Is home management of fevers a cost-effective way of reducing under-five mortality in Africa? The case of a rural Ghanaian District

Authors


Corresponding Author Moses Aikins, School of Public Health, College of Health Sciences, University of Ghana, Legon, Accra, Ghana. E-mails: mksaikins@ug.edu.gh; maikins57@yahoo.com

Abstract

Objective To assess the cost-effectiveness of two strategies of home management of under-five fevers in Ghana – treatment using antimalarials only (artesunate–amodiaquine – AAQ) and combined treatment using antimalarials and antibiotics (artesunate–amodiaquine plus amoxicillin – AAQ + AMX).

Methods We assessed the costs and cost-effectiveness of AAQ and AAQ + AMX compared with a control receiving standard care. Data were collected as part of a cluster randomised controlled trial with a step-wedged design. Approximately, 12 000 children aged 2–59 months in Dangme West District in southern Ghana were covered. Community health workers delivered the interventions. Costs were analysed from societal perspective, using anaemia cases averted, under-five deaths averted and disability-adjusted life years (DALYs) averted as effectiveness measures.

Results Total economic costs for the interventions were US$ 204 394.72 (AAQ) and US$ 260 931.49 (AAQ + AMX). Recurrent costs constituted 89% and 90% of the total direct costs of AAQ and AAQ + AMX, respectively. Deaths averted were 79.1 (AAQ) and 79.9 (AAQ + AMX), with DALYs averted being 2264.79 (AAQ) and 2284.57 (AAQ + AMX). The results show that cost per anaemia case averted were US$ 150.18 (AAQ) and US$ 227.49 (AAQ + AMX) and cost per death averted was US$ 2585.58 for AAQ and US$ 3272.20 for AAQ + AMX. Cost per DALY averted were US$ 90.25 (AAQ) and US$ 114.21 (AAQ + AMX).

Conclusion Both AAQ and AAQ + AMX approaches were cost-effective, each averting one DALY at less than the standard US$ 150 threshold recommended by the World Health Organisation. However, AAQ was more cost-effective. Home management of under-five fevers in rural settings is cost-effective in reducing under-five mortality.

Abstract

Objectif:  Evaluer le rapport coût-efficacité de deux stratégies de prise en charge à domicile des fièvres chez les enfants de moins de cinq au Ghana - un traitement utilisant des antipaludiques seuls (artésunate amodiaquine: AAQ) et des traitements combinés à l’aide d’antipaludiques et d’antibiotiques (artésunate amodiaquine plus amoxicilline: AAQ + AMX).

Méthodes:  Nous avons évalué les coûts et la rentabilité de AAQ et AAQ + AMX par rapport à un groupe témoins recevant des soins standard. Les données ont été recueillies dans le cadre d’un essai contrôlé randomisé en grappes avec un concept par étape. Environ 12.000 enfants âgés de 2 à 59 mois à Dangme West District dans le Sud du Ghana ont été couverts par l’étude. Des agents de santé communautaires ont administré les interventions. Les coûts ont été analysés selon une perspective sociétale, en utilisant comme mesures d’efficacité, les décès évités chez les enfants de moins de moins de cinq ans et les années de vie évitées, ajustées par l’incapacité (DALY).

Résultats:  Les coûts économiques totaux pour les interventions étaient de 204.394,72 US$ (AAQ) et 260.931,49 US$ (AAQ + AMX). Les coûts récurrents constituaient 89% et 90% du total des coûts directs pour AAQ et pour AAQ + AMX respectivement. Les décès évités étaient de 79,1 (AAQ) et 79,9 (AAQ + AMX), avec des DALY évitées atteignant 2.264,79 (AAQ) et 2.284,57 (AAQ + AMX). Les résultats montrent que le coût par décès évitéétait de 2.585,58 US$ pour AAQ et 3.272,20 US$ pour AAQ + AMX. Le coût par DALY évitée était de 90,25 US$ (AAQ) et de 114,21 US$ (AAQ + AMX).

Conclusion:  Les approches AAQ et AAQ + AMX étaient toutes deux rentables, chacune permettant d’éviter un DALY à moins du seuil standard de 150 US$ recommandé par l’OMS. Toutefois, AAQ était la plus rentable. La prise en charge à domicile des fièvres chez les moins de cinq ans en milieu rural est rentable en réduisant la mortalité chez les moins de cinq ans.

