Upload
kevin-l
View
217
Download
1
Embed Size (px)
Citation preview
© 2005 Nature Publishing Group
NATURE|Vol 437|8 September 2005 BRIEF COMMUNICATIONS ARISING
E3
Estimates of the disease burden caused bymalaria are crucial for informing malaria con-trol programmes. Snow and colleagues claimthat their estimate of 515 million cases ofmalaria caused by Plasmodium falciparumglobally is up to 50% higher than that reportedby the World Health Organization (WHO),and 200% higher for areas outside Africa1.However, this comparison refers to the WHO’sestimates from 1990 and 1998, and not to the range of 300 million to 500 million that the WHO has used since 2000 (ref. 2). Bothgroups agree that the burden of malaria dis-ease outside Africa, especially in South Asia, isgreater than was estimated in the 1990s.
A new global map of populations living atrisk of malaria transmission, produced in 2004and funded in part by the WHO’s Roll BackMalaria department (WHO/RBM), forms thebasis of new estimates both by Snow and col-leagues1 and the WHO/RBM3. Both currentsets of estimates reflect the consensus that casesrecorded and reported in national health infor-mation systems capture far less than the fullburden of malaria in most parts of the world.
In October 2004, a group of independentexperts reviewed the WHO/RBM’s new esti-mation method and concluded that, even withthe best available data, it is preferable to presentany malaria incidence estimate as a range4. Animportant reason for this conclusion is that theresearch studies used as input are normally
conducted in areas where malaria transmis-sion is greatest, and during the season of peakmalaria transmission. Extrapolation fromthese studies may therefore result in a picturethat is not truly representative of the entireregion. In addition, the population-at-riskmap may not be totally accurate, given fluctu-ations in malaria transmission patterns inresponse to environmental change, develop-ment and vector control4.
The WHO’s global burden estimate for 2004of 350 million to 500 million cases, of which270 million to 400 million are due to infectionby P. falciparum3, is generally consistent withthe 300 million to 660 million range for P. fal-ciparum proposed by Snow and colleagues1.The somewhat smaller range estimated by theWHO is likely to be due to our inclusion of theimpact of preventive interventions (insecti-cide-treated mosquito nets and indoor resid-ual spraying), the coverage of which hasincreased since 2000. Further, for selectedareas with highly unstable malaria transmis-sion, the WHO’s estimate takes into accountcase numbers reported through routine healthinformation systems, which often reflect bothpassive and active case detection and, in thesesettings, may provide a more reliable and up-to-date picture than extrapolations made frompublished research in sometimes distant sites.
The WHO is working with countries toimprove their capacity for collecting the data
required for future national-level incidenceestimates. Data will be obtained from nationalsurveys and sentinel surveillance sites, as wellas from health information systems. Theseefforts should improve the precision of burdenestimates at all levels and allow us to assess, inthe second half of this decade, whether theongoing increase in coverage of effective prevention and treatment measures are bring-ing us closer to the global goal of reducing theburden of malaria by half by the year 2010. Bernard L. Nahlen*, Eline L. Korenromp*, John M. Miller*, Kenji Shibuya†World Health Organization, *Roll Back MalariaDepartment and †Evidence for Information andPolicy Department, 1211 Geneva, Switzerlande-mail: [email protected]
1. Snow, R. W., Guerra, C. A., Noor, A. M., Myint, H. Y. & Hay, S. I. Nature 434, 214–217 (2005).
2. WHO Expert Committee on Malaria: 20th Report (WorldHealth Organization, Geneva, 2000).http://www.rbm.who.int/docs/ecr20.pdf
3. Korenromp, E. L. for the RBM MERG Task Force on MalariaMorbidity: Malaria Incidence Estimates at Country Level for theYear 2004 — Proposed Estimates and Draft Report (WorldHealth Organization, Roll Back Malaria, Geneva, 2005).http://mosquito.who.int/docs/incidence_estimations2.pdf.
4. Minutes of the First MERG Task Force Meeting on MalariaMorbidity (World Health Organization, Roll Back Malaria,Geneva, 2004). http://rbm.who.int/partnership/wg/wg_monitoring/docs/MorbidityTaskForce_meetingOct04_2004.pdf.
doi:10.1038/nature04178Reply: Snow et al. reply to this communication(doi:10.1038/nature04180).
MALARIA RISK
Estimating clinical episodes of malariaArising from: R. W. Snow, C. A. Guerra, A. M. Noor, H. Y. Myint & S. I. Hay Nature 434, 214–217 (2005)
Accurate estimates of the global burden ofmalaria are important for planning, monitor-ing and advocacy. Snow et al.1 attempt toaddress the shortcomings of previous estimatesof the incidence of malaria caused by Plas-modium falciparum by combining current andhistorical data. However, we believe that thedesign of their model and its inputs have led toa significant overestimate of the malaria bur-den outside Africa — as in the example of theWorld Health Organization (WHO) westernPacific region (WPR), for which their modelpredicts 60 times the 2002 incidence reportedby national malaria-control programmes2.
