• View in gallery

    Definitions of populations at risk differ between countries in the Western Pacific Region and sometimes change radically over time.21

  • View in gallery

    A national estimate of admissions for cerebral malaria in Papua New Guinea was made by imputing missing values for health facilities that failed to report. The increase in cases estimated nationally is similar to the trend observed in the set of facilities that submitted practically complete reports.23

  • View in gallery

    Estimates of morbidity in children obtained through immunization clinics may be sufficiently similar to population estimates for program management purposes. Each dot shows the value of anemia or fever prevalence for the 2003 Burkina Faso DHS in the full sample vs. the value obtained by focusing on children immunized against measles or DPT3.29

  • 1

    Roll Back Malaria/World Health Organization, 2000. Framework for Monitoring Progress and Evaluating Outcomes and Impact. Geneva: Roll Back Malaria/World Health Organization.

  • 2

    Korenromp EL, Williams BG, Gouws E, Dye C, Snow RW, 2003. Measurement of trends in childhood malaria mortality in Africa: an assessment of progress toward targets based on verbal autopsy. Lancet Infect Dis 3 :349–358.

    • Search Google Scholar
    • Export Citation
  • 3

    Snow RW, Guerra CA, Noor AM, Myint HY, Hay SI, 2005. The global distribution of clinical episodes on Plasmodium falciparum malaria. Nature 434 :214–217.

    • Search Google Scholar
    • Export Citation
  • 4

    Rowe AK, Rowe SY, Snow RW, Korenromp EL, Schellenberg JR, Stein C, Nahlen BL, Bryce J, Black RE, Steketee RW, 2006. The burden of malaria mortality among African children in the year 2000. Int J Epidemiol 35 :691–704.

    • Search Google Scholar
    • Export Citation
  • 5

    Bell D, Jorgensen P, Christophel E, Palmer KL, 2005. Estimation of the malaria burden. Nature 437 :E3–E4.

  • 6

    Omumbo JA, Hay SI, Guerra CA, Snow RW, 2004. The relationship between the Plasmodium falciparum parasite ratio in childhood and climate estimates of malaria transmission in Kenya. Malar J 3 :17.

    • Search Google Scholar
    • Export Citation
  • 7

    Carter R, Mendis KN, Roberts D, 2000. Spatial targeting of interventions against malaria. Bull World Health Organ 78 :1401–1411.

  • 8

    Lubanga RG, Norman S, Ewbank D, Karamagi C, 1997. Maternal diagnosis and treatment of children’s fever in an endemic malaria zone of Uganda: implications for the malaria control programme. Acta Trop 68 :53–64.

    • Search Google Scholar
    • Export Citation
  • 9

    Dunyo SK, Afari EA, Koram KA, Ahorlu CK, Abubakar I, Nkrumah FK, 2000. Health centre versus home presumptive diagnosis of malaria in southern Ghana: implications for home-based care policy. Trans R Soc Trop Med Hyg 94 :285–288.

    • Search Google Scholar
    • Export Citation
  • 10

    Einterz EM, Bates ME, 1997. Fever in Africa: do patients know when they are hot? Lancet 350 :781.

  • 11

    Korenromp EL, Armstrong-Schellenberg JRM, Williams BG, Nahlen BL, Snow RW, 2004. Impact of malaria control on childhood anemia in Africa—a quantitative review. Trop Med Int Health 9 :1050–1065.

    • Search Google Scholar
    • Export Citation
  • 12

    Zambia Ministry of Health, 2006. Zambia Malaria Indicator Survey 2006. Lusaka: Ministry of Health Zambia.

  • 13

    Monasch R, Reinisch A, Steketee RW, Korenromp EL, Alnwick D, Bergevin Y, 2004. Child coverage with mosquito nets and malaria treatment from population-based surveys in African countries: a baseline for monitoring progress in Roll Back Malaria. Am J Trop Med Hyg 71 :232–238.

