Cost-Effectiveness Analysis of the Lyme Disease Vaccine

Transcription

1 ARTHRITIS & RHEUMATISM Vol. 46, No. 6, June 2002, pp DOI /art , American College of Rheumatology Cost-Effectiveness Analysis of the Lyme Disease Vaccine Elizabeth C. Hsia, James B. Chung, J. Sanford Schwartz, and Daniel A. Albert Objective. A vaccine for Lyme disease was approved in 1998 for use in the US. Given the high cost of the vaccine, the low risk of Lyme disease in many areas, and the largely curable nature of the disease, the cost-effectiveness of the vaccine in various risk groups is uncertain. This study was undertaken to examine the cost-effectiveness of the Lyme disease vaccine and the factors that influence its cost-effectiveness. Methods. We constructed a Markov decision-analysis model to evaluate the clinical effectiveness and cost-effectiveness of the Lyme disease vaccine in populations at various levels of risk for the disease. The probabilities of clinical events and costs were estimated from reports in the literature. Sensitivity analyses assessed the impact of potential variations of parameters on model results. Results. At the average national incidence of Lyme disease (0.0067%), the incremental costeffectiveness of vaccination was $1,600,000 per case averted when a yearly booster was given for 10 years after the standard initial vaccination regimen of 3 inoculations at 0, 1, and 12 months. For populations with an annual Lyme disease incidence of 1% (the incidence in several well-defined geographical areas of the US), the incremental cost-effectiveness was $9,900 per case averted. Disease incidence had to exceed 10% before vaccination with yearly boosters became both more effective and more cost saving than no vaccination. Presented in part at the 63rd Annual Scientific Meeting of the American College of Rheumatology, Boston, MA, November Dr. Hsia s work was supported by a Physician Scientist Development Award from the Arthritis Foundation. Elizabeth C. Hsia, MD, James B. Chung, MD, J. Sanford Schwartz, MD, Daniel A. Albert, MD: University of Pennsylvania, Philadelphia. Address correspondence and reprint requests to Elizabeth C. Hsia, MD, Division of Rheumatology, University of Pennsylvania, Maloney Building, Suite 504, 3600 Spruce Street, Philadelphia, PA hsia@mail.med.upenn.edu. Submitted for publication March 26, 2001; accepted in revised form January 25, Conclusion. The Lyme disease vaccine is costeffective only for individuals who live in areas where Lyme disease is endemic and who are frequently exposed to ticks. Lyme disease is the most common tick-borne illness in the US, with 16,802 cases reported in 1998 (1). It is believed that the true incidence of Lyme disease is 6 12 times the reported number of cases, and the incidence of the disease in humans in the US appears to be increasing (2). Wide geographic variation is displayed in the occurrence of Lyme disease, with the Northeast, Mid-Atlantic, and North Central regions accounting for 90% of reported cases (Figure 1 and Table 1) (3). In December 1998, the Food and Drug Administration (FDA) approved a vaccine for Lyme disease. Consisting of recombinant Borrelia burgdorferi outer surface lipoprotein A (OspA), the vaccine induces human antibodies that are taken up by the Ixodes tick during a blood meal and that kill the spirochete in the tick vector (4,5). The vaccine consists of an initial 3-inoculation regimen administered at 0, 1, and 12 months. Clinical trials demonstrated a 1-year reduction of symptomatic Lyme infection of 49% after 2 injections and of 76% after the third injection (6). There were no cases of asymptomatic infection after the initial 3 injections. There are no firm data on the necessary frequency of boosters with the approved vaccine, but studies suggest that boosters every 1 3 years will probably be needed for continuing adequate protection (7,8). The vaccine is well tolerated, and no serious side effects have been reported to date (6,9). While Lyme disease is not contagious between people and is usually curable with well-tolerated and inexpensive antibiotics, complications can occur, primarily from neurologic, musculoskeletal, and cardiac sequelae. A recent cost-of-illness study estimated a 5-year US incidence of 56,000 cases of Lyme disease, resulting in national expenditures of $2.5 billion (10). The Lyme vaccine costs $ for the initial vaccination series and $ annually for boosters. We performed 1651

