Materials and Methods
A previously published Markov decision-analysis model that evaluated isoniazid prophylaxis for low-risk tuberculin reactors older than 35 years of age21 was adapted to estimate the costs and outcomes of three strategies for treatment of latent tuberculosis infection detected during pregnancy. Markov models allow estimation of the overall likelihood of various competing events (eg, death, development of a disease, compliance with therapy) when the probability of those events could change over time. In the adapted model, no therapy for latent tuberculosis infection was compared with isoniazid therapy during pregnancy (antepartum group) and to therapy delayed until after delivery (postpartum group). A cohort of women 20 years old with a positive (greater than 10 mm) tuberculin skin reaction and a negative chest radiograph obtained at 20 weeks' gestation served as the base case and were followed up until age 85 years. Treatment of latent tuberculosis was defined as isoniazid 300 mg daily for 6 months.18 The model was validated by comparing age-specific estimates of outcomes to the model of Salpeter et al.21
We used a health system perspective that considered the direct medical costs of treatment for latent tuberculosis infection, as well as costs of diagnosis and treatment of active tuberculosis. Because no validated data exist on quality of life for these states or on nonmedical costs associated with tuberculosis and hepatitis in a population of pregnant women at risk for latent tuberculosis activation, a broader societal perspective was not considered.
For each strategy, cases of active tuberculosis within the cohort, cases of tuberculosis secondary to horizontal transmission,21,22 cases of nonfatal and fatal hepatitis, and life expectancy were estimated. All hepatitis survivors were assumed to have the same degree of risk of tuberculosis as patients who received an incomplete course of therapy.
Possible teratogenic effects of isoniazid were not included in the model, because there is no evidence that isoniazid is a human teratogen,20 and in the absence of an identifiable syndrome, estimation of the consequences of any potential congenital anomalies associated with isoniazid is impossible. We assumed that there would be no fetal deaths in women who develop hepatitis, as isoniazid treatment would begin after 20 weeks' gestation, and the median time from initiation of therapy until development of symptomatic hepatitis is 9 weeks.16 Because maternal hepatitis is not directly detrimental to the fetus, and even the most fulminant course would allow time for delivery, we assumed that isoniazid-associated hepatitis would not result in excessive fetal mortality rates. Because of lack of data on health-related quality of life associated with tuberculosis and isoniazid-related hepatitis, quality-adjusted life years were not calculated.
Probabilities for different outcomes are shown and referenced in Table 1. When probabilities were unavailable, assumptions were made and varied widely in sensitivity analysis. In the base case, estimates were chosen that would bias our analysis against antepartum treatment (eg, a relatively low rate of loss to follow-up after delivery).
Published estimates of costs associated with treatment of latent and active tuberculosis and isoniazid-related hepatitis21 were converted to 1998 dollars using the medical care consumer price index.32 Costs associated with tuberculosis cases secondary to horizontal transmission were not calculated. Costs and life expectancy were discounted at an annual 3% rate28; the discount rate was varied from 0 to 5% in sensitivity analysis. All parameters were varied across the ranges shown in Table 1 in one-way sensitivity analyses.
We also performed a worst-case analysis using estimates at the extremes of the ranges, which would favor postpartum treatment. We specifically chose to use only direct medical costs because of lack of data on nonmedical costs, such as transportation, child care, loss from work, and the like. Because there are no reliable estimates on the costs associated with horizontal transmission in the pediatric population, and because including the costs associated with horizontal transmission would favor antepartum treatment, we chose not to include those costs.
The risk of developing active tuberculosis is dependent on duration since exposure, efficacy of treatment, and whether the patient had a complete or incomplete course of therapy. Table 2 shows the predicted cases of nonfatal and fatal tuberculosis within the cohort, the additional cases secondary to horizontal transmission, deaths from tuberculosis, nonfatal and fatal hepatitis cases, undiscounted and discounted total costs, and life expectancy for both treatment strategies. Both strategies were less costly and resulted in higher life expectancy than no treatment. The antepartum group had slightly higher life expectancies, despite a higher mortality rate from hepatitis.
