The Simple Spatial Survey Method (S3M) was developed from the Centric Systematic Area Sampling (CSAS) coverage survey method as a response to the widespread adoption of community-based management of acute malnutrition (CMAM) by ministries of health. Large-scale programmes need a large-scale survey method and S3M was developed to meet that need.

S3M was designed to:

  • Be simple enough for MoH, NGO and UNO personnel to perform.
  • Be able to survey areas up to ten times larger than the CSAS method for approximately twice the cost whilst maintaining the spatial resolution of CSAS surveys.
  • Provide a general survey method. S3M can be used to survey and map the coverage of selective entry programmes such as CMAM, universal programmes such as growth monitoring programmes (GMP) and expanded programme on immunisation (EPI), various indicators, and prevalence over wide areas.

Here we outline the use of S3M for the assessment of coverage of the CMAM programme and of infant and young child feeding (IYCF) practices in five regions of Niger namely Tillaberi, Dosso, Tahoua, Maradi and Zinder. The assessment was conducted by Valid International in partnership with UNICEF Niger, Ministerio de Salud and the Institut National de la Statistique (INS) of Niger from October 2011 to February 2012.

Methods

S3M sampling design

Find a map

The first step was to find a map of the entire five regions of Niger showing the locations of all towns and villages within. Two sets of maps were used. A small-scale map (a wide area map but with poor detail) of the entire survey area of five regions of Niger was used as a base map to identify all the sampling points across the whole survey area. This map did not show the location of all towns and villages in the survey area (Figure 1). In addition, a collection of larger-scale maps (a small area map but with good detail) of each of the five regions of Niger showing the locations of all towns and villages was utilised. Large-scale maps of the five regions of Niger used for the survey were provided by UNICEF and can be found in Figure 2.

Figure 1: Small-scale map of five regions of Niger

Dosso Maradi Tahoua Tillaberi Zinder
Figure 2: Large-scale maps of each of the five regions of Niger

Decide the area to be represented by each sampling point

The geographical area units in which CMAM coverage and IYCF practices will be assessed was determined. This was conceptually approached as a function of the intended maximum distance (\( d \)) of any community from the nearest sampling point (see Figure 3). Since one of the surveys was for the coverage of the CMAM programme that treats severe acute malnutrition (SAM), a rare condition, then it was important to select a small value for \(d\). A small \(d\) means more sampling points, which in turn translates to greater likelihood of finding a significant number of SAM cases. As for the survey on IYCF practices, the target sample were children aged 0 to 23 months which is a significantly larger number of children than SAM children. Hence, the value of d adequate for the coverage survey of CMAM is more than enough for the survey on IYCF. For the Niger survey, a \(d\) of 15 km was chosen as it was deemed small enough to assume homogeneity of survey results.

Figure 3: Conceptual presentation of the area represented by each sampling point image-center

Draw a grid over the maps

A grid was drawn over the maps. The size of the grid was determined by the distance \(d\) decided earlier which in this case was at \(d = 15 \text {kms}\). The grid drawn was rectangular.

The width of the grid in the east-west (\(x\)) direction was calculated using the following formula:

\[ x = \frac {3d}{2} \]

The height of the grid in the north-south (\(y\)) direction can be calculated using the following formula:

\[ y = \frac {d \sqrt{3}}{2} \]

For the case of the Niger survey with \(d = 15 \text {kms} \):


Figure 4 shows the small-scale map of the five regions of Niger laid over with rectangular grids of \(22.5 \text {kms} \) long and \(13 \text {kms} \) high. Figure 5 shows the large-scale map of each region with rectangular grids of the same dimensions.

Figure 4: Small-scale map of five regions of Niger with rectangular grids

Dosso Maradi Tahoua Tillaberi Zinder
Figure 5: Large-scale maps of each of the five regions of Niger with rectangular grids

During the design phase of the survey, both small-scale and large-scale maps were used. Maps were printed on large format (A1 size) paper and same size acetates were placed on top. Appropriate grids were drawn onto the overlying acetates using non-permanent acetate markers so as not to damage the surface of the map and to make marking easily erasable when a mistake was made or a different grid needed to be drawn (see Figure 6).

Figure 6: Survey technical team drawing grids on large-scale maps image-center

A mapping specialist from UNICEF joined the survey design exercise. Based on the discussion of the type of mapping required for S3M, the mapping specialist signified that producing the appropriate maps with the appropriate grids on a geographical information system (GIS) used by UNICEF will be possible in future iterations of the survey method.

