Health Policy and Planning Advance Access originally published online on September 9, 2005
Health Policy and Planning 2005 20(6):347-353; doi:10.1093/heapol/czi041
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Original article |
The burden of disease, economic costs and clinical consequences of tuberculosis in the Philippines
1 Institute for Global Health, University of California at San Francisco, 2 San Francisco Veterans Affairs Research Enhancement Award Program, San Francisco, 3 RAND, Santa Monica, 4 School of Public Health, University of California, Los Angeles, California, USA and 5 University of the Philippines School of Economics, Quezon City, Philippines
Correspondence: John Peabody, MD, PhD, Institute for Global Health, 50 Beale Street, Ste #1200, San Francisco, CA 94105, USA. Tel: +1 415 597-8200; Fax: +1 415 597-8299. E-mail: peabody{at}psg.ucsf.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionBiographiesReferences |
|---|
Objective: To provide a multidisciplinary and comprehensive analysis on the impact of tuberculosis (TB) in a high incidence country.
Methods: Data from several large scale surveys, carried out between 1997 and 2001 in the Philippines, were used to: (1) perform a burden of disease (BoD) analysis, (2) model the economic costs to society due to lost wages, and (3) determine the clinical outcomes, including the costs of care, for a hypothetical cohort of TB cases.
Results: Over 500 000 disability-adjusted life years (DALYs) are lost due to illness and premature mortality from TB in the Philippines annually. This is equal to 9% of all years of life lost (YLL) in the Philippines. The combined economic losses due to premature mortality and morbidity total PhP8 billion (approximately US$145 million). Clinically, only 28% of patients with incident active TB are diagnosed and successfully treated, while 20% of patients will die without ever being diagnosed and 6% more will die after they are diagnosed because they do not receive adequate care. The costs of treating all expected cases requires between PhP475–1625 million (approximately US$8–29 million) annually.
Conclusion: The high burden of disease from TB, large economic losses from mortality and morbidity from TB and the poor clinical outcomes all suggest that there is an urgent need for an increased investment in TB control. The costs of providing this treatment appear to be significantly lower than the current economic losses.
Key Words: tuberculosis, pulmonary, burden of illness, costs, disease, Philippines
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionBiographiesReferences |
|---|
Tuberculosis (TB) continues to be a leading cause of morbidity and mortality worldwide. It kills more than 2 million people per year, making it the single leading microbial killer of adults (Frieden et al. 2003
). According to the World Health Organization (WHO), one-third of the world's population is infected with M. tuberculosis. These cases, however, disproportionately occur in the developing world, particularly in Asia (Corbett et al. 2003), where, for example, it is estimated that approximately 40% of people living in the Indian subcontinent, China and the Pacific Rim are infected with TB (Dye et al. 1999). The Philippines is especially hard-hit by the disease, having the second highest incidence of TB in Asia (Corbett et al. 2003). Simple health sector interventions have been inadequate to stem this staggering toll here and elsewhere (Gupta et al. 2004). Thus, there exists a need for complete assessment of the impact of TB, not simply to inform policy-makers of the burden of disease but also to provide the basis for targeted interventions in TB control (Nunn et al. 2002). This paper describes an analysis that provides a more comprehensive understanding of the impact of TB in the Philippines. We evaluate TB in the Philippines from three disciplinary perspectives: the national disease burden, the economic costs, and the clinical consequences. The Philippines is a country well suited to introducing this approach: high quality data from a variety of sources facilitates a multidisciplinary analysis and there is an established tradition of high quality field research to inform policy. Just as importantly, the high prevalence of TB exists in an environment where policy-makers are eager to have an evidence base for decision-making.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionBiographiesReferences |
|---|
Data collection
Project team members had expertise in TB in the Philippines, fieldwork, and epidemiologic and econometric analysis. The team identified all recent datasets related to TB in the Philippines (Table 1) and reviewed them with respect to three initial eligibility criteria: completeness, accessibility and concurrence with other data. If they met these criteria, two more criteria were applied: did they support the analyses from more than one of the analytic perspectives, and were they suitable for concurrent analysis?
|
Seven unique datasets pertaining to TB prevalence and economic impact in the Philippines met the first set of criteria. Four of these datasets best met the last two criteria. They were the 1997 National TB Prevalence Survey (NTPS), the 1997 Philippine Health Statistics (PHS), the 1998 Annual Poverty Incidence Survey (APIS), and the 2001 Urban Health and Nutrition Program Evaluation Survey (UHNP). From these datasets we extracted the variables necessary to perform specific analyses; the remaining datasets were used to cross check or augment these analyses.
