Archive for the 'Surveillance' Category

Measuring malaria incidence

Why are disease burden estimates important? There are two sets of reasons: 1) technical, e.g. national priority setting etc and 2) political economy of health, e.g. advocacy and visibility, donor influence, etc. Thus, estimates are often quite contentious apart from the methodology used. The practical applications of burden estimation are on the other hand limited. Monitoring and evaluation systems can, with valid measurements of trend, help allocate regional resources, determine impact, and respond to outbreaks without necessarily measuring the magnitude of disease.

There are broadly two competing approaches for estimating malaria burden. One extrapolates from maps and surveys of population cross-sections, the other extrapolates from routinely collected surveillance data. Both have certain assumptions and biases: the former tends to overestimate, the latter generally underestimates. How should countries proceed? A new article from WHO in PloS Medicine (open access!) describes and contrasts each in a careful and informative way including actual results from 2009. While neither method is perfect, the authors present a compelling case for investing in surveillance-based approaches in the long run particularly outside of hyperendemic settings and in the context of intervention scale-up and declining transmission. Only that approach can improve the health system, and provide disaggregated and timely data amenable to frequent updates.

An accompanying editorial by Mueller, Slutsker, and Tanner broadly agrees with the author’s outlook though it includes a careful placative statement:

The two methods thus have their unique strengths and weaknesses, and rather than seeing them as competing approaches, they should be synergistically combined.

At first pass this seems a sensible assertion but on closer examination, and without other supporting text, it’s not clear what to make of it. Which distinct synergies do they envision? Should the two methods be used globally, i.e. one in countries without quality surveillance and the other elsewhere? Or should the two methods always be used in each nation? Why? Should they be used indefinitely – what about when surveillance systems mature? Are there not trade-offs both in time and space? Such discussion, drawing on the considerable program experience of the writers, would have been more helpful than conciliatory statements of questionable sincerity.

Innovation in malaria case detection

The paradox of malaria and many tropical diseases is that those most at-risk are also some of the least likely to access, or be able to access, health facilities. Active case detection, the screening of fever cases in the community itself, helps enable case management in such remote or inaccessible areas. But it’s also time and manpower intensive. Answering the questions of where to target, and whom to target is critical to making sure that effort will be worthwhile. In practice, the usual mechanical application of the strategy ensures that it will not be efficient.

So it’s beautiful to see an example of creative thinking. Last year in the infamous Jalpaiguri district of India, I met an unique, young block medical officer. Over some hot milk tea, the clean-shaven late 20s year old commented on how most of the cases at his primary health center labored in the dense jungle along the Bhutanese border. Living in secluded villages, many sought care only after prolonged illness and often arrived with severe complications. In fact,  he had described a well known phenomenon. From central India to throughout Southeast Asia stretches a vast epidemiological belt of  ‘forest malaria’. The ecotype is notorious for intense transmission due to the efficient mosquito species (A. fluviatilis, minimus, and dirus) particular to that habitat.

How did he respond? Simply, intensify active case detection in those areas. To his surprise, few of the blood smears collected turned up malaria parasites. Why?  Slides were collected during the day, exactly when most of the workers were away. Undeterred, the medical officer led his staff to the villages at dusk when workers were back from the bush. Travelling as late as 8 and 9pm by unlighted, broken roads, they again returned the following day to treat infected patients. He knew it to be unsustainable for the health workers themselves in the long run, ideally the villages could receive  a community health worker (ASHA) or later they could train community volunteers, but continued the practice for the rest of the high transmission season. It paid off. Inpatient admissions for severe malaria at the primary health center fell dramatically.

I was floored. Here he was young, new to the area, no special training in malaria, but already making a difference with the few resources at his disposal. It is the value of good management. Best of all it’s not a story about new technology or glossy strategy guides – just careful observation and dedication.

Interpreting malaria control program evaluations

About a year and a half ago I briefly discussed a WHO report (see comments here) claiming the success of scale-up of malaria control interventions. Now a group of CDC/ex-CDC scientists have published a superb commentary (Malaria Journal – open access) on the same evaluation and on using facility-based data more broadly (hat tip: Matt Price). The authors focus on technical pitfalls, which were aplenty, but these are often exacerbated by the incentives of the evaluators. In act of terrific political deftness, Rowe et al. avoid any discussion of possible conflicts of interest.

On an unrelated note, I found the first sentence of the abstract to be curiously phrased:

The global health community is interested in the health impact of the billions of dollars invested to fight malaria in Africa.

First,  the prima facie concern regarding the impact of malaria control is with endemic countries. They have skin in the game. It is unclear what “the global health community” really means – while it could be inclusive of endemic nations the connation of the phrase seems more aligned with a donor perspective. Thus, the rationale of this paper reads “accurate program evaluations are needed because donors want to assess their impact.” This is wrong. Quality evaluations are important first and foremost because they allow country programs to track and improve their progress in minimizing the suffering of their citizens.  Anything else is secondary and subjugate to this concern. The framing of the sentence reflects a subtle, and likely unintentional, appropriation of responsibility which may not impact practice but devalues local decision makers. Second, why “in Africa”? Neither malaria, large investments, nor the cautions highlighted in the commentary are specific to that continent.

Monitoring antimalarial drug resistance with molecular tools

Millions of people get treated for malaria every year so its important to use an antimalarial which works. Treatment failures result in a prolonged illness for the patient with an increased risk of severe malaria and death. In addition, they contribute to increased malaria transmission. Makes sense right? However, the parasite has an amazing ability to develop drug resistance and it’s hard to know what works ahead of time. Point-of-care diagnostics to identify drug resistance are not available and nor would they likely be practical given the low-cost and short course of malaria treatments. Rather, population level monitoring of drug resistance is used to inform the national drug policy which specifies a fixed treatment strategy.