Abstract

Objetivo:  Evaluar el coste-efectividad de dos estrategias de manejo domiciliario de menores de cinco años en Ghana – un tratamiento utilizando solo antimaláricos (artesunato amodiaquina – AAQ) y un tratamiento combinado utilizando antimaláricos y antibióticos (artesunato amodiaquina + amoxicilina − AAQ + AMX).

Métodos:  Hemos evaluado los costes y el coste-efectividad de la AAQ y AAQ + AMX comparados con un control recibiendo cuidados estándar. Se recolectaron datos como parte de un ensayo aleatorizados y controlado en conglomerados con un diseño escalonado. Se incluyeron aproximadamente 12,000 niños con edades entre los 2 y 59 meses en el distrito de Dangme del Oeste, al sur de Ghana. Las intervenciones fueron realizadas por trabajadores sanitarios comunitarios. Los costes se analizaron desde una perspectiva social, utilizando las muertes de menores de cinco años evitadas y los años de vida ajustados por discapacidad (AVADs) evitados como medidas de la efectividad.

Resultados:  Los costes económicos totales para las intervenciones fueron US$ 204,394.72 (AAQ) y US$ 260,931.49 (AAQ + AMX). Los costes recurrentes constituían un 89% y 90% del total de costes directos de AAQ y AAQ + AMX respectivamente. Las muertes evitadas fueron 79.1 (AAQ) y 79.9 (AAQ + AMX), con 2,264.79 (AAQ) y 2,284.57 (AAQ + AMX) AVADs evitados. Los resultados muestran que el coste por muerte evitada era de US$ 2,585.58 para AAQ y US$ 3,272.20 para AAQ + AMX. El coste por AVAD evitado era de US$ 90.25 (AAQ) y US$ 114.21 (AAQ + AMX).

Conclusión:  Tanto la opción de AAQ como AAQ + AMX eran coste-efectivas, cada una evitando un AVAD a menos del umbral estándar de US$ 150 recomendado por la Organización Mundial de la Salud. Sin embargo, la AAQ era más coste-efectiva. El manejo domiciliario de fiebres en menores de cinco años en zonas rurales es coste-efectivo para la reducción de la mortalidad en menores de cinco años.

Introduction

Access to prompt and appropriate treatment is crucial in reducing mortality in febrile children under five years. In case of malaria, current and more effective antimalarials (i.e. artemisinin-based combination therapy – ACT) and diagnostic techniques (i.e. rapid diagnostic test – RDT) have been introduced to help reduce the burden of the disease in developing countries. However, access to treatment is still a major problem hindering the control of the disease in most countries (Lubell et al. 2010). WHO recommended managing febrile conditions in children under five years old at home (WHO 2004). Home management of malaria (HMM) involves using the services of trained community volunteers who worked as community-based agents (here termed ‘community health workers’) in treating febrile children near the home with pre-packed antimalarials. By 2008, 18 African countries had taken steps to implement the approach (Ajayi et al. 2008). Currently, the implementation of HMM has expanded to other childhood fevers such as acute respiratory infections (ARIs) and diarrhoea.

Studies in Africa have provided evidence of the effectiveness of home management in treating fevers (Kidane & Morrow 2000; Sirima et al. 2003). However, evidence of the cost-effectiveness of the approach is limited. There are two published studies on the cost-effectiveness of HMM in Africa (Lubell et al. 2010; Chanda et al. 2011).

The Ministry of Health (MOH) of Ghana began implementing home-based care in selected districts throughout the country (MOH/NMCP 2010) in 2011 to target malaria, ARI and diarrhoea in children under five years old. The intervention will be scaled up to cover other districts in the country in subsequent years. To date, there is no (published) evidence of the cost-effectiveness of HMM in the specific Ghanaian context. Our aim was to assess the cost-effectiveness of two strategies of home management of fevers (HMF) in children under five years: treatment using antimalarials only (artesunate–amodiaquine – AAQ) and treatment using a combination of antimalarials and antibiotics (AAQ plus amoxicillin – AAQ + AMX). During a cluster randomised controlled trial, the two strategies – AAQ and AAQ + AMX – were implemented, and the impact of each strategy on morbidity and all-cause mortality in children under age five was assessed.

