|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||
Background We hypothesized that investments to improve the control of tuberculosis in selected high-incidence countries would prove to be cost saving for the United States by reducing the incidence of the disease among migrants.
Methods Using decision analysis, we estimated tuberculosis-related morbidity, mortality, and costs among legal immigrants and refugees, undocumented migrants, and temporary visitors from Mexico after their entry into the United States. We assessed the current strategy of radiographic screening of legal immigrants plus current tuberculosis-control programs alone and with the addition of either U.S.-funded expansion of the strategy of directly observed treatment, short course (DOTS), in Mexico or tuberculin skin testing to screen legal immigrants from Mexico. We also examined tuberculosis-related outcomes among migrants from Haiti and the Dominican Republic using the same three strategies.
Results As compared with the current strategy, expanding the DOTS program in Mexico at a cost to the United States of $34.9 million would result in 2591 fewer cases of tuberculosis in the United States, with 349 fewer deaths from the disease and net discounted savings of $108 million over a 20-year period. Adding tuberculin skin testing to radiographic screening of legal immigrants from Mexico would result in 401 fewer cases of tuberculosis in the United States but would cost an additional $329 million. Expansion of the DOTS program would remain cost saving even if the initial investment were doubled, if the United States paid for all antituberculosis drugs in Mexico, or if the decline in the incidence of tuberculosis in Mexico was less than projected. A $9.4 million investment to expand the DOTS program in Haiti and the Dominican Republic would result in net U.S. savings of $20 million over a 20-year period.
Conclusions U.S.-funded efforts to expand the DOTS program in Mexico, Haiti, and the Dominican Republic could reduce tuberculosis-related morbidity and mortality among migrants to the United States, producing net cost savings for the United States.
Implementation of the World Health Organization (WHO) strategy of directly observed treatment, short course (DOTS),15 can substantially reduce the incidence16,17 and prevalence18 of tuberculosis in countries with a high incidence of the disease. However, because of inadequate funding,19 global implementation of the DOTS program remains far from complete. One strategy to reduce the incidence of tuberculosis in high-income countries that has received little attention is to strengthen tuberculosis-control programs through the expansion of the DOTS program in key countries that are the source of migrants.
We estimated tuberculosis-related morbidity, mortality, and costs associated with radiographic screening and existing tuberculosis-control programs over a period of 20 years among Mexican-born migrants to the United States. We compared these results with those arising from the addition of either a U.S.-funded expansion of the DOTS program in Mexico or tuberculin skin testing of legal immigrants. We conducted parallel analyses for migrants entering the United States from Haiti and the Dominican Republic.
Methods
General Description of the Model
We devised a decision-analysis model incorporating multiple Markov processes (TreeAgePro 2005 Health Care release 0.4, TreeAge Software) to estimate the cumulative probability of active tuberculosis, tuberculosis-related death, and associated costs among migrants to the United States. We used the societal perspective, meaning that direct and indirect costs were included.20 Future expenditures and outcomes were discounted at a rate of 3 percent.21 Migrants were categorized as legal immigrants (persons applying for entry from abroad or within the United States, refugees, asylum seekers, and short-term laborers), undocumented migrants, or temporary visitors. Throughout the 20-year period, the number, age, and types of migrants entering the United States annually remained unchanged in this analysis. We considered three source countries: Mexico for the primary analysis and Haiti and the Dominican Republic for secondary analyses. A more detailed description of the structure of the model, underlying assumptions, and variables and additional results are provided in the Supplementary Appendix (available with the full text of this article at www.nejm.org).
Tuberculosis-Control Strategies
We evaluated three strategies of tuberculosis control: the current program of radiographic screening and tuberculosis control in Mexico, radiographic screening plus expansion of the DOTS program in Mexico, and radiographic screening together with tuberculin skin testing. With the use of the current strategy, legal immigrants and refugees would undergo radiographic screening when entering the United States and current outcomes would apply.5,9,22,23 Current epidemiologic and tuberculosis-control variables would not change in the United States or in Mexico over a 20-year period in this analysis.
In the second scenario, expansion of the DOTS program would be added to radiographic screening. The U.S. government would pay for all costs of expanding the DOTS program in Mexico in order to achieve WHO benchmarks of 100 percent coverage of the population, a 70 percent rate of case detection, and an 85 percent rate of treatment success15 within three years. The rates of treatment failure, mortality, and drug resistance would not change during the 20-year period. After the expansion of the DOTS program, the incidence of new, smear-positive cases would decline 6 percent annually,16 as would the prevalence of latent tuberculosis infection24 among departing migrants.
