies such as Medecins Sans Frontieres (MSF), Save the Children and UNICEF and is the most commonly reported. In order to obtain the most sensitive coverage calculation (i.e. leaving out as few cases as possible), a MUAC < 12.0cm cut off point was used to screen children eligible for full anthropometric assessment.
Constraints
The map was several years old (1994) and no kebele or Gandas names were included on it. The map also included areas belonging to a neighbouring woreda (Babile). In order to remedy this problem, local knowledge was used to plot the kebele and Gandas names on the map. In several cases, the precise locations of Gandas were uncertain which might have meant that some sampled villages were not necessarily those closest to the centre of the quadrat. There may also have been instances when carers provided inaccurate information about whether a child was or was not in the programme.
| Table 3 Source of knowledge about GOAL's CTC programme |
| Individual |
Percentage |
| Neighbour/Word of mouth |
74% |
| Community mobilisers |
12% |
| Health extension agents |
6% |
| Health Centre Staff |
5% |
| Kebele Leader |
3% |
| Table 4 Reason for not taking child to the OTP/SC site |
| Reason (carers could provide more than one response) |
n |
Percent of
responses |
| Carer thought child was sick but not malnourished |
5 |
45.5 |
| No other carer to look after other children |
3 |
27.3 |
| Mother carer too busy |
2 |
18.2 |
| OTP site too far away |
1 |
9.1 |
Results and Discussion
Point Coverage
The point coverage refers to the ratio of cases receiving treatment found in the sample to the total number of cases requiring treatment found in the sample11. Figure 1 presents the point coverage histogram for OTP/SC.
The point coverage in Fedis woreda was poor across the entire programme area. Thirteen quadrats had a point coverage of less than 10%, while only two quadrats had a point coverage of 100%. There was no correlation between the point coverage and distance from the OTP/ SC sites.
Period Coverage
Period coverage includes children who may not be eligible for entry into the programme on the day of the survey (i.e. children who are in the recovery phase and whose weight-for-height and/or MUAC is higher than that required for entry into the programme or who no longer exhibit nutritional oedema). These children, now in recovery, were recently severely malnourished. Period coverage is an estimator of recent coverage in a given period12.
Period coverage was high across the programme area with 10 quadrats having period coverage of 100%, and 15 quadrats having coverage of equal or greater than 50% (see figure 2). Only three quadrats had a period coverage equal or less than 40%. The average coverage for the woreda was 70.5%. The Sphere guidelines state that the coverage of therapeutic and supplementary feeding programmes should be greater than 50%13. GOAL's community based programme in Fedis woreda does meet these recommended guidelines when measured by period coverage.
A mother and her child
attending the OTP
Anthropometric assessment
Out of 35 severely malnourished children identified during the survey, five (14.3%) had oedema, one (2.9%) had a Weight for Height % of the median (WFH%) <70% and 29 (82.9%) had a MUAC <11cm aged between 12 to 59 months. Of the 25 severely malnourished children not in the programme at the time of the survey, 19 (76%) had a MUAC <11cm, five children (20%) had oedema, and one (4%) child had a WFH%<70%. The very high proportion (76%) of severely malnourished children identified through MUAC measurements might, in part, account for the low point coverage results found in this survey. Sixty percent (21/35) of the severe cases identified based on MUAC had a MUAC between 10.7 and 10.9 cm (inclusive).
In such a context, MUAC measurements need to be very precise as a slight inaccuracy in taking MUAC measurements can have a great impact on coverage estimates. It is possible that a number of the children with MUAC <11.0 cm with an age between 12 to 59 months would not have been detected by trained community health workers. In addition, difficulties establishing exact age may have led to MUAC based assessment being employed for infants under 12 months who were inappropriately identified as over 12 months of age14. This could also have contributed to a downward bias in the point coverage results.
Outreach/mobilisation strategy
Programme knowledge was on average very high in the programme area. Of the total children screened, 98% of the carers/mothers knew about the programme. Table 3 shows how people came to know about the programme. While the vast majority (74%) learned through word of mouth/neighbours, only 5% of carers were informed by their local health facility and 3% by kebele leaders. This suggests poor linkage between OTP activities and other health facility activities and a need to re-think the mobilisation strategy involving community leaders.
Mobilising the community on
communty based programme
Timing of the survey
The survey was carried out in the post harvest period which is a relatively food secure time of year in Fedis woreda. This correlates with the low programme admissions that may, in turn, bias the point coverage results downward as not many children were registered in the OTP programme. It is also possible that a number of the severely malnourished had experienced rapidly deteriorating nutritional status due to illness and that there had not been sufficient time to be admitted into OTP.
Awareness of the CTC programme
Throughout the survey, a questionnaire was completed by 23 carers whose severely malnourished children were not in the community-based programme, but who had heard about it. Fifteen carers (65.2%) had taken their child to an OTP site in the past but were not admitted due to various reasons e.g. child did not fit OTP admission criteria at that time. Two carers never took their child to an OTP ever. Reasons for not taking their child to the OTP/SC site are given in Table 4. These findings suggest that more community education about the signs and symptoms of malnutrition could increase coverage.
