In several countries, including Malawi’s neighbor Zambia, health facility data have indicated significantly declining trends in malaria (4–8,23–25). Investigators have attributed such declines (at least partially) to the rapid scaling up of malaria control interventions. To our knowledge, this study is one of the few hospital-based retrospective analyses to show no markedly declining trends in malaria during the past decade in a sub-Saharan African country (26). The apparently undiminishing malaria at QECH has occurred in the context of increasing efforts to increase malaria control interventions in the country and despite marked reductions in overall deaths among children <5 years of age (27). Progress has in fact been substantial in scaling up malaria control interventions in Malawi in recent years, but some parts of the country did not reach high coverage until 2010 (13). That evidence of a reduction in malaria in children was not observed in the period during 2001–2010 is therefore not entirely surprising; such a reduction might not become apparent until after 2010. Although use of insecticide-treated bed nets for children <5 years of age has increased quickly in recent years, reaching ≈52% in urban areas and 57% in the southern region (including Blantyre), coverage of other interventions, such as indoor residual spraying, remained low (≈3%) in Blantyre in 2010 (13). In areas where dramatic reductions of malaria attributable to rapid scale-up of interventions have been reported (such as in neighboring Zambia), high coverage of interventions with both insecticide-treated bed nets and indoor residual spraying was achieved (8,28–30). Therefore, because indoor residual spraying coverage is likely to play a critical role in reducing transmission intensity and, ultimately, the incidence and prevalence of malaria, in 2011, Malawi expanded its indoor residual spraying activities to 7 districts (13) to maximize the effects of malaria control interventions. Strengthened surveillance systems capable of monitoring short-term changes in disease patterns and evaluating in-country differences in uptake of interventions will become key for assessing these expected effects.

The data presented in this retrospective analysis show little change in the pattern of malaria in children during 2001–2010 in Blantyre. The proportion of malaria-positive slides among febrile children in the PAEU declined in 2003 from the preceding 2 years, but no further reductions were observed after 2005. Approximate parasite densities assessed from thick blood films also remained similar throughout the decade. The prevalence of anemia decreased significantly during the first 3 years of the study period, but no further changes of note occurred after 2004. Trends in parasite density and anemia prevalence in children in the PAEU were based on a systematic sample of records. Although a systematic sampling approach could have led to a potential underestimate or overestimation of monthly figures, these potential differences are likely to have been diluted after monthly estimates were pooled to obtain overall annual figures and are therefore unlikely to have distorted true trends of parasite density and anemia levels over time.

The unexpectedly high SPRs and anemia levels in 2001 might have resulted from the severe food crisis that affected the entire country during that period. There were reports at the time of increases in the prevalence of malaria and anemia in the Blantyre area (31). Because laboratory records from outpatient visits were not available before that period, we were unable to assess trends before 2001.

The number of children with cerebral malaria admitted each season to QECH has fluctuated moderately since 2001 but with no downward trend and with no significant reduction in any year since 2001. Variations in the rainfall pattern are unlikely to have played a major role in the variations in patients admitted with cerebral malaria. The amount of rainfall was fairly constant from year to year, and the months of peak rain coincided with the months of peak cerebral malaria admissions in all years except 2008 (N. Feasey et al., unpub. data).

The mean age of children with cerebral malaria was significantly higher after 2007 than in 2001. A shift of susceptibility to severe malaria toward older children is theoretically possible in areas where malaria control interventions are being rapidly scaled up and has been reported elsewhere (5). Continued monitoring of our patient population is needed to confirm whether this trend is maintained. Unlike the situation in other sites, in Blantyre this recent age trend in cerebral malaria patients has not been accompanied by equivalent changes in SPR or anemia prevalence among hospital outpatients.

The use of health facility–based data is subject to several limitations. These limitations include lack of a known community denominator, likely variations in access to or use of the facility, and uncertain quality and standardization of diagnoses. Such data are best interpreted in the context of community-based surveys, which can provide data on the extent of coverage by various interventions and provide indicators of infection and disease prevalences among the population. We used malaria indicators from children admitted for elective surgery to an adjacent pediatric orthopedic hospital (BCIH) as a proxy for indicators likely to prevail in the community surrounding both hospitals. These data showed no substantial reductions in SPR during 2003–2009 but some evidence of a reduction (p<0.05) in 2010 from 2003. The SPR obtained from these trauma patients is a good proxy for the malaria prevalence in the community (5) because children attending a free trauma hospital are unlikely to be a biased sample with regard to malaria risk. Although the catchment area of BCIH is wider than that of QECH (because it admits children from the southern and central regions), and therefore could include children exposed to higher or lower intensities of malaria transmission than those of Blantyre district, the catchment area and referral patterns have not changed over time. Individual SPR trends are therefore unlikely to have been biased in this respect. These data, based on a retrospective health-facility analysis that used a significance level of 0.001, suggest that the persisting levels of P. falciparum infection among children with fever at the hospital reflect a similar persistence of the infection in the community over the last decade.

Several factors might have led to underestimation of a decline in malaria prevalence, however. The increase in the number of outpatient visits over time probably reflects population growth of the area or changes in health-seeking behavior (from increased activities in health promotion that would have resulted in increased self-reported outpatient visits), and the number of slides taken remained roughly constant over time. This trend could represent a decrease in the proportion of attendees with a malaria-like illness; alternatively, the facility might have reached its capacity to conduct malaria tests, and when that capacity was exceeded, patients were treated for malaria on the basis of a presumptive diagnosis. If the latter occurred, we would have underestimated malaria in the later years of our study period. This potential underestimation is an intrinsic problem of retrospective health facility–based studies that might have resulted in a systematic bias in observed trends. Although a systematic bias could have occurred in the PAEU, we are confident that it did not affect the high-dependency ward because all cerebral malaria patients are admitted to this ward during the rainy season. However, previous studies reporting declining trends of hospital admissions for malaria (4,6–8,23–25) also were subject to similar limitations and should be interpreted with caution. Lastly, many children seen at QECH who were residents of Blantyre might have acquired their malaria in their journeys to rural areas because use of bed nets by short-term travelers is less common than for those sleeping in their permanent residence (32), which partially could have masked a true decline in malaria in Blantyre.

As malaria control efforts increase in Malawi and in similar areas, continued surveillance is needed to describe changes in malaria among a population exposed to lower intensities of malaria transmission. Significant increases in the mean age of outpatient and hospitalized children with positive slides and a rise in the proportion of cerebral malaria among all those admitted for severe malaria are among changes that might occur (33). Field-based surveillance and regular malaria indicator surveys are major methods by which to monitor the effectiveness of malaria control. However, to identify changes in disease patterns, mechanisms for the sustainable, prospective collection of surveillance data from health facilities will be increasingly needed to evaluate the effect of malaria control programs.