    • Search Google Scholar
    • Export Citation
  • 14

    Monograph Series INDEPTH, 2002. Demographic Surveillance Systems for Assessing Populations and Their Health in Developing Countries, Volume 1: Population, Health and Survival in INDEPTH Sites. Ottawa: IDRC/CRDI.

  • 15

    De Savigny D, Kasale H, Mbuya C, Reid G, 2004. In Focus: Fixing Health Systems. Ottawa: IDRC/CRDI.

  • 16

    Flew SJ, 2002. District health care in Tari until 1991. P N G Med J 45 :106–112.

  • 17

    Whiting DR, Setel PW, Chandramohan D, Wolfson L, Hemed Y, Lopez AD, 2006. Estimating cause-specific mortality from community- and facility-based data sources in the United Republic of Tanzania: options and implications for mortality burden estimates. Bull World Health Organ 84 :921–1000.

    • Search Google Scholar
    • Export Citation
  • 18

    Agyepong IA, Kangeya-Kayonda J, 2004. Providing practical estimates of malaria burden for health planners in resource-poor countries. Am J Trop Med Hyg 71 :162–167.

    • Search Google Scholar
    • Export Citation
  • 19

    Erhart A, Thang ND, Xa NX, Thieu NQ, Hung LX, Hung NQ, Nam NV, Toi LV, Tung NM, Bien TH, Tuy TQ, Cong LD, Thuan LK, Coosemans M, D’Alessandro U, 2007. Accuracy of the health information system on malaria surveillance in Vietnam. Trans R Soc Trop Med Hyg 101 :216–225.

    • Search Google Scholar
    • Export Citation
  • 20

    Chilundo B, Sundby J, Aanestad M, 2004. Analyzing the quality of routine malaria data in Mozambique. Malar J 3 :3.

  • 21

    Graves PM, 2006. Assessment of the Implementation of the WHO Western Pacific Region Malaria ‘Kunming’ Indicator Framework, 1999–2005. Consultant Report. Manila: WHO WPRO.

  • 22

    Mapping Malaria in Asia and the Pacific, 2006. Report of a WHO Technical Consultation, Bangkok, Thailand, September 14–16, 2006.

  • 23

    Ministry of Health, 1998. National Health Plan 1996–2000: Mid-Term Review. Waigani: Ministry of Health, Papua New Guinea.

  • 24

    Cibulskis RE, 2002. Trends in the Distribution of Health and Health Services in Papua New Guinea 1991–2000. Papua New Guinea Health Sector Review. Manila: Asian Development Bank.

  • 25

    Gething PW, Noor AM, Gikandi PW, Ogara EAA, Hay SI, Nixon MS, Snow RW, Atkinson PM, 2006. Improving imperfect data from health management information systems in Africa using space-time geostatistics. PLoS Med 3 :e271.

    • Search Google Scholar
    • Export Citation
  • 26

    Drakeley CJ, Carneiro I, Reyburn H, Malima R, Lusingu JPA, Cox J, Theander TG, Nkya WMMM, Lemnge MM, Riley EM, 2005. Altitude dependent and independent variations in Plasmodium falciparum prevalence in Northeastern Tanzania. J Infect Dis 191 :1589–1598.

    • Search Google Scholar
    • Export Citation
  • 27

    Delacollette C, Barutwanayo M, Mpitabakana P, 1990. Epidemiologie du paludisme au Burundi: observations preliminaires. Med Afr Noire 37 :718–721.

    • Search Google Scholar
    • Export Citation
  • 28

    Some ES, Koech DK, Ochogo JO, Ocholla F, Mumbi F, 1997. An evaluation of surveillance of malaria at primary health care level in Kenya. East Afr Med J 74 :573–575.

    • Search Google Scholar
    • Export Citation
  • 29

    Institut National de la Statistique et de la Démographie, Ministère de l’Économie et du Développement, Ouagadougou, Burkina Faso, 2004. Burkina Faso: Enquête Démographique et de Santé. Calverton, MD: ORC Macro.