Create an even spread of sampling points

Sampling points were located at the intersections of the rectangular grid in a staggered fashion. Alternate intersections of the grid were used such as that shown in Figure 7 below:

Figure 7: Identifying sampling points at alternating intersections of the grid image-center

Sampling points were selected right to the edge and even over the edge of the survey area. Sampling points were identified and marked on the acetates with grids laid on top of the maps. Again, non-permanent acetate markers were used to allow for the marks for sampling points to be erased and moved somewhere else as needed (as discussed in the following step). Figure 8 shows the identified sampling points on the small-scale map of the five regions of Niger. Figure 9 shows the corresponding sampling points on the large-scale maps of each each region.

Figure 8: Identified samplings points on the small-scale map of five regions of Niger

Dosso Maradi Tahoua Tillaberi Zinder
Figure 9: Identified sampling points on the large-scale maps of each of the five regions of Niger

Select the communities to sample

To select where to sample, communities closest to the sampling points were chosen. For the coverage assessment of the CMAM programme, it was estimated that about three (3) communities per sampling point would need to be selected so that it can be ensured that SAM cases will be found at each sampling point. Based on previous experience of case finding in Niger for coverage surveys utilising CSAS, on average about 6-8 SAM cases have been found for every 3 to 5 communities sampled. For the IYCF practices, a target sample size of eighteen (18) children 0-23 months per community was aimed for. This translated to about fifty-four (54) children 0-23 months per sampling point.

The position of the sampling point was moved into the middle of the three (3) communities selected. This can be noted in Figure 10 below in which some sampling points were moved to where the communities are in the map.

Dosso Maradi Tahoua Tillaberi Zinder
Figure 10: Selected sampling points on the large-scale maps of each of the five regions of Niger

Label each sampling point

Each sampling point was given a unique identifying number to identify which community belongs to which sampling point. The label was used when collecting, organising, and analysing data.

Within-community sampling

The sampling process that was used to select a sample from a community depended on survey. For Niger, there were two surveys nested into the S3M sample. The first was the coverage survey of the CMAM programme and the second was the investigation of IYCF practices. Both surveys were done simultaneously in each community to be sampled with each survey with its own within-community sampling methodology.

For the coverage assessment of CMAM, either active and adaptive case finding or door-to-door screening method were utilised depending on the context and setting. The method that finds all, or nearly all, cases in the sampled communities was the method of choice. Based on Valid International’s experience in case finding during coverage assessments, active and adaptive case finding generally works well (i.e. finds all or nearly all cases) for communities in rural settings as within-community social networks and links are still intact. On the other hand, door-to-door screening works best in urban towns where the assumption of intact social connections cannot be readily assumed or expected.

Each case was confirmed by applying the programme’s admission criteria (e.g. MUAC < 115 mm and/or bilateral pitting oedema). When a confirmed case was found, the surveyors checked whether the child is in the programme. If the child was not in the programme, a simple questionnaire was administered and the child was referred to the nearest clinic providing CMAM services. The number of cases found in and out of the programme was tallied accordingly.

For the investigation of IYCF indicators, a variant of the standard EPI sampling method was used to find eligible children. This was done through the following steps (also see Figure 11):

  • Selection of the first household: Stood at a central point in the community and chose a direction at random (e.g. by spinning a bottle). Counted the houses between the central point and the edge of the community in that direction. Selected one of these houses at random Checked for eligible children. Started interview. All eligible children found in a household were administered the IYCF questionnaire.

  • Selection of subsequent households: Chose a new random direction from the previous house. Selected the third house in the random direction. Checked for eligible children. Started interview.

  • Stopping sampling: Sampling stopped when required number of children sampled.

EPI1 EPI2
Figure 11: EPI3 sampling used in the IYCF practices survey

Recording the data

For the coverage survey of the CMAM programme, the number of cases found, the number of these cases that are in the programme, and the number of non-cases that are in the programme (recovering cases) found were recorded on a simple tally sheet.

For the assessment of IYCF practices, individual children data were recorded using an IYCF questionnaire.

In all questionnaires and data collection forms, the unique sampling point identifier was recorded.

Using a GPS locator, the coordinates of the location of the centres of the sampled communities were recorded on the tally sheet. These coordinates were used to calculate the centroid of the villages that constitute a single sampling point during data analysis.

EpiData1 was used to perform double data entry and routine data checks and validation to assure data quality. Once checked and verified, data was saved in comma-separated-value (CSV) format.

Analysing data

All data analysis was done using R language for statistical computing2. Bespoke statistical scripts were coded using R for data handling and analysis.

The sampling points were connected to create triangles using Delaunay triangulation3. This creates unique triangular tiles of about 292 sq km area size.

