Archive for the 'Research' Category

More quick malaria links

  • The mosquito menace in Chennai (the city’s response might be scarier)
  • J Kevin Baird picks apart a historical fallacy in what might be the most important review of the year
  • Malaria, swine flu, and understanding the political economy of control programmes in endemic countries
  • Is an elimination agenda increasing inequity (previously discussed here) and why so much funding for Equatorial Guinea (here)? Some evidence on the need for donor re-allocation from GFATM and WHO
Advertisements

Frank remarks on disease burden estimates

Disease burden estimates are contentious as I’ve previously discussed on the blog (here and here) and in print (here, here, and here). So I won’t say much but share some surprising new comments. They are surprising not for their content (which we’ve already said) but because they are from the Global Burden of Disease group, which includes the authors of the controversial malaria estimates, and they are published in the Lancet, which published and actively promoted these estimates.

We now joke to each other how we used to get away with murder in the past doing burden studies. We tended to make lots of not so replicable ad-hoc decisions and few people knew what we were doing anyway.

We have also had some experiences of disease experts taking on roles of advocates with aim to ‘boost the numbers’ rather than being ‘impassionate’ scientific advisers.

Countries should develop sufficient capability to independently assess the merits of different methodologies and interpretation of findings: simply because WHO has issued estimates of disease burden does not make them correct.

Let’s see if these get noticed. I should add that I admire the openness of the article and the intent to change business-as-usual expressed there within (thanks Matt for the link).

Malaria hotspots and targeting

Heterogeneity exists in the risk of malaria – both between different areas and between individuals in the same area. The old maxim states “malaria is a local and focal disease”. Targeting malaria control interventions (nets, spraying, etc) is thus, a time-honored activity. Countries which undertook malaria eradication efforts have stratified populations at-risk for intervention delivery since  in the 1950s-60s.

In recent years a spate of papers (here, here, and a policy piece here) have emerged on the idea of hotspots and ‘hitting them’ for high-impact control. What do these papers add? These studies quantified the heterogeneity in risk at the individual and household levels and their change with time. Another innovation in measuring the concentration of transmission was the use of serological in addition to parasitological data – although the two studies conflicted on which serological measurement was better. Finally, the papers recommend individual or household-level targeting of certain control interventions (vaccines, mass drug administration, specific vector control activities) at certain times based on annual surveys. What do these studies miss? First, as with many studies, particularly those conducted or focused in sub-Saharan Africa (all of the above), they suffer from a lack of historical perspective. Enamored with the idea of novelty, the authors fail to acknowledge decades of experience in ‘hitting hotspots’ using different data and different units of selection elsewhere in the world. Second, these studies do not assess, deductively or empirically, the direct and indirect costs of targeting. What are the training, delivery, coverage gaps and leakage implications of using serological data compared to parasitological data or of using household level stratification compared to village or area stratification? Without the comparative and programme perspectives such key questions remain unasked.

Using and improving targeting is an important area of work in malaria control. But the detection of these hotspots therefore, does not necessarily lead to their use in control work. Long-standing, sophisticated malaria control programmes face considerable difficulty in targeting activities using much coarser criteria. Conclusions and recommendations should follow from study results. This is a basic principle of scientific writing, and possibly, the most violated due to well-meaning intentions to ‘do something’ along with the less noble imperative to overstate one’s work. Academic research and publication will always favor the new over the credible, future possibilities over present realities. This is not necessarily a bad thing. We need to aim high and we need to look ahead. Programme managers and other practical consumers of the literature should be prudent in the promotion of such work. What worries me is the academic-donor nexus which prevents this caution and brings in varying degrees of distraction.

Paid writing of malaria trial results

The New England Journal of Medicine recently reported phase III trial results for artesunate + pyronaridine and tucked away in the acknowledgments I noticed this:

We thank Naomi Richardson (Magenta Communications) for developing the first draft of the manuscript and for editorial assistance.