Past two years Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 87 87 16
PDF Downloads 13 13 7
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 

 

 

Estimating Trends in the Burden of Malaria at Country Level

View More View Less
  • 1 World Health Organization, Global Malaria Program, Geneva, Switzerland; World Health Organization, Regional Office for the Western Pacific, Manila, Philippines; World Health Organization, Honiara, Solomon Islands; World Health Organization, Mekong Malaria Office, Bangkok, Thailand; World Health Organization, Regional Office for Africa, Harare, Zimbabwe

National disease burdens are often not estimated at all or are estimated using inaccurate methods, partly because the data sources for assessing disease burden—nationally representative household surveys, demographic surveillance sites, and routine health information systems—each have their limitations. An important step forward would be a more consistent quantification of the population at risk of malaria. This is most likely to be achieved by delimiting the geographical distribution of malaria transmission using routinely collected data on confirmed cases of disease. However, before routinely collected data can be used to assess trends in the incidence of clinical cases and deaths, the incompleteness of reporting and variation in the utilization of the health system must be taken into account. In the future, sentinel surveillance from public and private health facilities, selected according to risk stratification, combined with occasional household surveys and other population-based methods of surveillance, may provide better assessments of malaria trends.

WHY ARE COUNTRY AND SUB-NATIONAL LEVEL ESTIMATES OF MALARIA BURDEN NEEDED?

Information on the population at risk of malaria and incidence of disease is critical for malaria program design and implementation. Estimates of the population at risk are required to quantify the need for preventive interventions such as insecticide-treated bednets and indoor residual spraying and to guide disease surveillance strategies. Information on the incidence of disease, and its severity, is needed to determine the demand for treatment of uncomplicated and severe malaria. Once available, the data can be compared with existing levels of service provision to identify underserved populations and, in situations of resource constraint, to target interventions to high-priority areas. Information on the incidence of disease in relation to historical levels can also alert programs of epidemics so that intensified control measures can be applied. Finally, data on changes in disease incidence and mortality are needed to judge the success of program implementation and to determine whether programs are performing as expected or whether adjustments in the scale of or blend of interventions are required.

COUNTRIES EXPERIENCE DIFFICULTIES IN QUANTIFYING BURDEN: WHY?

The World Health Organization (WHO), along with other development partners, recommends that the number of malaria cases and malaria-attributed deaths should be used as core indicators by all malaria-endemic countries.1 There has been some effort at estimating the number of malaria cases and deaths on an international scale.24 However, the applicability of international estimates, using wide area geographical models, to country level assessments is uncertain.5,6 Malaria transmission intensity and disease incidence can vary within a country,7 with different areas having malaria seasons of different peak amplitudes and different lengths. Malaria transmission may also vary from year to year. These factors complicate the assessment of disease burden and suggest that information should be obtained from all endemic areas within a country and continuously over time, but the data sources normally available to countries rarely allow this.

Nationally representative surveys.

Demographic and health surveys (DHSs) and multiple indicator cluster surveys (MICSs) produce reliable estimates of all-cause mortality rates in children < 5 years old, which are useful in countries with high malaria burdens, although the estimates are usually centered on a period 3–5 years before the survey. The DHSs and MICSs also include a 2-week period-prevalence of fever in children < 5 years old, an indicator that is difficult to interpret as a measure of malaria morbidity.810 An increasing proportion of surveys now assesses the prevalence of low birth weight and hemoglobin levels,11 and countries are now undertaking malaria indicator surveys (MISs), which can include estimates of parasite prevalence.12 Such household surveys are seldom representative at a district level, because the two-stage cluster sampling technique does not always include all districts, and the number of clusters within a district is small. This, along with the long interval between surveys, reduces their direct relevance for monitoring changes in disease burden at country level; however, the information provided on mortality, morbidity, parasitemia, and care-seeking behavior can be valuable in interpreting data from other sources.

