Corresponding Author Christiane Horwood, Centre for Rural Health, University of KwaZulu-Natal, George Campbell Building, Howard College campus, Durban 4013, South Africa. E-mail: firstname.lastname@example.org
Objective To describe the presenting complaints and disease profile of children attending primary health care (PHC) clinics in two provinces of South Africa.
Methods Participants were sick children 2–59 months old presenting for care at PHC clinics in KwaZulu-Natal (KZN) and Limpopo provinces from 2006–2007. Children were assessed by an expert Integrated Management of Childhood Illnesses (IMCI) practitioner. Children for whom parental/guardian consent was obtained were tested for HIV.
Results A total of 1357 children attending one of 74 clinics were assessed. HIV seroprevalence overall was 7.1%, but was significantly higher in KZN than Limpopo (7.5 vs. 2.4%; OR = 3.3, 95%CI 1.9–5.8%). Commonest presenting complaints were cough (72%), skin conditions (22%) and diarrhoea (19%). Of 1349 children, 120 (8.9%) had a weight below the third percentile; 108/1357 (8.0%) children required urgent referral, most commonly for severe pneumonia (53.7%) and severe malnutrition (16.7%). In multivariate analyses, severe pneumonia, growth faltering and urgent referral were independently associated with younger age, residence in KZN and HIV infection (P < 0.05).
Conclusions Many children with severe illnesses and undiagnosed HIV infection present to PHC facilities. PHC staff require skills to correctly manage these conditions and undertake HIV testing. Although IMCI provides evidence-based guidelines, implementation must be improved to achieve adequate coverage of life-saving interventions.
Objectif: Décrire les symptômes et le profil de la maladie chez les enfants présentés dans les cliniques de soins de santé primaire dans deux provinces d’Afrique du Sud.
Méthodes: Les participants étaient des enfants malades âgés de 2 à 59 mois présentés pour des soins dans les cliniques de soins de santé primaire dans les provinces du KwaZulu-Natal (KZN) et du Limpopo, de 2006 à 2007. Les enfants ont étéévalués par un expert praticien de l’IMCI (Gestion Intégrée des Maladies de l’Enfant). Ceux pour lesquels le consentement des parents/tuteurs a été obtenu ont été testés pour le VIH.
Résultats: Un total de 1357 enfants fréquentant une des 74 cliniques ont étéévalués. La séroprévalence globale du VIH était de 7,1%, mais était significativement plus élevée dans le KZN que dans le Limpopo (7,5% vs 2,4%; OR = 3,3; IC95% 1,9–5,8). Les symptômes les plus fréquentes étaient la toux (72%), les affections de la peau (22%) et la diarrhée (19%). 120 sur 1349 enfants (8,9%) avaient un poids inférieur au troisième percentile. 108 sur 1357 enfants (8,0%) nécessitaient un transfert d’urgence, le plus souvent pour une pneumonie sévère (53,7%) et une malnutrition sévère (16,7%). Dans une analyse multivariée, une pneumonie sévère, le retard de croissance et un transfert d’urgence étaient indépendamment associés au plus jeune âge, à la résidence dans le KZN et à l’infection par le VIH (P < 0,05).
Conclusions: Beaucoup d’enfants atteints de maladies graves et non diagnostiqués pour l’infection VIH sont présentés dans les services de soins de santé primaire. Le personnel des soins de santé primaire nécessite des compétences pour gérer correctement ces conditions et pour entreprendre le dépistage du VIH. Bien que l’IMCI fournisse des directives fondées sur des preuves, l’implémentation doit être améliorée afin d’atteindre une couverture adéquate des interventions sauvant la vie.
Objetivo: Describir las principales dolencias y perfiles de enfermedad de los niños atendidos en centros sanitarios de primer nivel (CSPN) de dos provincias de Sudáfrica.