Methods

Study area

The study was conducted between January 2006 and December 2009 in the Dangme West District, one of the rural districts of the Greater Accra Region, Ghana with an area of ca. 1700 km2. Approximately, 12% of the district’s total population of 109 459 consists of children under age five (Nonvignon et al. 2010). The district experiences an all-year malaria transmission with peaks during the rainy seasons in April and October (M. A. Chinbuah, P. A. Kager, M. Abbey, M. Gyapong, E. Awini, J. Nonvignon, M. Adjuik, M. Aikins, F. Pagnoni, J. O. Gyapong, in preparation). Residents are predominantly farmers and fishermen.

Study design and interventions

The home management trial was a cluster controlled trial with the interventions introduced in a step-wedge manner. At the beginning of the trial, all clusters belonged to the control arm until they were randomised to either of the two intervention arms at different time periods. During the control period, no intervention was put in place; febrile children 2–59 months continued to seek standard care at health facilities, over-the-counter and other sources previously used. In clusters belonging to the AAQ (antimalarial only) arm, febrile children 2–59 months received artesunate–amodiaquine when taken to community health workers (CHWs) by their caregivers and diagnosed of having uncomplicated fever. In clusters belonging to the AAQ + AMX (antimalarial plus antibiotic) arm, febrile children received artesunate–amodiaquine plus amoxicillin when diagnosed of fever by CHWs. In each arm, the treatment first dose was observed. A detailed description of the study design and interventions are reported elsewhere (M. A. Chinbuah, P. A. Kager, M. Abbey, M. Gyapong, E. Awini, J. Nonvignon, M. Adjuik, M. Aikins, F. Pagnoni, J. O. Gyapong, in preparation).

Study population

The study populations were all children 2–59 months, their caregivers and CHWs in the study district. By the end of the trial, 660 CHWs had been trained and worked.

Data collection

Data on recurrent and capital costs were obtained from the programme records kept at the Dodowa Health Research Centre (DHRC). CHWs also recorded the number of children treated and the drugs given. In addition, some CHWs were trained to record the caregivers’ travel time and the CHW’s time spent on treatment and with field supervisors. Numbers of deaths were obtained from the mortality records of the trial and the demographic surveillance system at the DHRC. The number of anaemia cases was determined during biannual cross-sectional surveys undertaken during the trial by taking the blood samples from surveyed children using a haemocue.

Estimating costs

Costing was done from the societal perspective, which considered households costs (transport, food and related expenses) and health system costs. Financial costs (actual financial expenses) and economic costs were estimated and are reported in the results. Direct and indirect costs are reported.

Direct costs. Direct costs had capital and recurrent components. Capital costs included vehicles and equipment that were annualised using the useful life of items and the recommended discount rate of 3% (Gold et al. 1996). The useful life of capital was estimated on the basis of the discussion with the accountant and the stores manager at the DHRC. The capital items were vehicles, motorcycles and helmets, television sets, flip chart stands, generators, digital camera, DVD players, VCDs and extension equipment. Recurrent costs included medical supplies (e.g. drugs), stationery supplies, allowances and salaries of staff involved in programme activities, and cost of vehicle operation and maintenance. Market rental price was used for building spaces. Capital costs and recurrent costs were summed up to obtain the total costs for each study year. These costs were then apportioned to the two study arms using the proportion of CHWs and number of communities under each study arm. However, the cost of drugs that were specific to a particular study arm was allocated to the respective study arm.

Indirect costs. Indirect costs constitute productivity losses. This included the time of CHWs (who were not remunerated) volunteered to the trial. Also, the productive losses to caregivers (travel and treatment time) were included. Indirect costs were analysed using the average local agricultural labour wage per day.

Costs in the local currency (Ghana Cedis) for each year were adjusted to 2009 (final year) cost using the consumer price indices (CPIs) for health goods for the respective years. These values were then converted into a United States dollar equivalent using the 2009 average interbank exchange rate of 1.42 Ghana Cedis to 1.00 dollar (as quoted by the Bank of Ghana). The calculation of costs was done on a Microsoft Excel (2007) spreadsheet.

