Nutritional status and disease severity in children acutely presenting to a primary health clinic in rural Gambia

Quality health information is a prerequisite for improving clinical practice, informing public health approaches and guiding health policy. Health information systems draw data from a range of sources including: epidemiological studies, birth and death registers and health facility records. John Fry’s work on common diseases provides a historical example of how primary care data can provide a window into the health of a population [1]. Migration from paper to electronic health records has further facilitated the utilisation of health facility data [2], which can be used for auditing, quality improvement and epidemiological purposes. In the absence of well-designed longitudinal surveys–which are generally required to quantify the burden of infectious disease morbidity [3]–health facility data can provide useful insights into population morbidity profiles. Given the expense and level of technical capacity to manage such systems, few low-income countries have implemented one [4‐6].

There are limited examples in the published literature where routinely collected primary care data from sub-Saharan African countries has been used for epidemiological purposes (see additional file 1, section 1 for literature search criteria). A well-defined cohort study in South Africa used primary care data to describe the disease profiles of 1357 children under 5 years presenting to primary care clinics between 2006 and 2007 [7]. Data from 4026 prescriptions given on the ‘Phelophepa’ primary care train in South Africa were also used to determine prescribing characteristics [8]. However, we are not aware of any analysis using primary care data to describe the morbidity profiles of a population in a West African setting.

This is the first study utilising clinical data collected via an electronic medical records system for all children under 60 months presenting to a primary health care facility in the Kiang West district, The Gambia [9]. We sought to provide an insight into illness severity and mortality of clinic presentations in relation to nutritional status of children under 60 months making acute presentations to the clinic.

Between 1st January 2010 and 31st December 2014, the cumulative under 60 months old population of Kiang West was 5021. Of these, 3839 (77%) made at least one acute visit to the Keneba clinic, with a total of 21,278 visits (median of 5 visits per individual over the reporting period (IQR = 7). In addition, 7452 non-acute visits (follow-up and welfare clinic appointments) were made by 1495 children, of those only 39 children made non-acute visits to the clinic. The median age at presentation for acute clinic visits was 20.2 m (IQR = 25.9) with 47% of visits made by girls (Table 1).

A total of 160 deaths in children under 60 m were recorded in Kiang West over the period (median age at death was 1.3 m, IQR = 12.3) representing an under 60 m mortality rate of 51 deaths per 1000 live births. Of these, 31% (50/160) presented to the clinic at any point prior to death, with a median time of 10 days (IQR = 151) between final presentation to the clinic and death. Table 1 shows the characteristics of both those who did and did not attend the clinic prior to death. Children were less likely to present to the clinic at any point prior to death if death occurred in the neonatal period (OR = 0.10, 95% CI = 0.04–0.24 p < 0.001) or infancy (OR = 0.15, 95% CI = 0.07–0.32 p < 0.001). The distance to the clinic for those who did not present prior to death (median = 21 km (IQR = 12) was significantly greater than for those who did present (median = 13 km (IQR = 9, p = 0.02), irrespective of age at death. (Table 1).

The median number of presenting complaints per visit was two (IQR = 2) with 1242/21,278 visits having no complaint recorded. The most common complaints were fever and cough, noted in 73% (15,478/21,278) and 52% (10,960/21,278) of visits respectively. Of the 26,001 diagnoses (median of 1 diagnosis per visit (IQR = 1), 22,456 (86%) were for infectious diseases. Antibiotics were prescribed in 67% (14,327/21,278) of all visits, (for full details of presenting complaints, diagnoses and prescriptions see Additional file 1, section 4), including to 31% of children with common cold as their sole diagnosis.

In 4.5% of visits (961/21,278) patients were classified as being severely ill. Disease categories differed significantly between those with and without severe illness (Table 2). Pneumonia was the most common diagnosis in patients with severe illness (287/961, 30%) (see additional file 1, section 4). The risk of severe disease declined with age (Table 3). Of the 50 children to present to the clinic prior to death 72% (18/25) were severely ill at their last clinic visit.