Métodos: Los participantes eran niños enfermos con 2–59 meses de edad que se presentaron en los CSPN de las provincias de KwaZulu-Natal (KZN) y Limpopo entre 2006–2007. Los niños fueron evaluados por un sanitario experto en el Manejo Integrado de Enfermedades Infantiles (MIEI). En aquellos niños cuyos padres/tutores dieron su consentimiento, se realizó la prueba del VIH.
Resultados: Se evaluó a un total de 1357 niños que atendieron uno de los 74 centros sanitarios de la zona de estudio. La seroprevalencia para VIH era del 7.1%, siendo significativamente más alta en KZN que en Limpopo (7.5% vs. 2.4%; OR = 3.3, 95%IC 1.9%–5.8%). Los síntomas más comunes eran tos (72%), problemas de la piel (22%), y diarrea (19%). 120/1349 (8.9%) niños tenían un peso por debajo del tercer percentil. 108/1357 (8.0%) niños requirieron de traslado urgente, principalmente debido a una neumonía severa (53.7%) o a una desnutrición severa (16.7%). En un análisis multivariado, la neumonía severa, el retraso del crecimiento, y el requerir un traslado urgente estaban asociados de forma independiente con una menor edad, residir en KZN, y tener VIH (P < 0.05).
Conclusiones: Muchos niños con enfermedades severas y VIH sin diagnosticar se presentan en los CSPN. Los sanitarios que atienden en los CSPN necesitan adquirir habilidades para manejar correctamente estas condiciones y realizar la prueba del VIH. Aunque el MIEI provee guías basadas en la evidencia, la implementación debe mejorar para alcanzar una cobertura adecuada de aquellas intervenciones capaces de salvar vidas.
The millennium development goal for child mortality (MDG4) commits nations to reduce child deaths by two-thirds by 2015 (UN 2000). However, in South Africa, child mortality has increased over the past decade (Chopra et al. 2009a), despite the introduction of free health care services for children. This increase is largely driven by HIV, and there is a correspondingly high HIV prevalence among pregnant women (SA National Department of Health [DoH] 2008). HIV causes over 50% of deaths in children under 5 years (Chopra et al. 2009a), but diarrhoeal disease, lower respiratory infections and malnutrition are still important causes of child mortality (Garrib et al. 2006). Most child deaths could be prevented by the implementation of good quality primary health care (PHC) services (Bradshaw et al. 2003).
Integrated Management of Childhood Illness (IMCI) was introduced in South Africa in 1997 and is the gold standard for management of sick children at PHC level. IMCI is a child survival strategy developed by WHO and UNICEF (Tulloch 1999), which aims to improve case management skills of first-level health workers, strengthen the health system for effective management of sick children and promote good family and community child care practices (Bryce et al. 2005). IMCI provides guidelines for health workers in the management of those conditions causing most child deaths. Guidelines are adapted to suit local conditions, and in South Africa, IMCI guidelines focus on acute respiratory infections, diarrhoeal disease, meningitis, malaria, ear infections, HIV and malnutrition.
Although there are published reports describing causes of child deaths in South Africa (Grandin et al. 2006; Bradshaw et al. 2008) and some hospital-based data on morbidity in children (Zwi et al. 1999; Yeung et al. 2000), we are not aware of any published data describing morbidity among children at primary care level. It is important to understand the profile of diseases with which children present to PHC facilities to inform child health policies and to ensure health workers have the resources and training to provide appropriate care for children.
In this article, we describe the disease profile of children under 5 years attending PHC clinics in two provinces of South Africa, with particular reference to the prevalence of paediatric HIV in this population.
Study sites and population
The study was conducted between May 2006 and January 2007 in two provinces in South Africa: KwaZulu-Natal (KZN) and Limpopo. Limpopo province has an estimated population of 5.2 million (Statistics South Africa 2009) and is predominantly rural, with high rates of poverty and unemployment, and poor access to basic services (SA DoH 2004; SA DoH and Medical Research Council 2007). KZN has relatively lower rates of poverty and several large urban centres, but almost 50% of the population still live in rural areas (Van Aardt & Schacht 2004), and with a population of over 10 million, KZN is the most populous of South Africa’s provinces (Statistics South Africa 2009). Antenatal HIV seroprevalence in 2006 was 39.1% in KZN and 20.7% in Limpopo (SA National DoH 2007).