Estimating programme effectiveness

Anaemia cases averted, deaths averted and disability-adjusted life years (DALYs) averted were the effectiveness measures. The number of deaths observed during the trial period were 181 (control), 72 (AAQ) and 57 (AAQ + AMX). Records of these deaths were obtained from the trial records and the Health and Demographic Surveillance System (HDSS), which is a biannual census that obtains data on vital events such as deaths, births and pregnancies. However, for the purpose of estimating the cost-effectiveness of the trial interventions, the indirect standardization method (Kirkwood & Sterne 2005, pp. 268 & 269) was used to calculate the expected number of deaths in AAQ and AAQ + AMX with the control as the reference population. The method made use of the observed age-specific deaths and person years in each study arm. The expected deaths in AAQ and AAQ + AMX were estimated to be 101.9 and 101.3, respectively. Originally developed in 1990, DALYs are a measure of the burden of diseases and are expressed as number of years of life lost to illness and years lived with disability (YLD). Years of life lost (YLL) is a component of DALY and constitutes the majority of the burden of fever (especially malaria). Using life expectancy at a mean age of death of 65.4 years (Ghana Statistical Service 2011), no age weighting and a discount rate of 3% (Murray & Lopez 1996a,b), each death was estimated to be equivalent to 28.6 YLLs. The differences between the effects (i.e. anaemia cases, deaths and DALYs) under each study arm and those under the control constituted the final effects (i.e. anaemia cases averted, deaths averted and DALYs averted) used in the cost-effectiveness analysis. The anaemia cases were 2289 (control), 928 (AAQ) and 1142 (AAQ + AMX). The anaemia cases were not standardised because prevalence was assessed by determining the proportion of surveyed children with haemoglobin levels lower than 11 mg/dl.

Estimating cost-effectiveness of interventions

Cost-effectiveness ratios (CERs) were estimated for AAQ and AAQ + AMX, in each case using the control as the comparator. Cost per death averted for each intervention arm was compared to the threshold guideline used by the WHO Commission on Macroeconomics and Health (also used by WHO-CHOICE) that an intervention is ‘highly cost-effective’ if CER is less than three times gross domestic product (GDP) per capita, ‘cost-effective’ if CER is between one and three times three times GDP per capita and ‘not cost-effective’ if CER is greater than three times GDP per capita (WHO 2001). The World Bank estimated Ghana’s GDP per capita for 2009 to be US$ 1098 (data.worldbank.org/indicator/NY.GDP. PCAP.CD). Cost per DALY averted was also compared to the threshold guideline that an intervention is ‘highly attractive’ if cost per DALY is below US$ 25–30 and ‘attractive’ if cost per DALY is below US$ 150 (WHO 1996).

Sensitivity analysis

First, a one-way sensitivity analysis was performed using a lower (2%) and a higher discount rate (5%) and varying the useful life of vehicles (to 8 years instead of 5 years) used in calculating capital cost. The national daily minimum wage rate was also used in place of the local daily wage rate in estimating indirect cost. Further, other drugs (arthemeter-lumefantrine and co-trimoxazole), which are also recommended by the MOH, were used in the sensitivity analysis. During the trial, CHWs served as volunteers who were not remunerated for their services. However, as part of the sensitivity analysis, it was assumed that CHWs were given allowances equal to their indirect costs. Further, an assumption of higher (50%) attrition rates (reported by Olang’o 2010 in Nigeria) among CHWs was also used. In addition, assumptions of mortality rates being 10% lower and between 1% and 20% higher than the observed rates were used alternatively. Finally, a multi-way sensitivity analysis was performed varying multiple parameters at the same time.

Ethical clearance

The ethics committees of Ghana Health Service and the WHO granted ethical clearance for the study.

Results

Programme costs

The total financial costs of the study arms were US$ 99 690.32 for AAQ (Table 1a) and US$ 138 321.21 for AAQ + AMX (Table 1b). Recurrent costs constituted 84% (AAQ) and 86% (AAQ + AMX) of the total financial costs. The total financial cost of the interventions per 1000 population was US$ 7431.26 for AAQ (Table 1a) and US$ 9796.81 for AAQ + AMX (Table 1b). Table 1 also present the annual financial costs and annual financial costs per 1000 population.

Table 1.   Financial costs of (a) AAQ study arm and (b) AAQ + AMX study arm
 Year 1*Year 2*Year 3*Year 4*Total (2009 US$)†
  1. *Average US$-GHC exchange rates were 0.92, 0.94, 1.07 and 1.42 for Year 1 (2006), Year 2 (2007), Year 3 (2008) and Year 4 (2009), respectively.

  2. †Total costs in year 2009 were arrived at by, first, converting Year 1, Year 2 and Year 3 costs into final year (Year 4) costs up using the formula inline image

  3. Where, Cost yeari is cost of year (1, 2 and 3) in final year value; CPIfinal year is the average health component of consumer price index for the final year; CPIyear i is the average health consumer price index for the year (1, 2 and 3). Then, these were summed up.