Analysis of the burden of disease
We applied the WHO burden of disease (BoD) methodology to produce a combined measure of TB mortality and morbidity. The BoD methodology and the assumptions underlying it are described in detail in the results of the Global Burden of Disease study (Murray and Lopez 1996a,b). Years of life lost (YLL) were calculated from TB mortality statistics. The remaining years of potential life at any age of death were calculated from a life table based on a life expectancy at birth of 82.5 years for females and 80.0 years for males (Murray and Lopez 1996a). For efficiency of calculation, counts of deaths were aggregated into 5-year groups of age at death for each gender, and the years of potential life lost were calculated based on the midpoint of the age range. The YLLs were summed across age groups, and rates per 100 000 were calculated for each age/gender group and the population as a whole. Years lived with disability (YLD) due to TB were calculated from the number of incident cases in a year, times the estimated duration of the disease, times the disability weight. The weights for TB range from 0.264 to 0.294 for different age and gender groups (Murray and Lopez 1996a), and the rates were calculated per 100 000 as for YLLs. Disability-adjusted life years (DALYs) are the sum of YLLs and YLDs. Since YLDs were calculated for broader age groups than for YLLs, data are presented by these broader groups. The DALY calculation also incorporates two adjustment factors: discounting (at 3%) and age weighting (which assigns a higher social value to lost years of life in the early and middle adult years than those in the earliest or latest years of life) (Murray and Lopez 1996a).
Population data for the study's base year, 1997, by age and gender, were derived from the 1997 Philippine Health Statistics report. TB incidence was estimated using the 1997 NTPS. Incidence was calculated from prevalence by assuming that 45% of cases are sputum smear positive and using disease duration of 2.2 years; incidence was assumed to be zero for persons aged under 10 years. The 1997 incidence was calculated as 321 per 100 000 nationwide. A detailed reconciliation of NTPS prevalence rates, and incidence rates calculated from them, was conducted to ensure that NTPS-based values were consistent with WHO estimates for the Philippines of 314 per 100 000 in 1997 (Dye et al. 1999). We also performed a sensitivity analysis of the morbidity and mortality effects of improving DOTS (directly observed treatment, short course) detection and reducing the duration of illness.
Analysis of the economic consequences
Analyses of the economic consequences of TB relied on data from both the 1998 Annual Poverty Incidence Survey and the 2001 Urban Health and Nutrition Program Evaluation Survey. The social and economic consequences of TB were analyzed along four dimensions: (1) the prevalence of TB by age and income quintiles, (2) the social and demographic predictors of TB prevalence, (3) wage differentials, representing economic losses experienced by those with TB, and (4) foregone income from premature deaths and disability due to TB.
To determine social and demographic variations, TB prevalence was stratified across age, gender, income quintiles and place of residence (urban versus rural). To further investigate the influence of socioeconomic variables on TB prevalence, we used a logit regression model to obtain estimates of the marginal effect of these predictors on TB prevalence. The logit model included age, gender and other monetary benefits for those with TB and those without, to estimate wage differentials between these groups. The daily full-wage rate was computed as the ratio between total income during a quarter and the total number of full days worked during the same (third) quarter (APIS only collects data on these variables during the third quarter, July to September). In turn, the total number of full days worked is the sum of full days worked plus total number of hours worked during less than full days worked divided by 8. The daily full-wage rates across age groups was then used to compute the weighted average annual wage. To present these daily wage rates in 2002 prices, a 1998–2002 inflation factor of 22% was calculated using the general or national consumer price index (CPI).
The analysis of wage loss for those with TB required a number of modelling techniques (Behrman and Wolfe 1984; Greene 1993). For individuals who choose not to work, we used a probit regression to account for the endogeneity of this decision. Equations for wage and decision to work were estimated jointly using Heckman's two-step sample selection estimator (Heckman 1979). Since self-reported TB prevalence is endogenous by construction, the use of ordinary least squares methods would yield biased estimates. Hence, we used a treatment effects model that estimated the treated self-reported TB illness variable as an endogenous variable consequent from the survey respondent's decision to report the occurrence of TB. The statistical remedy to this selectivity problem is akin in character to the unobserved wage problem. Consequently, the daily wage rate was regressed against the following variables: age, gender, educational attainment, urban and regional dummy variables, unpaid family worker dummy variable, decision to work (work versus not work) inverse Mills ratio, and self-reported TB prevalence inverse Mills ratio (Heckman 1979).