I wondered, why? On one hand, for a well-defined type of study such as a phase III randomized controlled trial the writing can be fairly formulaic and easy to outsource. The data analysis was also conducted by another corporation. As an industry-sponsored study, no doubt the funders wanted a quick, and efficient process once the trial was complete. The additional cost of these services (anyone know how much?) which I suspect are expensive may not be much compared to the entire study budget. Still, it is somehow disappointing to me to see a paper which was not analyzed or written by the scientists who conducted the work. Is the process not important and what are we losing to the CRO culture which dominates these days? By all accounts (number of previous publications), Dr Ronnatrai Rueangweerayut who is the first-author, could have benefited from the writing experience. How else will we build scientific capacity?

A flood of artemisinin resistance

Articles that is. The spread of resistance to artemisinin drugs, the main-stay of modern Plasmodium falciparum (and even P. vivax in some places) malaria therapy, would endanger control programs globally (previously discussed here, here, here, and here). Last week saw a series of high-profile publications which received an impressive amount of coverage in the general media, here’s the list:

  • Evidence for a genetic basis for much of the observed resistance – are molecular markers in sight?
  • Another combination therapy trial with prolonged clearance in both treatment arms among patients from Western Cambodia
  • Longer clearance half-lives among patients with a high parasite load treated with just artemisinin on the Thai-Burma border, it’s already spread
  • An accessible narrative providing a glimpse of the operations of the Thai-Cambodia containment program (thanks Matt)

New results for intermittent preventative therapy in children

Truly beautiful studies – well designed, well thought, even examined cost and service delivery – were recently conducted for regular, presumptive antimalarial treatment (using SP and amodiaquine) of children in Mali and Burkina Faso in settings where treated bed-nets are already in use (PLoS Medicine – open access!). The intervention was effective at reducing clinical burden – from malaria incidence, the primary target, to secondary endpoints such as anemia, all-cause mortality, and stunting.

There is one important caveat here – IPTc is only “effective” where the transmission is quite high. In the communities in Burkina Faso and Mali where the study was conducted transmission was  very intense (3-13 infective mosquito bites per person per month). At medium and low levels of transmission (last two rows of the table) the strategy becomes rather untenable, expending a lot of drug (which wastes money and risks side effects and resistance), for preventing a single case. Caveat to my caveat – the interpretation of rates differences in number needed to treat calculations is not always straightforward though I believe valid in this case.

Table: Number need to treat (NNT) and post-intervention incidence rate across varying baseline transmission and IPTc efficacy

IPTc Efficacy

85%

75%

65%

Baseline*

Rate

NNT

Rate

NNT

Rate

NNT

1000.0

150.0

0.1

250.0

0.1

350.0

0.2

100.0

15.0

1.2

25.0

1.3

35.0

1.5

10.0

1.5

11.8

2.5

13.3

3.5

15.4

1.0

0.2

117.6

0.3

133.3

0.4

153.8

0.1

0.0

1176.5

0.0

1333.3

0.0

1538.5

*Baseline transmission and rates with the intervention are expressed per 100 persons per season

Interestingly, similar to the famous Garki project the reductions in incidence appear to be much greater than reductions in prevalence – likely due to the seasonal nature of the intervention against a high vectorial capacity and thus risk of exposure.  Since the focus here is burden reduction, and not transmission reduction as in Garki, it doesn’t matter though.

Malaria deaths in India

Malaria mortality in India caused much controversy last fall. The study estimated almost 10 times the number malaria deaths in India during 2001-2003 compared to the estimates of the Government of India and WHO. The key strength of the work by Dhingra et al. was the use of a nationally representative sample of deaths during that period. However, the method for assigning the cause of death, verbal autopsy, is known to be problematic particularly for non-specific illness such as malaria and has come under much criticism. Some good early critiques appeared in Nature and Tropika and now The Lancet has published our reply as well as four others. The authors have also responded and the Government of India is setting up a panel to examine the findings and reconcile the available data.

To enable a proper assessment of research involving burden estimates, The Lancet and other journals should require authors to: 1) perform a validation exercise of the study instrument 2) conduct sensitivity analysis using validation results and other parameters of uncertainty and 3) make the underlying sample, or at least fully disaggregated tables, available for external verification. Otherwise we risk appearing to promote sensationalism over science. It is surprising such standards are not already in place.


Archives