Demographic surveillance sites.

Some malaria-endemic countries have demographic surveillance sites (DSSs) that enable trends in malaria mortality and morbidity to be measured at a population level.13 Because of the intensive nature of the data collection activities, DSSs cannot generally be applied over wide geographic areas or in large populations. If a DSS site exists within a district, it may provide pertinent information for a district manager. At a national level, the DSS cannot necessarily be used to paint a broader picture because they represent only a small fraction of a country’s population, and the sites may not be typical of all areas in the country. A key advantage of DSS is that they produce cause-specific death rates, as assessed through verbal autopsy, and some sites have followed trends over 10 years or more. The lack of sensitivity and specificity of verbal autopsies has been highlighted,14 but even with some imprecision, they can be a useful guide for resource allocation decisions.15

Health management information systems.

Every country has a system for routinely recording the morbidity and mortality seen at health facilities, sometimes with parasitologic confirmation. Such information is collected continuously, from every district in a country, and for most districts, is the only readily available source of information on malaria. When such systems are working well, they show consistent seasonal variation in numbers of cases, coinciding with the pattern of malaria transmission that suggests, with other factors being constant, that they might be able to detect changes in morbidity over time. Moreover, when data from routine systems are compared with other sources of information (such as verbal autopsies in sentinel surveillance sites), it seems that health facilities capture the major causes of mortality in rural areas, although some individual causes of death may be over-or under-represented.16,17 A major limitation of such systems is that they represent a biased and incomplete sample of the morbidity and mortality experienced by communities. They often do not consider private clinics or other non-government facilities or morbidity treated at home. Hence, they can underestimate the total burden of disease by a considerable fraction.18,19 They are also influenced by changes in health service utilization, which may be general (such as introduction of user fees or construction of roads) or specific to malaria (such as availability of new therapies or introduction of laboratory confirmation). Moreover, such systems seldom function optimally; there is often inconsistent application of malaria case and risk definitions and irregularity in reporting from health facilities and districts to central levels. Trends in morbidity and mortality are therefore particularly prone to variations in reporting rates.20

Other sources.

Systems of vital registration can sometimes provide relevant information, and many countries have locally conducted surveys that consider entomologic inoculation rates, parasite prevalence, or self-reported fever. Often such studies are primarily to answer locally specific questions and conducted in areas where malaria is perceived to be a problem, rather than to build up a national picture of disease burden or assess trends over time.

HOW CAN WE USE AND IMPROVE AVAILABLE INFORMATION IN COUNTRIES?

Estimating population at risk of malaria.

Countries use a variety of definitions of risk either by looking at 1) the geographic and climatic suitability for transmission or 2) the actual transmission occurring. The resulting estimates are not always comparable between countries or years (Figure 1).

An important step forward would be a more consistent and informative quantification of population at risk across countries, preferably using an up-to-date assessment of transmission; definitions of risk based on geographical or climatic suitability are a useful reference point but need to be supplemented by contemporary data as treatment and control become more widely available, and urbanization and habitat destruction increases. Because of the need for data from a wide geographical area, such assessment is most likely achieved using health facility data on the presence or absence of confirmed cases in combination with locally conducted parasitologic surveys and expert opinion. (There are some notable exceptions such as forested areas in Asia where health services are absent, but transmission occurs.) One suggestion is that endemic areas would be those that had confirmed cases recorded in each of the past 3 years, whereas epidemic-prone areas would have confirmed cases recorded some time in the last 10 years.22 If updated annually, such risk maps could, in themselves, provide information on the changing burden of malaria.

Estimating case incidence.