We estimated foregone income as the amount of earnings lost due to premature mortality and TB disability using multiple logit models (Schmidt and Strauss 1975). The first mortality loss component is defined as the product of years of life lost (YLL) multiplied by the weighted average annual wage. Since wage rates are expected to vary through time because of changes in productivity and inflation, a number of assumptions were made to account for these issues. The first was that wages are expected to change at the same rate as inflation. The second was that the daily wage rate profile across age groups reflects the productivity time path (i.e. wages increase until an age threshold level where wages then decrease). With these assumptions, an average daily wage rate was calculated as the average of the prospective wages in the remaining life of a person. To calculate foregone income due to premature mortality, this average annual wage was multiplied by the YLLs.
The second economic loss component was computed in a manner analogous to that described above. The estimated TB population disaggregated by age group and sex is derived from the DALY calculation table along with the expected length of disease and disability factor. The product of these three variables is multiplied by the TB wage differential (defined as the wage difference between the with-TB and without-TB states) to compute for the TB disability loss by age group and gender. The sum of TB disability losses across age groups and gender constitutes the economic loss due to TB disability. All other costs associated with omitting treatment of TB were not included in these calculations.
Analysis of the clinical consequences
To evaluate the clinical consequences of TB, we adopted both a system-level perspective and a patient-level perspective, in which the latter considers utilization and costs of treatment. The system-level perspective entailed ‘following’ the sequence of clinical events for 100 hypothetical or ‘typical’ patients who develop clinically active TB. We classified these 100 patients depending on whether they were sputum smear positive (SS+) or negative (SS–) and whether they would have lived or died. Using published estimates that 45% of global TB cases are SS+ and 55% are SS– (Dye et al. 1999), we combined these with estimates from the 2003 WHO report on TB for Philippine-specific data on treatment effectiveness (World Health Organization 2003). Accordingly, for the SS+, 58% would be detected; 77% of those detected are in a DOTS programme, of whom 77% are registered; and 89% of those in a DOTS programme receive effective treatment. For the 40% of SS+ not detected (and hence not treated), we use the published case fatality of 0.70 (Dye et al. 1999). For those SS+ who are detected but not treated in a DOTS programme, we have assumed a case fatality to be 0.30. Finally, for the SS–, we conservatively assumed that 40% would be diagnosed by other means, such as CXR, and we assumed that SS– and SS+ groups have the same proportions of cases who are detected, registered for DOTS, and receiving effective therapy. Lastly, for the SS– who are not detected, case fatality is estimated to be 0.20 (Dye et al. 1999).
The patient-level perspective was used to estimate utilization and costs of care. These data were taken from the 2001 Urban Health and Nutrition Program Evaluation Survey and the 1991 Department of Health – Philippine Institute for Development Studies (DOH-PIDS) survey of outpatient clinic users, respectively. Costs of TB care, borne by the consumer, were estimated across all service providers. These estimates were weighted by the number of cases seen by each provider group. All calculations were based upon conservative assumptions. For example, a case-fatality rate for untreated or undetected cases is included, and the wage loss estimates include a provision for a basal unemployment rate or work in the informal wage sector.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionBiographiesReferences |
|---|
Burden of disease
Over 500 000 years of healthy life (DALY) are lost due to illness (YLD) and premature mortality (YLL) from TB in the Philippines annually (Table 2). Over two-thirds of this burden is due to premature mortality (YLL), reflecting TB deaths in the prime of life. TB causes 9% of all YLLs in the Philippines, which at a minimum is equivalent to 26 000 deaths. The burden of TB is dramatically higher among men than women (Table 2), and TB prevalence and mortality rise sharply with age (Tables 3 and 4). DALYs peak at over 3600 per 100 000 in males aged 60–69 years, and at over 2000 per 100 000 in females aged 70–79 years.
|
|
|
Sensitivity analyses indicate that the burden of TB control could decline significantly if detection or duration of illness could be altered. If, for example, TB control programmes achieved the WHO target of detecting 70% of SS+ cases in DOTS programmes while maintaining a successful treatment rate of 85%, in this instance, the number of deaths due to TB would drop by 22.7%, reducing the number of YLLs by 80 655. Similarly, if the average duration of disease was reduced from 2.2 years to 1.5 years, this would reduce the number of YLDs by 30.9%, or 49 213.