In the third scenario, tuberculin skin testing would be added to radiographic screening.5,8 The tuberculosis-control programs would not change in Mexico, nor would relevant variables related to tuberculosis and human immunodeficiency virus (HIV) infection change among migrants. We assumed that tuberculin skin testing would have a sensitivity of 99 percent25 and a specificity of 88 percent.8,26 Migrants with a positive test (defined by induration of at least 10 mm) would be prescribed nine months of isoniazid therapy, with an efficacy of 90 percent for isoniazid-sensitive cases of latent tuberculosis27 and 0 percent for isoniazid-resistant cases28 (Table 1). Overall, only 21 percent of all screened migrants with latent tuberculosis would complete nine months of isoniazid therapy — the average percentage in several large-scale programs that use tuberculin skin testing9,10,11,38,39 (see Table S4 of the Supplementary Appendix).
|
We included four tuberculosis-related health states: no infection; latent tuberculosis infection, subdivided into recent infection (acquired within the past two years) and long-standing infection (acquired at least two years ago); active tuberculosis; and healed active tuberculosis (treated or spontaneously resolved). Latent and active tuberculosis were modeled as drug-sensitive, resistant to a single drug, or multidrug-resistant. The prevalence of recent and long-standing latent tuberculosis infection reflected migrants' mean age at entry12 and the annual risk of infection,24 derived from the incidence of smear-positive cases in the source country.48 Among HIV-negative migrants with a recent diagnosis of latent tuberculosis infection, we estimated that the infection would become active in 5 percent in the first two years after entry into the United States,49 with an annual rate thereafter of 0.1 percent if the chest x-ray film showed no abnormalities23,50 and of 0.6 percent if the film showed abnormalities.23 We also accounted for the possibility that migrants could become infected during return visits to their countries of origin.
We assumed that the outcomes among treated migrants would equal those among U.S.-born persons.2 The mortality rate among persons with undiagnosed smear-positive tuberculosis was assumed to be 33 percent annually,51 and 25 percent of cases were assumed to resolve spontaneously.52
Effect of HIV Infection
We modeled three HIV-related health states: uninfected, early infection (asymptomatic), and late infection (clinical acquired immunodeficiency syndrome). The number of HIV-infected legal immigrants entering the United States is negligible, but the prevalence of seropositivity for HIV among undocumented migrants and visitors was assumed to be similar to that in the general population of the source countries.30 For migrants with both tuberculosis and HIV infections, the risk of active tuberculosis depends on the duration of latent tuberculosis and the stage of HIV infection,53,54 and the mortality rate is higher, but cure and relapse rates for active tuberculosis are similar to those among HIV-negative migrants55,56 (see Table S1 of the Supplementary Appendix).
Costs
Costs, expressed in 2003 U.S. dollars, were categorized as direct (borne by the U.S. government and health care system for the expansion of the DOTS program and tuberculosis-related screening and health care) or indirect (out-of-pocket expenditures by patients and their families and lost wages due to disability, death, or provision of care). Costs for expansion of the DOTS program, including those related to infrastructure, equipment, materials, personnel, and training, were derived from a DOTS-expansion project in Ecuador46 overseen by the Canadian Lung Association (see the Supplementary Appendix for details). Projected drug expenditures reflected WHO incidence estimates48; projected rates of failure, relapse, and defaults requiring retreatment; and drug prices in the Global Drug Facility.47
Within the United States, the most important direct cost was $36,045 for the treatment of active tuberculosis, a value based on a comprehensive, 17-state survey of all payers29 (adjusted for inflation57). Productivity losses from premature deaths related to tuberculosis were calculated on the basis of the expected annual income and the number of years remaining in the 20-year simulation. To estimate other indirect costs, we used a standardized questionnaire to interview 50 patients with tuberculosis and their families in Montreal, New York, and Miami. All respondents provided written informed consent, and the survey was approved by the institutional review boards of all participating centers.
Sensitivity and Secondary Analyses
We changed variables individually and then jointly to assess changes in all outcomes among migrants from Mexico. We also examined the effect of the same three tuberculosis-control strategies among migrants from Haiti and the Dominican Republic.
Results
According to our model, over the 20-year period of analysis, 35.4 million migrants are projected to enter the United States from Mexico. With the current strategy of radiographic screening and the current tuberculosis-control program in Mexico, we estimated that there would be 47,610 cases of tuberculosis and 5245 deaths related to tuberculosis in this population, resulting in direct and indirect costs of $1.985 billion and $632 million, respectively (Table 2). If the U.S. government invested $34.9 million to expand the DOTS program in Mexico without changing screening or control programs in the United States, there would be 2591 fewer cases of tuberculosis and 349 fewer deaths related to tuberculosis in the United States. These numbers reflect the projected reduction in latent tuberculosis infection, particularly recent infection, among newly arrived migrants owing to the reduction in incidence after the expansion of the DOTS program in Mexico. By preventing these cases of tuberculosis, the expanded DOTS program would result in a net reduction in direct (government) spending of $84 million and a reduction of another $24 million in indirect costs. Of the cases of tuberculosis averted with the implementation of the expanded DOTS program, 88 percent would have occurred among undocumented migrants and visitors.