Presently, in vivo efficacy trials are the gold standard for anti-malarial drug resistance monitoring. In vivo trials involve treating a cohort of malaria patients and subsequently checking their blood at regular intervals to detect parasites. Thus, in vivo studies are not direct measures of parasite resistance per se – as other factors such as host immunity and pharmacokinetics play a role, but the outcome (treatment failure) is directly relevant to programs. However, implementing in vivo studies presents several challenges in endemic settings including cost, well-trained staff, lengthy periods of patient follow-up (28 to 63 days), and the need to distinguish treatment failures as recrudescences or reinfections. Since most malaria control programs have limited resources, only a few in vivo trials can be conducted each year. An alternate method to extend the surveillance of antimalarial efficacy is to measure molecular markers of drug resistance.

We employed a molecular surveillance system in Cambodia (Emerging Infectious Diseases Journal – open access) to help detect ACT resistance. This new tool does not replace the standard in vivo trial but rather enables a more efficient use of existing resources. Using this system, a high level of artesunate-mefloquine failures (which were previously believed to be confined to the Thai border) was discovered in the center of the country. Here’s a quick summary of the benefits:

Monitoring changes in antimalarial efficacies is essential for control programs in an era of multi-drug resistance. However, such studies are resource intensive. In Cambodia, for example, the National Malaria Control Program can manage to conduct in vivo studies at only 2-3 sites per year because of limited funds and trained staff. Molecular markers can help target in vivo studies to where they are needed the most. Molecular surveillance is high-throughput and can be performed in a central laboratory on dried blood spots, which are easily collected in the field. Expanded molecular surveillance could also accurately help map anti-malarial resistance and enable sub-national treatment policies in countries with marked geographic variation in drug susceptibility.

What excites me about this work is the translation of molecular knowledge into a monitoring tool for informing policy and improving malaria control operations. I think the last two sentences of the paper say it all:

pfmdr1 assays are now routinely performed in Cambodia by National Malaria Control Program staff. National and regional molecular surveillance by malaria-endemic countries is a real possibility.

Monitoring insecticide resistance in malaria control

As large scale distributions of insecticide treated bed-nets continue, the selection pressure for insecticide resistance increases. Insecticide resistance, particularly to DDT, helped destroy the malaria control efforts of many countries in the 1960s-70s. Monitoring insecticide resistance is part and parcel of any indoor residual spraying (IRS) program, and is arguably even more important for bed-nets. Currently, bed-nets are made with only one type of insecticide – pyrethroids. Pyrethroids are the only widely used insecticides which are both effective and safe for human contact. Since pyrethroid resistant mosquitoes are known to emerge, large scale resistance is likely inevitable.

The only way to detect insecticide resistance and be able to respond with adequate public health measures is by maintaining strong surveillance programs. In Lancet Infectious Diseases, Kelly-Hope et al. draw upon lessons provided by past campaigns and propose sound recommendations for the future. The authors call for insecticide resistance monitoring systems and outline actual specifics for such plans. In addition to country teams, regional centers are important. By conducting advanced biochemical and molecular assays they can detect emerging resistance and provide crucial lead time to plan for policy change. Many groups pay lip service to the importance of such work but few programs possess a concrete strategy. The costs for a quality surveillance program are minimal but the costs due to widespread insecticide resistant mosquitoes would be catastrophic. We cannot afford to continue flying blind.

How many malaria deaths in Botswana this year?

The answer is two according to Ministry of Health officials as reported by AllAfrica.com. Now, no one (hopefully) believes only two deaths have occurred and yet the figure is presented as fact. This is either a case of imprecise language on the part of a public health official or bad reporting by the news outlet. Either way it points to a larger problem – malaria deaths (or cases) are hard to count and our capacity to do so is very poor. Two recorded deaths could mean two actual deaths (unlikely), or 20, or 200. We have no idea. Since we really don’t know, we should make that clear when we are relaying the information. Unfortunately, as illustrated above, that does not happen. Accurate measurements are essential in public health to deploy and evaluate interventions. Accurate communications are essential in public health to inform stakeholders and maintain trust.

Mapping malaria – a new analysis of spatial limits and transmission intensity

The KEMRI-Oxford group has published a new map of malaria transmission. This is a terrific accomplishment providing a global picture of falciparum malaria and answers some basic questions: where is malaria a risk and how intense is that risk? Using results from 4,278 surveys, the researchers found approximately 2.37 billion people live in areas at risk of malaria infection of which 1 billion live in areas of unstable or low transmission. Furthermore, areas of high parasite rates (>50%) are largely in sub-Saharan Africa while other malaria endemic areas are largely hypoendemic (<10%).

While mapping transmission is easier than estimating the number of worldwide cases, it is an undertaking fraught with significant obstacles. In the accompanying editor’s summary, Stephen Rogerson duly notes the accuracy of the findings are dependent on the author’s assumptions and the accuracy of the underlying data. Quite honestly, some of the methodology used in the paper is a bit beyond my skills, but most of the assumptions presented seemed very reasonable and the authors are terrific researchers. I think the spatial limits are more likely to be accurate than transmission intensities due to challenges with the input data. Cross sectional surveys are subject to multiple biases including death bias, seasonal bias, and missing asymptomatic infections. Furthermore, as the authors note, the data is not population representative which is a deficiency plaguing the surveillance of most tropical diseases and a crucial challenge to address for effective control.

Regardless of the precise numbers, the larger trends described can support a logical conclusion, one which many people have previously advocated. We can eliminate malaria outside of sub-Saharan Africa as these regions are of low and unstable transmission and yet comprise almost half of the world’s population at risk for malaria. It’s an exciting proposition and a growing possibility.


Archives