Economic costs
After controlling for age, gender and location, we found that for every 10% increase in income, TB prevalence declines by 2%. The model also shows that Southern Luzon and the National Capital Region (NCR), the two most densely populated areas of the country, are ranked first and second by prevalence. Reported TB prevalence is higher for the lower household-income quintiles in comparison with the higher quintiles and this difference increases with age (Table 3).
Men with TB earn PhP451 (US$8.15) less per day than those who do not have TB. Women earn PhP216 (US$3.90) less per day. When these losses are aggregated at the national level, the annual income lost due to morbidity from TB is over PhP6 billion (approximately US$108 million) per year. Similarly, when we calculate the loss due to premature mortality from TB, PhP1.8 billion (approximately US$32 million) are lost annually. The combined loss, based on these estimates, in the Philippines was PhP8 billion (approximately US$145 million) due to TB morbidity and mortality in 1997 (Table 4).
Clinical consequences
To evaluate the clinical sequelae of TB, we ‘followed’ 100 hypothetical TB cases on their passage through the health care system. In total, just 28 of 100 patients with incident active TB are diagnosed and successfully treated. By contrast, 20 patients die without ever being diagnosed and 6 more will die even though they are diagnosed because they never get treated. Of the remaining 46 patients, some are diagnosed and not treated and some are partially treated, many will subsequently go on to be carriers while others will eventually develop multiple-drug-resistant TB (Table 5).
|
Estimates of utilization showed that 38% of respondents with TB used the health centre in their own area, 27% used private clinics, 26% used the government hospital, 6% used private hospitals, and 3% visited a health centre outside their own area for TB treatment. There was a wide range of out-of-pocket costs to patients for TB care across providers; private hospitals had the highest costs when compared with other providers (Table 6). Out-of-pocket patient costs associated with TB treatment costs varied from as low as PhP626 (US$11.32) in health centres to PhP6192 (US$111.97) in private hospitals. Because publicly-funded health centres often run out of drugs and private insurance does not cover these costs, patients have to pay for drugs themselves. The cost of TB drugs per patient for a 6-month treatment was between PhP1000 (US$18.08) and PhP4000 (US$72.33); physician consultation fees ranged between PhP200 (US$3.61) and PhP700 (US$12.66); and laboratory tests were between PhP700 (US$12.66) and PhP1800 (US$32.55).
|
We used these figures to calculate the total costs needed to pay for care of all incident cases, all untreated cases, and all partially treated cases. The complete costs for diagnosis and treatment are divided into low, middle and high estimates to give a range of estimates, from a low of PhP1900 (US$34.36) to a high of PhP6500 (US$117.54) per patient (Table 7). The costs of treating all known, albeit partially treated, cases, therefore, would be between PhP115–393 million (approximately US$2–7 million) in annual direct costs. Treating all expected cases would require between PhP475–1625 million (approximately US$8–29 million) annually.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionBiographiesReferences |
|---|
TB control has received insufficient attention from policy-makers and health services researchers worldwide (Nunn et al. 2002
). The great tragedy of TB is that an effective system for TB control, the WHO DOTS strategy, has existed for almost a decade (Kochi 1991). A number of factors inhibiting the expansion of DOTS have previously been identified: lack of political will, poor financial support, poor organization of health services, inadequate human resources, irregular drug supply, the HIV epidemic and the increase of multidrug-resistant TB (Nunn et al. 2002). One additional reason for the inconsistent attention to TB control is the fragmentary way in which the consequences and costs of TB are presented to the public and to policy-makers. This paper presents a comprehensive population, economic- and clinical-level analysis of TB in the Philippines. We use multiple data sources and a multidisciplinary, quantitatively based analytic approach. The resulting picture is discouraging. There were about a quarter of a million new TB cases and at least 26 000 TB deaths annually in a country of approximately 80 million people. The individual economic losses translate into a staggering PhP7.9 billion (approximately US$145 million) in lost wages from premature deaths and lower earning from illness or disability. The current system of care is inadequate: 20% of new cases die without ever being diagnosed and another 6% die after receiving inadequate treatment. Juxtaposed against the huge economic loss are relatively modest costs for diagnosis and treatment. Even using the highest input cost estimates, these range from PhP392 million (approximately US$7 million) for complete treatment for partially treated patients to PhP1623 billion (approximately US$29 million) to treat all cases annually. Clearly, the net benefit argues for increased resources to reduce the personal and economic consequences.