The primary objective of this study was to evaluate IMCI implementation, particularly the HIV component (Horwood et al. 2009b). To identify health workers for inclusion in the study, we randomly selected IMCI-trained health workers by a simple random sample, using computer-generated random numbers, from a list of all IMCI-trained health workers currently working in PHC clinics in the two provinces. Seventy-seven health workers, all registered nurses, were selected (31 in Limpopo and 46 in KZN); data were therefore collected in the 74 PHC clinics (29 in Limpopo and 45 in KZN) where these nurses were working. At the time of the study, IMCI was being implemented at 60% (283/474) of PHC clinics in Limpopo, and 64% (387/604) clinics in KZN.
At each clinic, children 2–59 months old who presented for care because of illness (not for immunisations or weighing only) and their accompanying caregivers were recruited to participate. Once informed consent was obtained, consultations between IMCI-trained nurses and sick children were observed by an IMCI expert who recorded the presenting complaints reported by the caregiver. Thereafter, a second IMCI expert assessed the same child independently. Selected nurses were observed for 20 consecutive consultations with study-eligible children, or for 3 days if 20 observations had not been completed in that time. All IMCI experts were experienced IMCI practitioners who received training in use of data collection tools, which were piloted prior to start of data collection.
Integrated Management of Childhood Illness experts used standardized data collection tools to record children’s clinical presentations and classification of disease. Clinical features identified in the assessment are clearly defined according to IMCI guidelines (WHO 2002). The findings of the assessment by the second IMCI expert were considered to be the gold standard for the present analyses.
Children presenting symptoms were assessed from caregiver reports to the observed health worker and to the second IMCI expert to ensure that comprehensive information about caregivers’ complaints was included. A coding system was developed for presenting complaints; these were coded into groups of related symptoms, which were considered as relating to a single condition. Coding of presenting complaints was conducted by two doctors independently, according to agreed criteria, discrepancies were discussed and agreement reached. Coding is described in Table 2.
Findings were classified according to IMCI guidelines, which are a set of evidence-based guidelines for management of the leading causes of mortality and severe morbidity in children under 5 years of age (WHO 2008).
All children were routinely weighed, and the weight-for-age plotted on the South African growth monitoring chart (SA National DoH 2002). Children were defined as being underweight if the weight was below the third percentile (weight-for-age Z-score [WAZ] <−2 standard deviations), and severe malnutrition was defined as those children with weight below the marasmic line (WAZ <−3 standard deviations). If the child’s weight was above the third percentile, the weight gain was assessed from the curve plotted on the growth chart: weight gain parallel to the growth curve was defined as good weight gain, weight gain falling away from the curve was defined as inadequate weight gain, or if there was documented weight loss, this was recorded.
Children over 18 months old were screened for HIV antibodies using the rapid test Determine™ HIV-1/2 (Abbott, Wiesbaden, Germany). All positives were confirmed by a second HIV rapid test using the Smart Check™ HIV1/2 (World Diagnostics Inc, Miami, USA). If the screening HIV rapid test was positive in children younger than 18 months, HIV status was confirmed with PCR testing. Fifty microlitres of whole blood was collected by heel prick using a lancet and dropped onto a filter paper for the dried blood spot (DBS). The filter paper was transported to Inkosi Albert Luthuli Central Hospital in Durban for PCR testing. The DBSs were tested for HIV-1 DNA using the HIV-1 DNA test, version 1.5 (Roche, Branchburg, USA). All blood samples were discarded when HIV testing was complete.