For the coverage assessment, data on cases in the programme and total cases found from the three communities in each sampling point were aggregated by summation and a coverage proportion calculated per sampling point. Then, the mean of the coverage proportions at each of the triangulated sampling points was taken the result of which provided the coverage estimate for each of the triangular tiles.

For the IYCF practices, data was analysed to provide results on the following indicators:

  1. Proportion of exclusively breastfed children 0 - 6 months

\[ \frac {\text {Children 0 - 6 months exclusively breastfed}}{\text {All children 0 - 6 months}} \]

  1. Infant and child feeding index (ICFI) score The Infant and Child Feeding Index (ICFI) developed by Arimond and Ruel4 5 was applied to the data. The ICFI has 3 component indicators:

Age-appropriate continued breastfeeding

Age-appropriate dietary diversity \[ \frac {\text {Children 6 - 24 months with dietary diversity score of 2}}{\text {All children 6 - 24 months}} \]

Age-appropriate meal frequency

In addition to these three component indicators, a summary indicator of good ICFI was analysed as well. \[ \frac {\text {Children 6 - 24 months with ICFI score of 6}}{\text {All children 6 - 24 months}} \]

  1. Standard infant and young child feeding (IYCF) indicators These standard indicators are those recommended by WHO6 7. Of this set of 15 indicators, the following were the ones analysed and reported in this survey:

Proportion of exclusively breastfed children 0 – 6 months – as above

Minimum dietary diversity \[ \frac {\text {Children 6 - 24 months who received foods from} \ \geq 4 \ \text {food groups during the previous day}}{\text {All children 6 - 24 months}} \]

Minimum meal frequency

Minimum acceptable diet

Proportion of children practicing good infant and young child feeding This indicator was analysed and reported in 2 ways:

Proportion of children 0 – 24 months who are exclusively breastfed (for children 0 – 6 months) and children with ICFI score of 6 (for children 6 – 24 months)

Proportion of children 0 – 24 months who are exclusively breastfed (for children 0 – 6 months) and children minimum acceptable diet (for children 6 – 24 months)

For all the indicators above, proportions were calculated accordingly at each sampling point. The mean of the coverage proportions at each of the triangulated sampling points was taken the result of which provided the indicator proportion per triangular area.

Aggregated results at district and regional level and overall were calculated through bootstrapping8 approach utilising re-sampling with replacement of the sampling points replicated 2000 times. Then, weighted mean is applied to each of the 2000 replicates of the bootstrapped data and the 2.5th, 50th and 97.5th quantiles of the 2000 weighted means is calculated to provide the estimates and the upper and lower confidence intervals.

S3M implementation process

Design stage

A survey technical team was formed composed of representatives from the Nutrition Department, Institute of National Statistics and UNICEF. The survey technical team worked through the steps presented discussed above in preparation for the survey proper. The design stage took 1 week.

Training stage

This stage was aimed at training teams of surveyors in the conduct of the survey. Thirty persons were initially hired as surveyors and underwent this stage of the survey. After training, 20 trainees were retained and seven (7) teams were formed. The training of surveyors was done through learning-by-doing approach. Some classroom sessions conducted but on-the-job training was given emphasis during this phase. There was no separate training period per se as the survey started alongside the training phase. However, the training phase of the survey was done slower to allow for the surveyors to be trained and learn the survey skills and process.

The training stage was conducted in Dosso region because it had the least number of sampling points. The smallest region was chosen to begin with so that the teams can be closely supervised during the training. It also allowed for end-of-day debriefs of the teams with the lead trainer and surveyors (Figure 12).

EPI1 EPI2
Figure 12: Surveyor’s during their end-of-day debriefs

The survey mechanics were adjusted as appropriate during the early stages based on learnings from the training phase:

  • Adapted and refined the case-finding question for active and adaptive case finding

  • Shifted from EPI3 to QTR + EPI3 approach

  • Door-to-door approach added as a sampling method for very small villages

  • Team structure and hierarchy changed based on optimal team composition based on surveyor dynamics

  • Re-training and refresher training for surveyors done routinely

Survey manuals were produced for the supervisors and the surveyors which served as guides for the survey personnel for the duration of the survey process.

Implementation stage

Regions were segmented into 7 by route accessible by road (see Figure 13) and each segment was assigned to 1 team. A list of sampling communities in each segment was provided to the team covering that segment. Teams started out in the farthest sampling point in the segment and then worked their way back to the regional capital as they finish their sampling points.

Figure 13: Segmentation done for Dosso region

The choice of the next region to survey was made based on the following principle of increasing size of survey area, number of sampling points, and level of difficulty over time to match increase in confidence, competency and capacity of surveyors.