Data on wasting (WHZ) were available for 20,478/21,278 visits (96%) (outside plausible range n = 51, missing n = 749) and on stunting (HAZ) for 20,545/21,278 visits (97%) (outside plausible range n = 122, missing n = 611) (Table 3, for demographics of children with and without missing anthropometric data see Additional file 1, section 5). Children over 12 m old had significantly higher prevalence of mild, moderate and severe stunting than infants, whereas the prevalence of wasting was similar between the age groups (additional file 1, section 6). The coexistence of stunting in children classified as wasted was 33% (788/2405) and the presence of wasting in children classified as stunted was 18% (788/4389). Nutrition-related diagnoses accounted for 14% (188/1330) of all severe illness diagnoses, compared with 3% (663/24,670) of non-severe illness (Table 2): OR 5.97 95% CI:5.02–7.09, p < 0.001. Both mean HAZ and WHZ where significantly lower in children with severe illness (Table 3). Of the 50 children who presented to the clinic prior to death 33 (66%) presented in the 101 days preceding death, with WHZ data available for 30 of these individuals (outside plausible range n = 1, missing n = 2). The mean WHZ of these 30 children at their last clinic visit before death (mean = − 1.45, SD = 1.65) was significantly lower than that of children during other acute clinic visits (mean = − 0.70, SD = 1.15; W = 3.68, p = 0.04).

Wasting was significantly associated with a higher risk of severe illness in a dose-dependent manner (‘risk of wasting’ aOR 1.68, 95% CI:1.43–1.98, p < 0.001, ‘moderate wasting’ aOR 2.78, 95% CI:2.31–3.36, p < 0.001 and ‘severe wasting’ aOR 7.82, 95% CI:6.40–9.55, p < 0.001). The association between different grades of wasting was observed for all specific infectious diseases, with the exception of severe bronchiolitis (Table 4). The fraction of severe illness attributable to severe wasting, calculated as PAF, was 0.11 (95% CI:0.09–0.13) and for wasting overall 0.21 (95% CI:0.18–0.24). Including those at risk of wasting increased the PAF to 0.35 (95% CI:0.29–0.40). Stunting on the other hand, even in its severe form was not significantly associated with illness severity. Both severe stunting (aOR 6.04, 95% CI: 1.94–18.78, P = < 0.001) and severe wasting (aOR 9.33, 95% CI:3.39–25.70, P < 0.001) at the last clinic presentation were however associated with a significantly increased risk of subsequent death (Table 4).

This is the first study to analyse data collected routinely by an electronic medical records system in West Africa, and is one of only a small number of studies to use primary health care data to assess disease patterns and nutritional status in a sub-Saharan African context [7, 8].

The main strengths of our study lie in the magnitude of our data set, with over 21,000 acute clinic visits over a 5-year period, and the robust electronic capture methodology. For a rural primary care setting, our data are highly complete, with clinic outcome data for 99% of all visits. Diagnoses in our study were solely made by paediatricians or general physicians as opposed to nurse-led clinics which are more common in such settings [16] and although misclassifications may be present [17], the use of ICD-10 coding makes our results easily comparable to other studies.

Infectious diseases formed the majority of the morbidity burden in our study regardless of clinical severity reaffirming the need for focused efforts aimed at both preventing and treating these illnesses [18]. Both severe wasting and severe stunting were associated with an increased risk of death, similar to previous findings [19‐21]. Wasting was a strong risk factor for severe illness, overall and for the most common infectious diseases in a dose dependent manner. In the adjusted model, stunting–even when severe– was not associated with an increased risk of severe illness. When estimating the fraction of severe illness attributable to WHZ the inclusion of those at risk of wasting increases this from around one-fifth to almost one-third. Our findings that children at risk of wasting (WHZ ≤ -1 to > − 2) have an almost 70% increased risk of severe illness compared to children who are not classified as at risk (WHZ > -1), indicates that children who would not traditionally be identified as being acutely malnourished are still more vulnerable to severe illness than their better nourished counterparts. This reemphasises the need for a range of actions that promote optimal nutrition in young children, while ensuring that therapeutic measures are in place for those in need.

Community management with ready to use therapeutic food has proven highly effective in treating children with severe wasting [22, 23]. Coverage of therapeutic programmes can however be low, due to programmatic constraints, or a lack of awareness of malnutrition or the therapeutic services in communities [24]. Recent advances in the formulation of therapeutic foods could have important implications for the cost and logistics currently associated with these programme, [25] which may lead to increased provision of these services. The effectiveness of community treatment for moderate wasting is unclear from both programmatic and policy perspectives [26‐28]. Currently the routine provision of supplementary foods to children with moderate wasting is not recommended [29].

