14
Targeted Tuberculosis Contact Investigation Saves Money Without Sacrificing Health Maria Pisu, PhD, Assistant Professor, Medicine/Division of Preventive Medicine, University of Alabama at Birmingham. Joe Gerald, MD, PhD, Analyst, Department of Health System Information Services, University of Alabama at Birmingham. He is now with the Division of Community, Environment and Policy, Mel and Enid Zuckerman College of Public Health, University of Arizona. James E. Shamiyeh, MD, MSPH, Clinical Assistant Professor, University of Tennessee Hospital, Knoxville. William C. Bailey, MD, and Medical Director of the Lung Health Center, Professor of Medicine, and Eminent Scholar Chair in Pulmonary Diseases, University of Alabama at Birmingham. Lynn B. Gerald, PhD, MSPH Director of the Lung Health Center, and Professor of Medicine, Division of Pulmonary, Allergy, and Critical Care Medicine, University of Alabama at Birmingham. Dr Gerald is now with the Division of Health Promotion Sciences, Mel and Enid Zuckerman College of Public Health, University of Arizona. Abstract Background—Health departments require an efficient strategy to investigate individuals exposed to Mycobacterium tuberculosis. The contact priority model (CPM) uses a decision rule to minimize testing of low-risk contacts; however, its impact on costs and disease control is unknown. Methods—A cost-effectiveness analysis compared the CPM with the traditional concentric circle approach (CCA) in a simulated population of 1000 healthy, community-dwelling adults with a 10% background rate of latent tuberculosis (TB) infection. The analysis was conducted from the perspective of the Alabama Department of Public Health. Model inputs were derived from the literature and the Alabama Department of Public Health. Lifetime costs (2004 dollars) and outcomes were discounted 3 percent annually. Incremental cost-effectiveness ratios were used to compare the strategies. Results—Over the lifetime of 1000 simulated contacts, the CPM saved $45 000 but led to 0.5 additional TB cases and 0.24 fewer years of life. The CCA is more effective than the CPM, but it costs $92 934 to prevent one additional TB case and $185 920 to gain one additional life year. Conclusions—The CPM reduces costs with minimal loss of disease control and is a viable alternative to the CCA under most conditions. Copyright © 2009 Wolters Kluwer Health Corresponding Author: Maria Pisu, PhD, Medicine/Division of Preventive Medicine, University of Alabama at Birmingham, MT 628, 1530 3rd Ave South, Birmingham, AL 35294 ([email protected]). NIH Public Access Author Manuscript J Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19. Published in final edited form as: J Public Health Manag Pract. 2009 ; 15(4): 319–327. doi:10.1097/PHH.0b013e31819c3ef2. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

Targeted Tuberculosis Contact Investigation Saves Money Without Sacrificing Health

Embed Size (px)

Citation preview

Targeted Tuberculosis Contact Investigation Saves MoneyWithout Sacrificing Health

Maria Pisu, PhD,Assistant Professor, Medicine/Division of Preventive Medicine, University of Alabama atBirmingham.

Joe Gerald, MD, PhD,Analyst, Department of Health System Information Services, University of Alabama at Birmingham.He is now with the Division of Community, Environment and Policy, Mel and Enid ZuckermanCollege of Public Health, University of Arizona.

James E. Shamiyeh, MD, MSPH,Clinical Assistant Professor, University of Tennessee Hospital, Knoxville.

William C. Bailey, MD, andMedical Director of the Lung Health Center, Professor of Medicine, and Eminent Scholar Chair inPulmonary Diseases, University of Alabama at Birmingham.

Lynn B. Gerald, PhD, MSPHDirector of the Lung Health Center, and Professor of Medicine, Division of Pulmonary, Allergy, andCritical Care Medicine, University of Alabama at Birmingham. Dr Gerald is now with the Division ofHealth Promotion Sciences, Mel and Enid Zuckerman College of Public Health, University ofArizona.

AbstractBackground—Health departments require an efficient strategy to investigate individuals exposedto Mycobacterium tuberculosis. The contact priority model (CPM) uses a decision rule to minimizetesting of low-risk contacts; however, its impact on costs and disease control is unknown.

Methods—A cost-effectiveness analysis compared the CPM with the traditional concentric circleapproach (CCA) in a simulated population of 1000 healthy, community-dwelling adults with a 10%background rate of latent tuberculosis (TB) infection. The analysis was conducted from theperspective of the Alabama Department of Public Health. Model inputs were derived from theliterature and the Alabama Department of Public Health. Lifetime costs (2004 dollars) and outcomeswere discounted 3 percent annually. Incremental cost-effectiveness ratios were used to compare thestrategies.