2 1652 HSIA ET AL Figure 1. Geographic distribution of reported cases of Lyme disease in the US (12). Each dot represents 1 case, placed randomly within the county of residence. The total number of cases is 16,802. a decision analysis to evaluate the clinical effectiveness and cost-effectiveness of the Lyme disease vaccine for the US population at various levels of disease risk to inform clinical practice and health policy. MATERIALS AND METHODS Model overview. A Markov state transition decisionanalysis model (modified from a model previously described by Magid et al [11]) comparing the clinical effectiveness and cost-effectiveness of a vaccine strategy with those of a novaccine strategy was constructed using DATA 3.5 software (TreeAge, Williamstown, MA). A simplified version of our model is displayed in Figure 2. A hypothetical cohort of year-old persons at risk for Lyme disease (the population in which Lyme vaccine clinical trials were conducted) was followed up for 10 years. Individuals moved between health states at yearly intervals. With each yearly cycle of the model, individuals either contracted Lyme disease or remained free of infection. Clinical and economic outcomes were aggregated over a 10-year period, with benefits and costs discounted for time effects, and Table 1. US states and counties with the highest incidence rates of Lyme disease (12,22) Highest statewide incidence Highest county-wide incidence State Incidence* County, state Incidence* Connecticut 100 Nantucket, MA 1,510 Rhode Island 76 Dutchess, NY 587 New York 30 Hunterdon, NJ 460 New Jersey 24 Washington, RI 454 Pennsylvania 23 Columbia, NY 383 * Cases per 100,000. Figure 2. Simplified and truncated version of the Markov decisionanalysis model comparing vaccination with no vaccination for one cycle. Only the vaccine strategy subtree is illustrated. The square represents a decision node, circles represent chance nodes, and triangles represent terminal nodes. Asterisks denote continuation of tree branches as seen after the node labeled Late Lyme ;[ ] denotes continuation of tree branches that are not shown. Late Lyme IV represents late-stage Lyme disease manifestations treated with intravenous antibiotics. If persons with facial palsy, first-degree atrioventricular (AV) block, and arthritis did not respond to oral antibiotic treatment, they would be treated with intravenous antibiotics. Individuals who developed Late Lyme could experience major or minor reactions to antibiotic therapy (not shown). Persistent sequelae were subdivided into neurologic, cardiac, and arthritic sequelae (not shown). one-way and multi-way sensitivity analyses were conducted for key variables (Table 2). Data and assumptions. Incidence of Lyme disease. Estimates of Lyme disease incidence were obtained from the annual number of cases reported to the Centers for Disease Control and Prevention (CDC) by geographic area of the US (12). An average incidence across all states of 6.7 cases per 100,000, as estimated by the CDC in 1998, was used in the base-case analyses to represent the national incidence of Lyme disease (1). An incidence of 1% was used to represent the risk of Lyme disease in endemic areas. Analyses were repeated for the entire range of Lyme disease incidences (0 10%) ever reported for various geographic areas. Vaccine administration and efficacy. Estimates of vaccine administration and efficacy were based on the study by the Lyme Disease Vaccine Study Group (6). The vaccine was assumed to be administered as 3 injections at intervals of 0, 1, and 12 months (6). Vaccination was assumed to reduce clinical Lyme disease by 49% in the first year and by 76% thereafter. In light of data indicating that Lyme antibody titers decline over time (8), the impact of vaccine booster doses administered at 1- and 3-year intervals was assessed. In sensitivity analyses, vaccine efficacy was varied from 50% to 100%. Natural history of Lyme disease and sequelae. Estimated probabilities of clinical presentation of Lyme disease were

3 COST-EFFECTIVENESS OF LYME DISEASE VACCINE 1653 Table 2. Estimated base-case and range probabilities for model parameters* Variable Probability estimate, base-case (range) Reference Incidence of Lyme disease Average national ( ) 3 Endemic areas 0.01 ( ) 3 Vaccine efficacy, % First year 49 (50 95) 6 Second year 76 (50 95) 6 Efficacy of oral antibiotics For early infection 0.95 ( ) 11 For isolated facial palsy and first-degree AV block 0.95 ( ) 11 For arthritis 0.70 ( ) 11 Efficacy of IV antibiotics 0.90 ( ) 11 For arthritis 0.50 ( ) 11 Presenting with early Lyme disease 0.85 ( ) 6, 9, 10, 11 Neurologic sequelae 0.17 ( ) 11, 13 Isolated facial palsy 0.05 ( ) 11, 13 Meningitis, encephalopathy, radiculopathy, cranial neuritis 0.12 ( ) 11, 13 Cardiac sequelae 0.06 ( ) 11, 13 First-degree AV block 0.04 ( ) 11 High-grade AV block 0.02 ( ) 11 Arthritis sequelae 0.60 ( ) 11, 13 Adverse reactions to oral doxycycline Minor 0.04 ( ) 11, 13, 19 Major ( ) 11, 13, 19 Adverse reactions to IV ceftriaxone Minor 0.06 ( ) 11, 13, 19 Major ( ) 11, 13, 19 *AV atrioventricular; IV intravenous. Applies to patients in whom oral therapy has already been unsuccessful. Values indicate the probability of presenting with given symptoms if patients