Estimates of the population at risk of malaria also serve as the denominators for estimating malaria case incidence. (Care should be taken to ensure that population data are derived from sources officially recognized locally, such as projections from the national census, if estimates of incidence are to have local credibility.) Estimates of disease incidence have been inferred from parasite prevalence data, but the relationship is, at best, approximate. Incidence is best measured directly to genuinely determine whether the scaling-up of malaria interventions results in a reduction in morbidity and mortality. Direct measurement becomes particularly important as parasite prevalence declines and the proportion of infections that result in disease increases. Some countries might be able to estimate incidence by extrapolating the results of DSSs to provide national estimates of disease burden using the population at risk stratification. However, because the number of DSSs is usually small, or non-existent, it will usually be necessary to examine data obtained from routine health information systems. In doing so, at least two problems must be tackled: deficiencies in the reporting and collation of routine data and the under-representation that cases in health facilities represent.

Overcoming incompleteness of reporting.

One strategy is to concentrate on good reporting centers; if disease trends are estimated using a sample of facilities where reporting rates are high (e.g., > 90%), there is less chance that changes in disease incidence are distorted by fluctuation in reporting rates. Another strategy is to replace any missing value with an expected value for the health facility adjusted for the time of year and the year in question. In practice, it is best to use both methods because they complement each other. By concentrating on facilities with high reporting rates, one can be confident about a trend, but the facilities represent an incomplete selection from all facilities in the country. Conversely, the imputation of missing values will summarize trends for all facilities in the country, but the procedure uses incomplete temporal data. If the trends exhibited by both methods are consistent, one can have some confidence that the conclusions drawn are correct. Such a method was used in the mid-term review of the National Health Plan in Papua New Guinea in 1998 (Figure 2) and provided confirmation that the upward trend in cases of cerebral malaria was not caused simply by changes in reporting rates or selection of facilities reporting.23,24

More sophisticated methods for interpolating missing values that take into account the proximity of health facilities have also been developed that are much more difficult to apply but may prove valuable in situations with low reporting rates.25

Overcoming incomplete coverage.

Although trends might be inferred from consistently reported health facility data, it is more difficult to provide estimates of disease incidence that are comparable between sub-national administrative areas or can be aggregated to form a national estimate. Only a small proportion of fever cases may seek treatment in formal health facilities, and areas with better coverage may report unduly high levels of malaria incidence compared with those with less extensive health infrastructure. Some adjustment to incidence rates may be provided by considering health service utilization, as revealed by nationally representative household surveys, and by taking into account differential patterns of utilization by educational or socioeconomic status or sex. Such a procedure would benefit from greater consistency between the data items requested from household surveys and routine information systems. In particular, it would be useful for DHSs and MICSs to ask questions about total outpatient and inpatient utilization and reported diagnosis, as well as for selected conditions such as fever, diarrhea, and respiratory infections.

In many situations, reported malaria cases, if not confirmed parasitologically, should also be adjusted downward based on positivity rates from those health facilities that undertake laboratory confirmation. The adjustment required is likely to vary by levels of endemicity,26 and in areas of uncertainty, there may also be value in tracking confirmed and unconfirmed cases separately. Assessment of trends also requires validation that procedures for diagnosis and recording have not changed, such as laboratory confirmation or revision of the HMISs, and should also examine trends in other conditions to confirm that trends are not simply caused by changes in availability, accessibility, affordability, or quality of services.

HOW CAN WE GATHER BETTER INFORMATION IN THE FUTURE?

Where information systems are less developed, there may be advantages of focusing on strategies that can yield better results with smaller investments.

Sentinel sites.

Trend estimation could be limited to selected health facilities, including some in the private sector, that are assisted in obtaining good quality information, with burden estimation based on parasitologically confirmed cases for a defined catchment population that has good access to the facility. Other indicators, including anemia, low birth weight, severe malaria, convulsions, and deaths, could also be tracked, and these routinely collected data supplemented by occasional household surveys. Sentinel sites can be chosen according to the results of malaria risk stratification to enable easier extrapolation of results nationally. A disadvantage of sentinel sites is that they can become unrepresentative if the information they generate leads to more intensive local interventions. One way to overcome this is to “rotate” sentinel sites (periodically replace a number of sites with other locations).