Results—Over the lifetime of 1000 simulated contacts, the CPM saved $45 000 but led to 0.5additional TB cases and 0.24 fewer years of life. The CCA is more effective than the CPM, but itcosts $92 934 to prevent one additional TB case and $185 920 to gain one additional life year.

Conclusions—The CPM reduces costs with minimal loss of disease control and is a viablealternative to the CCA under most conditions.

Copyright © 2009 Wolters Kluwer HealthCorresponding Author: Maria Pisu, PhD, Medicine/Division of Preventive Medicine, University of Alabama at Birmingham, MT 628,1530 3rd Ave South, Birmingham, AL 35294 ([email protected]).

NIH Public AccessAuthor ManuscriptJ Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19.

Published in final edited form as:J Public Health Manag Pract. 2009 ; 15(4): 319–327. doi:10.1097/PHH.0b013e31819c3ef2.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Keywordscontact investigation; cost-effectiveness analysis; pulmonary tuberculosis; southeastern UnitedStates

Mycobacterium tuberculosis is transmitted from individuals with active disease (cases) toexposed individuals (contacts) via aerosolized respiratory droplets. Contacts may becomeinfected with tuberculosis (TB) following exposure and develop either active disease or latenttuberculosis infection (LTBI).1 Many health departments rely on tuberculin skin testing toidentify infections resulting from exposure. A positive skin test result without symptoms, signs,or a suggestive chest radiograph is presumptive of LTBI and an increased risk of developingactive disease in the future.2 This risk can be reduced by isoniazid (INH) prophylaxis.3However, not all exposures are equally likely to result in transmission, so the decision of whomto test can be difficult.4 Contact investigations are resource intensive but integral componentsof successful tuberculosis control programs; therefore, using an efficient contact investigationstrategy is a public health priority.

The concentric circle approach (CCA), a commonly used contact investigation strategy,categorizes contacts as high, medium, or low risk on the basis of an assessment of the likelihoodof transmission (eg, person, time, and place factors). Skin testing begins in high-risk contactsand is continued until the rate of positive skin test results equals the background prevalence ofLTBI in the community.4 Deciding when the background rate has been reached is difficult.This uncertainty has led the Centers for Disease Control and Prevention to recommend a morestandardized approach to contact investigations.5

The contact priority model (CPM) developed by the University of Alabama at Birmingham incollaboration with the Alabama Department of Public Health (ADPH) uses a decision rule toexplicitly categorize contacts as high risk requiring testing or low risk not requiring testing(Figure 1).6,7 It has been estimated that using the CPM in Alabama would lead to 20 percentfewer skin tests annually. It is also estimated that the rate of positive skin test results in low-risk contacts would be similar to the background rate in the general Alabama population (7%–10%).6,7 However, if these low-risk contacts were to be tested using the CCA, those testingpositive would be offered treatment. Because the CPM would potentially miss these contacts,some loss of disease control may occur. This article attempts to explicitly define this trade-offbetween fewer skin tests and loss of disease control.

MethodsA decision-analytic model was used to simulate the lifetime costs and outcomes using the CCAand the CPM in a hypothetical cohort of 1 000 otherwise healthy, community-dwelling adultsrepresentative of contacts investigated by the ADPH. Children younger than 15 years,individuals with HIV infection, individuals taking immunosuppressive medications, andinstitutionalized individuals were not included because they would be skin tested using theCPM. The background prevalence of LTBI in the simulated population was estimated to be 10percent.6,7 The life course of each contact was simulated until death or age 100. Because thestudy was a simulation and did not involve actual human subjects, it was granted an exemptionby the Institutional Review Board of the University of Alabama at Birmingham.

The analysis was performed from the perspective of the ADPH and included all costs associatedwith skin testing, follow-up of skin test–positive contacts, treatment of LTBI, and treatmentof active TB disease. Costs were adjusted to 2004 dollars using the medical care componentof the Consumer Price Index, and future costs and outcomes were discounted 3 percent

Pisu et al. Page 2

J Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

annually. Outcome measures included the number of active TB cases that developed in thesimulation and the number of life years attained. Results are presented using incremental cost-effectiveness ratios (ICERs). The ICER represents the average cost of preventing oneadditional TB case or achieving one additional life year using the CCA strategy as comparedwith the CPM strategy. Analyses were performed with TreeAge Pro 2004 (TreeAge Software,Inc, Williamstown, Massachusetts). The ICERs were calculated as follows:

Model specifications: Decision treeA decision tree was used to model various probabilities including the likelihood of performinga skin test, obtaining a positive result, completing INH prophylaxis, and developing hepatitis(Figure 2). The decision tree sorted contacts into one of three groups on the basis of thepredicted health state following skin testing: active TB disease, LTBI, or no TB infection. Onceassigned to a group, Markov models were used to simulate TB-related costs and outcomes ofindividual contacts until death or age 100.

