This study was conducted at the University of Nigeria Teaching Hospital, Enugu. Enugu is the capital of Enugu State, and is located in south-east Nigeria. It has an estimated population of about 3.3 million inhabitants according to the 2006 national census figures 9 . The main social and economic activities include small and medium scale trading, artisan works while a significant number of people are also employed in the various government institutions. The inhabitants are mainly Igbos and cut across all social-economic class. The main health-related problems seen among children in Enugu are diarrhea diseases, malaria and respiratory tract infections. The old site of the Teaching Hospital (where the hospital was cited during the period under review) had 700 bed capacity and provided specialty care in paediatrics, internal medicine, surgery and obstetrics & gynaecology as well as subspecialty care in over 50 other areas. The paediatric wards have 80 bed capacities. On the average 640 children (aged 0–18 years) were admitted into the paediatric wards annually and malnutrition accounted for nearly 3% of those admissions.

This was a 10 year (1996–2005) retrospective quantitative study. The source documents for retrieving information were the admission and mortality registers of the hospital during the period under review. Available case notes/folders were also retrieved. A proforma was used to obtain relevant information. These information included date of admission and discharge, bio-data, clinical features, history of breast feeding, socio-economic status of the Caregiver, classification of malnutrition using Modified Wellcome Classifications, co-morbidities noted and, eventual outcome. Outcome was discharged from hospital, died while still in the hospital or discharged against medical advice. The advantage of using retrospective quantitative study was that reasonable sample size could be achieved in a relatively short time.

A total of 7703 children were admitted during the period under review. All children aged six to 59 months admitted into the paediatric wards of UNTH Enugu between 1996 and 2005 with diagnosis of any form of PEM were included in this review. Their case files and/or documentations on hospital registers were retrieved from the Medical Records Unit of the hospital. Children with diagnosis of PEM but had in addition, other chronic conditions such as congenital heart diseases and cerebral palsy were excluded.

The outcome variables were recovery and discharged; death and discharge against medical advice. Recovery was defined as children, whose appetite has returned, gaining weight with resolution of clinical features. Death was defined based on hospital records and exclude those that may have died at home. Discharged against medical advice were those that did not meet the discharge criteria but whose Caregivers insisted on going home.

Relevant information was extracted from each retrieved case file and/or hospital registers and transferred into the proforma. Diagnosis of PEM was based on the Modified Wellcome Classification because it was the method used for clinical diagnosis by the clinicians. This classified PEM into kwashiorkor, underweight kwashiorkor, underweight, marasmus, marasmic kwashiorkor and there was also provision for unclassified PEM. Marasmus and the various forms of kwashiorkor are part of the recently defined Severe Acute Malnutrition (SAM) by the World Health Organization (WHO). The WHO defined SAM by a very low weight for height (below -3z scores of the median WHO growth standards), visible severe wasting or the presence of nutritional oedema 11 12 . Modified Wellcome classification uses weight for age and the presence or absence of oedema to classify PEM. The weights were measured using infant weighing scales (Waymaster) and stadiometers (Health Scale) depending on the age of the child. A total of 212 proforma were completed covering the entire period of the study.

Diagnosis of HIV was made using Enzyme Linked Immunosorbent Assay [ELISA] and Westerblot. In children aged less than 18 months, positive antibody test was combined with clinical features to make presumptive diagnosis of HIV infection. Diagnosis of malaria was confirmed using blood film and bronchopneumonia using chest X-ray. Diarrhea was defined as passage of watery or loose stools or an increase in frequency above normal for a child. Severe anaemia was defined using a packed cell volume of less than 15%. Sepsis was defined as clinical features of systemic inflammatory response (fever, tachycardia, tachypnea, leukocytosis or leukopenia) associated with infection. Diagnosis of tuberculosis was made in the presence of chronic cough that have lasted for more three weeks supported by varied combination of the following: positive family history of tuberculosis, positive mantoux, suggestive chest X-ray and elevated erythrocyte sedimentation rate. Diagnosis of scabies was clinical based on the typical itching papular rash located at the intertrigous areas. Chronic suppurative otitis media and rickets were suspected clinically and confirmed by culture of ear swab and X-ray of the limbs respectively.

