Disaggregation of health and nutrition indicators by ageand gender in Dadaab refugee camps, Kenya

By Henry Mark

Henry recently graduated with a BSc in Food and Human Nutrition from Newcastle University. He has conducted research in The Gambia and interned with UNHCR at the regional office in Nairobi, Kenya.

The author would like to give special thanks to UNHCR staff Allison Oman, Senior Regional Nutrition and Food Security Officer and Caroline Wilkinson, Senior Nutrition Officer, for their technical input and guidance. With thanks also to Ismail Kassim and Gloria Kisia.

Adequate nutrition in the early years of life is vital for optimal growth and development. However, within the under 60 month (less than 5 years) age group, the nutritional needs vary widely. Despite this, health and nutrition data are often not disaggregated throughout this age range.

The United Nations High Commissioner for Refugees’ (UNHCR) Health Information System (HIS) was developed to be a standardised tool with the aim of helping to guide, monitor and evaluate all public health programmes within UNHCRs operations. The data derived from this system are essential in forming evidence based decisions on projects and interventions¹. HIS is operational in more than 90 refugee camps in 18 countries worldwide, providing monthly data on a number of critical indicators across several sectors (e.g. camp population, growth monitoring, nutrition programmes).

The Dadaab refugee camps in the North East of Kenya were originally opened in 1991. The camps have remained open since, due primarily to continued insecurity, violence and on-going food crises in Somalia. In Dadaab, most data are disaggregated in terms of broad age groups (less than 5 years, 5 years and older) and gender. However, it is uncommon for health and nutrition programmes in Dadaab to disaggregate health and nutrition indicator data for the under 60 month age group. In an exception to routine practice, for a 16 month period spanning 2007-2008, data for a number of key health and nutrition indicators were disaggregated into sub groups for children less than 5 years. This article summarises the results of examining thus disaggregated data for children under 60 months in Dadaab in terms of mortality (allcause), morbidity (watery diarrhoea only) and admissions to feeding programmes.

Objectives

The objectives of the research were:

• To determine whether gender- and agedisaggregated data in children under 60 months of age were more effective at measuring morbidity and mortality rates than data aggregated across the age range.
• To investigate how representative aggregate age data for children less than 60 months of age compared with each disaggregated age group individually.
• To investigate whether disaggregated data have the potential to be a more accurate tool to guide and monitor health and nutrition programmes.

Methods

During the 16 month period, data were collected by UNHCR and their implementing partners using HIS protocols. The under 60 month age range was disaggregated into four sub-groups for age (< 6 months, =6-<12 months, =12 -<24 months, =24 -<60 months) and aggregated as < 6 months.

Morbidity and mortality data were classified by gender as is standard practice in all UNHCR programmes utilising the HIS. Feeding programme admission data was not disaggregated by gender (standard practice also).

Monthly data for each indicator were provided for all three of the camps that were operational at the time of data collection: Dagahaley, Hagadera and Ifo. Data across the age range were aggregated retrospectively (referred to as ‘age aggregate’ data) for comparison with the individual age group data in this study. The aggregate data depict how data for these indicators would have appeared had disaggregation not been conducted during this time period.

Morbidity data were collected at health centres, within each of the camps. Mortality data were collected in health centres and by community health workers throughout the camps. Accurate monthly population data were provided by UNHCR for each of the individual age groups and genders allowing mortality and morbidity rates to be calculated. The data displayed for both mortality and morbidity represent deaths (all causes) or cases of watery diarrhoea per 1000 of the specific population per month respectively. All data represented in this article are an aggregate of the three camps and are referred to as ‘Dadaab’.

Nutrition programmes available in the camps are supplementary feeding programmes (SFP) for moderate acute malnutrition (MAM) and therapeutic (TFP) and outpatient programming (OTP) for severe acute malnutrition (SAM) management. Feeding programme admissions were collected on a monthly basis by health workers in each of the camps and were disaggregated for each of the four age groups, but not for gender. Data for the three camps were compiled retrospectively to provide aggregate ‘Dadaab’ data. Feeding programme admission rates were calculated as number of admissions per 1000 of the specific population per month.

Results

Sample characteristics

Table 1 shows the mean sample characteristics for the data collection period. The sex ratio shows that within each age group, the proportion of males to females was comparable.

Mortality variations between genders

General Linear Model (GLM) ANOVA was conducted on total pooled data to assess the overall gender variation in mortality. The analysis suggested there was a significant difference in mortality rate between genders (p=0.003). However, one-way ANOVA conducted on data grouped by age suggested there was no significant difference (p=0.401).

Mortality variations between age groups

Figure 1 shows the mean mortality rate for each age group in Dadaab. One-way ANOVA was conducted to assess the variation between each individual age group and the aggregate age mortality data, within each gender. The findings of the analysis are summarised in Table 2. One-way ANOVA was also conducted to assess the mortality rate variation between each of the age groups individually.

The mortality rate for female infants under 6 months of age was significantly higher than the aggregate age mortality data. A significant difference also existed between the =12 < 24 months and =24 <60 months age groups and the aggregate age mortality data in both genders. Both the =12 <24 months and =24 < 60 months age groups had a significantly lower mortality rate than the aggregate age data.

There was no significant difference in the mortality rate between the <6 month and =6 <12 months age groups. However, there was a significant difference in mortality rate between both of these age groups and the =12 <24 month and =24 < 60 month age groups observed in both genders (see Figure 1).

Morbidity (watery diarrhoea) variations between genders

GLM ANOVA was conducted to assess the variation in morbidity (watery diarrhoea) rates between genders. No significant variation was detected (p=0.578).