In Figure 2, the first node on the left reflects the probability that a contact would be skin testedusing the CCA or the CPM. Skin testing was presumed to be 100 percent sensitive and specificfor LTBI because (1) the ADPH requires INH prophylaxis for all skin test–positive contacts,(2) no other testing (eg, T-SPOT.TB) is used to confirm negative skin test results in Alabama,and (3) skin testing has a high positive predictive value in high-risk contacts of smear-positivecases.8 All contacts are skin tested using the CCA strategy, but only 80 percent are tested usingthe CPM because the decision rule would identify 20 percent of contacts as low risk.6,7

The “Do Not Investigate” branch accounts for this fundamental difference between the twostrategies. Contacts assigned to this branch represent low-risk contacts not tested using theCPM who have a real, albeit small, risk of being skin test positive. The branch has three possibleoutcomes modeling what would have happened if these contacts were skin tested: skin testnegative and no TB, skin test positive with LTBI, or skin test positive with active disease.Based on prior ADPH data, 10 percent of the low-risk contacts would have a positive skin testresult.6,7 The majority of the skin test–positive contacts would have LTBI (99%), and a smallnumber (1%) would have active disease at the time of the skin test.4,5,9 Those with LTBI havea positive skin test result with the chest radiograph showing no abnormality and no signs orsymptoms of TB. The lifetime costs and outcomes of contacts with active disease and LTBIare simulated in Markov model 1 (M1) and Markov model 2 (M2), respectively (discussed inthe following text). The remaining 90 percent of low-risk contacts are predicted to be skin testnegative.6,7 Their lifetime costs and outcomes are simulated in Markov model 3 (M3) withoutfuture risk of developing TB due to their exposure.

The “Investigate” branch models contacts who are skin tested and has the same outcomes asdiscussed above: skin test negative and no TB, skin test positive with LTBI, or skin test positivewith active disease. In this branch, skin test–positive contacts come to the attention of the healthdepartment and are treated. The probability of having a positive skin test is slightly higherusing the CPM (26%) than the CCA (22%) because in the CPM, low-risk contacts are notincluded in the denominator.6,7 Again, the majority of skin test–positive contacts have LTBI(99%) and a small number have active disease (1%).4,5,9 As discussed above, contacts withactive disease, LTBI, and no TB are assigned to Markov models M1, M2, and M3, respectively(discussed in the following text).

Pisu et al. Page 3

J Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

On the basis of ADPH data, 83 percent of contacts with LTBI are expected to accept INHprophylaxis, which is near the recommended Centers for Disease Control and Prevention target.5 Of those who start INH and who do not develop hepatitis, 73 percent are expected to complete6 months of treatment. The risk of developing hepatitis increases with the age as follows: 0.1percent, 0.2 percent, and 0.3 percent risk for contacts younger than 50 years, 50 to 70 years,and older than 70 years, respectively.10–17 The risk of dying from hepatitis is 0.004 percentand does not vary with age.17

Model specifications: Markov modelsThree Markov models were used to simulate the lifetime TB-related costs and outcomes ofcontacts (Figure 3). In the figure, ovals represent health states and arrows represent transitionsamong health states. Contacts may stay in the same health state for more than one cycle.Transitions occur annually until death or age 100 on the basis of probabilities found in actuarialtables18 or in Table 1.

Markov model 1 simulates contacts with active disease at the time of skin testing. The risk ofdying from active disease increases with age such that the risk is 1 percent, 5 percent, and 10percent for contacts younger than 50 years, 50 to 70 years, and older than 70 years, respectively.19 After the first year, survivors transition to the post-TB well state where their life course issimulated using life tables without risk of sequela or developing TB again.10 A mortalitypenalty (2.4 times increase in the risk of dying from TB in the first year) is levied in M1 of the“Do Not Investigate” branch to account for higher acuity of illness resulting from delayedrecognition and treatment of active disease.20 A transmission penalty of 1.035 is also leviedupon each contact with active disease to represent transmission of TB to 0.035 individualsoutside of the simulated cohort. This penalty is based on an average of 10 contacts per case,with 20 percent of contacts becoming skin test positive (n = 2).2,4,6 Approximately 1 percentof skin test–positive contacts will immediately develop active disease4,5,9 (n = 0.02) and anadditional 0.74 percent (n = 0.015) will develop active disease within a year.21

