Article Text

Download PDFPDF

Less diarrhoea but no change in growth: 15 years’ data from three Gambian villages
  1. E M E Poskitta,
  2. T J Coleb,
  3. R G Whiteheadb
  1. aMRC Dunn Nutrition Group, PO Box 273, Keneba, Banjul, The Gambia, bMRC Dunn Nutrition Unit, Downham’s Lane, Cambridge CB4 1XJ, UK
  1. Dr E M E Poskitt, MRC Human Nutrition Research Centre, Downham’s Lane, Cambridge CB4 1XJ, UK.
  1. Department of Child Health
  2. Royal Hospital for Sick Children
  3. Yorkhill, Glasgow G3 8SJ, UK


    AIMS To review diarrhoea presentations and nutritional status in young rural Gambian children over a 15 year period as a test of an earlier hypothesis that reduced diarrhoea prevalence would lead to improved growth and a reduced prevalence of malnutrition.

    SUBJECTS AND METHODS Growth and morbidity were documented routinely in 1190 children under 2 years of age attending the Dunn Nutrition Group clinic at Keneba between 1979 and 1993.

    RESULTS The numbers of presentations with diarrhoea (1069 in 1979; 220 in 1993) and the proportion of clinic attendees with diarrhoea (30% in 1979; 8% in 1993) fell steadily between 1979 and 1993. However, at both 1 year old, mean weights (Z scores: 1979, −1.8; 1993, −1.8) and mean lengths (Z scores: 1979, −1.3; 1993, −1.7), and at 2 years old, mean weights (Z scores: 1979, −2.0; 1993, −1.9) and mean lengths (Z scores: 1979, −2.0; 1993, −2.1) did not change noticeably over the 15 year period.

    CONCLUSION Major progressive reductions in clinic presentations with diarrhoea have not been associated with improved nutritional status in this population of young rural Gambian children.

    Key messages In a rural West African community:

    • Diarrhoea is less common in young children than 15 years ago

    • Episodes of diarrhoea are shorter

    • Reduction in diarrhoeal disease has not been associated with improved young child growth

    • diarrhoea
    • growth
    • nutrition

    Statistics from

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Diarrhoeal disease is widely recognised as a leading cause of morbidity and mortality among young children, particularly in developing countries. More than 880 000 children under 5 years of age died of diarrhoeal illness in sub-Saharan Africa in 1990.1Despite continuing high mortality in the developing world, the outlook for diarrhoea has improved over the past 20 years. Bern and colleagues2 reviewed studies in the decade to 1990 and estimated a similar incidence for diarrhoeal disease (2.6 episodes/child/year) as 10 years previously,3 but a lower mortality; worldwide, 3.3 million children died of diarrhoeal disease each year compared with 4.6 million 10 years previously. More widespread knowledge and use of oral rehydration solution (ORS) are likely to account for at least some of the improvement in mortality rates.4 5

    For more than 30 years, the concept that malnutrition and acute infections interact to the disadvantage of young children has provided an explanation for persistent undernutrition and high young child mortality in many disadvantaged and/or developing societies.6 Among acute infections, diarrhoeal diseases are those considered most likely to lead to malnutrition.7 8 The children most at risk of persistent diarrhoea seem to be those already malnourished. Black and colleagues9 showed that Bangladeshi children developing diarrhoea had more prolonged illness if they were already undernourished. This concept of negative interaction between malnutrition and diarrhoea is enhanced by clinical experience. For example, in the late 1970s, regression analysis of data on disease incidence in young children attending the MRC Dunn Nutrition Group clinic in Keneba indicated a close relation between growth faltering and the incidence of diarrhoea.8 The data appeared to suggest that if the incidence of diarrhoeal disease could be reduced to low levels, then rates of growth in early childhood could be expected to follow normal centiles, except during the annual hungry season (during the rains).