Morbidity (watery diarrhoea) variations between age groups

Figure 2 shows the mean morbidity rate for each age group in Dadaab. One-way ANOVA was conducted to assess the variation between each individual age group and the aggregate age mortality data, within each gender. Oneway ANOVA was also conducted to assess the morbidity rate variation between each of the age groups individually.

All of the individual age groups were significantly different to the aggregate age morbidity data, as summarised in Table 3. The only two age groups that did not have a significantly different morbidity rate compared with each other were infants <6 months and =6 <12 months of age. Figure 2 indicates that the morbidity rate was significantly higher in the two youngest age groups, with a sharp decrease in morbidity rate in children =12 month of age.

Feeding programme admission rates

Admission rates to all of the nutrition programmes provide an insight into the nutrition situation within each of the age groups during the study period (see Figure 3). There is low availability of treatment for MAM and SAM in infants <6 month of age that affects the admission rate for this age group which is very low (rate/1000 of population/month is SFP 2.9, TFP/SC 7.6 and OTP 1.7). The lack of treatment options for this age group is a concern, given the high rates of morbidity and mortality observed in this age group (as highlighted above).

With the exception of the <6 month age group, Figure 4 shows a clear trend in decreasing admissions to all nutrition programmes with age from 12 months. This trend mirrors the decrease seen in the mortality and morbidity data for the same age groups. There was a significant fall in the admissions rates between the =6 <12 month and =12 <24 month age groups, for both SFP (p=0.021) and TFP/SC (p=0.001).

These variations mean that aggregate data for the <60 month age group provides a significant underestimation of the admission rates for the =6 <12 months and =12 < 24 month age groups for all of the feeding programmes. Equally, the aggregate data provide an overestimation of admissions to all of the feeding programmes in the =24 <60 months age group.

Discussion

Although the refugee camp context is unique and may not reflect the situation in which many health programmes operate, the findings presented here pose interesting considerations that apply to the wider humanitarian sector, as well as to other refugee settings.

When mortality data were pooled for age, a significant difference in mortality rate was detected between genders. However, this association was not seen when data were grouped for age, suggesting that significant amounts of data may be needed in order to detect such differences. This trend has also been noted in previous research in non-refugee populations in Kenya². To determine if gender disaggregated indicators for mortality are beneficial, data collection over a longer time period is needed. This would ascertain if there are any significant gender differences in mortality within each age group. There is also a need to determine if any scenarios, such as a quick onset emergency, may impact more significantly upon a specific gender, perhaps through a change in health seeking behaviours of caregivers.

In female infants <6 months of age, there was a significant difference in mean mortality rate compared with the aggregate age data. The mortality rate in male infants < 6 months of age was not significantly different from the aggregate age mortality data. However, the analysis showed there may be a weak association (p=0.087) and this is worth further investigation. Furthermore, the aggregate age data gave a significant overestimation of the mortality rates in all age bands for all children over 12 months of age for both genders.

A significant difference was seen in the morbidity rate between all of the individual age groups and the aggregate age data, for both genders. In this instance, the aggregate morbidity data gave a significant underestimation of morbidity in the youngest two age groups and a significant overestimation in the eldest two age groups.

Aggregate data may detect differences in the overall morbidity and mortality trends in a population. However this analysis strongly suggest that aggregate age data for infants and children less than 60 months of age does not accurately represent morbidity and mortality throughout the age range. Perhaps most importantly in relation to health and nutrition programmes, the aggregate data fail to reflect the high morbidity and mortality rates in infants <12 months of age and the strong, clinically meaningful decrease in morbidity and mortality rate with increasing age. This trend is mirrored in nutrition programme admissions that decrease with increasing age. Disaggregated data may also allow for more accurate and timely responses to changes in the health and nutrition status of specific population groups, guide and monitor specific health and nutrition interventions and also help improve targeted resources allocation.

Public health and nutritional interventions may target specific age groups (e.g. complementary feeding interventions to children =6-<24 months of age, exclusive breastfeeding support to infants < 6 months) but detection of impact is masked by aggregate data. Given the urgent need to strengthen the management of acute malnutrition in infants <6 months, the collection and presentation of health and nutrition data in infants < 6 months helps to define the caseload and spotlight the need.

The age categories used for disaggregation in this study are closely matched to significant changes in the nutritional requirement and development of infants and young children. In some situations it may be suggested that the >12 month age groups could be aggregated. However, given the significant difference in both the morbidity rate between the =12 <24 and =24 <60 month age groups and the admission rates to the feeding programmes, the four age groups used in this study would seem appropriate for future data collection.

This study presents a strong case for age disaggregation of data in children <60 months. Further research is needed to determine the cost:benefit (financial, capacity, time) of such disaggregated data collection, the additional burden this would place on existing health and nutrition programmes and their implementing partners and how to address these. The implications will partly depend on prevailing practice that does vary. For example, some implementing partners in Dadaab routinely record only aggregate data at data collection point (e.g. a child is only classified as <5 years), while others collect data by smaller age groups (e.g. < 6 months) that are aggregated for reporting. An agreed standardised approach to age disaggregation would greatly enable comparisons between UNHCR programmes, as well as with non-UNHCR programmes. There is a need to conduct further research with a larger sample size and within different settings in order to better understand the dynamics and strengthen efficacy of gender and age disaggregated data collection in this age range in different humanitarian contexts.

For more information, contact: Allison Oman, email: oman@unhcr.org or Henry Mark, email: hmark3483@gmail.com

¹UNHCR. Health Information System. 2011 [cited 2012 09/03/2012]; Available from: http://www.unhcr.org/pages/49c3646ce0.html. See also research piece on HIS in this issue of Field Exchange.

²Hill, K. and D.M. Upchurch, Gender differences in child health - evidence from the demographic and health surveys. Population and Development Review, 1995. 21(1):p.127-151.