Often the dietary guidance provided for carers of children with moderate wasting is no different to that of well-nourished children, while they need guidance that is locally adapted and promotes nutritionally adequate diet for their physiological state [30].

A range of efficacious nutrition interventions are available, that when delivered at high coverage could significantly reduce the burden of wasting and stunting [31]. Nutrition educational strategies and mass media interventions can be effective at improving infant and child feeding practices, yet we need to know more about the specific design components of these interventions that deliver the greatest impact [32]. There is also evidence multi-faceted nutrition specific interventions deliver greater benefit over singular focused programmes, [33]. Additionally, to address both the proximal and distal determinants of malnutrition and ensure sustained impact, nutrition interventions need to be implemented in conjunction with other nutrition sensitive public health measures, including social safety nets [34].

Despite a long history of nutritional intervention studies in Kiang West [35] the level of malnutrition in our study population equates to a serious public health problem [36], although this is not dissimilar to the situation in many low-income countries [37, 38]. Our results highlight the importance of early identification of poor nutritional status in children. The collection of multiple anthropometric measurements enabling the assessment of growth velocity rather than one off measurements could be a more effective strategy [39], however in many contexts this approach may prove impractical [40]. This challenge could partly be overcome by the use of simple measurements that caregivers can conduct at home regularly [41]. There is growing evidence for the effect of mHealth interventions [42], and we believe technology innovations, such as an SMS based platform like Rapid Pro [43], could be used for self-reporting and to facilitate focused guidance for caregivers and to enable remote follow up.

As well as identifying specific interventions, striking a balance between the different delivery platforms for nutrition services – health facility, community and out-reaches [44] – with respect to effectiveness for nutrition outcomes and cost remains challenging. However, as a contact point where help is sought for common issues, such as cough and fever, rural clinics present an ideal place where targeted assessment, care and referrals can be provided. This is further supported by the observations that the outcomes of community-wide preventative nutrition supplementation programmes in children have been conflicting, particularly in terms of morbidity, which may relate to a lack of targeting for those most in need [45, 46]. Equally, nutritional supplementation in those with severe illness can also have variable effects on subsequent malnutrition [47] and morbidity [48] supporting targeted intervention at primary health care level. For such an approach to be successful there is a need to build the capacity and knowledge of front line health workers around nutrition [49], to strengthen community structures and build better linkages between primary care and community services [50].

Effective and accurate management of the primary complaint at presentation is however paramount. This study not only highlights the increased risk of severe illness with undernutrition, leading to our call for targeted preventative measures in primary care, but also highlights again the lack of adherence to clinical guidelines [51]. Over two-thirds of all children received antibiotics, including one-third of children with common cold. Inappropriate prescribing of drugs remains a global issue in the era of antimicrobial resistance [51‐53]. Interventions aimed at both supply and demand side of prescription medications are required to redress the overuse of these medications, with the identification of local barriers to change important [54].

We acknowledge that our study has a number of limitations. While Kiang West has many similarities to other rural sub-Saharan African settings, the provision of high-quality free health services is unlikely to be one of them, although health and medication provision is also free in other government run health facilities in The Gambia. Information on a number of key factors related to nutritional status were also not available, including exclusive breastfeeding in infants under 6 months of age. Verbal autopsies were not available for the registered deaths and the majority of children who died did not present to the clinic prior to death, limiting our ability to analyse the majority of child deaths. However, our study highlights the need for approaches to reduce neonatal deaths and further examine the impact of improving access to care, especially in the neonatal period and for those furthest from primary care sites. The PAF is based on an assumption of causality between the exposure and outcome, however we recognise that the interactions between nutritional status and infectious disease are not unidirectional, and that our results do not prove causality. Further investigation is required in order to unravel the sequence of events in complex nutrition-infection interactions. [55]

The association between nutritional status and illness is complex, yet we believe our findings provide a strong case for further investment in improving child nutrition, especially in those at risk or suffering from wasting, as a key strategy for preventing acute disease and promoting child survival. While there are a number of efficacious nutrition interventions, knowledge gaps remain, including specific guidance on the treatment of moderate wasting. Early identification of poor or worsening nutrition status is an important factor for addressing malnutrition, the role of technology in facilitating this is worthy of further exploration. Operational research is also required to strengthen the evidence of how to take proven interventions to the scale required for maximum impact.