Markov model 2 simulates contacts with LTBI at the time of skin testing. The probability ofdeveloping active disease following exposure is 0.74 percent per year for years 1–2, 0.31percent per year for years 3–5, 0.16 percent per year for years 6–7, and then declinesexponentially.21,22 Completing a full or partial course of INH prophylaxis reduces the risk ofactive disease by 70 percent and 16 percent, respectively.3,10,19,21,22 As discussed above,contacts with LTBI who develop active disease have an age-adjusted risk of death from TBbased on their age at the time of reactivation (age at start + elapsed simulation years). Nomortality penalty is applied to reactivation disease. The transmission penalty of 1.035representing transmission outside of the simulated cohort is applied to all cases of reactivationdisease and not just to those in the “Do Not Investigate” branch because the simulation accountsfor a differential reactivation rate with and without prophylaxis. Applying the penalty to onlythose contacts in the “Do Not Investigate” branch would overestimate transmission outside ofthe cohort. Survivors of reactivation disease transition to the post-TB well state where theirremaining life course is simulated without risk of sequela or developing TB again.

Markov model 3 simulates the life course of skin test–negative contacts using life tables withoutfuture risk of TB due to their exposure.

Model specifications: CostsThe costs of contact investigations were provided by the ADPH (Table 1). Each contact skintested was estimated to generate $247 in costs including wages, transportation, and actual skintesting. Follow-up of skin test–positive contacts was estimated to cost an additional $82including a chest radiograph ($21) and baseline liver function test ($56). A full course (6

Pisu et al. Page 4

J Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

months) and a partial course of INH treatment were estimated to cost $13.80 and $6.90,respectively. The treatment of active TB disease was estimated to cost $12 160, whichrepresented a weighted average of inpatient ($14 777) and outpatient ($4 774) costs.13,20,23–25

Sensitivity analysisUnivariate sensitivity analyses were performed for each modeled input to evaluate the impactof uncertainty surrounding the true input value (Table 1). In sensitivity analysis, each input ismodeled at its lowest and highest value while all other inputs are held constant. This helpsdetermine how influential changes in that variable are to the calculation of the ICER. Thesimulation was run with a mean population age of 35, 45, and 70 to examine the relationshipbetween age, TB mortality, and INH hepatitis. Sensitivity analyses were not performed for theprobability of skin testing or the probability of a positive skin test result because these estimatesare highly correlated and intrinsic to the original validation data obtained from the ADPHpopulation of contacts. Adjusting one value would have an unpredictable impact on the other.Because of this uncertainty, we are unable to model changes in either.

Best-and worst-case scenarios were also analyzed. In the best-case (favoring the CPM) andworst-case (favoring the CCA) scenarios, selected inputs were estimated at their maximum orminimum value depending on their predicted effect (Table 1). In the best-case scenario, theprobability of developing active disease in contacts with LTBI was low, INH treatment wasnot very effective, INH treatment was associated with a high risk of hepatitis, and the cost ofcontact investigations was high. With the inputs purposely biased for or against the CPM, it ispossible to estimate the robustness of the overall simulation.

ResultsThe CCA is more effective but more costly than the CPM. The CCA prevents one additionalcase of active disease at a cost of $92 934 and gains one additional life year at a cost of $185920 (Table 2). The total costs per 1 000 contacts are $339 896 and $294 596 for the CCA andthe CPM, respectively. For every 1 000 contacts, the CPM would yield approximately $51 000in immediate savings from fewer skin tests, which would be offset by approximately $6 000in additional costs for the treatment of active disease. Using the CCA, 5.73 cases of activedisease are expected to arise over the lifetime of the 1 000 contacts. Using the CPM, 6.22 casesare expected to arise leading to an additional 0.5 cases of active disease and 0.24 fewer yearsof life over the lifetime of 1 000 contacts.

Two inputs, the cost of investigating contacts and the likelihood of reactivation disease in the2 years following exposure, were associated with relatively large effects on the ICER duringsensitivity analysis (Table 3). The CCA strategy was more cost-effective if reactivation diseaseduring the first 2 years following exposure was higher than estimated (3% instead of 0.74%).Under conditions of highest reactivation risk, the CCA costs $35 804 to prevent one additionalcase of active disease and $106 811 to gain one additional life year. If the cost to investigatea contact was higher than estimated ($500 instead of $250), then the CCA was less cost-effective. Under conditions of highest cost, the CCA costs $196 974 to prevent one additionalcase of active disease and $394 045 to gain one additional life year. The model was relativelyinsensitive to changes in the other inputs.