    The relevance of diarrhoeal illness to persistent growth faltering in early childhood among children in developing countries has been challenged. Briend et al,10 studying children aged 6–35 months in Bangladesh, found that the medium to long term effects of diarrhoea on growth—both length and weight—were insignificant. Deficits acquired during acute diarrhoeal illness were transient and no longer apparent a few weeks later. Their conclusion was that efforts to control diarrhoea would be unlikely to improve children’s nutritional status. Studies of young Indonesian children in the late 1980s also showed no cumulative effect of disease on growth performance.11Nevertheless, the concept of an important interaction between diarrhoea and growth faltering remains seminal to much public health planning.

    We present prospectively collected data on diarrhoeal illness, which have enabled us to re-examine the interaction between this and early childhood growth in our three Gambian “study villages”.



    Since 1974 the Medical Research Council Dunn Nutrition Centre has had a field station in Keneba, The Gambia, studying nutrition in childhood, pregnancy, and lactation. All children under 2 years of age from Keneba and two neighbouring villages have been part of the unit’s research cohort since that time. Morbidity has been documented through attendance at daily clinics, well baby (call) clinics, and emergency presentations. The nearest alternative health centre to Keneba is 15 km away, and not readily accessible by public transport. Adults, particularly men, visit traditional healers readily, but young children are taken to traditional healers only very rarely and then usually when Western medicine is deemed to have failed. Thus, we have almost complete documentation of morbidity in children under 2 years of age resident in the three study villages.

    Typically, young children in this part of The Gambia have a low average birth weight compared with international standards (mean, 2700–2900 g, depending on season), but there is rapid catch up growth in the first 3–4 months of exclusive breast feeding, so that 3 month old babies have average weights close to those of international standards.12 Commonly, after 4 months, there is a faltering in weight and length increments, which persists until around 21 months, when growth gradually assumes velocities more or less equivalent to international standards.12


    Clinics are held at Keneba morning and evening, seven days a week, and medical advice is available for emergencies at any time of the day or night. Children under 2 years are seen by one of the unit paediatricians or by one of two experienced nurses/midwives, and diagnosis and treatment are recorded at each attendance. Diagnostic criteria are based largely on clinical findings because microbiological services are limited.

    All newborns are visited by a paediatrician in the 1st few days of life so dates of birth are known precisely. Children are “called” to well baby clinics for immunisation and growth/development assessment nine times in the 1st year of life and four times in the 2nd year of life. Children failing to attend these “call” clinics are followed up so that all are likely to be reviewed regularly, provided they remain resident in one of the three study villages.

    The same basic clinical criteria have been used over the years to record diagnoses. This paper is confined to presentation of diarrhoeal disease, which has been defined as four or more stools/day and/or a significant change in stool consistency. Other diseases commonly recorded in the clinic are upper and lower respiratory tract infections, eye and skin infections and, during the rainy season and shortly after, malaria.

    Medical personnel usually spend two years at the unit. Some variation in coding of diagnoses as a result of individual diagnostic and therapeutic practices has been inevitable over the years, but we have found no evidence to suggest that changes in the frequency of diagnoses are significantly related to the residency of particular clinicians.

    We present data on the number of attendances with diarrhoea at all clinics as proxy for the incidence of diarrhoea at any one time. We make an estimation of the duration of diarrhoea by reviewing the next attendance of each case at the clinic and whether he or she complained of diarrhoea at the second attendance. We use the documentation of diarrhoea at call clinics as proxy for prevalence of diarrhoea because children were brought to the clinic on those occasions for review and did not attend specifically for treatment of symptomatology.


    Weight unclothed was recorded in kilograms (to two decimal places) by trained technicians each time a child attended the clinic. In the early years of the study, manually operated baby scales (Todd Scales, Cambridge, UK) but, later, Seca electronic baby scales (CMS, London, UK) were used. Lengths were measured on Holtain infantometer (Holtain, Crymmych, UK) but more recently on Kiddimetre (Raven Equipment, Great Dunmow, Essex, UK). Lengths were recorded only at call clinics. Weight and lengths were related to international standards12 to develop Z or standard deviation scores.