Table 2 also shows the results of a one-way sensitivity analysis in which the tuberculosis case-fatality rate was varied. With a case-fatality rate for tuberculosis of 0.1%, which is tenfold lower than the base-case estimate, no treatment resulted in a slightly higher life expectancy than either antepartum or postpartum treatment. However, marginal gains were small: 3.5 hours for postpartum compared with antepartum treatment and less than 1 hour for no treatment compared with postpartum treatment, with resulting large incremental cost-effectiveness ratios.
In other one-way sensitivity analyses, there were no thresholds in any of the variables tested where antepartum treatment did not result in the fewest tuberculosis cases, including a scenario where the possibility of tuberculosis activation during the index pregnancy was excluded. With regard to costs, only if the mean cost per case of treating active tuberculosis was less than $2064 did no treatment become the least expensive strategy, otherwise antepartum treatment had the lowest expected costs when compared with no treatment or with postpartum treatment. In contrast, two variables, the case-fatality rate for active tuberculosis and the risk of hepatitis, when varied beyond a threshold, resulted in increases in life expectancy. If the case-fatality rate for active tuberculosis was varied from the base-case estimate of 1% to below 0.13%, then no treatment resulted in the greatest life expectancy; if the case-fatality rate was between 0.13% and 0.45%, then postpartum treatment had the greatest life expectancy, and if the rate was greater than 0.45%, then antepartum treatment resulted in the greatest life expectancy. Second, antepartum treatment provided the longest life expectancy unless the risk of hepatitis was varied from the base-case of 1% to less than 0.078%.
In sensitivity analysis using the worst-case scenario, biased against antepartum treatment, antepartum treatment still resulted in the fewest cases of tuberculosis (1986 per 100,000 compared with 2226 per 100,000 for postpartum treatment and 3316 per 100,000 for no treatment) and lowest aggregate costs. However, life expectancy when tuberculosis risks were low and hepatitis risks high was substantially greater with no treatment (26.72 years at a discount rate of 3%, compared with 24.8 years for antepartum and 24.92 years for postpartum treatment). The incremental cost-effectiveness ratio for no treatment was also favorable ($56 per life-year saved).
The number of cases of tuberculosis increased dramatically among reproductive-aged men and women in the late 1980s and early 1990s,1,2 and remains disproportionately high in certain communities. The Advisory Council for the Elimination of Tuberculosis recently recommended to the CDC, Congress, and the National Institutes of Health, an aggressive course toward tuberculosis elimination in the United States, recognizing that a cornerstone of any successful strategy must include targeted screening and efficient delivery of preventive therapy.33 Because prenatal care is often the only contact that many high-risk women have with the health care system, screening for tuberculosis is recommended during pregnancy.19 Our findings indicate that antepartum isoniazid therapy for latent tuberculosis detected during pregnancy would result in the fewest cases of tuberculosis and would be less costly than no treatment or treatment delayed until postpartum. These findings are attributable to the possibility of activation of tuberculosis during the index pregnancy and to failure of women to return for treatment after delivery. The current recommendation to delay isoniazid therapy for latent tuberculosis until postpartum is based on a retrospective analysis that suggested that isoniazid-induced hepatitis might be more frequent in the puerperium.17 However, our model suggests that life expectancy could be longer and costs lower if antepartum therapy is initiated for latent tuberculosis, unless the case-fatality rate for tuberculosis is much lower than a realistic 1% and the risk of fatal hepatitis is tenfold higher than reported.