Sensitivity analysis of the relationship between age, TB mortality, and INH hepatitis revealedthat the CCA was more cost-effective in older cohorts when the outcome of interest is cost perlife year gained. This is due to the higher risk of death from active disease in older individuals.However, the CCA is less cost-effective in older cohorts when the outcome of interest isprevented cases of active disease. This is because of the higher risk of INH hepatitis in older

Pisu et al. Page 5

J Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

individuals. Sensitivity analysis of the transmission and mortality penalties showed that neitherhad an appreciable impact on the ICER for prevented cases or life years.

In the best-case scenario, the CPM was estimated to save $97 154 annually, miss approximately0.3 cases of active disease, and result in 0.2 less years of life over the lifetime of 1 000 contacts.Under the best-case scenario, the CCA costs $313 470 to prevent one additional case of activedisease and $523 664 to gain one additional life year. In the worst-case scenario, the CPM wasmore costly and less effective than the CCA.

DiscussionThe CCA is more effective but more costly than the CPM. Given savings of $45 000 per 1 000contacts investigated, the ADPH could save $135 000 annually using the CPM because theADPH investigates approximately 3 000 contacts per year. These savings would be realizedimmediately because they result from performing fewer skin tests. The trade-off in diseasecontrol would be one to two additional cases of active disease sometime during the lifetime ofthose 3 000 contacts. This includes active disease that would develop one generation outsideof the cohort. The trade-off in mortality would be approximately 263 fewer days of life overthe lifetime of those 3 000 contacts.

If resources were unlimited, spending approximately $93 000 to prevent each additional caseof active disease and $186 000 to gain one additional year of life using the CCA would not bea concern. Unfortunately, resources are constrained. For health departments struggling withdiminishing resources, the CPM may offer a viable strategy to realize substantial savings whilemaintaining a similar level of disease control. This article is intended to help decision makersdetermine under what conditions adopting the CPM would be appropriate.

The CPM is designed to evaluate otherwise healthy, community-dwelling adults who havebeen exposed to M tuberculosis. The CPM has not been evaluated in children, institutionalizedindividuals, those with HIV infection, or those taking immunosuppressive drugs. The averageage of our simulated population was 45 years; however, sensitivity analysis demonstrated thatthe CPM yielded similar results in older and younger populations. Approximately 10 percentof the low-risk contacts identified by the CPM would have a positive skin test result if tested.This rate is consistent with the estimated background prevalence of LTBI (7%–10%)6,26 inAlabama, suggesting that the CPM recommends skin testing until the rate is near thebackground rate in the community.5

Previous research has indicated that there are considerable differences among TB fieldworkersin the interpretation of definitions used in the CCA to determine a contact’s risk (eg, size andventilation of the exposure environment).27,28 These differences lead to considerable varianceamong fieldworkers about what constitutes high, medium, or low risk.27 Even workers withyears of experience in TB classified a contact’s risk of disease differently when using the CCA.27 The greater standardization achieved using the CPM decision rule allows fieldworkers tomore efficiently identify high-risk contacts.

Sensitivity analysis demonstrated that our results are most influenced by the cost ofinvestigating contacts and the likelihood of reactivation disease in the first 2 years followinginfection. Our estimated cost of investigating a contact ($250) is slightly lower than otherestimates (~$300).23,24 If the investigation costs are higher than our estimated costs, the CPMwould be even more cost-effective. Our estimated reactivation rate in the first year (0.74%) issubstantially higher than would be expected if the date of the skin test conversion wereunknown (0.07%)29 and slightly lower than that found at the time of the contact investigation(1%).4,30 We believe that our estimate is conservative; however, if it was higher, the CCA

Pisu et al. Page 6

J Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

would be more cost-effective. Other than these two inputs, our model was robust to changesin our modeled inputs.

Our study has several limitations and our results should be applied cautiously to populationsthat are substantially different from the one we describe, particularly those with high endemicrates of multidrug-resistant TB and/or HIV. In 2003, Alabama began using a four-drug regimenas initial or empiric therapy, although INH resistance was less than 4 percent (the usualthreshold for using four drugs instead of three).26 Thus, the costs for the treatment of activedisease could be more expensive than our estimated costs; however, sensitivity analysisdemonstrated that changes in this cost had little impact on the results. We also did not accountfor HIV infection in our population of contacts, although HIV-positive patients are much morelikely than non–HIV-positive patients to develop active disease following LTBI.5 While theCPM calls for testing of all known HIV-positive persons, it is possible that some of the low-risk contacts who are not investigated using the CPM strategy may have undiagnosed HIVinfection and LTBI. Thus, our results may be biased against the CCA, particularly inpopulations with a high endemic prevalence of HIV.