Using SQUEAC to investigate in detail barriers to service uptake and access

Semi-quantitative evaluation of access and coverage (SQUEAC) was used for more focused and in-depth investigation of factors to coverage in one of the districts that had a relatively high coverage (i.e. district with predominant triangular tiles of yellow to light green) and one of the districts that had a relatively low coverage (i.e. district with predominant dark to light red triangular tiles). This investigation approach follows the algorithm shown in Figure 14.

Figure 14: Algorithm for conducting SQUEAC after a S3M survey

The SQUEAC assessment was started by using the information on barriers to coverage gained from the S3M survey across the five regions as leads for further investigation. In-depth investigation of these barriers to coverage was then conducted through the typical semi-quantitative approach of the SQUEAC toolbox9. This included analysis of routine programme monitoring data and collection of qualitative information from the target communities, beneficiaries and the health staff. Through this process, a mind map10 of the barriers and boosters to coverage was produced.

Results

A total of 1435 villages were sampled from 477 sampling points spread evenly across five regions. Delaunay triangulation of the sampling points produced 937 unique triangular tiles each with an area of about 292 sq kms.

CMAM Coverage

The survey found 6,073 SAM cases of which 1,016 cases were assessed to be in the programme. Of these total cases, 594 (54 in programme) were from Dosso, 988 (225 in programme) from Maradi, 1145 (181 in programme) from Zinder, 1281 (252 in programme) from Tahoua and 2065 (304 in programme) from Tillaberi (see Table 1 for district, regional and overall summary of cases found).

Mapping of CMAM coverage results at the level of 937 triangular tiles across the five regions surveyed is shown in Figure 15.

Figure 15: Map of CMAM coverage in Tillaberi, Dosso, Tahoua, Maradi and Zinder regions of Niger

The overall coverage estimate for the five regions of Niger was 19%. Of the regions, Maradi region had the highest coverage estimate at about 24% and Dosso region had the lowest coverage estimate at about 12%.

SQUEAC investigation

Using coverage map from S3M (Figure 15), Dosso district in Dosso region with its dark red to light red triangular tiles (Figure 16) was chosen as the district with relatively low coverage while Tera district in Tillaberi region with its yellow to light green triangular tiles (Figure 17) was chosen as the district with relatively high coverage.

Figure 16: Map of coverage in Dosso region

<a name=”FIG17”</a> Figure 17: Map of coverage in Tillaberi region

SQUEAC was conducted in each of these districts to probe deeper into the factors leading to low or high coverage respectively. The detailed process for the Tera and Dosso SQUEAC is described in separate reports11 12. For this report, key findings are reported.

Tera district SQUEAC

The SQUEAC in Tera district was implemented in August 2012. In general the same positive and negative factors were identified throughout the whole district of Tera, without any major difference between the high and low coverage areas. Overall the majority of population knew about the CMAM programme and appreciated it. The signs of malnutrition are well-known and the community choose the health centres (CSI or CS) to seek health assistance.

Among the barriers discovered during the investigation, the most important ones were distance to the health centre (for the villages far away from the health centres) and lack of means to pay for transportation. Also as the investigation took place during the rainy season, rivers and other smaller bodies of water often prevented mothers from reaching the centres. Additionally there was a cholera epidemic to the north of the district and mothers would chose not to go to the centre due to fear of getting sick with cholera. In some areas the number of health centres was not adequate in relation to the distribution of the population with very few centres in the south of the district were most of the population was to be found. There was also a significant number of problems noted at the health centre level. Among the most significant ones were:

  • long wait for the CMAM consultation

  • stock outs (sometimes stock available at the district level but not transported to the health centres)

  • health centre staff being overworked and not well organised

  • MUAC was not an independent criterion for admissions

  • staff lacked motivation

  • community mobilisation was very limited and only in the villages closest to the health centres were volunteers active and known by the community

  • screening was also limited and done mostly at the health centre level or during mass screening in the villages 3 or 4 times a year

Overall it was noted that villages closest to the health centre were better informed about the programme and frequented the health centre more often. During the S3M survey there was a long rupture of RUTF (up to 6 weeks in some CSI) and this might have lowered significantly coverage in the area at that time. The communication between NGOs and district was also lacking. HELP, an NGO, was supporting all of the 35 health centres in the area, however with a very small team (only 8 people) they were not able to visit each health centres often enough to supervise their activities adequately. Additionally there was very little communication with the district staff and none with World Vision who is also working in the area and supporting a few health centres.