did not present with and were not treated successfully for early Lyme disease. obtained from reports in the literature (6,9 11,13). Patients who developed Lyme disease presented in either early or late stages. Early-stage Lyme disease manifested as erythema migrans with possible virus-like symptoms. Five percent of early Lyme disease patients did not respond to antibiotic treatment and progressed to late-stage Lyme disease. Approximately 15% of patients initially presented with late-stage Lyme disease, which included early disseminated Lyme disease, manifested by certain neurologic symptoms (facial palsy and meningitis) and cardiac sequelae (variable degrees of atrioventricular block), as well as late Lyme disease, manifested by encephalopathy, neuropathy, and musculoskeletal (chronic inflammatory arthritis) sequelae. Late-stage Lyme disease manifestations were usually cured with antibiotics; however, some cases resulted in persistent neurologic, cardiac, and arthritic sequelae. Reports in the literature were used to estimate the risk of developing various sequelae after treatment failure (11,13). In the model, a proportion (17%) of persons could be asymptomatic despite infection; however, these persons had the same risk of symptomatic recurrent infection as those who never acquired disease or who were successfully treated (14,15). Treatment of Lyme disease. As recommended by the Infectious Diseases Society of America (16), early Lyme disease, corresponding to erythema migrans with or without virus-like symptoms, was treated with a 14-day course of oral doxycycline (100 mg twice a day), while patients with facial palsy and first-degree atrioventricular block were treated with a 21-day course of oral doxycycline, those with arthritis were treated with a 28-day course of oral doxycycline, and other late-stage manifestations were treated with intravenous ceftriaxone (2 gm/day). If adverse reactions developed, oral doxycycline was switched to amoxicillin (500 mg 3 times a day), or ceftriaxone was changed to penicillin G (20 million units/day). If oral antibiotics failed, intravenous antibiotic therapy was administered. Adverse reactions (Table 2) (11,13,19). Major antibiotic side effects included pancytopenia, renal impairment, fever and chills, and anaphylactoid reaction. Minor antibiotic side effects included diarrhea, upper gastrointestinal symptoms, and rash. Monte Carlo simulations were performed using a hypothetical cohort of 10,000 persons and a 1% incidence of Lyme disease to estimate the number of antibiotic reactions in the vaccine and no-vaccine strategies. Costs (Table 3). For vaccine and administration costs, we used a base-case estimate of $75 per booster (average wholesale price [AWP] $61.25 [17]), an average estimate of charges of several local-area medical practices. Costs of doxycycline, ceftriaxone, amoxicillin, and penicillin G were calculated by using the 1999 AWP of the medicines and adding the costs of medical care and administration obtained from previous studies (11,13,18). The costs of both major and minor adverse drug reactions and sequelae of Lyme disease were based on the estimates in reported studies (11,13,19). All costs were adjusted to 1999 dollars using the non seasonally ad-

4 1654 HSIA ET AL Table 3. Estimated costs of model parameters* Cost variable Cost (range), dollars Reference One vaccine booster 75 (50 100) 17 administration Initial vaccination (3 225 ( ) booster cost of $75) Doxycycline 14 days 107 (5 200) 11, days 110 (5 200) 11, days 112 (10 200) 11, 18 Amoxicillin 14 days 15 (5 100) days 23 (5 75) days 31 (10 200) 18 Ceftriaxone for 3 weeks 4,000 (1,500 6,500) 13, 18, 19 Penicillin G for 3 weeks 3,200 (500 5,000) 13, 18, 19 Minor oral antibiotic reaction 315 (5 2,000) 13, 19 Major oral antibiotic reaction 2,515 (200 5,000) 13, 19 Minor IV antibiotic reaction 315 (5 2,000) 13, 19 Major IV antibiotic reaction 2,515 (200 5,000) 13, 19 Neurologic sequelae 10,290 (3,000 11, 13 16,000) Cardiac sequelae 11,520 (3,000 11, 13 17,500) Rheumatologic sequelae 3,817 (1,000 6,000) 11, 13 Lyme disease vaccine falls below $50,000 per case averted as the incidence of Lyme disease increases to 0.2% (2 cases per 1,000) when yearly boosters are given and at a disease incidence of 0.1% (1 case per 1,000) when vaccine booster doses are administered at 3-year intervals. With yearly boosters, the incremental cost-effectiveness of the vaccine strategy compared with the no-vaccine strategy is $100,000 per case averted at an incidence of 0.1%, $20,000 at an incidence of 0.5%, $10,000 at an incidence of 1%, $2,500 at an incidence of 3%, and $1,000 at an incidence of 5% (see Figures 3A and B). * All costs obtained from reports of previous studies were adjusted to 1999 dollars using the medical care services component of the Consumer Price Index. Future costs were