  4. ‡Total costs per 1000 population equals (Total cost × 1000)/population for the study arm in the respective year – in the case of the 2009 total cost, the total population for the period was used.

  5. §Interventions had not been introduced so there was no population for this year, although pre-intervention costs were apportioned to the respective arms as these costs were incurred on selecting CHWs and sensitizing communities, which eventually belonged to the interventions.

(a)
Direct cost
 Capital cost741.485127.043447.201313.6116 330.29
 Recurrent cost1585.4724 019.0721 424.778994.6483 360.03
 Total2326.9529 146.1124 871.9710 308.2599 690.32
 Total cost per 1000 pop‡N/A§6408.565512.402366.997431.26
(b)
Direct cost
 Capital cost898.506008.774225.241605.5219 440.73
 Recurrent cost1921.2133 999.1631 611.0412 502.82118 880.49
 Subtotal2819.7140 007.9335 836.2714 108.34138 321.21
 Total cost per 1000 pop‡N/A§8574.357511.273013.329796.81

The total direct economic costs of the study interventions were US$ 154 411.12 and US$ 203 172.75 for AAQ (Table 2a) and AAQ + AMX (Table 2b), respectively. Recurrent costs accounted for 89% (AAQ) and 90% (AAQ + AMX) of the direct economic costs. The total indirect costs of the interventions to CHWs and caregivers were US$49 983.59 for AAQ (Table 2a) and US$57 758.74 for AAQ + AMX (Table 2b). Further, the total economic costs for the study interventions were estimated to be US$204 394.72 for AAQ (Table 2a) and US$260 931.49 for AAQ + AMX (Table 2b). Figure 1 shows that 24% of the total economic cost of AAQ was indirect costs. For AAQ + AMX, 22% of the total economic costs were indirect costs (Figure 2).

Table 2.   Total economic costs of (a) AQ study arm and (b) AAQ + AMX study arm
 Year 1*Year 2*Year 3*Year 4*Total (2009 US$)†
  1. Notes *, †, ‡ and § on Tables 1 also apply to Table 2.

  2. ¶The local wage rates per hour were GHC 0.875 for 2007, GHC 1 for 2008 and GHC 1.5 for 2009. The numbers of CHWs were AAQ – 179, 298 and 234 for Year 2, Year 3 and Year 4, respectively; AAQ + AMX – 204, 362 and 386 for 2007, 2008 and 2009, respectively. Finally, on average, the numbers of hours that CHWs worked on the trial were AAQ – 62.17, 48.67 and 36.05 for Year 2, Year 3 and Year 4, respectively; AAQ + AMX – 62.59, 46.72 and 35.25 for Year 2, Year 3 and Year 4, respectively.

(a)
Direct cost
 Capital cost809.305526.993708.641400.7817 590.45
 Recurrent cost3836.2840 904.0532 413.3313 596.67136 820.67
 Subtotal4645.5846 431.0436 121.9714 997.45154 411.12
 Indirect cost¶11 269.1114 969.7610 341.6849 983.59
 Total cost4 645.5857 700.1551 091.7325 339.13204 394.72
 Total cost per 1000 pop‡N/A§12 686.9311 323.525818.4015 236.28
(b)
Direct cost
 Capital cost980.686479.474546.231712.0620 944.95
 Recurrent cost4648.6753 665.4445 046.5118 127.52182 227.80
 Subtotal5629.3560 144.9149 592.7419 839.58203 172.75
 Indirect cost¶12 813.7517 488.5712 161.9357 758.74
 Total cost5629.3572 958.6767 081.3232 001.51260 931.49
 Total cost per 1000 pop‡N/A§15 636.2314 060.226835.0118 480.88
Figure 1.

 Total economic cost of AAQ study arm.

Figure 2.

 Total economic cost of AAQ + AMX study arm.

Programme effects

Anaemia cases averted by AAQ were 1361 and by AAQ + AMX, 1147. The deaths averted by AAQ and AAQ + AMX were calculated as 79.1 and 79.7, respectively. These translate into 2264.79 DALYs averted for the AAQ arm and 2284.57 for the AAQ + AMX arm.

Cost-effectiveness ratios

The cost-effectiveness ratios are presented in Table 3. The cost per anaemia case averted was US$ 150.18 (AAQ) and US$ 227.49 (AAQ + AMX). Cost per death averted was US$ 2585.58 for AAQ and US$ 3272.20 for AAQ + AMX, while the cost per DALY averted was US$ 90.25 for AAQ and US$ 114.21 for AAQ + AMX.