The current Centers for Disease Control and Prevention guidelines recommend that contactinvestigations be undertaken in high-risk contacts as determined by an objective process.5 Thisanalysis demonstrated that one such approach, the CPM, would be cost saving withoutsacrificing disease control or population health under most conditions. For health departmentsfaced with budget cuts, the use of the CPM for the evaluation of otherwise healthy, community-dwelling adult contact is a viable alternative to the CCA.

REFERENCES1. Comstock, GW. Epidemiology of tuberculosis. In: Reichman, L.; Hershfield, HE., editors.

Tuberculosis: A Comprehensive International Approach. Vol. 2nd ed.. New York, NY: Marcel Dekker;2000. p. 129-156.

2. Reichler MR, Reves R, Bur S, et al. Evaluation of investigations conducted to detect and preventtransmission of Tuberculosis. JAMA 2002;287:991–995. [PubMed: 11866646]

3. Cohn, DL.; El-Sadr, WM. Treatment of latent tuberculosis infection. In: Reichman, L.; Hershfield,HE., editors. Tuberculosis: A Comprehensive International Approach. Vol. 2nd ed.. New York, NY:Marcel Dekker; 2000. p. 471-502.

4. Etkind, S.; Veen, J. Contact follow-up in high- and low- prevalence countries. In: Reichman, L.;Hershfield, HE., editors. Tuberculosis: A Comprehensive International Approach. Vol. 2nd ed.. NewYork, NY: Marcel Dekker; 2000. p. 377-399.

5. National Tuberculosis Controllers Association; Centers for Disease Control and Prevention (CDC).Guidelines for the investigation of contacts of persons with infectious tuberculosis. Recommendationsfrom the National Tuberculosis Controllers Association and CDC. MMWR Recomm Rep 2005;54:1–47.

6. Gerald LB, Tang S, Bruce F, et al. A decision tree for tuberculosis contact investigation. Am J RespirCrit Care Med 2002;166:1122–1127. [PubMed: 12379558]

7. Bailey WC, Gerald LB, Kimerling ME, et al. Predictive model to identify positive tuberculosis skintest results during contact investigations. JAMA 2002;287:996–1002. [PubMed: 11866647]

8. Menzies, R. Tuberculin skin testing. In: Reichman, L.; Hershfield, HE., editors. Tuberculosis: AComprehensive International Approach. Vol. 2nd ed.. New York, NY: Marcel Dekker; 2000. p.279-322.

9. Jereb J, Etkind SC, Joglar OT, Moore M, Taylor Z. Tuberculosis contact investigations: outcomes inselected areas of the United States. Int J Tuberc Lung Dis 2003;7(12):S384–S390. [PubMed:14677827]

10. Tsevat J, Taylor WC, Wong JB, Pauker SG. Isoniazid for the tuberculin reactor: take it or leave it.Am Rev Respir Dis 1988;137:215–220. [PubMed: 3276255]

Pisu et al. Page 7

J Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

11. Rose DN, Schechter CB, Silver AL. The age threshold for isoniazid chemoprophylaxis. A decisionanalysis for low-risk tuberculin reactors. JAMA 1986;256:2709–2713. [PubMed: 3773178]

12. Kopanoff DE, Snider DE Jr, Caras GJ. Isoniazid-related hepatitis: a U.S. Public Health Servicecooperative surveillance study. Am Rev Respir Dis 1978;117:991–1001. [PubMed: 666111]

13. Salpeter SR, Sanders GD, Salpeter EE, Owens DK. Monitored isoniazid prophylaxis for low-risktuberculin reactors older than 35 years of age: a risk-benefit and cost-effectiveness analysis. AnnIntern Med 1997;127:1051–1061. [PubMed: 9412307]

14. Rose DN, Schechter CB, Fahs MC, Silver AL. Tuberculosis prevention: cost-effectiveness analysisof isoniazid chemoprophylaxis. Am J Prev Med 1988;4:102–109. [PubMed: 3134928]

15. Jasmer RM, Snyder DC, Saukkonen JJ, et al. Short-course rifampin and pyrazinamide compared withisoniazid for latent tuberculosis infection: a cost-effectiveness analysis based on a multicenter clinicaltrial. Clin Infect Dis 2004;38:363–369. [PubMed: 14727206]

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;281:1014–1018. [PubMed:10086436]

17. Saukkonen JJ, Cohn DL, Jasmer RM, et al. An official ATS statement: hepatotoxicity ofantituberculosis therapy. Am J Respir Crit Care Med 2006;174:935–952. [PubMed: 17021358]