Quasi-population estimates of morbidity.

A recent development in monitoring intervention coverage has been to consider using children attending immunization clinics as a sample in which to ask questions from caregivers on whether their child slept under a bednet the night previously (or how children with fever were treated; M. Otten, personal communication). It has been proposed to add questions to the routine EPI report forms when children attend for the third dose of DPT (typically 10 weeks to 6 months old) or measles (9 months and over). Such a way of measuring intervention coverage is potentially biased, but coverage rates of malaria interventions have proven to be similar between vaccinated and unvaccinated children (unpublished data). It is possible that child morbidity could be monitored in a similar way, because analysis of DHS data suggests that estimates of prevalence of anemia and fever, measured in immunized children, are similar to those obtained for the population as a whole (Figure 3). Several questions regarding the acceptability and effectiveness of monitoring for malaria at immunization clinics need to be considered, but children attending well-baby clinics have previously provided a continuous assessment of malaria morbidity at a local level, including parasite prevalence.27,28

CONCLUSION

No single data source is entirely satisfactory for monitoring malaria disease trends; rather, a combination is needed. The relative weight given to different sources will vary by country, and the challenge will be to arrive at an appropriate balance between different methods. Greater emphasis may be given to surveys in countries such as Indonesia, which are highly decentralized and which already conduct annual household surveys, whereas countries with more closely controlled health services such as the Solomon Islands will depend almost entirely on routinely collected information. Where routine data are used, they must be carefully screened and adjusted to take into account missing values and different levels of service use before they can provide a reasonably informative picture of morbidity and mortality in a country. Such analysis will benefit from the insight provided by DHSs and DSSs into the coverage and completeness of health facility data.

Exploration of sentinel site methodology and the use of indirect methods may enhance future estimates of disease burden and trends. However, even if better data were available, it is uncertain whether it would make a difference to the operation of programs. Even now, the availability of data often exceeds the use made of it. The lack of use reflects more general weaknesses in planning and monitoring in the health sector, which needs to be addressed if monitoring is to influence decisions regarding resource allocation and use. Investment in data collection without a commensurate effort in decision-making processes is unlikely to yield significant returns; ultimately, it is quality of decision rather than quality of data that will help to reduce the malaria burden.

Figure 1.
Figure 1.

Definitions of populations at risk differ between countries in the Western Pacific Region and sometimes change radically over time.21

Citation: The American Journal of Tropical Medicine and Hygiene 77, 6_Suppl

Figure 2.
Figure 2.

A national estimate of admissions for cerebral malaria in Papua New Guinea was made by imputing missing values for health facilities that failed to report. The increase in cases estimated nationally is similar to the trend observed in the set of facilities that submitted practically complete reports.23

Citation: The American Journal of Tropical Medicine and Hygiene 77, 6_Suppl

Figure 3.
Figure 3.

Estimates of morbidity in children obtained through immunization clinics may be sufficiently similar to population estimates for program management purposes. Each dot shows the value of anemia or fever prevalence for the 2003 Burkina Faso DHS in the full sample vs. the value obtained by focusing on children immunized against measles or DPT3.29

Citation: The American Journal of Tropical Medicine and Hygiene 77, 6_Suppl

*

Address correspondence to Richard E. Cibulskis, 20 Avenue Appia, CH-1211 Geneva 27, Switzerland. E-mail: cibulskisr@who.int

Authors’ addresses: Richard Cibulskis and Maru Aregawi, Global Malaria Program, World Health Organization, 20 Avenue Appia, CH-1211 Geneva 27, Switzerland. Telephone: 41-22-791-2667, Fax: 41-22-791-4824, E-mail: cibulskisr@who.int. David Bell and Eva-Maria Christophel, World Health Organization, Regional Office for the Western Pacific, P.O. Box 2932, 1000 Manila, Philippines. Jeffrey Hii, World Health Organization, P.O. Box 22, Honiara, Solomon Islands. Charles Delacollette, World Health Organization, c/o Ministry of Public Health, Tiwanon Road, Muang, Nonthaburi 11000, Thailand. Nathan Bakyaita, World Health Organization, Regional Office for Africa, Cité du Djoué, P.O. Box 06, Brazzaville, Congo.