discounted at a rate of 3% per year. IV intravenous. Includes office visit and antibody test. Does not include medical care costs, since these are included in the costs of treatment of antibiotic reactions. Includes medical care and administration costs. justed medical care services component of the Consumer Price Index (20). Costs of long-term sequelae accrued with each model cycle were discounted at a rate of 3% per year. The primary outcome measure was the incremental cost-effectiveness, defined as the difference in time-discounted direct costs divided by the difference in time-discounted cases of Lyme disease between the vaccine strategy and the no-vaccine strategy. Sensitivity analysis. One-way sensitivity analysis was performed on all model variables over the full range of estimates as determined from reports in the literature and from expert opinion (Tables 1 and 2). Two-way sensitivity analysis was also conducted with Lyme disease incidence and key variables. RESULTS Effect of the incidence of Lyme disease. The incidence of Lyme disease drives the cost-effectiveness of vaccination. Figures 3A and B demonstrate the relationship between the annual incidence of Lyme disease and the incremental cost-effectiveness of the vaccine strategy compared with the no-vaccine strategy. As the incidence of Lyme disease increases, the incremental cost-effectiveness of vaccination exponentially decreases. The incremental cost-effectiveness of the Figure 3. Incremental cost-effectiveness of the vaccine strategy compared with the no-vaccine strategy as a function of the annual incidence of Lyme disease (base-case analysis). The significant relationship between incremental cost-effectiveness of vaccination and the annual incidence of Lyme disease is examined A, at lower incidences of Lyme disease and B, at higher incidences of Lyme disease. In A, the vertical dotted line indicates the point on the x-axis where the incidence of Lyme disease equals (6.7 cases per 100,000), the average reported national incidence. In B, where the incremental cost-effectiveness of vaccination decreases below zero, the vaccine strategy becomes less costly and more effective than the no-vaccine strategy.

5 COST-EFFECTIVENESS OF LYME DISEASE VACCINE 1655 Figure 4. Sensitivity analysis of the annual incidence of Lyme disease comparing the vaccine strategy with the no-vaccine strategy for A, yearly boosters and B, boosters every 3 years. The y-axis represents the average cost in dollars per person over 10 years as determined for each strategy. The points where the two lines intersect indicate the thresholds of 10% (A) and 5% (B) above which the no-vaccine strategy becomes more costly overall. These points correspond to the thresholds shown in Figure 3B (the points where incremental costeffectiveness falls below zero). Figures 4A and B illustrate the results of sensitivity analysis on the annual risk of Lyme disease and the impact on cost of vaccination compared with no vaccination as risk is varied from 0% to 12%. When the incidence of Lyme disease exceeds 10%, the vaccine strategy with yearly boosters becomes both more effective and less costly than the no-vaccine strategy (Figure 4A). When a 3-year booster regimen is used (Figure 4B), this threshold declines to 5%. At the estimated national incidence of Lyme disease (0.0067%), the incremental cost-effectiveness of vaccination with annual boosters for Lyme disease is $1,600,000 per case averted (Table 4). Thus, it would cost $200 billion to vaccinate all 250 million persons in the US, and 2,041 persons would need to be vaccinated over a 10-year period to prevent 1 case of clinical Lyme disease (number-needed-to-treat [NNT] analysis). If boosters are given at 3-year intervals, the incremental cost-effectiveness is $830,000 per case averted. At a Lyme disease incidence of 1%, the incremental cost per case averted is $9,900 with a yearly booster regimen and $4,500 with a 3-year booster regimen (Table 4), with 1 case of clinical Lyme disease prevented per decade for every 14 people vaccinated (NNT analysis). When the effectiveness measure was changed from all cases averted to only late-stage (i.e., disseminated ) Lyme disease cases averted, the incremental cost per late-stage Lyme disease case averted at a 1% Lyme disease incidence was $36,000 for yearly boosters and $16,500 for 3-year boosters. Similarly, when effectiveness was measured as the number of cases with persistent sequelae (despite treatment) averted, the incremental cost per case with sequelae averted at a 1% incidence of Lyme disease was $80,000 with yearly boosters and $36,000 with 3-year boosters. Monte Carlo simulations using a hypothetical cohort of 20,000 patients (10,000 patients each for the vaccine and no-vaccine strategies) and a 1% incidence of Lyme disease provided estimates of the number of