18. Arias E. United States life tables, 2001. Natl Vital Stat Rep 2004;52:1–38. [PubMed: 15008552]19. American Thoracic Society. Medical Section of the American Lung Association: treatment of

tuberculosis and tuberculosis infection in adults and children. Am Rev Respir Dis 1986;134:355–363. [PubMed: 3527010]

20. Hansel NN, Merriman B, Haponik EF, Diette GB. Hospitalizations for tuberculosis in the UnitedStates in 2000: predictors of in-hospital mortality. Chest 2004;126:1079–1086. [PubMed: 15486367]

21. Ferebee SH. Controlled chemoprophylaxis trials in tuberculosis. A general review. Bibl Tuberc1970;26:28–106. [PubMed: 4903501]

22. Taylor Z. The cost-effectiveness of screening for latent tuberculosis infection. Int J Tuberc Lung Dis2000;4:S127–S133. [PubMed: 11144542]

23. Brown RE, Miller B, Taylor WR, et al. Health-care expenditures for tuberculosis in the United States.Arch Intern Med 1995;155:1595–1600. [PubMed: 7618981]

24. Menzies, D.; Oxlade, O.; Lewis, M. Costs for Tuberculosis Care in Canada. Ottawa: Public HealthAgency of Canada; 2006 [Accessed November 1, 2008]. Published 2006http://www.phac-aspc.gc.ca/tbpc-latb/costtb/index-eng.php

25. Taylor Z, Marks SM, Rios Burrows NM, Weis SE, Stricof RL, Miller B. Causes and costs ofhospitalization of tuberculosis patients in the United States. Int J Tuberc Lung Dis 2000;4:931–939.[PubMed: 11055760]

26. University of Alabama and Alabama Department of Public Health. Tuberculosis Primer. Birmingham:University of Alabama/ Montgomery. Alabama Department of Public Health; 2004.

27. Gerald LB, Bruce F, Brooks CM, et al. Standardizing contact investigation protocols. Int JTubercLung Dis 2003;7(12):S369–S374. [PubMed: 14677825]

28. Sprinson JE, Flood J, Fan CS, et al. Evaluation of tuberculosis contact investigations in California.Int J Tuberc Lung Dis 2003;7(12):S363–S368. [PubMed: 14677824]

29. Salpeter SR, Salpeter EE. Screening and treatment of latent tuberculosis among health care workersat low, moderate, and high risk for tuberculosis exposure: a cost-effectiveness analysis. Infect ControlHosp Epidemiol 2004;25:1056–1061. [PubMed: 15636292]

30. Marks SM, Taylor Z, Qualls NL, Shrestha-Kuwahara RJ, Wilce MA, Nguyen CH. Outcomes ofcontact investigations of infectious tuberculosis patients. Am J Respir Crit Care Med 2000;162:2033–2038. [PubMed: 11112109]

Pisu et al. Page 8

J Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

FIGURE 1.Contact Priority Model Decision Rule for Tuberculin Skin Testing (TST) of Healthy,Community-Dwelling Adults Exposed to Mycobacterium Tuberculosis

Pisu et al. Page 9

J Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

FIGURE 2.Decision Tree for the Evaluation of the Concentric Circle Approach (CCA) and the ContactPriority Model (CPM)aaTST indicates tuberculin skin testing; LTBI, latent tuberculosis infection; INH, isoniazid; M1,Markov model 1; M2, Markov model 2; and M3, Markov model 3.

Pisu et al. Page 10

J Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

FIGURE 3.Markov Models for Contacts With Active Disease (Model 1, M1), Latent TuberculosisInfection (LTBI) (Model 2, M2), and Negative Skin Test Results (Model 3, M3) FollowingCompletion of Skin TestingaaTB indicates tuberculosis.

Pisu et al. Page 11

J Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Pisu et al. Page 12TA

BLE

1B

ase

case

inpu

ts, r

ange

s, be

st- a

nd w

orst

-cas

e es

timat

es o

f pro

babi

litie

s, an

d co

sts,

with

lite

ratu

re re

fere

nces

Inpu

tB

ase

case

Ran

geB

est c

ase

Wor

st c

ase

Ref

eren

ces

Prob

abili

ty o

fB

eing

skin

test

ed

CC

A10

0%…

……

AD

PH, 6

C

PM80

%…

……

AD

PH, 6

Posi

tive

skin

test

resu

lt

CC

A22

%

CPM

, inv

estig

ated

26%

……

…A

DPH

, 6

CPM

, not

inve

stig

ated

10%

Acc

epta

nce

of IN

H83

%65

%–9

3%65

%95

%A

DPH

Dev

elop

men

t of h

epat

itis

By

age

0%–1

%0.