Dosso district SQUEAC

Dosso district was chosen for this investigation as the coverage result from S3M showed it had coverage of 5.35% which was the lowest among the districts in the 5 regions. In Dosso, the SQUEAC investigation took place in December 2012, more than a year after the S3M survey was undertaken in that district. Given this time gap, it was judged necessary to not only investigate barriers and boosters to coverage but also to check the actual coverage achieved in Dosso a year after the S3M. This meant adding steps in the SQUEAC that would allow for coverage estimation using Bayesian inference1 techniques. These steps included the building of a prior estimate and a likelihood survey.

In Dosso district where the coverage was very low, even more barriers to coverage were revealed.

  • Even though a large majority of the communities visited did know about the CMAM programme, there were still a few villages were communities did not know about the programme and did not recognize the signs of malnutrition.

  • Communities did not know the volunteers and community mobilisation was almost non-existent.

  • Often the health centre staff did not know how to implement correctly the CMAM protocol: how many RUTF packages to give to SAM children and for how long they should be in the programme to allow recovery.

  • MUAC was also not an independent criterion of admission.

  • The screening was not systematic at the health centre level and even if screened for malnutrition, the staff would often not explain to the mothers the treatment and the schedule of appointments to follow.

  • On occasion staff would ask for money for what should be a free of charge treatment in Niger.

  • The RUTF stock outs were not frequent; however they occurred and caused interruption of CMAM services for many weeks.

Information gathered through the qualitative methods and through interviews with key stakeholders was further strengthened when the likelihood survey was conducted. Among 49 SAM children found only 9 were covered in the programme. The questionnaires administered to the mothers of SAM children further confirmed the initial findings. Besides the reasons such as distance and lack of means to pay for transportation and a heavy workload, mothers were not frequenting the centres because they were not well treated, had to pay for treatment or were not screened for malnutrition. Bayesian analysis of the prior information from the investigation and the results of the likelihood survey, a coverage of 16.4% was estimated.

Comparison of Dosso and Tera district

Overall coverage in Tera district was higher than in Dosso district because the population was better informed and even though limited, the community mobilisation was more visible in Tera district. The treatment in the health centres was also more positive in Tera and mothers, as well as the community, were overall more satisfied. There were no other significant differences which can be highlighted to show that CMAM programme was better functioning in Tera than in Dosso district.

IYCF Indicators

The survey on IYCF practices sampled 24,434 children. There were 9 children aged greater than 24 months. There were 72 children who did not have records of their birth date or whose mothers could not or did not report their child’s age. Both sets of children (total of 81) were excluded from the IYCF sample. We kept in the sample children whose age based on birth date or self-report was 24 months13. Hence, a total of 24,353 children aged 0 – 24 months were included in the IYCF survey sample. This was the sample population for the good IYCF practices indicator. Of this total sample, there were 6,754 children aged less than 6 months. This was the sample population for the exclusive breastfeeding indicator. The remaining 17,599 children aged 6 – 24 months was the sample population for the rest of the IYCF indicators.

Exclusive breastfeeding

Mapping of the proportion of children less than 6 months who were exclusively breastfed across the five regions is shown in Figure 18.

Figure 18: Proportion of children 0 – 6 months exclusively breastfed

The overall proportion of children less than 6 months who were exclusively breastfed was 36%. This proportion was highest in Maradi region at 52% followed closely by that in Tahoua region at 50%. Dosso region had the lowest proportion of exclusively breastfed children aged less than 6 months at 13%.

Continued breast feeding

Mapping of the proportion of children aged 6 – 24 months who continue to breastfeed across the five regions is shown in Figure 19.

Figure 19: Proportion of children 6 – 24 months continuing breastfeeding

The overall proportion of children 6 – 24 months who continue breastfeeding was 42%. This proportion was highest in Tahoua region at 44% followed closely by that in Tillaberi region at 43% and that in Zinder region at 42%. Dosso region had the lowest proportion of children 6 – 24 months who continue breastfeeding at 36%.

Age-appropriate dietary diversity

Mapping of the proportion of children 6 – 24 months with age-appropriate dietary diversity across the five regions is shown in Figure 20.

Figure 20: Proportion of children 6 – 24 months with age-appropriate diet diversity

The overall proportion of children 6 – 24 months with age-appropriate dietary diversity was 10%. This proportion was highest in Zinder region at about 13% closely followed by that in Maradi region at 12% while Tillaber region had the lowest among the regions at about 8%.

Age-appropriate meal frequency

Mapping of the proportion of children 6 – 24 months with age-appropriate meal frequency across the five regions is shown in Figure 21.