    The total population of the three villages increased from 1500 in 1979 to 2700 in 1993. Annual births averaged 98 in the quinquennium 1979–83; 92 in 1984–88; and 108 in 1989–93. Overall, 1190 children aged 0–2 years are involved in our analysis.


    Infant mortality has not changed appreciably over the 15 year period (1979–83, 44/1000; 1984–88, 39/1000; and 1989–93, 48/1000).


    Table 1 shows the total number of clinic attendances for children under 2 years over the period 1979–93, together with the number and percentage of all attendances in this age group caused by diarrhoea. It is obvious that there have been large changes in both sets of figures. However, the distribution of diarrhoea over the years follows similar annual patterns (fig 1), with consistent peak levels in the rainy season (July to September) and a small peak early in the year, previously shown to be caused by rotavirus epidemics.13This seasonality is consistent with an overall reduction of diarrhoeal illness, rather than a reduction in particular epidemic pathogens.

    Table 1

    Annual clinic attendances (all clinics) and attendances with diarrhoea (“incidence”) for children less than 2 years old, 1979–93

    Figure 1

    Five year averages for clinic attendance with diarrhoea in children under 2 years of age for the periods 1979–83, 1984–88, 1989–93.

    Table 2 shows a very similar pattern for diarrhoea among call clinic attendees, suggesting that both the incidence of diarrhoea presenting at the clinic and the prevalence have declined. Figure 2 shows that diarrhoea is less prolonged. In 1979, 26% of children returned within 10 days still complaining of diarrhoea, whereas in 1993 the same figure was 17% of children, and most returned in the early days of the disease.

    Table 2

    Annual call clinic attendances and attendances with diarrhoea (“prevalence”) for children less than 2 years old, 1979–93

    Figure 2

    Percentage of children under 2 years of age returning to clinic within 10 days of first presentation still suffering from diarrhoea, 1979–93.


    Figure 3 shows mean weight Z scores at 1 and 2 years of age according to the year of birth, based on an annual average of 80 weights at 1 year and 55 at 2 years. There is some movement away from the villages as mothers may “transfer” after childbirth to live with their husbands in urban areas, and in the 2nd year, as children are weaned from the breast and sent to grandmothers. Despite a dramatic reduction in the incidence and prevalence of diarrhoeal disease, there was no improvement in weight gain in the study children over the first 2 years of life for the period 1979–93. Mean weight Z scores (SD) at one year were −1.8 (0.9) in 1979 and −1.8 (1.1) in 1993. At 2 years, these figures were −2.0 (0.7) in 1979 and −1.9 (0.9) in 1993. Likewise, length measurements showed no significant change over this period. Mean length Z scores at 1 year were −1.3 (1.0) in 1979 and −1.7 (1.0) in 1993. Corresponding mean length Z scores at 2 years were −2.0 (0.9) in 1979 and −2.1 (0.9) in 1993. The similarity in SD of the Z scores over the survey period rules out the possibility that the lack of change in mean Z scores is a result of some children doing better, but others doing very badly yet surviving, during the latter part of the study, whereas they might have died in the early years of study.

    Figure 3

    Mean weight Z score for children at 1 year and 2 years of age, 1979–93.


    The fall in the number of children presenting to our clinics since 1979 with symptoms of diarrhoea is impressive. It could be argued that this reflects greater ability of parents to manage diarrhoea at home since the introduction of ORS,14 with its availability from the unit or from individuals within the villages altering the management of diarrhoea. However, the decline in the number of children attending with diarrhoea has been fairly steady over the years, rather than occurring in the early 1980s when the introduction of ORS might have been expected to have had an impact. Thus, we believe that the fall in cases cannot be fully explained by the use of ORS. Furthermore, we continue to see most children with diarrhoea in the first days of illness, whereas if most cases were managed at home, we would expect that only those whose home treatment had failed would attend the clinic. In this case, presentation would be later into the illness and diarrhoea more likely to be protracted. Our evidence is that not only have the numbers of presentations with diarrhoea fallen but so have the number of days children are ill with each episode of diarrhoea. Moreover, diarrhoea prevalence as suggested by call clinic data also indicates that diarrhoea is less common than in the late 1970s.