Because minority populations of reproductive age have the highest rate of tuberculosis in the United States, obstetricians treating them have an important opportunity to detect and potentially prevent activation of latent tuberculosis. Pregnancy represents a unique opportunity to provide monitored isoniazid treatment, and community services are maximized, decreasing socioeconomic barriers to care. Scheduled visits occur frequently and facilitate monitoring for compliance and side effects. Delay of therapy until postpartum can incur unnecessary risk, because a significant proportion of pregnant women with a positive tuberculin skin reaction might not return to receive preventive therapy.17,23 For comparison, 10–30% of women with abnormal Papaniculou smears detected during pregnancy are lost to follow-up,34,35 with follow-up rates lowest in populations similar to those most at risk for tuberculosis. If clinicians could predict which women were likely to return for postpartum isoniazid treatment, therapy might be delayed until postpartum. Unfortunately, accurate predictors of compliance do not exist, and even if such predictors existed, it seems likely that most women eligible for tuberculosis screening in pregnancy would be considered at high risk for noncompliance. For pregnant women with identifiable risk factors, such as intravenous drug use, immigration from high-prevalence countries, or membership in certain minority groups (Hispanic, black, Native American), this missed opportunity for treatment of latent tuberculosis infection could place them at unnecessary risk for activation of tuberculosis.
Although conclusions based on a theoretical cohort of subjects are artificial, these findings suggest that antepartum treatment of latent tuberculosis during pregnancy might be advantageous. The use of isoniazid during pregnancy for the treatment of latent tuberculosis deserves further study, especially in light of the increase in tuberculosis in reproductive-aged women. From a public health perspective, antepartum treatment strategy would result in fewer cases of tuberculosis and be the most cost-effective treatment. In a population of pregnant women, some of the secondary cases prevented by treatment would be in infants or children, where tuberculosis is more likely to disseminate and the mortality rate is relatively high.18 Including these cases in our model would result in substantial gains in average population life expectancy. For an individual patient, the lifetime risk of death from tuberculosis in the base case was roughly equivalent to the more immediate risk of death from isoniazid-related hepatitis. Careful monitoring during isoniazid therapy and exclusion of women with evidence of liver dysfunction or alcohol abuse would decrease the likelihood of hepatitis-related morbidity without increasing costs substantially. Given the risk potential of antepartum isoniazid treatment, strategies to improve postpartum follow-up should be implemented. Until then, our results indicate that consideration should be given for initiation of isoniazid treatment for latent tuberculosis detected during pregnancy.
1. Jereb JA, Kelly GD, Dooley SW Jr, Cauthen GM, Snider DE Jr. Tuberculosis morbidity in the United States: Final data, 1990. MMWR CDC Surveill Summ 1991;40:23–7.
2. Anonymous. Tuberculosis among pregnant women—New York City, 1985–1992. MMWR Morb Mortal Wkly Rep 1993;42:605.
4. Zuber PL, McKenna MT, Binkin NJ, Onorato IM, Castro KG. Long-term risk of tuberculosis among foreign-born persons in the United States. JAMA 1997;278:304–7.
5. Ferebee S. Controlled chemoprophylaxis trials in tuberculosis. A general review. Adv Tuberc Res 1969;17:29–106.
6. Ferebee S. A controlled trial of isoniazid prophylaxis in mental institutions. Am Rev Resp Dis 1963;88:161–75.
7. Horwitz O, Payne P, Wilbek E. Epidemiological basis of tuberculosis eradication. The isoniazid trial in Greenland. Bull World Health Organ 1966;35:509–29.
8. Comstock GW, Baum C, Snider DE Jr. Isoniazid prophylaxis among Alaskan Eskimos: A final report of the bethel isoniazid studies. Am Rev Respir Dis 1979;119:827–30.
9. Snider DE Jr, Caras GJ, Koplan JP. Preventive therapy with isoniazid. Cost-effectiveness of different durations of therapy. JAMA 1986;255:1579–83.
10. United States Preventive Task Forces. Guide to preventive clinical services. 2nd ed. Alexandria, Virginia: International Medical Publishing, 1996.
11. Groth-Petersen E, Gad U, Ostergaard F. Mass chemoprophylaxis of tuberculosis. Am Rev Resp Dis 1960;81:643–52.
12. Ferebee S, Mount F. Tuberculosis morbidity in a controlled trial of the prophylactic use of isoniazid among household contacts. Am Rev Resp Dis 1962;81:490–521.
13. International Union Against Tuberculosis Committee on Prophylaxis. Efficacy of various durations of isoniazid preventive therapy for tuberculosis: Five years follow-up in the IUAT trial. Bull World Health Organ 1982;60:555–64.