Disclosure: The authors are staff members of the WHO. The authors alone are responsible for the views expressed in this publication and they do not necessarily represent the decisions, policy, or views of the WHO.

REFERENCES

  • 1

    Roll Back Malaria/World Health Organization, 2000. Framework for Monitoring Progress and Evaluating Outcomes and Impact. Geneva: Roll Back Malaria/World Health Organization.

  • 2

    Korenromp EL, Williams BG, Gouws E, Dye C, Snow RW, 2003. Measurement of trends in childhood malaria mortality in Africa: an assessment of progress toward targets based on verbal autopsy. Lancet Infect Dis 3 :349–358.

    • Search Google Scholar
    • Export Citation
  • 3

    Snow RW, Guerra CA, Noor AM, Myint HY, Hay SI, 2005. The global distribution of clinical episodes on Plasmodium falciparum malaria. Nature 434 :214–217.

    • Search Google Scholar
    • Export Citation
  • 4

    Rowe AK, Rowe SY, Snow RW, Korenromp EL, Schellenberg JR, Stein C, Nahlen BL, Bryce J, Black RE, Steketee RW, 2006. The burden of malaria mortality among African children in the year 2000. Int J Epidemiol 35 :691–704.

    • Search Google Scholar
    • Export Citation
  • 5

    Bell D, Jorgensen P, Christophel E, Palmer KL, 2005. Estimation of the malaria burden. Nature 437 :E3–E4.

  • 6

    Omumbo JA, Hay SI, Guerra CA, Snow RW, 2004. The relationship between the Plasmodium falciparum parasite ratio in childhood and climate estimates of malaria transmission in Kenya. Malar J 3 :17.

    • Search Google Scholar
    • Export Citation
  • 7

    Carter R, Mendis KN, Roberts D, 2000. Spatial targeting of interventions against malaria. Bull World Health Organ 78 :1401–1411.

  • 8

    Lubanga RG, Norman S, Ewbank D, Karamagi C, 1997. Maternal diagnosis and treatment of children’s fever in an endemic malaria zone of Uganda: implications for the malaria control programme. Acta Trop 68 :53–64.

    • Search Google Scholar
    • Export Citation
  • 9

    Dunyo SK, Afari EA, Koram KA, Ahorlu CK, Abubakar I, Nkrumah FK, 2000. Health centre versus home presumptive diagnosis of malaria in southern Ghana: implications for home-based care policy. Trans R Soc Trop Med Hyg 94 :285–288.

    • Search Google Scholar
    • Export Citation
  • 10

    Einterz EM, Bates ME, 1997. Fever in Africa: do patients know when they are hot? Lancet 350 :781.

  • 11

    Korenromp EL, Armstrong-Schellenberg JRM, Williams BG, Nahlen BL, Snow RW, 2004. Impact of malaria control on childhood anemia in Africa—a quantitative review. Trop Med Int Health 9 :1050–1065.

    • Search Google Scholar
    • Export Citation
  • 12

    Zambia Ministry of Health, 2006. Zambia Malaria Indicator Survey 2006. Lusaka: Ministry of Health Zambia.

  • 13

    Monasch R, Reinisch A, Steketee RW, Korenromp EL, Alnwick D, Bergevin Y, 2004. Child coverage with mosquito nets and malaria treatment from population-based surveys in African countries: a baseline for monitoring progress in Roll Back Malaria. Am J Trop Med Hyg 71 :232–238.