Our primary analysis compared presenting complaints, disease classifications, nutritional status and hospital referral in children presenting for care in PHC clinics in Limpopo and KZN. HIV seroprevalence by age group and province was calculated, and 95% confidence intervals (CIs) were calculated using the Wilson approach. Logistic regression was performed to adjust for confounding and assessed for potential effect modification by age, province and HIV infection status. HIV status was included to account for differences in outcomes attributable to HIV infection. Province was also included to account for additional differences (e.g. population density, water sources, socio-economic status, etc.) between provinces not accounted for by other model covariates such as HIV status. Because the sample included children presenting for care at different clinics through each of two provinces, it could not be assumed that their outcomes were independent of province. Therefore, we used the generalized estimating equations approach with an exchangeable correlation structure to account for within-clinic correlations of responses (Liang & Zeger 1986; Zeger & Liang 1986). All potential explanatory variables were retained in regression models. Analyses were conducted using sas 9.2 (SAS Institute, Cary, NC).
Permission to undertake the study was obtained from the South African National DoH, and the provincial Departments of Health in Limpopo and KZN. Written informed consent was obtained from the caregiver who accompanied the child to the clinic. Separate consent was requested for HIV testing of children. For children accompanied by a caregiver who was not the legal guardian, consent was requested for observation and assessment by the IMCI expert only. When consent for HIV testing was obtained, pre- and post-test counselling was provided by HIV counsellors on the study team, and the mother received the result. Ethical approval was obtained from the Biomedical Research Ethics Committees of the University of KwaZulu-Natal Medical School, Durban, and the World Health Organization, Geneva.
A total of 1357 children between 2 and 59 months of age [median = 15 months, interquartile range (IQR) 7–29 months] presenting for care at one of 74 PHC clinics in KZN (n = 858) and Limpopo provinces (n = 499) between May 2006 and January 2007 were assessed (Table 1).
Table 1. Characteristics of 1357 children age 2–59 months presenting to primary health care facilities in KwaZulu-Natal (KZN) (n = 858) and Limpopo provinces (n = 499), South Africa
HIV infection status
HIV seroprevalence among children
Carers of 10 (0.74%) children reported that the child had had a prior positive HIV test, which was confirmed from clinic records. Of 1347 children with unknown HIV status, consent for HIV testing was obtained from the carers of 1060 (78.7%) children; evaluable results were available for 1054 children, 66 (6.3%) were confirmed HIV-infected and 988 (93.7%) tested negative for HIV, resulting in an overall HIV seroprevalence of 7.1% (76/1064). HIV seroprevalence by province is presented in Table 1, and by province and age group in Figure 1. HIV seroprevalence was significantly higher among children attending clinics in KZN compared to Limpopo (7.5 vs. 2.4%; OR = 3.3, 95%CI 1.9–5.8, P < 0.001), but was not significantly associated with months of age (OR 1.0, 95%CI 0.99–1.02, ptrend = 0.24).
The most common presenting complaints were cough or difficult breathing (72%), skin conditions (22%) and diarrhoea (19%) (Table 2). Of 1357 enrolled children, 551 (41%) presented with two or more complaints (median number of complaints = 1, IQR = 1–2, absolute range = 1–4). In analyses adjusted for age (months), province and HIV status, the odds of presenting with two or more complaints were significantly higher among children presenting to clinics in KZN [adjusted odds ratio (AOR) 1.8, 95%CI 1.4–2.3, P < 0.001] and among children who were HIV-infected compared to those who were HIV-uninfected or whose HIV status was unknown (AOR 2.3, 95%CI 1.4–3.9, P = 0.001). The number of presenting complaints did not differ significantly by age (AOR 1.007, 95%CI 1.0–1.02, P = 0.06).
Table 2. Presenting complaints of 1357 children age 2–59 months presenting to primary health care facilities in KwaZulu-Natal province (n = 858) and Limpopo province (n = 499), South Africa*
Number presenting with complaint
*Complaints as presented by carer prior to direct questioning about symptoms.