5%0%

10– 17

Com

plet

ion

of IN

H73

%50

%–9

0%50

%90

%A

DPH

INH

eff

ectiv

enes

s

Full

cour

se70

%30

%–9

0%30

%90

%3, 10

, 19, 22

, 23

Parti

al c

ours

e16

%10

%–3

0%10

%30

%3, 10

, 19, 22

, 23D

eath

from

hep

atiti

s0.

004%

0%–0

.003

%17

Act

ive

TB in

con

tact

s1.

0%…

9D

evel

opm

ent o

f TB

if la

tent

TB

infe

ctio

n

1–2

y af

ter e

xpos

ure

0.74

%

3–5

y af

ter e

xpos

ure

0.31

%0%

–2.5

%0.

4%1.

4%22

, 23

6–7

y af

ter e

xpos

ure

0.16

%D

eath

from

TB

35

yea

rs o

ld1.

0%0%

–2%

50

yea

rs o

ld5.

0%0%

–9%

……

19

70 y

ears

old

10.0

%0%

–30%

Cos

ts (i

n 20

04 d

olla

rs) o

fC

onta

ct in

vest

igat

ion

$247

$0–$

500

$500

$0A

DPH

Follo

w-u

p fo

r tub

ercu

lin sk

in te

stin

g +

cont

act

$82

$0–$

500

……

AD

PHIN

H p

roph

ylax

is…

Com

plet

e$1

3.8

……

…A

DPH

Pa

rtial

$6.9

……

…A

DPH

INH

hep

atiti

s…

Non

fata

l$1

05…

……

13

Fata

l$1

0 09

6…

……

13A

ctiv

e TB

$12

162

$0–$

40 0

00…

…13

, 20, 24

, 25, 26

O

utpa

tient

car

e$4

774

……

…13

, 20, 24

, 25, 26

In

patie

nt c

are

$14

777

……

…13

, 20, 24

, 25, 26

Mor

talit

y pe

nalty

2.4

1–3

13

20Tr

ansm

issi

on p

enal

ty1.

035

1–1.

071

1.07

2, 4– 6, 9, 21

Abb

revi

atio

ns: A

DPH

, Ala

bam

a D

epar

tmen

t of P

ublic

Hea

lth; C

CA

, con

cent

ric c

ircle

app

roac

h; C

PM, c

onta

ct p

riorit

y m

odel

; IN

H, i

soni

azid

; TB

, tub

ercu

losi

s.

J Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Pisu et al. Page 13

TABLE 2Costs (2004 Dollars), outcomes, and incremental cost-effectiveness ratios (ICERs) comparing the concentric circleapproach (CCA) and the contact priority model (CPM) in a simulated population of 1 000 contacts

CCA CPM Difference (CCA – CPM) ICER

Cost (2004 dollars) Investigation/latent TB infection treatment $270 156 $218 924 $51 231 … Active disease treatment $69 740 $75 671 −$5 930 … Total $339 896 $294 596 $45 300 …Outcomes Active TB cases 5.73 6.22 −0.49 $92 934a Life years attained 21 084.2 21 083.9 0.24 $185 920b

Abbreviation: TB, tuberculosis.

aCost to prevent one additional TB case using the CCA rather than the CPM.

bCost to save one additional life year using the CCA rather than the CPM.

J Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Pisu et al. Page 14

TABLE 3Sensitivity analysis for the cost per prevented case of active disease and cost per life year saved using the concentriccircle approach versus the contact priority model

Input RangeCost per additional

TB case avertedCost per additional

life year saved

Baseline result Cohort age 35–70 $91 585–$104 215 $226 861–$57 356Probabilities INH acceptance 65%–95% $100 892–$88 227 $198 703–$178 167 INH hepatitis 0%–1% $92 929–$92 951 $185 899–$186 004 Complete INH 50%–90% $101 227–$87 546 $199 263–$177 018 INH effectiveness (full) 30%–90% $114 911–$84 508 $220 433–$171 847 Active TB (years 1–2) 0%–3% $162 031–$35 804 $231 797–$106 811Costs Investigation $0–$500 <$0–$196 974 <$0–$394 058 Latent TB infection treatment $0–$500 $92 729–$100 113 $185 511–$200 282 Active disease treatment $0–$40 000 $105 101–$65 086 $210 261–$130 210 Discount rate 0%–6% $80 541–$103 616 $96 640–$311 702Penalties Transmission 1–1.007 $92 200–$93 678 $185 757–$186 084 Mortality 1–3 $92 914–$92 942 $246 847–$168 135

Abbreviations: INH, isoniazid; TB, tuberculosis.

J Public Health Manag Pract. Author manuscript; available in PMC 2009 October 19.