Past history/referrals
Carers of severely malnourished children who were not in the OTP but knew about it, were asked if their child had ever been in the GOAL CTC programme before. The vast majority, 73.9% (17/23), said that the child had not been in the programme in the past. Five out of 23 carers (21.7%) said that their child had been in the programme and had been discharged. Children who had relapsed were enrolled in the SFP programme. One child was in the programme but defaulted as the carer thought the child had recovered. With sustained outreach efforts and by strengthening the integration of community based management of malnutrition activities with the existing outreach system, this proportion of severely malnourished children who have never been enrolled in the programme is likely to decrease. Stronger linkages between the SFP and OTP programmes need to be established to ensure that children in the SFP can be referred to the OTP if their nutritional status deteriorates.
Acceptance/rejection rates - malnourished children
Carers of malnourished children found who were not in the OTP/SC but knew about it, were asked if the child was ever referred to a GOAL OTP/ SC site. Fourteen (60.9%) of 23 respondents had while 39.1% (9/23) had not. On referral, over half (57.1%) were rejected because they did not fit the programme admission criteria. A number of carers who brought their child to the OTP site might have done so at a time when their child did not fit the admission criteria and was therefore rejected. Subsequently, when their child's situation deteriorated, carers may have been unaware that they could bring their child again for re-assessment. This phenomenon could bias coverage downward and could be easily remedied by having all programme staff and health/community workers explain to carers they should re-present children in the event of further nutritional deterioration.
Recommendations
Improved and sustainable high coverage of the CTC programme could be achieved by:
- Strengthening the training of CHWs with regard to the signs and symptoms of malnutrition and encouraging CHWs to include an educational message as part of their community mobilisation and outreach activities.
- Providing health and nutrition education to health staff and community health resource persons that will have a positive impact on changing the behaviour of the community with regard to childcare and feeding practices, with special emphasis on children with special needs, e.g. mentally and physically handicapped.
- Developing effective techniques for identifying and managing cases with psychosocial causes.
- Strengthening community mobilisation messages to enhance the communities' understanding of the community-based programme, e.g. admission and discharge criteria, importance of completing treatment, and of returning if their child deteriorates after discharge or was rejected and subsequently deteriorates further.
- Strengthening linkages at the health facilities through ongoing capacity building of the woreda's health officials in order to maximise the likelihood that all carers bringing in sick children to their local health facility are referred onto the OTP if their child is malnourished.
Conclusion
Generally, the programme coverage is high and meets the Sphere standard set out for coverage of therapeutic feeding programmes in rural areas. The high programme coverage of 66% was achieved mainly through sustainable methods of community mobilisation. The mobilisation efforts were integrated into existing Ministry of Health outreach system and other community based institutions. Community volunteers, TBAs, Community Health Advocates (CHAs), Community Hleath Promoters (CHPs) and malaria agents, all active in routine health facility outreach activities, have played a significant role in community mobilisation, absentee and defaulter tracing. There is however room for further improvement and the coverage survey has helped identify how this might be achieved.
For further information, contact: Simon Karanja, email: skaranja@goalkenya.org
1A term used to describe both urban dwelling associations and peasant associations in rural areas.
2Conducted by the Federal Disaster Preparedness and Prevention Commission (April 2005)
where a Global Acute Malnutrition rate of 19.2% and Severe Acute Malnutrition rate of 2.9%
were found.
3Mark Myatt. New method of estimating programme coverage. Community Based Approaches
to Managing Severe Malnutrition. ENN report on the proceedings of an inter-agency workshop.
Dublin 8-10th October 2003. Available at www.ennonline.net
4Area surveyed is approximately 87% of total area.
5Valid International, 2006. Notes on using Capture-recapture techniques to assess the sensitivity
of rapid case finding methods.
6Feleke T. 2004. Selective Nutrition Programme Coverage Survey Report. Hulla and Arbegona
Districts. Ethiopia. Valid International.
7Monica Zanchettin. Selective Nutrition Programme Coverage Survey Report for Awassa Zuria
Woreda, Ethiopia. Feb 2006. Valid International Ltd and GOAL Ethiopia.
8The data gathered during the validation study was not part of the coverage calculations.
9Two key references used to inform the validation study were: Fekele T., Myatt, M., Collins,
S., Sadler, K., 2003, Feeding Programme Coverage Survey for Severely Malnourished Children:
Dowa and Mchinji Districts, Malawi, Valid International, and Fekele, T., 2003, Feeding Programme
Coverage Survey: Kalu and Desie Zuria Districts, Ethiopia, Valid International.
10The overall percent coverage is the average of all quadrat percentages coverage.
11In coverage survey, the cases are the denominator. This means that quadrats with zero cases
are treated as "N/A" (not applicable) data point. The 'N/A' quadrats do not contribute to the
overall coverage estimate (personal communication; Mark Myatt)
12Myatt M, Feleke T, Sadler K, & Collins S. 2005. "A Field Trial of a Survey Method for
Estimating the Coverage of Selective Feeding Programme". Bulletin of the World Health
Organization. Vol. 83, no. 1, pp.20-26.
13The Sphere Project, 2004, The Sphere Project, Oxfam Publishing, p.148
14According to current programme admission criteria, MUAC should not be used for infants
under 12 months of age.
Taken from Field Exchange Issue 31, September 2007
http://fex.ennonline.net/31/surveyinethiopia.aspx