†Vomiting was considered to be related to the cough if no other abdominal symptoms were reported. Vomiting without cough or with other abdominal symptoms was coded as diarrhoeal disease or abdominal complaint.
Cough or difficult breathing
With/without fever, runny nose, vomiting†, sneezing. Includes carer reports the child has asthma (n = 6)
Includes runny/blocked nose, ‘flu’, sneezing, without cough
With/without abdominal pain, vomiting, fever. Includes diarrhoea with blood (n = 13)
Abdominal complaints (without diarrhoea)
Includes abdominal pain, vomiting & constipation
Carer reports seeing worms
Includes oral thrush reported by the carer
Includes rash, skin sores, or abscess (10) on any part of the body
Includes ear pain or ear discharge with/without fever
Includes discharging eyes, sore eyes or red eyes
Fever with no other associated symptoms
Carer reports that child has a sore throat
Carer reports an injury
Includes convulsions (2), nose bleed (5), headache (10), grinding teeth (3), urinary symptoms (3), itchy ears without pain or discharge (10), poor appetite alone, mumps, swelling, TB follow-up, thrush, penile discharge or sores, joint pains, bleeding gums, delayed development, teething etc.
Cough and difficulty breathing
Cough and difficulty breathing was mentioned by 976/1357 (71.9%) carers when asked about their child’s problems, but by 1076 (79.3%) when asked directly if the child had a cough or difficult breathing. Of these children, 59 (5.5%) had lower chest wall in-drawing during inspiration and 3 (0.28%) had stridor. These 62 children were classified as having severe pneumonia (Table 3) and referred urgently to hospital. Pneumonia was defined as the presence of tachypnoea (>50 breaths per minute if 2–11 months old; >40 breaths per minute if 12–59 months old). A total of 358 (33.3%) children with cough had pneumonia and treated at home with oral antibiotics (Table 3). Most children with cough (59.9%) had no signs of respiratory distress and received no specific treatment.
Table 3. Disease classifications among children presenting at primary health clinics in Limpopo province and KwaZulu-Natal province, South Africa
Presentation with cough or difficulty breathing (n = 1076)*
Presentation with diarrhoea (N = 310)*
All children (N = 1357)
Cough or cold
Some or severe dehydration
Poor weight gain or weight loss
Yes N = 62
No N = 1014
Yes N = 358
No N = 718
Yes N = 645
No N = 43
Yes N = 40
No N = 269†
Yes N = 402
No N = 753
*Number of carers who reported symptom after being directly asked about their child’s problems by the nurse.
†Missing data for 1 child.
Age (months), per 1 month increase (median, IQR)
HIV infection status
In analyses adjusted for age (months), province and HIV infection status, diagnoses with pneumonia and severe pneumonia increased significantly with younger age (P < 0.001) (Table 4). HIV-infected children had seven times the odds of being diagnosed with severe pneumonia compared to HIV-uninfected children (AOR 7.1, 95%CI 3.4–14.8).
Table 4. Determinants of respiratory disease, dehydration and growth faltering among children presenting at primary health clinics in Limpopo province and KwaZulu-Natal province, South Africa
Presentation with cough or difficulty breathing (n = 1076)†
Presentation with diarrhoea (n = 310)†
All children (n = 1357)
Severe Pneumonia N = 62
Pneumonia N = 358
Cough or cold N = 645
Some or severe dehydration N = 40
Poor weight gain or weight loss N = 402
*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.
†Number of carers who reported symptom after being directly asked about their child’s problems by the nurse.
‡Adjusted for all other variables in column.
§For explanatory variables with three or more levels, the omnibus Wald chi-square test was performed to calculate the P value.