The data were analyzed using SPSS version 19. Chi-square was used to test significant association of categorical variables: sex, age group, mortality rate and number of complications while One-way Analyis of Variables (ANOVA) was used to test for significant association of the continuous variable (mean hospitalization duration and mean number of complications). A p-value of less than 0.05 was regarded as significant and 95% confidence interval reported where indicated.

Ethical approval was obtained from the Health Research and Ethics Committee of the University of Nigeria Teaching Hospital, Enugu, Enugu State, Nigeria.

A total of 7703 children were admitted into the paediatric wards and 212 of them were cases of PEM during the period under review. This represented about 2.8% of the total paediatric admissions. One hundred and twenty seven (59.9%) were males while 85 (40.1) were females giving a male: female ratio of 1: 0.7. The age group studied was 6 to 59 months (under-5). The mean age of the participants was 15.4 ± 9.3 months.

PEM was most common among the age groups 6 to 12 and 13 to 24 months, and these accounted for 55.7% and 36.8% of the study population respectively. There was however, no statistically significant difference between the age groups and various forms of PEM as shown in Table 1(χ² = 19.38, df =16, p = 0. 249). The most common form of PEM noted in this review was marasmus (34.9%). Except for marasmic-kwashiokor, more males than females had more of all the various types although this was not statistically significant (χ² ⁼ 8.382, df =4, p = 0. 079) as shown in Table 2. Admissions for PEM were recorded more in 1996, 1999 and 2004 (15.1, 13.7 and 12.3% respectively), but there were no consistent pattern in the yearly admissions of children with PEM during the period under review (Figure 1).

χ² = 19.38, df =16, P = 0. 249.

χ² = 8.382, df =4, P = 0. 07.

Record for associated co-morbidities was available in only 66 subjects. Diarrhea and malaria were the most common associated conditions in 72.7% and 43.9% of the children respectively. Other notable associated conditions were sepsis (37.9%), severe anaemia (24.2%), bronchopneumonia (16.7%) and tuberculosis (12.1%) as shown in Table 3. Information on the educational level and occupation of the parents was available in only 33 of the reviewed cases. Among them, 69.4% belonged to the lower social class, 19.4% to middle class while 5.6% was of the upper class. The predominant infant feeding in the first six months was breast milk and water for 0 to 3 months (48.6%), breast milk and water for 4 to 6 months (24. 3%), exclusive breast feeding for up to 3 months (18.9%) and use of breast milk substitutes (8.1%) as shown in Table 4. These feeding methods were followed by weaning with pap gruel that was variably fortified for the children.

The table shows the associated co-morbidities noted in the patients.

The table shows the prevalence of the various pattern of feeding for the children during their early infancy. The 95% confidence interval is also reported.

The duration of hospitalization was available in only 84 subjects and ranged from 0 to 62 days. The mean duration of hospitalization was 16 ± 15 days. Kwashiokor patients had the highest mean hospitalization days of 19.15 days while marasmic and underweight patients had the least days of 14.52 and 14.55 days respectively. There was no statistically significant difference in the mean hospitalization days for the various types of PEM (F = 0.317, df =4, P = 0. 866). A total of 85 (40.1%) children died while on admission, 124 (58.5%) recovered and were discharged home while 3 (1.4%) were discharged against medical advice. Mortality was higher among the males (50.9%) than females (34.1%) although this was not statistically significant (χ² = 0.723, df =2, P = 0. 697). Most of the deaths were recorded in the age groups 0–12 (55.3%) and 13–24 (36.5%) months although this difference was not statistically significant (χ² = 10.98, df =8, p = 0. 203). The marasmic-kwashiokor and unclassified groups had higher mortality rates (53.3% and 54.5% respectively) than the marasmus (37.8%) or kwashiorkor groups (30%). There was a statistically significant difference in the mortality rates of the various types of PEM as shown in Table 5 (χ² = 17.26, df =4, p = 0. 002) The number of complications ranged from none to four. Kwashiokor has the highest mean number of complications (2.06) while unclassified had the least number of 1.26. There was a statistically significant difference in the number of complications and the various PEM (F = 8.92, df =4, P <0.05)

Kwash = Kwashiokor; UWM = Underweight malnutrition, MK = Marasmic kwashiokor. The table shows some prognostic indicators: mean number of complications, mean hospitalization days and mortality rates of the various types of PEM.

