Until relatively recently malaria has always been considered a rural disease in Africa mainly because of the perception that suitable malaria vector breeding sites in towns and cities were rare. This may explain why the transmission of malaria is generally lower in the urban environments, when compared to that of rural areas; and together with a greater availability of health care providers and a generally higher socio-economic status in the urban situation  results in a relatively lower malaria morbidity and mortality . However, data from Accra and Kumasi, Ghana demonstrated that malaria can be a major public health problem in the urban setting and that prevalence can vary markedly between communities and between cities .
The studies described in this paper are implementation research studies, which (by definition) have to follow national policies and be embedded in 'real life' local ways and cultures. This accounts for the diverse type of CMDs operating in different countries ('traditional CHWs in some countries vs. HSA on the government payroll in Malawi) and may explain either the diversity of the effects observed on health seeking behaviour and some of the differences in performance or impact between studies.
Community medicine distributers in urban CCMm
Notwithstanding differences between sites, the results show that CCMm in African urban environments is a feasible option, which is well accepted by the populations. In line with findings in rural areas, the overall quality of care provided by CMDs is quite satisfactory (99.2% of correct prescriptions), as well as adherence by caregivers to the correct treatment schedule (92.6%) and the promptness of administering the first dose (68.3%). However, the coverage obtained by urban CCMm is generally more modest than in rural environments, with "only" 40% of caregivers consulting CMDs for the treatment of their febrile children compared to the overall figure of 59% noted by Ajayi .
In Jimma, about 80% of caregivers reported that their preferred provider for childhood malaria treatment was the CMDs because services were available in the neighbourhood and were quickly performed, however only 7.3% of caregivers used the services as they preferred to utilize the 'established' health services. In fact, this apparent disconnect between perceptions and practice was also evident in Bolgatanga and Ouagadougou, although the differences were not as great as in Jimma. In Kumasi, only 31% said that they would prefer to use the services of the CMDs, but actually 41% used them. These differences between perceptions and practice can generally be accounted for by caregivers' preference for the existing systems (which they know); the proximity of established pharmacies, clinics, etc.; lacking confidence in their CMD; not finding the CMD at the first attempt; professing to be unaware of the quality of their training; stock-outs, and inability to pay. Of interest is the situation in Lilongwe, where 64% of respondents said they would prefer the CMDs, and the same proportion actually used them. The point about the study area in Lilongwe (Kauma) is that the CMDs were the only health service providers, and as has been stated previously, the eight CMDs had a much wider role than those in the other sites under investigation. The fact that the Lilongwe CMDs had, in addition to their normal duties, malaria diagnosis and treatment, meant that they were filling a void in the area. Thus it is not surprising that more people sought their services than going elsewhere.
The main reason that CMDs were used was because of proximity; they were trained, and they provided quick treatment.
It would appear that the effectiveness of the CCMm is directly linked to the efforts of the CMDs to become familiar with the caregivers in their areas, and to engender trust in their ability to diagnose and treat childhood malaria quickly, safely and effectively.
Perceptions and practice; measures of performance and indicators of treatment coverage
A comparison of those parameters that were measured by both the household survey and the CMDs registers shows a close agreement between what the CMDs recorded and that which the caregivers recalled in the last two weeks of the implementation phase in terms of quality of care and utilization of services.
The CMD registers showed that 99.4% of the children were treated with the correct dose for their age, and the household survey showed that 92.6% were treated with the correct dose and for the correct number of days.
The CMDs recorded that 54.2% of the children were treated within 24 hours of being ill, whereas the household survey recorded 67.6%. We acknowledge that the data are not strictly comparable, as CMD registers were maintained over a period of one year (the intervention phase) and the household survey was conducted at the end of the intervention phase with a recall period of two weeks. In a similar study carried out in rural areas of sub-Saharan Africa  adherence was measured in terms of caregivers' report of the number of doses administered to the child, the number of days over which treatment was given and the promptness of treatment after the onset of symptoms. Overall 85% of children were treated correctly in terms of drug dose and duration of administration. This figure is in line with those of the present study in urban communities, where the overall adherence was 92.6% and greater than 90% in three of the five sites.