Age, per 1 month increase
HIV infection status§
χ2(2) = 8.7**
χ2(2) = 8.9**
χ2(2) = 1.9
χ2(2) = 3.7
χ2(2) = 14.6***
χ2(2) = 17.3***
χ2(2) = 1.2
χ2(2) = 1.9
χ2(2) = 16.7***
χ2(2) = 15.8***
Children with diarrhoea were assessed for dehydration according to the presence or absence of four signs: child’s general condition, how the child is drinking, skin turgor and sunken eyes. When questioned directly, 310/1357 (22.8%) carers reported the child had diarrhoea (Table 3). Presentation with diarrhoea was independently associated with younger age (AOR = 0.97, 95%CI 0.96–0.98, P < 0.001), residence in KZN (AOR = 1.7, 95%CI 1.3–2.3, P < 0.001) and HIV infection (AOR 1.6, 95%CI 1.001–2.4, P = 0.05).
Of 310 children with diarrhoea, 3 (0.97%) had severe dehydration (defined as diarrhoea with two of the following signs: lethargic or unconscious, unable to drink or drinking poorly, sunken eyes, skin pinch goes back very slowly >2 s) and 37 (11.9%) had some dehydration (defined as diarrhoea with two of the following signs: restless or irritable, drinking eagerly or thirsty, slow skin pinch and sunken eyes) (Table 3). Younger age was independently associated with having some or severe dehydration (P < 0.01) (Table 4).
Among children with diarrhoea, 33 (10.6%) had dysentery (defined as blood in the stool), and 12 (3.9%) had persistent diarrhoea (defined as diarrhoea for >14 days). After controlling for age (months) and HIV infection status, only residence in KZN was associated with classification of dysentery (AOR 2.4, 95%CI 1.04–5.5, P = 0.04). There was no evidence for an association between age, province or HIV infection status and classification of persistent diarrhoea (P > 0.05 for all).
Weight-for-age Z-scores were available for 1349 children. With reference to the South African national growth monitoring charts (SA National DoH, 2002), 120 (8.9%) children were underweight, including 12 (10.0%) with severe malnutrition (WAZ<−3 standard deviations). Seven children with severe malnutrition were HIV-infected. In analyses adjusted for age (months), province and HIV infection status, HIV-infected children had significantly higher odds of being underweight than children who were HIV-uninfected or whose HIV status was unknown (AOR = 5.4, 95%CI 3.1–9.3, P < 0.001). There was no evidence for an independent association between being underweight and age (AOR 1.0, 95%CI 0.99–1.0) or province (AOR 1.1, 0.60–2.1).
Growth charts were available for 1155 children; 753 (65.2%) children were gaining weight well, 210 (18.2%) had inadequate weight gain and 192 (16.6%) had documented weight loss (Table 3). Younger age, residence in KZN and HIV infection were independently associated with inadequate weight gain or loss (Table 4). The odds of inadequate weight gain or loss were more three times greater among HIV-infected children than HIV-uninfected children (Table 4).
When directly asked, caregivers reported weight loss in 329/1357(24.2%) children, but this was never mentioned as a presenting complaint. There was moderate correlation between reported and documented weight loss; 164/192 (85.4%) children with documented weight loss were reported as having lost weight (P < 0.001, correlation coefficient (Φ = 0.64). Similarly, there was moderate correlation between reported weight loss and growth faltering; carers of 252/401 (62.8%) children with documented inadequate weight gain or weight loss reported that the child had lost weight (P < 0.001, Φ = 0.66).
Following IMCI guidelines, 108/1357 (8.0%) children required urgent referral to the district hospital (Table 5). Thirteen (12.0%) children had more than one condition requiring referral: six children had both severe pneumonia and severe malnutrition; two children with severe malnutrition also had stridor and one had persistent diarrhoea; one child with severe pneumonia also had blood in the stools. Of three children who were lethargic or unconscious, two had severe dehydration and one had severe pneumonia. Of 28 HIV-infected children requiring urgent referral, 11 (39.3%) had severe malnutrition and 14 (50.0%) had severe pneumonia, including two with both conditions. Younger age, residence in KZN and HIV infection were independently associated with urgent referral (Table 6). HIV-infected children had 10 times the odds of being referred that HIV-uninfected children had.