The study's major limitation is that the individual datasets that are available do not contain all of the necessary information nor can they be merged along either demographic or diagnostic characteristics. For example, the 1997 National TB Prevalence Survey, which uses sputum smears for diagnosis, has the most valid estimate of TB prevalence but lacks detailed socioeconomic variables to further investigate the economic burden of TB; the 1998 Annual Poverty Incidence Survey contains the necessary data for evaluating the economic consequences of TB, but TB prevalence was only determined by self-report. The four primary datasets were based on surveys conducted over a span of 4 to 5 years (1977–2001), limiting our ability to make comparisons across a narrower time-frame. To minimize this we have conducted parallel analyses across datasets whenever possible to confirm the reasonableness and consistency of our estimates. For example, conducting sensitivity analyses and providing ranges for input parameters helps mitigate these data limitations. Another limitation is the set of assumptions underlying the DALY method for calculating the burden of disease, which includes standardized life expectancies, discounting, and the use of disability weights based on the person tradeoff method. Lastly, we were unable to estimate the indirect costs associated with TB (apart from forgone wages due to premature death and morbidity from TB), such as travel to the hospital or clinic and food required while in the hospital. We also did not estimate intangible costs of pain and suffering (Kamolratanakul et al. 1999).
One consequence of these shortcomings is that we have potentially underestimated the burden of disease and economic consequences. The 26 102 TB deaths officially reported from the 1997 NTPS data, for example, is used as the basis for our analyses and should be considered a lower bound estimate. Many TB deaths are known to be either unreported or misreported as other causes. For example, Dye et al. (1999) estimate that in the same year (1997) there were 48 000 TB deaths in the Philippines, and Corbett et al. (2003) estimate that there were 44 000 deaths in 2000. Neither report provided sufficient age or gender detail to do a BoD or econometric analysis, and thus we used the NTPS figures. Even these higher mortality figures may be low estimates for 2003. If we use the 2003 incidence rate of 297/100 000 (World Health Organization 2003) and the 25.6% case fatality rate we calculated herein, we estimate that there would have been 58 644 deaths last year. As a result, our estimates of DALYs and the consequent economic losses are based on extremely conservative estimates of the number of deaths due to TB in the Philippines.
Effective TB control is a public good: because TB is a contagious disease, curative care for the individual makes the population healthier. TB treatment and control is, therefore, a responsibility of government and public policy. Maintaining a healthy workforce and the health of adult men, who are often the primary wage earners, is another societal benefit. Our own estimates show a return of 5 to 18 fold if resources were made available.
Because the burden of TB remains high and the economic benefits of TB control easily outweigh the costs, there is a strong impetus for developing policies to curb the disease. Infectious disease control programmes have begun to shift perspective from strictly biological to multidisciplinary with the multisectoral collaborations that are required for successful TB control, particularly in developing countries (Porter et al. 1999; Raviglione and Pio 2002). The multidisciplinary work we have described could have new value in informing such comprehensive policies for TB control.
Nonetheless, it remains important to ensure the effectiveness of TB policies through continual evaluation (Peabody et al. 1999). Researchers and policy-makers rarely have the resources to find out whether well-intentioned policy interventions are effective in reducing TB morbidity and mortality, even though it is badly needed. There is no compelling evidence, for example, that public awareness campaigns decrease the stigma of TB, let alone increase utilization or patient compliance. Similarly, we are only now learning how to precisely measure how much government regulation and incentives change provider behaviour and therefore quality of care (Peabody et al. 1999, 2005). This, however, is a critical area for research on TB policy. Obtaining estimates of cost also remains a frustrating exercise, and expenditure data, in our experience, are seldom precise and another area where more investigation is needed (Tan 1996; Tan and Capuno 2004).
We believe, therefore, that a comprehensive, multidisciplinary analysis can be a crucial starting point for policy-makers. How many people are dying, how much does this cost society, and how much would it take to ‘fix’ the problem, are basic questions that health policy research should try to answer. Equipped with these answers, policy-makers and civil society may be able to more effectively reduce the clinical and economic burden of TB.
|
Biographies |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionBiographiesReferences |
|---|
John W Peabody, MD, PhD, is Deputy Director of the Institute for Global Health and Professor, University of California at San Francisco, and formerly Co-Director of the San Francisco Veterans Affairs Research Enhancement Award Program, San Francisco. He is also affiliated with RAND, Santa Monica, and the School of Public Health, University of California, Los Angeles, California, USA.