Figure 21: Proportion of children 6 – 24 months with age-appropriate meal frequency

The overall proportion of children 6 – 24 months with age-appropriate meal frequency was 44%. Zinder region had the highest proportion at 47% and Tahoua region closely following at 46%. Dosso region was lowest at 38%.

ICFI score of 6

Mapping of the proportion of children 6 – 24 months with ICFI score of 6 across five regions of Niger is shown in Figure 22</a>.

Figure 22: Proportion of children 6 – 24 months with ICFI score of 6

The overall proportion of children 6 – 24 months with ICFI score of 6 was about 8%. Maradi and Zinder regions had the highest proportion at 9% and Tillaberi region the lowest at 6%.

Age-appropriate IYCF

Mapping of proportion of children 0 – 24 months practising age-appropriate IYCF across five regions of Niger is shown in Figure 23.

Figure 23: Proportion of children 0 – 24 months practising good IYCF

The overall proportion of children 0 – 24 months practising age-appropriate IYCF was 16%. Maradi region had the highest proportion of children 0 – 24 months practising age-appropriate IYCF at 21% followed closely by Tahoua region at 20%.

Minimum dietary diversity

Mapping of proportion of children 6 – 24 months with minimum dietary diversity across five regions of Niger is shown in Figure 24.

Figure 24: Proportion of children 6 – 24 months with minimum dietary diversity

The overall proportion of children 6 – 24 months with minimum dietary diversity across five regions of Niger was 6%. Dosso and Zinder region had the highest proportion of children 6 – 24 months with minimum dietary diversity at 8% while Tillaberi region was the lowest at 5%.

Minimum meal frequency

Mapping of the proportion of children 6 – 24 months with minimum meal frequency across five regions of Niger is shown in Figure 25.

Figure 25: Proportion of children 6 – 24 months with minimum meal frequency

The overall proportion of children 6 – 24 months with minimum meal frequency across five regions of Niger was at 76%. Tahoua region had the highest proportion at 78% while Tillaberi and Zinder region had the lowest both at 75%.

Minimum acceptable diet

Mapping of proportion of children 6 – 24 months with minimum acceptable diet across five regions of Niger is shown in Figure 26.

Figure 26: Proportion of children 6 – 24 months with minimum acceptable diet

The overall proportion of children 6 – 24 months with minimum acceptable diet was about 6%. Dosso region had the highest proportion among the regions at 8% while Tillaberi region had the lowest at 4%.

Consumption of iron-rich flesh foods

Mapping of proportion of children 6 – 24 months consuming iron-rich flesh foods across five regions of Niger is shown in Figure 27.

Figure 27: Proportion of children 6 – 24 months consuming flesh foods

The overall proportion of children 6 – 24 months consuming iron-rich flesh foods across five regions of Niger was about 19%. Dosso region had the highest proportion among regions at 26% while Tillaberi and Zinder region the lowest at 16%.

Consumption of Vitamin A-rich plant foods

Mapping of the proportion of children 6 – 24 months consuming vitamin A-rich plant foods across five regions of Niger is shown in Figure 28.

Figure 28: Proportion of children 6 – 24 months consuming vitamin A-rich foods

The overall proportion of children 6 – 24 months consuming vitamin A-rich plant foods across five regions of Niger was at 13%. Tillaberi region had the highest proportion at 17% with Dosso region a close second at 16%. Tahoua region had the lowest proportion at 9%.

Discussion

CMAM coverage

Spatial distribution

The spatial distribution of CMAM coverage in the five regions of Niger surveyed is homogenous towards low coverage with majority of the triangular tiles having coverage below what is expected of a CMAM programme. If we are to base it on Sphere14 which set standards for coverage of therapeutic feeding programmes in a rural setting to be greater than 50%, then only 1% of the 937 triangular tiles meet standards.

It is also important to note that there are significant “hot” areas of very low coverage (i.e. coverage of 0%) such as in Tillaberi region particularly Filingue district and in Zinder region in Goure district on the border to Diffa region. However, there is also relatively “cooler” areas of moderate level coverage (i.e. greater than 20%) such as parts of Tera and Ouallam district in Tillaberi region, small area in Dogonduchi district in Dosso region, parts of Keita district in Tahoua, Tessaoua district in Maradi region and parts of Miriah and Goure district in Zinder region.