|
|
As shown in Table 3, if the expansion of the DOTS program produced a slower-than-expected decline in the incidence of tuberculosis in Mexico, cost savings for the United States would be anticipated unless the rate of decline was less than 1.2 percent annually. Cost savings would be anticipated even if the U.S. government doubled its initial investment for expansion of the DOTS program or paid for antituberculosis drugs for all new and retreated cases in Mexico for all 20 years or if the number of migrants was only 33 percent of current levels. If the number of migrants entering the United States or the prevalence of HIV infection, latent tuberculosis infection, or drug resistance were higher or increased over the 20-year period, savings from expansion of the DOTS program would be even greater (data not shown). Finally, if the frequency or duration of return visits by legal immigrants and undocumented migrants were doubled, expansion of the DOTS program would avert an additional 249 cases of tuberculosis and save an additional $5.2 million (data not shown).
|
Secondary Analyses Involving Haiti and the Dominican Republic
According to our model, over a 20-year period, nearly 2 million migrants are expected to enter the United States from Haiti and the Dominican Republic, and active tuberculosis will develop in 11,809 of them, resulting in 1288 deaths. These higher rates of morbidity and mortality reflect the higher prevalence of tuberculosis and HIV infection in these countries. As shown in Table 4, a U.S. investment of $9.4 million to expand the DOTS program in these two countries would result in 590 fewer cases of tuberculosis and net savings of $20 million in the United States. Adding tuberculin skin testing of legal immigrants would result in 315 fewer cases of tuberculosis in the United States, but at an added total cost of $128 million.
|
Our comparison of three strategies for the control of tuberculosis among migrants to the United States suggests that U.S. government–funded expansion of the DOTS program in Mexico could result in the lowest net costs to the United States and the greatest reduction in the number of cases of tuberculosis in the United States. This finding was robust in sensitivity analyses. The alternative strategy of adding tuberculin skin testing to current radiographic screening of legal immigrants would have less effect and would substantially increase costs.
The DOTS strategy is a cost-effective method of controlling tuberculosis in low-income countries61 that has reduced the mortality,62,63 prevalence,18 and incidence16 of tuberculosis in several countries with a large burden of disease. Despite these demonstrated benefits, global implementation of the DOTS program remains hampered by insufficient funding. Our analysis demonstrates the potential domestic economic and public health gains afforded by U.S. underwriting of expansion of the DOTS program in three nearby countries. These domestic gains would complement the humanitarian, economic, and public health benefits from improved control of tuberculosis abroad.
A recent analysis predicted that the implementation of tuberculin skin testing of immigrants to the United States from various parts of the world, with treatment of latent infection tailored to regional patterns of drug resistance, would result in cost savings.8 Our findings regarding the implementation of tuberculin skin testing are at variance with this analysis for several reasons. We focused on Mexico because of its geographic proximity and the high numbers of Mexicans who migrate to the United States. The cost-effectiveness of tuberculin skin testing of migrants from countries such as Mexico, with an intermediate incidence of tuberculosis, will be less than that among migrants from countries with a higher incidence and therefore a higher prevalence of latent tuberculosis infection.26 We also accounted for the anticipated failures of providers and patients to comply with recommendations for screening and therapy for latent infection. These operational problems have substantially reduced the impact of several large-scale screening programs.9,10,11,23,38,39
There remains uncertainty regarding several key assumptions in our analysis, particularly the assumption that the incidence of tuberculosis would decline 6 percent annually within the migrants' source countries after the expansion of the DOTS program. This rate of decline was documented in Peru after the countrywide implementation of the DOTS program16 and falls midway between the 4.3 percent annual decline in prevalence attributable to the implementation of the DOTS program in China18 and the projected 7.5 percent reduction in the annual incidence in countries meeting DOTS targets.17 However, the epidemiologic effect in Mexico of the expansion of the DOTS program from 70 percent to 100 percent coverage of the population might be lower than the effect of expanding coverage from the very low levels that existed in Peru before the DOTS program was implemented or might be lower than assumed because the incidence in Mexico is already declining. However, in a threshold analysis, expansion of the DOTS program in Mexico would remain cost saving for the United States as long as the annual decline in the incidence after the expansion of the program was at least 1.2 percentage points greater than the decline expected in the absence of changes in the tuberculosis-control strategy.
Other important limitations of our analyses were our assumption that the patterns of migration would remain constant and our estimates of the number of undocumented migrants in the United States. If the number of migrants decreases in future years, then the savings afforded by the DOTS program would decrease, although this strategy would remain cost saving even if migration fell to a third of current levels. Conversely, if the number of migrants increases, expansion of the DOTS program would produce greater savings. Because of the large number and epidemiologic importance of undocumented migrants, inaccurate estimation of their numbers will affect overall projections. However, even if this group were excluded from our analysis, expansion of the DOTS program in Mexico would still result in net savings for the United States.