The reliance by Snow et al. on three broadcategories of malaria risk — hypo-, meso-, andhyper/holoendemic malaria — would seem to
be the likely basis for this overestimate. The1968 map3 that Snow et al. used to obtain theseendemic zones was developed from estimatesof maximum malaria prevalence between thelate nineteenth century and the 1960s andincluded some P. vivax infections. Snow et al.laid this over a current (2002) WHO map ofmalaria distribution, modified according totravel advice and some national statistics.However, the scale of this WHO map indicatesonly broad areas where transmission may befound. The reduction of prevalence in highlyendemic areas by one step from its historicalmaximum to account for control methods anddeforestation, and the further reduction of onestep in urban areas of more than 1 millioninhabitants1, seems arbitrary and assumes that
MALARIA RISK
Estimation of the malaria burdenArising from: R. W. Snow, C. A. Guerra, A. M. Noor, H. Y. Myint & S. I. Hay Nature 434,214–217 (2005)
transmission in smaller cities and towns isequal to that in forested areas. Hypoendemicareas outside major cities remain unchangedfrom the pre-1968 map, unless national figureswere known to show absence of transmission.Despite the stated methods, holo/hyper-endemic zones still also appear in calculations1.
The predominant vectors of malaria insoutheast Asian countries of the WPR areassociated with forests4. In the past 50 years,deforestation has removed much of this habi-tat5. Furthermore, although the proportion ofthe population living in urban areas has greatlyincreased, urban and peri-urban malaria isuncommon in Asian countries, and virtuallyabsent in the Philippines, Malaysia and theMekong region. Many people currently livingin areas that were highly endemic for malariaon the Lysenko and Semashko map3 are there-fore at low or no risk of malaria. Malaria inci-dence in remaining transmission areas isgeographically highly heterogeneous6–9.
We conclude that the prediction by Snow et al. of 15.03 million cases in the WPR, basedpredominantly on an estimated population of
© 2005 Nature Publishing Group
BRIEF COMMUNICATIONS ARISING NATURE|Vol 437|8 September 2005
E4
63.4 million at risk of 171 P. falciparum malariacases per 1,000 per year (“mesoendemic”)1, isnot consistent with the epidemiology ofmalaria in the region. Apart from Vanuatu, theSolomon Islands and Papua New Guinea(where vectors differ in terms of habitatrequirements), populations at such high risk are expected to be confined to relativelyisolated, forested areas with poor health-service access.
Snow et al. conclude that the burden ofmalaria outside Africa is far higher than previ-ous estimates, a prediction that should bebroadly verifiable. It would be helpful to haveverification of predictions if they are inconsis-tent with recorded data, such as the 60-times-higher incidence in the WPR, the 0.55 millioncases per year among the 6.5 million people inTajikistan, where transmission is seasonal andP. vivax predominates, or the extent to whichofficial figures in Pakistan underestimate inci-dence by 1,000 times1. We agree that healthinformation systems miss many malaria cases,particularly among remote populations and
where cases are managed by the private healthsector. However, the predictions of Snow et al.are undermined by the magnitude of their discrepancy with recent country data based on passive and active case detection providedto the WHO by countries in the WPR.
If the epidemiological characteristics of theWPR are considered, and if similar adjustmentsare required in other non-African regions, thenthe global estimates of incidence by Snow et al.would be close to current WHO estimates10.Rather than relying on historical estimates,improved national surveillance is necessary toprovide contemporary data on which to basemore accurate and verifiable estimates.David R. Bell, Pernille Jorgensen,Eva Maria Christophel, Kevin L. PalmerMalaria, Other Vector-borne and ParasiticDiseases Unit, World Health OrganizationRegional Office for the Western Pacific,PO Box 2932, Manila, Philippinese-mail: [email protected]
1. Snow, R. W., Guerra, C. A., Noor, A. M., Myint, H. Y. &
are uncertain of the subnational estimates ofintervention use. Furthermore, the verynature of malaria illness dictates that method-ologies institutionally wedded to routinenational reporting systems will consistentlyunderestimate malaria burden.
Bell and colleagues4 highlight the impor-tance of mapping the extent and intensity ofP. falciparum transmission at national scales.We agree. Higher-resolution information onforest, land use, human settlement and inter-vention, and how these relate to dominant vec-tor-species parasite transmission, will providea more robust framework — with the provisothat this was developed using scientificallydefensible models and data. This forms thebasis of ongoing work by our group as part of a global Malaria Atlas Project. Unlike theWHO/RBM9, we were careful to present onlyregional comparisons — given the uncertain-ties in the resolution of the only currentlyavailable, global malaria distribution map. Forthe benefit of the WHO Regional Office forthe Western Pacific, the Philippines andMalaysia contributed less than 10% of the clin-ical burden within the region. As the authorsalso indicate4, Papua New Guinea, theSolomon Islands and Vanuatu have malariatransmission characteristics similar to those inlarge parts of Africa12–14. As such, we maintainthat our delineation of risk in the WPR (west-ern Pacific region) has been conservative.