Riti Shimkhada, MPH, is a Research Assistant in the Department of Epidemiology, School of Public Health, University of California, Los Angeles, California, USA.
Carlos Tan Jr is a PhD candidate at the University of the Philippines School of Economics, Quezon City, Philippines.
Jeff Luck, MBA, PhD, is Associate Professor of Health Services in the Department of Health Services, School of Public Health, University of California, Los Angeles, California, USA.
|
Acknowledgments |
|---|
This study was carried out with support from CAMRIS International (Clapp and Mayne Division of Renaissance Information System, Inc.) with funding from USAID/Philippines under the Philippines Tuberculosis Initiatives in the Private Sector Project managed by Chemonics International. We would like to thank Dr Orville Solon and Dr Larry Day for their participation on the project.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionBiographiesReferences |
|---|
Behrman JR, Wolfe BL. 1984. Labor force participation and earnings determinants for women in the special conditions of developing countries. Journal of Development Economics 15: 259–88.[CrossRef]
Corbett EL, Watt CJ, Walker N et al. 2003. The growing burden of tuberculosis: global trends and interactions with the HIV epidemic. Archives of Internal Medicine 163: 1009–21.
Dye C, Scheele S, Dolin P, Pathania V, Raviglione NC. 1999. Global burden of tuberculosis: estimated incidence, prevalence and mortality by country. Journal of the American Medical Association 282: 677–86.
Frieden TR, Sterling TR, Munsiff SS, Watt CJ, Dye C. 2003. Tuberculosis. The Lancet 362: 887–99.[CrossRef][Web of Science][Medline]
Greene W. 1993. Econometric analysis (2nd edition). New York: Macmillan Publishing Company.
Gupta R, Espinal MA, Raviglione MC. 2004. Tuberculosis as a major global health problem in the 21st century: A WHO perspective. Seminars in Respiratory and Critical Care Medicine 25: 245–53.[CrossRef][Web of Science][Medline]
Heckman J. 1979. Sample selection as a specification error. Econometrica 4: 153–61.
Kamolratanakul P, Sawert H, Kongsin S et al. 1999. Economic impact of tuberculosis at the household level. International Journal of Tubercle Lung Disease 3: 596–602.
Kochi A. 1991. The global tuberculosis situation and the new control strategy of the World Health Organization. Tubercle 72: 1–6.[CrossRef][Web of Science][Medline]
Murray CJL, Lopez AD (eds). 1996a. The Global Burden of Disease, Volume 1. Cambridge, MA: Harvard School of Public Health, World Health Organization, World Bank.
Murray CJL, Lopez AD (eds). 1996b. The Global Burden of Disease, Volume 2. Cambridge, MA: Harvard School of Public Health, World Health Organization, World Bank.
Nunn P, Harries A, Godfrey-Faussett P et al. 2002. The research agenda for improving health policy, systems performance, and service delivery for tuberculosis control: a WHO perspective. Bulletin of the World Health Organization 80: 471–6.[Medline]
Peabody JW, Rahman MO, Gertler PJ et al. 1999. Policy and health: implications for development in Asia. Cambridge: Cambridge University Press.
Peabody J, Taguiwalo M, Robalino D, Frenk J. 2005. Improving the quality of care in developing countries. In: Jamison D (ed.) Disease control priorities in the developing world (2nd edition). New York: Oxford University Press for the World Bank, Chapter 70.
Porter J, Ogden J, Pronyk P. 1999. Infectious disease policy: towards the production of health. Health Policy and Planning 14: 322–8.
Raviglione MC, Pio A. 2002. Evolution of WHO policies for tuberculosis control, 1948–2001. The Lancet 359: 775–80.[CrossRef][Web of Science][Medline]
Schmidt P, Strauss R. 1975. The prediction of occupation choice using multiple logit models International Economic Review 16: 471–86.[CrossRef]
Tan CA. 1996. Planning model for health care services and financial resources: a health care market simulation model. Manila: Philippine Institute for Development Studies, manuscript.
Tan CA, Capuno JJ. 2004. The economic burden of tuberculosis and the need for private sector initiatives. Quezon City, Philippines: University of the Philippines – School of Economics (UPSE), Discussion Paper, November 2004.
World Health Organization. 2003. Global tuberculosis control: surveillance, planning, financing. Geneva: World Health Organization, document WHO/CDS/TB/ 2003.316.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||