Table 3.   Cost-effectiveness ratios for study arms
Costs (US$)AAQ armAAQ + AMX arm
Total economic cost204 394.72260 931.49
Cost per anaemia case averted150.18227.49
Cost per death averted2585.583272.20
Cost per DALY averted90.25114.21

Sensitivity analysis

The results of the one-way and multi-way sensitivity analyses indicate that the costs and cost-effectiveness estimates were not sensitive to changes in most of the parameters used, with the exception of changes in mortality rates. While lower mortality rates enhanced the cost-effectiveness of both interventions, a 1% increase in the mortality rates renders AAQ + AMX not cost-effective. On the contrary, AAQ remains cost-effective even if mortality rates increase by between 1% and 16%. Any higher percentage increase in the mortality rate would render AAQ not cost-effective.

Discussion

The study results indicate that HMF in children under age five using antimalarials plus antibiotics is financially and economically more expensive than using antimalarials only. This result was mainly attributable to the additional drugs used in treating the combined group and the different numbers of children treated and CHWs in the study arms. Further, a greater portion of the costs were recurrent, which indicates that the home management trial used more recurrent inputs than capital inputs. The study findings also indicate that a substantial portion (more than one-fifth) of the total economic cost of each study arm was indirect costs. Thus, the contribution of indirect costs to the total economic cost of home management programmes must not be overlooked. The use of the provider perspective in analysing costs and cost-effectiveness of home management interventions is likely to lead to important costs being ignored, because the provider perspective excludes indirect costs to caregivers.

Using the recommended threshold criteria, AAQ and AAQ + AMX are both cost-effective compared to standard practice even though the difference in deaths between AAQ and AAQ + AMX was not statistically significant. However, the results show that AAQ represents a more attractive use of society’s resources. This is because while AAQ + AMX averts more deaths and DALYs than AAQ, it does so at a higher economic cost to society. The findings of this study confirm the findings of other studies that home management of childhood illnesses is cost-effective (Goodman et al. 2006; Conteh et al. 2010; Lubell et al. 2010; Tozan et al. 2010; Chanda et al. 2011). This is the first study in Ghana and the third in sub-Saharan Africa – after Chanda et al. (2011) and Lubell et al. (2010)– to provide evidence that the delivery of pre-packed antimalarials (specifically ACTs) through the home management approach recommended by WHO is cost-effective. However, our study is the first to provide evidence that treatment of fevers in children under age five near the home using pre-packed antimalarials in combination with antibiotics is also a cost-effective way of reducing the burden of fevers and under five mortality in low-income countries, especially in sub-Saharan Africa.

The cost-effectiveness of child health interventions is important in determining which interventions should be scaled up or have increased resources devoted to them. Nevertheless, decisions to scale up or not to scale up given child health interventions are in some cases also dependent on other factors, such as the value a policymaker places on a child’s life in the specific context.

It is worth noting that our study was undertaken in a rural setting and that its findings may not be applicable to urban settings. We used special dispersible amoxicillin tablets produced by a local pharmaceutical company for different age groups and these drugs were more expensive than currently marketed pills. Thus, the total cost would have been differed if the market price had been used.

In conclusion, we emphasise the point that given the resource constraints in delivering child health interventions through the formal health-facility settings in developing countries, home management strategies could be used to complement service delivery through health facilities, especially in rural settings. These two ways – delivery through health facilities and home management strategies – together will accelerate progress towards meeting the Millennium Development Goal of reducing child mortality.

Acknowledgements

The study received technical and financial support from the UNICEF/UNDP/World Bank/WHO Special Programme for Research and Training in Tropical Diseases, the Knowledge Enriches Programme of the Netherlands Organization for the Advancement of Tropical Research, WOTRO (NWO) and the African Doctoral Dissertation Research Fellowship, a collaborative programme of IDRC and African Population and Health Research Centre. We further acknowledge contributions from the community health workers, staff of Dodowa Health Research Centre (especially the malaria–pneumonia team), Dr. Evelyn Ansah, Director of Health Services, Dangme West District, Professor Piet A. Kager, Amsterdam Medical Centre, and Professor Bart van den Borne, University of Maastricht, the Netherlands. We also acknowledge technical inputs from Dr. Catherine Goodman of the London School of Hygiene and Tropical Medicine.

Ancillary