    • Search Google Scholar
    • Export Citation
  • 14

    Monograph Series INDEPTH, 2002. Demographic Surveillance Systems for Assessing Populations and Their Health in Developing Countries, Volume 1: Population, Health and Survival in INDEPTH Sites. Ottawa: IDRC/CRDI.

  • 15

    De Savigny D, Kasale H, Mbuya C, Reid G, 2004. In Focus: Fixing Health Systems. Ottawa: IDRC/CRDI.

  • 16

    Flew SJ, 2002. District health care in Tari until 1991. P N G Med J 45 :106–112.

  • 17

    Whiting DR, Setel PW, Chandramohan D, Wolfson L, Hemed Y, Lopez AD, 2006. Estimating cause-specific mortality from community- and facility-based data sources in the United Republic of Tanzania: options and implications for mortality burden estimates. Bull World Health Organ 84 :921–1000.

    • Search Google Scholar
    • Export Citation
  • 18

    Agyepong IA, Kangeya-Kayonda J, 2004. Providing practical estimates of malaria burden for health planners in resource-poor countries. Am J Trop Med Hyg 71 :162–167.

    • Search Google Scholar
    • Export Citation
  • 19

    Erhart A, Thang ND, Xa NX, Thieu NQ, Hung LX, Hung NQ, Nam NV, Toi LV, Tung NM, Bien TH, Tuy TQ, Cong LD, Thuan LK, Coosemans M, D’Alessandro U, 2007. Accuracy of the health information system on malaria surveillance in Vietnam. Trans R Soc Trop Med Hyg 101 :216–225.

    • Search Google Scholar
    • Export Citation
  • 20

    Chilundo B, Sundby J, Aanestad M, 2004. Analyzing the quality of routine malaria data in Mozambique. Malar J 3 :3.

  • 21

    Graves PM, 2006. Assessment of the Implementation of the WHO Western Pacific Region Malaria ‘Kunming’ Indicator Framework, 1999–2005. Consultant Report. Manila: WHO WPRO.

  • 22

    Mapping Malaria in Asia and the Pacific, 2006. Report of a WHO Technical Consultation, Bangkok, Thailand, September 14–16, 2006.

  • 23

    Ministry of Health, 1998. National Health Plan 1996–2000: Mid-Term Review. Waigani: Ministry of Health, Papua New Guinea.

  • 24

    Cibulskis RE, 2002. Trends in the Distribution of Health and Health Services in Papua New Guinea 1991–2000. Papua New Guinea Health Sector Review. Manila: Asian Development Bank.

  • 25

    Gething PW, Noor AM, Gikandi PW, Ogara EAA, Hay SI, Nixon MS, Snow RW, Atkinson PM, 2006. Improving imperfect data from health management information systems in Africa using space-time geostatistics. PLoS Med 3 :e271.

    • Search Google Scholar
    • Export Citation
  • 26

    Drakeley CJ, Carneiro I, Reyburn H, Malima R, Lusingu JPA, Cox J, Theander TG, Nkya WMMM, Lemnge MM, Riley EM, 2005. Altitude dependent and independent variations in Plasmodium falciparum prevalence in Northeastern Tanzania. J Infect Dis 191 :1589–1598.

    • Search Google Scholar
    • Export Citation
  • 27

    Delacollette C, Barutwanayo M, Mpitabakana P, 1990. Epidemiologie du paludisme au Burundi: observations preliminaires. Med Afr Noire 37 :718–721.

    • Search Google Scholar
    • Export Citation
  • 28

    Some ES, Koech DK, Ochogo JO, Ocholla F, Mumbi F, 1997. An evaluation of surveillance of malaria at primary health care level in Kenya. East Afr Med J 74 :573–575.

    • Search Google Scholar
    • Export Citation
  • 29

    Institut National de la Statistique et de la Démographie, Ministère de l’Économie et du Développement, Ouagadougou, Burkina Faso, 2004. Burkina Faso: Enquête Démographique et de Santé. Calverton, MD: ORC Macro.

Save