It may have been unrealistic to assume that the prevalence of seropositivity for HIV would not change for 20 years in the three source countries or among persons leaving these countries. However, in sensitivity analyses, expansion of the DOTS program would yield greater cost savings if the prevalence of seropositivity for HIV was higher, because only undocumented migrants or visitors enter the United States with both tuberculosis and HIV infection. Enhanced screening would have no effect on these groups, since they are not screened, whereas expansion of the DOTS program would reduce the prevalence of latent tuberculosis infection among all migrants, including those with HIV infection.
We did not model the secondary spread of tuberculosis within the United States. However, the inclusion of secondary cases would further favor expansion of the DOTS program, because of the greater reduction in cases in the United States among all types of migrants with the implementation of this strategy.
The true costs of expanding the DOTS program are also uncertain, although costs similar to the ones we used have been documented in India.63 We used Ecuadorian cost data because of the similarity of socioeconomic conditions, health, and infrastructure in Ecuador46 to those in Mexico and the Dominican Republic. The applicability of these data to Haiti is less certain, given the possible need for increased infrastructure. Yet, the actual infrastructure-related costs of expanding the DOTS program in India63 (which has income levels very similar to those in Haiti31) were lower than the costs we estimated for Haiti. Moreover, in sensitivity analyses, the finding of net U.S. savings with expansion of the DOTS program was robust even with substantial increases in the costs of initial expansion.
Our analysis had a number of strengths. We considered all types of migrants, including undocumented migrants and visitors, who constitute the vast majority of entrants to the United States.12,13 We also considered the effect of recent infection and return visits to migrants' countries of origin, since the latter variable accounts for 20 percent of cases of tuberculosis among foreign-born permanent residents.14 Most values in the model were derived from cohort studies or randomized trials, reducing uncertainty. We estimated direct and indirect costs in an effort to determine the total costs to society of tuberculosis. Ultimately, the DOTS-expansion strategy remained the most effective and economical approach in multiple sensitivity analyses. Findings were similar for migrants from Haiti and the Dominican Republic, despite differences from Mexico in the current infrastructure of the DOTS program, DOTS coverage, and the epidemiologic characteristics of tuberculosis and HIV infection. This suggests the findings are potentially applicable to migrants from other countries and possibly also for other high-income countries that receive large numbers of migrants.
We conclude that U.S. government's underwriting of the expansion of the DOTS strategy in Mexico, the Dominican Republic, and Haiti is the most effective long-term approach to reducing tuberculosis-related morbidity and mortality among migrants from those countries and would produce net savings in the United States. These projected domestic benefits should encourage the governments of developed countries to provide substantial and sustained funding for the control of tuberculosis abroad.
Supported by a grant from the Rockefeller Foundation. Drs. Schwartzman and Menzies are recipients of research career awards from the Fonds de Recherche en Santé du Québec. Dr. Barr is the recipient of a Robert Wood Johnson Generalist Physician Faculty Scholar Award.
We are indebted to Joe Burzynski, Marie Desrosiers, Edward Ellis, Marcos Espinal, Anne Fanning, Del Garcia, Brian Gushulak, Kamran Khan, Mike Lauzardo, Kathy Moser, Terry Tannenbaum, Charles Wallace, and Militza Zencovich for sharing information and assisting in the conception, conduct, or analysis of this project.
Source Information
From the Respiratory Epidemiology Unit, Montreal Chest Institute (K.S., O.O., D.M.), and the Department of Economics (F.G.), McGill University, Montreal; the Departments of Medicine and Epidemiology, Columbia University Medical Center, New York (R.G.B.); the National Tuberculosis Control Program, Santo Domingo, Dominican Republic (I.A., R.E.M.); Centro Nacional de Investigaciones en Salud Materno Infantil, Santo Domingo, Dominican Republic (J.B.); the National Tuberculosis Action Program, Mexico City, Mexico (E.F., A.C.S.); the National Tuberculosis Control Program, Port-au-Prince, Haiti (W.M., V.J.); the Divisions of Global Migration and Quarantine (S.M.) and Tuberculosis Elimination (K.L.), Centers for Disease Control and Prevention, Atlanta; and the World Health Organization, Geneva (A.P.M.).
Address reprint requests to Dr. Menzies at the Respiratory Epidemiology Unit, Montreal Chest Institute, 3650 St. Urbain, Rm. K1.24, Montreal, QC H2X 2P4, Canada, or at dick.menzies{at}mcgill.ca.
References
| |||||||||||||||||||||||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