There was no significant difference between the two genders from our review. This agrees with an earlier review in Maputo by Cartmell et al. 13 . The age group 6 to 24 months accounted for 92.5% of the total number of children admitted for PEM. In previous Nigerian and Zambian studies, an approximate 64% of admitted cases of PEM were less than two years 4 14 while Cartmell et al. 13 documented a mean age of 21.7 months in the Maputo study. The reasons for the high number of cases of PEM among the age group 6 to 24 months could be due to a number of factors including low rate of exclusive breast feeding as documented in our review as well as poor weaning and feeding practices. Marasmus was the most common type of PEM noted in our review (34.9%) beside the unclassified cases. Gernaat et al. 4 also found more of marasmic patients among males aged less than a year in their review. Kwashiorkor and marasmus were the most common types among admitted children in Maputo in 1983 but changed to Kwashiokor and marasmic-kwashiokor in 2001 according to the review by Cartmell et al. 13 . This may imply that there is no consistent geographical pattern that predicts the type of PEM a child will manifest but rather depends on interplay of factors.

The overall mortality in our study was 40.1% which although lower than the WHO estimated 60% 2 is still very high. Studies conducted in various parts of Africa have documented unacceptable high mortality rates among children admitted for PEM. In Oshogbo, South West Nigeria, Ibekwe and Ashworth 6 documented an average mortality rate of 22% over a five year period among 803 children admitted for PEM in a Nutritional Rehabilitation Center. Similarly, in a hospital based study in north-eastern Zambia, involving children below the age of five years, Gernaat et al. 4 documented an overall mortality rate of 25.8% among 288 children admitted for various types of severe/complicated malnutrition . Higher mortality rate for marasmic kwashiorkor than marasmus or kwashiorkor was noted in this review. Gernaat et al. 4 noted similar finding in their review among Zambian children admitted and managed for PEM. This reason for this is unclear. However, Ibekwe and Ashworth 6 did note that PEM associated mortality among oedematous patients was significantly higher compared to those with marasmus. It can be argued therefore, that presence of oedema in a malnourished child connotes poor prognosis. The mean duration of hospitalization was 16 days which is similar to 13.1 and 14.3 days reported by Cartmell et al. 13 but differs from the 35 days reported by Ibekwe and Ashworth 6 . Both this review and the study by Cartmell et al. were hospital based while that of Ibekwe and Ashworth was conducted in a Nutrition Rehabilitation Center. The pressure on bed spaces in a hospital setting could have contributed to earlier discharges in hospital settings.

Our review noted that PEM was more common among children from the lower social class (69.4%) and those predominantly breast fed for three months or less (48.6%) compared to exclusively breast fed children (18.9%). The reason for this may not be unconnected to the fact that poor families have low purchasing power for adequate nutritious foods for their families. Illiteracy on the other hand, may influence feeding practices. The low rate of exclusive breast feeding noted in this review despite the Baby Friendly Initiatives is also very worrisome. Poverty and illiteracy as risk factors for PEM have been documented in the literature. . In a case control study conducted in Dhaka, Bangladesh which involved children aged six to 24 months, Nahar et al. 15 compared 507 children with weight-for-age z-score (WAZ) < −3 matched for age, sex and place of residence with 500 children whose weight-for-age z-score (WAZ) were > −2.5 . They documented that severely-underweight children were more likely to have: undernourished poorly educated teenage mothers, history of shorter duration of predominant breastfeeding, and fathers who were poorly educated and unskilled day-labourers 15 .

Diarrhea, malaria, sepsis and severe anaemia were the most prevalent associated co-morbidities from our review in that order. In Maputo, the most prevalent co-morbidities associated with PEM by Cartmell et al. were anaemia, bronchopneumonia, malaria and diarrhea. The prevalence of human immune deficiency virus (HIV) from our review was 13.6% and this compares to a prevalence of 12% in the Maputo study. This finding underscored the high rate of HIV infection among children with severe forms of PEM and the need to routinely screen such children for HIV when they present at a health facility.

This is a retrospective study and was not designed to optimize the risk factors and outcome of PEM but relied on data from previous documentations. At best, it may be a fair representation of some of the factors associated with PEM in a typical Nigerian teaching hospital. It is however the first paper to review PEM managed at a Nigeria tertiary health facility over a 10 year period.