    The trend towards less severe diarrhoea fits in with more widespread surveys. The overall decline in diarrhoea incidence is less well documented. As mentioned above, Bern and colleagues2 found the prevalence of diarrhoea worldwide to be about the same as 10 years before, but the overall morbidity appeared significantly less. It would be heartening to feel that the decline in diarrhoea was a reflection of greater resistance to the complications of diarrhoeal infections owing to improved nutrition. Clearly, with our children this cannot be substantiated. We have to accept that the position we adopted in the 1970s,8 that removal of diarrhoea from the population of young children would result in almost normal growth, is no longer tenable. Despite an ∼ 75% drop in diarrhoeal illness and a reduction in the number of days ill with diarrhoea, the average weights and lengths at 1 and 2 years of age in our population have not increased over the years 1979–93.

    The change in the number of children presenting with diarrhoea is dramatic. Why has this occurred? Poor quality water sources and inadequate sanitation are often considered as contributing to the prevalence of diarrhoea.15 16 Have changes in these accounted for the changes in the incidence of diarrhoea in our villages over the 15 years of study? In the early days of the Dunn Nutrition Group in Keneba, most water was obtained from hand drawn wells on the compounds.17 There were fewer latrines and specific areas of the bush were often used for sanitary purposes. Now one village has four taps delivering water from a borehole and all villages have covered hand pumped wells. Open wells on the compounds have largely run dry owing to a falling water table and lack of use because of other water sources. Most villagers have shallow pit latrines on their compounds. These changes have all been present for the past 10 years and do not adequately explain the progressive fall in diarrhoea. A study here in 1990, involving a health promotion programme which stressed the importance of hand washing, showed decreased skin and eye infections and reduced rates of diarrhoea but no decrease in respiratory tract infections in the children of families receiving the hand washing advice.18 However, this project cannot account for the decline in diarrhoeal incidence and prevalence before 1990. After 1990, the decline seems to continue a steady downward, rather than sudden stepwise, trend. Studies elsewhere have suggested that it is only with substantial changes in sanitation and water sources (flush toilets as well as tapped water supply) that falls in diarrhoea prevalence are associated with water/hygiene developments.19 20

    We are disappointed by the lack of association between reduction in diarrhoea cases and nutritional status of the children. However, our findings are not unique in this respect.10 11 A recent retrospective study of Aboriginal children in north west Australia21 from 1969 to 1993 demonstrated that declines in the incidence of low birth weight, mortality, and infectious disease were not accompanied by improvements in the growth patterns of children 0–60 months old. Reductions in the incidence and prevalence of diarrhoea are undoubtedly important for child health and morbidity; but it is clear that policies to reduce diarrhoea cannot alone be expected to achieve the hoped for improvements in growth and in rates of malnutrition in young children.

    It could be interpreted that our study shows the importance of inadequate dietary intake, as opposed to infection, in leading to growth faltering and failure of catch up growth in poor village children. Yet we have shown previously that communal macronutrient supplementation of village weanlings does not improve overall growth.22 Catch up growth has been induced in our young child population after diarrhoea by ensuring energy intakes 50% in excess of recommendations and protein intakes 100% more than recommended.23 But such intakes only achieved rapid catch up growth during the period of supplementation and even then the rates of growth fell far below the catch up expected in recovery from acute severe protein energy malnutrition: kwashiorkor and marasmus.24 That catch up growth can be induced with these very high intakes is not in itself evidence that poor growth is caused by intakes below recommended levels. Energy and protein supplementation can induce improved growth in conditions such as cystic fibrosis25 and in high output congenital heart disease.26 This does not indicate that affected children are necessarily having intakes less than recommended, but that extra food can overcome some of the consequences of illness.