The unit dosed pre-packed anti-malarial medication for the CCMm is widely accepted amongst mothers  and this study found over 90% and 84% acceptable levels for the use of pre-packs at the baseline and after the intervention. Willingness to pay for pre-packs amongst caregivers was slightly over 50%. In Bolgatanga, a moderate price (US$0.137) was affordable to most caregivers and as it was much lower than the prevailing price at the municipal health facility it was highly acceptable. This was cheaper than the prices charged for ACT in Kumasi (US$0.182 and US$0.455) or in Ouagadougou (US$0.23 and US$0.34), where the costs were acceptable. A future debate may therefore revolve around the affordable charges that programmes could make for the confirmatory diagnosis and treatment of malaria. Given the figures presented here, it is possible to conjecture that diagnosis may cost US$1.00  and ACT up to US$0.50, making a charge of US$1.50 per malaria confirmed under-five year old child. The question then becomes "Is CCMm affordable in the urban setting, or would caregivers find it cheaper to seek care in the private sector?" Self-treatment continues to be common, and in urban areas it is commonly due to the large numbers of medicine vendors (formal and informal) that exist. Various reasons why people patronize drug shops, pharmacies, and even illegal drug sellers instead of health centres and hospitals, indicate that it is the ease of buying and obtaining immediate treatment that encourages them. In a study in Maiduguri in Nigeria, it was reported that immediate attention for both consultations and treatment was the most important reason why pharmacies and drug shops were frequently patronized, rather than the health centres . Thus, it would appear that self-treatment is a common practice in malaria endemic regions of Africa, and this may be even more prevalent in urban areas. A recent study in urban Kampala in which anti-malarial drugs were stored by mothers at home and given to children if they developed fever, showed that there was substantial over-treatment and little effect on clinical outcome . While we agree with the authors that ACT provided in the home might not be appropriate for large urban areas or settings with fairly low malaria transmission, the findings of our study show that when drugs are distributed by appropriately trained CMDs, the quality of prescription and adherence to treatment schedules by caregivers are as high as in rural areas with high malaria transmission.
Qualitative research findings of the study indicate that the acceptability of pre-packed ACTs for treating malaria in children under 5 was very high. Compliance with the recommended dosage, as well as treating children with malaria or fever within 24 hours by caregivers was said to have improved significantly across sites. These data indicate that the use of the CCMm in urban settings is feasible and acceptable to caregivers, and contributes to an improvement of malaria case management.
It is vital for the success of CCMm in urban situations to implement an effective and sustained health educational programme aimed at caregivers, coupled with an equally important educational and training programme specifically for health personnel. It was extremely difficult to mobilize urban communities for health education programmes. In Kumasi, only about 25% of caregivers sought care for their children either from clinics or hospitals depending on distance and recommendations from relatives and friends (Kumasi interviewed respondents).
The main challenge to urban CCMm is the motivation of CMDs in order to sustain the pre-packed ACTs . At this stage it is not possible to state whether it was a failing of the IEC programmes, socio-cultural barriers or that it was the 'distraction' of competing health care providers that seemingly prevented a higher utilisation rate of CMDs. IEC strategies that are effective in rural settings may not be appropriate in urban environments, and sensitization programmes should be tailored to the urban setting. The experience gathered in these studies shows, for instance, that most IEC messages conveyed through local radio and television are more effective than through the traditional use of religious meetings, which has the highest impact in rural communities.
Considering that the IEC campaigns were not as successful as anticipated, did the caregivers gain any knowledge about malaria signs and symptoms during the period of the intervention? In Kumasi, Lilongwe and Ouagadougou there were marked improvements (Table 4). In Lilongwe there was an improvement of knowledge of nearly all the signs and symptoms listed; in Kumasi fewer signs and symptoms were listed as showing some improvement and in Ouagadougou there were less still. In Lilongwe and Ouagadougou the important signs: vomiting, convulsions, inability to feed/suckle and high temperature showed marked improvement, whilst in Kumasi knowledge of vomiting, high temperature and inability to feed/suckle showed marked improvement. In both Bolgatanga and Jimma, the caregivers tended to lose knowledge, with only one or two signs and symptoms showing any improvement.
CMDs will not always work as volunteers. They need remuneration or incentives in order for them to make an intervention sustainable and effective, e.g. in the Sudan the volunteer Malaria Control Assistants worked on a consultation fee of US$0.50  and in the Ouagadougou study reported here, the vendors of ACT were allowed to keep 10% of the sales as a personal incentive. In order for the CCMm approach to be valued in urban situations, the concept should be supported by the local health personnel, who would then see the CMDs as official health workers who spearhead an outreach service.
By using collaborative approaches that include the community, the private sector and existing health, urban planning, agricultural and governance structures, urban malaria is uniquely amenable to prevention and control .