Table 5. Reasons for referral to hospital of children age 2–59 months presenting to primary health clinics, N = 108
Reason for referral
Number of children referred* n (%)
*Total does not sum to total number of children because some children referred for two or more reasons.
†Defined as the lower chest wall moving in when the child breathes out.
‡Includes four children where the mother only reported convulsions when asked during the consultation, rather than as a presenting complaint.
Cough with chest in-drawing† ± fast breathing
Weight-for-age Z-score >−3 standard deviations below expected weight
Blood in the stool
Blood in stools AND dehydration OR child aged below 12 months
Diarrhoea for 14 days or more AND dehydration OR reported weight loss
History of convulsions with this illness
Vomits everything or unable to drink/breastfeed
Child either unable to take fluids or unable to keep any fluids down
Diarrhoea with severe dehydration
Diarrhoea with two of the following: Lethargic or unconscious; Unable to drink or drinking poorly; Sunken eyes; Skin pinch goes back >2 s
Lethargic or unconscious
Child who is unresponsive or drowsy
Harsh noise heard on inspiration
Stiff neck or bulging fontanelle
Table 6. Determinants of urgent referral among children presenting at primary health clinics in Limpopo province and KwaZulu-Natal (KZN) province, South Africa (n = 1357)
Unadjusted OR (95% CI) P value
Adjusted OR† (95% CI) P value
***P ≤ 0.001.
†Adjusted for all variables in column.
‡ For explanatory variables with three or more levels, the omnibus Wald chi-square test was performed to calculate the P value.
Age, per 1 month increase
HIV infection status‡
χ2(2) = 14.6***
χ2(2) = 14.6***
We examined presenting complaints, disease classification, nutritional status and HIV seroprevalence in a large sample of sick children attending PHC facilities in South Africa. To our knowledge, this is the first study to describe the disease profile of this population of children. Our findings show that serious illnesses, including emergency conditions and undiagnosed HIV infection, are common among children at PHC level. MDG4 requires a two-thirds reduction in child mortality by 2015, whereas South Africa is one of only 12 countries globally where child mortality is increasing (Chopra et al. 2009b). Our findings highlight the important role of good quality PHC services in managing life-threatening conditions in children, including identification and management of HIV infection, and we confirm that the overwhelming majority of serious illness is caused by a small number of conditions. This gives clear direction for interventions to improve childhood morbidity and mortality at primary level.
Most children presented with a cough or difficult breathing, and pneumonia was the commonest cause of severe illness. Pneumonia is the largest cause of child deaths globally, causing an estimated 2 million deaths annually (Mulholland 2007). Health workers must have skills to identify and manage children with severe pneumonia and distinguish them from the many children with coughs and colds. IMCI provides the current standard for pneumonia case management at primary level in most developing countries, but coverage and quality of IMCI implementation is frequently poor (Armstrong Schellenberg et al. 2004; Pariyo et al. 2005; Rowe et al. 2009), and may not reach the poorest communities (Victora et al. 2006). In South Africa, IMCI implementation is sub-optimal, and IMCI-trained health workers perform most poorly in the assessment of seriously ill children (Horwood et al. 2009a). Training in assessment and management of acute respiratory infections is a simpler alternative to IMCI, which could achieve rapid coverage and give health workers skills to manage most of the severely ill children in our setting. However, if clinics are to provide integrated and holistic care to children, innovative solutions are required to improve IMCI implementation and maintain health worker skills at primary level. There is a need for improved evaluations of methods to develop and sustain adequate coverage of child survival interventions (Victora et al. 2004).