This spatial pattern of CMAM coverage observed in the five regions seem to be partly explained by the level and length of involvement of various international NGOs in particular areas. Particular districts of Tahoua, Maradi and Zinder region have long been benefiting from CMAM programming with the support of various international NGOs as compared to Tillaberi and Dosso regions where NGO involvement and support is present but not in the same intensity and length of time as the other three. This is probably why certain parts of Tillaberi and Dosso regions have shown relatively better coverage levels than the rest of the region. For example, parts of Tera district in Tillaberi that are showing relatively better coverage levels have recently experienced an influx of NGOs responding to the on-going crisis on the border to Mali. With this level of attention associated with the on-going emergency at the time of the survey, it was to be expected to observe peaks in coverage levels during the acute phase of the emergency9.

Barriers to service uptake and access

Information on barriers to service uptake and access collected from mothers of SAM children not in the programme through a simple questionnaire provides additional insight as to the low coverage achieved by CMAM across the five regions of Niger. Issues with service delivery were consistently the most common reason for coverage failure in all the five regions surveyed. These issues included RUTF stock outs, SAM cases not being identified routinely in daily health clinic child survival activities and SAM cases treated as moderate acute malnourished children using corn soya blend (CSB) rather than RUTF. Other problems raised by mothers of SAM children not in the programme were issues with screening and referral of SAM cases for appropriate treatment and a lack of awareness and understanding of the mechanics of the programme. The problems with 1) RUTF stock outs; 2) screening and referral of SAM cases; and, 3) lack of awareness and understanding of how the programme works contributed to at least 50-60% of the reasons for non-covered SAM cases in each of the regions and overall.

RUTF stock out

RUTF is provided by UNICEF at the district level. From there, district MoH and NGOs are responsible for ensuring that the RUTF gets to the health centres. It is at this stage of the RUTF distribution chain that problems arise which leads to RUTF being available at the district level but not at the health centre level. Hence, the problem is not that there is no RUTF but that there is a break in the distribution chain going to the health centres. This problem is believed to be common throughout all the regions but in varying degree of frequency. This is also most likely the main reason why Filingue district in Tillaberi and parts of Dosso region had significantly low levels of coverage.

The issue of RUTF stock out is a relatively new problem faced by CMAM programmes everywhere but is now becoming a significant issue impacting on its coverage. This is most likely due to the current push for global and national CMAM scale up particularly in contexts such as Niger where acute malnutrition is highly prevalent throughout the country. During the early days of CMAM in Niger (c. 2005 to 2010), coverage surveys done of the small area and small scale CMAM programmes implemented by international NGOs mostly in districts in Maradi and Zinder regions didn’t elicit any reason of RUTF shortage or stock outs for non-coverage. However, by late 2010 to 2011 at the time of wider national scale up of CMAM in Niger, coverage surveys done during this period started showing RUTF supply issues as a barrier 15 16 17. It is very likely that the issue of RUTF stock outs is because the mechanism in place for RUTF distribution down to the the health centres and the capacity of the relevant regional / district MoH and / or NGOs to deliver this mechanism has not been able to keep in touch with the rate of the scale up of the programme. Hence, in places where there are historically stronger NGO support to the MoH such as Tahoua, Maradi and Zinder, coverage has been relatively better while areas such as Dosso and Tillaberi region where initiation of CMAM has been primarily led by MoH and has only recently been supported by NGOs. This is not to say that NGO-led CMAM performs better. On the contrary, there is some evidence in areas such as Tahoua district where communes served solely by a MoH-run programme are achieving coverage levels as good if not better than their NGO counterparts in the same district. What this illustrates is the fact that any scale up process of CMAM at a wide scale such as that in Niger is challenging and riddled with difficulties and what we are seeing are the birthing pains and challenges typical of CMAM start-up coming to the fore. A good comparison is the case of Sierra Leone which has scaled up CMAM throughout the country around about the same time period as Niger. Although the two countries are incomparable in terms of their geographic size, a national coverage survey done in Sierra Leone in 2011 identified similar issues impacting coverage such as RUTF stock outs18.

Screening, identification, referral and admission for treatment of SAM cases

The process of screening, identification, referral and admission for treatment of SAM cases is problematic. The current setup uses a two-stage screening and admission process in which screening, identification and referral at the community level uses MUAC < 115 mm as the standard measure while for admission to treatment at the health centre level, WHZ < -3 z-score is the measure that is applied. For majority of SAM cases, this is not an issue as both MUAC and WHZ identify these same children as SAM19 20. However, there will be a group of children that will fall under the category of being SAM by MUAC but not by WHZ and those who are SAM by WHZ but not by MUAC. Myatt et al19 in their review of methods to detect severely malnourished children in the community found that if a MUAC-only criteria for detecting SAM cases is used, the number of excluded low WHZ children is small (6.53%) relative to the overall need. However, if a WHZ-only case definition is used, the number of excluded low MUAC children is large (37.31%) relative to the overall need.