    We are still left with the question: why do young children in so many underdeveloped rural communities grow so poorly? In Keneba we have studied the effects of providing macronutrient supplementation22 23 and micronutrient supplementation27 28; we have succeeded in reducing the prevalence of severe illness and mortality in young children29; we have carried out vigorous programmes of nutrition education18; and currently we are in the process of investigating the role of enteropathy in young Gambian children.30 31 Yet we continue to see patterns of growth that define half the village children as moderately malnourished (< 80% expected weight for age) by the end of their 1st year, despite above average rates of growth in the first 4 months of life.

    It is sometimes suggested to us that the “growth faltering” we are studying is “normal growth” for these children. This is not likely, given that they show such good growth in the first 4 months. The shape of the growth curve is not equivalent to any known normal childhood growth pattern. Moreover, within our population there are some children who do pursue expected patterns of growth. We must accept that the average growth pattern of Keneba children is abnormal and pathological.

    Again, it could be argued that if these children grow poorly because of malnutrition this might be a natural way of using the limited amounts of food available to best advantage. However, small is not beautiful in terms of childhood nutrition. There is good evidence that even mildly malnourished children are more at risk of death from infection32; low physical fitness as older children and adults33; neurodevelopmental impairment34 35; and a greater propensity to conditions such as hypertension36 and diabetes37 in later life. We might be unable to explain nutritional stunting in disadvantaged populations, but we believe that it is a problem that we must continue to explore for the benefit of present and future generations.


    The authors thank all the clinicians—paediatricians and midwives—who have documented diagnoses for “Keneba” children over the years; the laboratory staff for parasitological back up; and the tolerance of the local community for our research unit in their midst. We also thank two referees for their comments.



    Poskitt et al report that, despite a decline in the prevalence and duration of diarrhoeal disease in children younger than 2 years in three Gambian villages served by a single clinic, the prevalence of undernutrition has remained unchanged over 15 years. This paper adds valuable data to the debate, highlighted by Briend,1-1 that childhood undernutrition is likely to result as much from poor food intake as from recurrent diarrhoea.

    The continuous collection of data on diarrhoeal prevalence through clinic attendance rates, and nutritional status by serial anthropometry, offers a rare opportunity to examine this question, but the absence of data about dietary intake makes it difficult to reach a firm conclusion.

    Growth is a function of food intake, energy–nutrient retention, and utilisation, and undernutrition must be caused by one or more of a combination of factors—insufficient intake, defective digestion or absorption, increased metabolic demands, or excessive losses (largely from the gastrointestinal tract).1-2 While it may be true that it does not follow that if growth can be achieved by dietary “super supplementation”1-3 undernutrition is caused by inadequate intake alone (as the authors assert); their findings focus attention on the “intake” and “utilisation” components of the energy–balance equation.

    The enteropathy present in most of these Gambian children1-4may well lead to suboptimal absorption, and recurrent or chronic infections contribute to a systemic inflammatory response that increases energy and nutrient needs. The incidence of chronic infection is not trivial. We have recently shown in the same population of children that Helicobacter pylori infection acquired in early life contributes to the growth faltering.1-5 There is a significant association betweenH pylori and undernutrition,6 by 3 years old the vast majority of children are colonised with the microorganism,7 and breast milk with high titres of anti-H pylori IgA may be protective.1-8 What effect this chronic infection has on gastrointestinal function and growth is the subject of much interest worldwide.1-9 1-10

    The conclusion of Briend1-1 that childhood undernutrition is caused as much by inadequate food intake as to recurrent diarrhoea remains very plausible but unconfirmed. Poskittet al, by demonstrating that poor growth persists when the burden of diarrhoea lessens, focuses the debate on nutrient intake and utilisation rather than loss, and challenges us to investigate the adequacy of the diet young children receive (not just after episodes of diarrhoea), how that is handled by the gut, and the ability of children—undernourished and with chronic microbial colonisation—to utilise it thereafter.


    1. 1-1.
    2. 1-2.
    3. 1-3.
    4. 1-4.
    5. 1-5.
    6. 1-6.
    7. 1-7.
    8. 1-8.
    9. 1-9.
    10. 1-10.