Despite availability of interventions for the prevention of mother-to-child HIV transmission (PMTCT), undiagnosed HIV infection is common among children attending PHC clinics in South Africa and is frequently associated with severe disease and malnutrition (Horwood et al. 2009b). In our study, one-third of children requiring referral were found to be HIV-infected. Further, we found a statistically significant association between HIV infection and the presence of severe pneumonia, as well as faltering growth. The obtained findings for effects, though based on relatively small group sizes, are potentially useful in suggesting future targets of focused research. Follow-up of infants enrolled in PMTCT programmes is frequently poor (Doherty et al. 2005; Jones et al. 2005) and depends on health workers at PHC clinics identifying HIV-exposed children (Sherman et al. 2004). Early initiation of antiretroviral therapy in HIV-infected children substantially improves outcomes (Violari et al. 2008), but only a minority of children needing treatment currently receive it (World Health Organization, United Nations Children’s Fund, UNAIDS, 2009). Increasing provider-initiated voluntary counselling and testing for children at PHC clinics should therefore be a priority, and in high HIV prevalence settings, screening all children for HIV should be considered (Rollins et al. 2007). Training in the specific counselling and communication skills required when dealing with children and families in the context of severe or terminal illness would improve health workers’ confidence to initiate HIV testing and provide ongoing care for HIV-infected children (Horwood et al. 2009c). All children under 5 years referred from PHC clinics to hospital should routinely be tested for HIV.
A comparison of disease profiles of children in the higher HIV prevalence area of KZN with those in the lower prevalence area of Limpopo showed that even when controlling for HIV infection, living in KZN was an independent risk factor for having multiple complaints, pneumonia (including severe pneumonia) and growth faltering. There is little systematic evidence examining the health of HIV-exposed but uninfected children, but increased mortality has been described, particularly where the mother has advanced disease or has died (Newell et al. 2004; Marinda et al. 2007). A high proportion of children in KZN are HIV-exposed, and as HIV transmission rates from mother-to-child are reduced by increasing coverage and effectiveness of PMTCT programmes, the health problems of these children are likely to become of major public health importance. Further research is required to determine the health needs of these children and to evaluate interventions (Filteau 2009).
Many sick children were malnourished, with more than a third of children having either poor weight gain or loss of weight. Weight-for-age below the third percentile, a later sign of malnutrition, was less common. If growth monitoring is to improve outcomes, it should be combined with feeding advice and with food supplements and vitamin A (Bhutta et al. 2008). Health workers must plot the weight-for-age on the growth chart at every visit, intervene early if weight gain is poor, and directly ask mothers about a history of weight loss in the child. Counselling mothers about infant feeding should be routine in a child health consultation. Health workers require adequate training and resources for the management of children with growth faltering. Although zinc supplementation improves outcomes, particularly in children with diarrhoea, zinc formulations are not currently available in PHC facilities in South Africa.
If children’s lives are to be saved and quality of care for children in PHC clinics improved, health workers need to have the skills to manage common presenting problems, identify severely ill children and those with unidentified HIV infection, and to provide appropriate, effective interventions. IMCI is currently the gold standard for management of sick children at primary level but implementation needs to be improved if adequate coverage is to be achieved.
We are grateful for the support of the child and adolescent health sub-directorate of the South African Department of Health, in particular Ntombi Mazibuko. Thanks also go to: Beatrice Mlati and Janet Dalton, maternal and child health programme managers from Limpopo and KZN provinces for their valuable help with logistics, Pravi Moodley from the Virology Department at IALCH, and Olga Sechocho from Polokwane hospital laboratory, Steve Reid and Patrick McNeill from the Centre for Rural Health at UKZN, and Ann Robertson and Steve Donohue from Limpopo Department of Health. We thank the data collection teams who worked so hard, the staff of the participating clinics for their support, and all the mothers and infants who agreed to participate. We are grateful to George Rutherford at the University of California San Francisco, Department of Epidemiology and Biostatistics, Global Health Sciences, for his review of this manuscript. This study was funded by the Department of Child and Adolescent Health and Development of the World Health Organization. Shamim Qazi and Nigel Rollins are staff members of the World Health Organization. The expressed views and opinions do not necessarily express the policies of the World Health Organization.