Endnotes



  1. Lauritsen, J M. “EpiData Data Entry, Data Management and Basic Statistical Analysis System,” Odense, Denmark: EpiData Association, n.d. http://www.epidata.dk. 

  2. R Core Team. R: a Language and Environment for Statistical Computing, Vienna, Austria: R Foundation for Statistical Computing, 2014. http://www.R-project.org/. 

  3. Isaaks, E H, and R M Srivastava. An Introduction to Applied Geostatistics, Oxford: Oxford University Press, 1989. 

  4. Arimond, Mary, and Marie T Ruel. Generating Indicators of Appropriate Feeding of Children 6 Through 23 Months From the KPC 2000+, Washington, D.C.: Food and Nutrition Technical Assistance, Academy for Educational Development, November 2003 

  5. Arimond, Mary, and Marie T Ruel. “Summary Indicators for Infant and Child Feeding Practices: an Example From the Ethiopia Demographic and Health Survey 2000 (Pub. 2002),” March 29, 2004, 1–78. 

  6. World Health Organization. Indicators for Assessing Infant and Young Child Feeding Practices: Conclusions of a Consensus Meeting Held 6–8 November 2007, Washington, DC, 2008. 

  7. World Health Organization. Indicators for Assessing Infant and Young Child Feeding Practices Part 2: Measurement, Geneva: World Health Organization, 2010. 

  8. In statistics, bootstrapping is a method for assigning measures of accuracy to sample estimates. This technique allows estimation of the sampling distribution of almost any statistic using only very simple methods. Generally, it falls in the broader class of resampling methods. 

  9. Myatt, Mark, Ernest Guevarra, Lionella Fieschi, Allison Norris, Saul Guerrero, Lily Schofield, Daniel Jones, Ephrem Emru, and Kate Sadler. Semi-Quantitative Evaluation of Access and Coverage (SQUEAC)/ Simplified Lot Quality Assurance Sampling Evaluation of Access and Coverage (SLEAC) Technical Reference, Washington, DC: FHI 360/FANTA, 2012.  2

  10. A mind map is a diagram used to visually outline information. A mind map is often created around a single word or text, placed in the centre, to which associated ideas, words and concepts are added. Major categories radiate from a central node, and lesser categories are sub-branches of larger branches. 

  11. Benda, B., 2012. Evaluation Semi-Quantitative de l’Accessibilité et de la Couverture (SQUEAC) District Sanitaire de Tera, République du Niger Du 23 Juillet au 3 Août 2012, Oxford: Valid International. 

  12. Benda, B., 2013. Evaluation Semi-Quantitative de l’Accessibilité et de la Couverture (SQUEAC) District Sanitaire de Dosso, République du Niger Du 3 au 14 Décembre 2012, Oxford: Valid International. 

  13. Children aged 24 months were included in the sample (27 children). Their ages were all self-reported and we assumed that if mothers are asked the age of their children, the tendency will be to round off to the nearest age in years. So, these 27 children may have had real ages of either 23 or 25 months give or take. Given that there were only 27 children of this age, including them in the sample was deemed acceptable. 

  14. The Sphere Project. The Sphere Project: Humanitarian Charter and Minimum Standards in Humanitarian Response, The Sphere Project, 2011. 

  15. Norris, A., 2010. Evaluation Semi-Quantitative de l’Accessibilité et de la Couverture (SQUEAC): Districts de Tessaoua (Maradi) et Kantché (Zinder) Niger. ed. Oxford: Valid International, Ltd. 

  16. Norris, A., 2011. Evaluation Semi-Quantitative de l’Accessibilité et de la Couverture (SQUEAC): CSI appuyés par World Vision ADP de Kornaka West, Gobir Yamma, Chadakori et Goulbi Kaba Région de Maradi République du Niger. ed. Oxford: Valid International, Ltd. 

  17. Mounier, B., 2012. Évaluation Semi-Quantitative de l’Accessibilité et de la Couverture (SQUEAC): Départment de Keita, Région de Tahoua, Niger. ed. London: Action Against Hunger-UK. 

  18. Guevarra, E., Guerrero, S. & Myatt, M., 2012. Using SLEAC as a wide-area survey method. Field Exchange, (42), p.40. 

  19. Myatt, M., Khara, T. & Collins, S., 2006. A review of methods to detect cases of severely malnourished children in the community for their admission into community-based therapeutic care programs, World Health Organization.  2

  20. World Health OrganizationUnited Nations Children’s Fund, 2009. WHO Child Growth Standards and the Identification of Severe Acute Malnutrition in Infants and Children, Geneva: WHO and UNICEF.