14. Kopanoff DE, Snider DE Jr, Caras GJ. Isoniazid-related hepatitis: A U.S. Public Health Service cooperative surveillance study. Am Rev Respir Dis 1978;117:991–1001.
15. Snider DE Jr, Caras GJ. Isoniazid-associated hepatitis deaths: A review of available information. Am Rev Respir Dis 1992;145:494–7.
16. Nolan CM, Goldberg SV, Buskin SE. Hepatotoxicity associated with isoniazid preventive therapy: A 7-year survey from a public health tuberculosis clinic. JAMA 1999;281:1014–8.
17. Franks AL, Binkin NJ, Snider DE Jr, Rokaw WM, Becker S. Isoniazid hepatitis among pregnant and postpartum Hispanic patients. Public Health Rep 1989;104:151–5.
18. Bass JB, Farer LS, Hopewell PC, O'Brien R, Jacobs RF, Ruben F, et al. Treatment of tuberculosis and tuberculosis infection in adults and children. American Thoracic Society and the Centers for Disease Control and Prevention. Am J Respir Crit Care Med 1994;149:1359–74.
19. American College of Obstetricians and Gynecologists. Pulmonary disease in pregnancy. ACOG technical bulletin no. 224. Washington DC: American College of Obstetricians and Gynecologists, 1996.
20. Briggs G, Freeman R, Yaffe S, eds. Drugs in pregnancy and lactation. 4th ed. Baltimore: Williams & Wilkins, 1994.
21. Salpeter SR, Sanders GD, Salpeter EE, Owens DK. Monitored isoniazid prophylaxis for low-risk tuberculin reactors older than 35 years of age: A risk-benefit and cost-effectiveness analysis. Ann Intern Med 1997;127:1051–61.
22. Salpeter EE, Salpeter SR. Mathematical mode for the epidemiology of tuberculosis, with estimates of the reproductive number and infection-delay function. Am J Epidemiol 1998;142:398–406.
23. Davidson PT. Managing tuberculosis during pregnancy. Lancet 1995;346:199–200.
24. Moulding TS, Redeker AG, Kanel GC. Twenty isoniazid-associated deaths in one state. Am Rev Respir Dis 1989;140:700–5.
25. Bloom BR, Murray CJ. Tuberculosis: Commentary on a reemergent killer. Science 1992;257:1055–64.
26. Haas D, Des Prez R. Mycobacterium tuberculosis. In: Mandell G, Bennett J, Dolin R, eds. Principles and practices of infectious diseases. Vol. 2. New York: Churchill Livingstone, 1995.
27. Anderson R. United States abridged life tables, 1996. Natl Vital Stat Rep. Hyattsville, MD: National Center for Health Statistics, 1996.
28. Gold MR, Siegel JE, Russell LB, Weinstein MC. Cost-effectiveness in health and medicine. New York: Oxford, 1996.
29. Fitzgerald JM, Gafni A. A cost-effectiveness analysis of the routine use of isoniazid prophylaxis in patients with a positive Mantoux skin test. Am Rev Respir Dis 1990;142:848–53.
30. Mohle-Boetani JC, Miller B, Halpern M, Trivedi A, Lessler J, Soloman SL, et al. School-based screening for tuberculous infection. A cost-benefit analysis. JAMA 1995;274:613–9.
31. Brown RE, Miller B, Taylor WR, Palmer C, Bosco L, Nicola RM, et al. Health-care expenditures for tuberculosis in the United States. Arch Intern Med 1995;155:1595–600.
33. Advisory Council for the Elimination of Tuberculosis. Tuberculosis elimination revisited: Obstacles, opportunities, and a renewed commitment. MMWR Morb Mortal Wkly Rep 1999;48:1–13.
34. Londo R, Bjelland T, Girod C, Glasser M. Prenatal and postpartum Pap smears: Do we need both? Fam Pract Res J 1994;14:359–67.
35. Jain A, Higgins R, Boyle M. Management of low-grade squamous intraepithelial lesions during pregnancy. Am J Obstet Gynecol 1997;177:298–302.