An epidemiological framework is requiredto define the spatial extent of malaria risk, howthese risks relate to disease burdens, and howthese might change by 2010 and 2015 when theWHO/RBM and the United Nations proposeto have halved malaria mortality or halted theincreasing incidence of malaria, respectively.Defining these targets will demand a credible,science-driven approach to risk mapping. The
MALARIA RISK
Snow et al. replyReply to: B. L. Nahlen et al. (doi:10.1038/nature04178) and D. R. Bell et al.(doi:10.1038/nature04179)
We have provided an evidence-based frame-work for estimating the public-health burdenof P. falciparum malaria1 by using a uniformseries of structured, transparent and repro-ducible data and methods to define the carto-graphy of this parasite’s clinical impact at aglobal scale. This contribution to global disease modelling is questioned by Nahlen et al.2 for United Nations agencies responsiblefor measuring development targets (see alsoref. 3), and our estimation of the malaria bur-den outside Africa is challenged by Bell et al.4.
Nahlen et al.2 point out that the WorldHealth Organization (WHO) estimates theglobal malaria morbidity burden as beingbetween 300 million and 500 million clinicalmalaria cases a year5. In the single sentence onthis subject5, we find this estimate representsthe combined Plasmodium falciparum andP. vivax clinical burden and is not sourced. Inthe following sentence concerning the propor-tion of global mortality in Africa, a 1993WHO publication is cited, which does notprovide any further citation for how these fig-ures were derived6. As with so many quotes ofglobal disease burdens, the empirical basis forthis estimate is unclear. The Global Burden ofDiseases (GBD) programme of the WHO during the late 1990s sought to replace suchguesswork with evidence so that internationalpublic-health policy could be better informed7.We have therefore chosen to use these 1999estimates8 as the WHO’s only global bench-
mark of clinical incidence of P. falciparum forwhich the method was transparent.
In 2004 we provided scientific advice to theWHO’s Roll Back Malaria (WHO/RBM)department so that they might begin toimprove the estimation of malaria risk9. TheWHO/RBM chose to use malaria transmis-sion classifications in their original 1968form10 and continued to use our modelled,actively detected estimates of malaria morbid-ity for Africa11; however, they adopted a different approach for outside Africa by trian-gulating disease reporting and interventioncoverage data provided by national govern-ments4. By their calculation, there were311 million clinical cases of P. falciparummalaria worldwide in 2002, only about 12%higher than the 273 million estimated by theGBD program8, but some 40% lower than ourestimate of 515 million cases, which wasderived using a more conservative approach toinfection risks and active case-detection datathat were independent of government statis-tics11. We therefore cannot agree with Nahlenand colleagues that these two estimates are“consistent”2.
The WHO/RBM approach to down-regu-lating clinical risks according to assumednational reported intervention coverage figures will overcorrect the attributable disease burden when the input incidence estimates are measured under prevailingnational-prevention coverage and where we
Hay, S. I. Nature 434, 214–217 (2005).2. World Health Organization Malaria, other vectorborne
and parasitic diseases: Regional trend in cases and deaths,2003. http://www.wpro.who.int/sites/mvp/data/ (2005).
3. Lysenko, A. J. & Semashko, I. N. in Medicinskaja Geografija:Itogi Nauki 1966 (ed. Lebedew, A. W.) 25–145 (VINITI,Moscow, 1968).
4. Gillies, M. T. in Malaria: Principles and Practice of Malariology(eds Wernsdorfer, W. H. & McGregor, I.) 453–485(Churchill Livingstone, Edinburgh, 1988).
5. United Nations Environment Programme Asia-PacificEnvironment Outlook 2: Regional Resource Centre for Asia andthe Pacific (Pathumthani, Thailand, 2001).www.rrcap.unep.org/reports/apo2.cfm
6. Carter, R., Mendis, K. N. & Roberts, D. Bull. World HealthOrgan. 78, 1401–1411 (2000).
7. Torres, E. P., Salazar, N. P., Belizario, V. Y. & Saul, A.Acta Trop. 63, 209–220 (1997).
8. Anthony, T. G. et al. J. Infect. Dis. 191, 1558–1564 (2005).
9. Socheat, D. et al. Southeast Asian J. Trop. Med. Public Health34 suppl 4, 1–102 (2003).
10. WHO Expert Committee on Malaria: 20th Report(World Health Organization, Geneva, 2000).http://www.rbm.who.int/docs/ecr20.pdf
doi:10.1038/nature04179Reply: Snow et al. reply to this communication(doi:10.1038/nature04180).