reactions to antibiotics. Transient minor reactions (e.g., rashes) to oral antibiotics occurred at a frequency of 8 per 10,000 in the vaccine strategy group compared with 37 per 10,000 in the no-vaccine strategy group. The frequency of minor reactions to intravenous antibiotics was 2 per 10,000 in the vaccine strategy group and 10 per 10,000 in the non-vaccine strategy group. The frequency of major reactions to antibiotics (oral and intravenous) was negligible in both strategy groups ( 1 per 10,000). Results of sensitivity analyses. Other factors that influenced the results included the proposed number of years of Lyme disease susceptibility, the required frequency of boosters, the cost of vaccination, and the vaccine efficacy. Lyme disease vaccination is more costeffective as the years of susceptibility, the duration of vaccine protection, and the efficacy of the vaccine increase. It is less cost-effective when the cost of vaccination increases (see Table 5). In sequential two-way sensitivity analyses using Lyme disease incidence and other model variables, the cost of ceftriaxone treatment, cost of rheumatologic sequelae, efficacy of antibiotics, probability of arthritis sequelae, and probability of presenting with erythema migrans were found to affect the Lyme disease incidence thresholds for dominance of the vaccine strategy by 1 or

6 1656 HSIA ET AL Table 4. Costs, cases averted, and cost-effectiveness of the vaccine strategy versus the no-vaccine strategy over a 10-year analysis* Strategy Cost, dollars Incremental cost, dollars Effectiveness, no. of cases Incremental effectiveness, no. of cases averted Incremental cost-effectiveness, dollars, incremental cost per case averted Incidence of Lyme disease at national average ( ) Yearly booster Vaccine ,600, No vaccine year booster Vaccine , No vaccine Incidence of Lyme disease in endemic area (0.01) Yearly booster Vaccine , No vaccine year booster Vaccine , No vaccine * Cost of one booster is $75. Costs and cases averted are the average results per person. 2 percentage points when varied over their estimated ranges. For all other parameters, the incremental costeffectiveness of the vaccine strategy compared with the no-vaccine strategy was not substantially affected throughout the range of variable estimates tested. Best-case and worse-case scenarios (Table 5). Using a 1% incidence of Lyme disease, we examined the effect of biasing our estimates to maximize and minimize the cost-effectiveness of vaccination. When selected variables that we determined to be pivotal from sensi- Table 5. Effect of adjustment of select variable estimates in the best-case and worst-case scenarios* Variable Best-case estimate Incremental C/E, dollars/case averted Worst-case estimate Vaccine efficacy 95% $7,357 50% $14,879 Efficacy of antibiotics (oral 80% $7,281 99% $10,852 and IV) Efficacy of oral antibiotics for 60% $9,682 90% $10,493 arthritis Efficacy of IV antibiotics for 40% $9,675 60% $10,221 arthritis Probability of presenting with 60% $8,571 95% $10,556 early Lyme disease Probability of arthritis 70% $9,754 50% $10,138 sequelae Incremental C/E, dollars/case averted Cost of arthritis sequelae $6,000 $9,704 $1,000 $10,252 Cost of 3 weeks of ceftriaxone $6,500 $9,767 $1,500 $10,119 Cost of a vaccine booster $50 $6,310 $100 $13,577 Incremental cost per case averted at a 1% Lyme disease incidence (yearly boosters) $213 $23,281 * Best-case parameter estimates are those biased in their plausible range to improve the cost-effectiveness of the vaccine strategy. In the worst case, parameter estimates are biased to reduce cost-effectiveness. Incremental cost-effectiveness (C/E) ratios calculated at a 1% Lyme disease incidence and with yearly boosters are presented for best- and worst-case estimates for each of the selected variables. IV intravenous. When selected variables determined to be pivotal from sensitivity analyses were adjusted to favor the vaccine strategy, the vaccine strategy dominated when Lyme disease incidence exceeded 1.1%. When these same variables were adjusted to bias against the vaccine strategy, the incidence of Lyme disease needed to be 59% for the vaccine strategy to be cost saving.

7 COST-EFFECTIVENESS OF LYME DISEASE VACCINE 1657 tivity analyses were adjusted to favor the vaccine strategy, the incremental cost-effectiveness of the vaccine strategy with yearly boosters was reduced to $200 per case averted, and the vaccine strategy dominated when Lyme disease incidence exceeded 1.1%. When selected variables were adjusted to bias against the vaccine strategy, the incremental cost-effectiveness of vaccination with yearly boosters increased to $23,000 per case averted, and the incidence of Lyme disease needed to be 59% for the vaccine strategy to be cost saving. DISCUSSION Our analysis reveals that the annual risk of contracting symptomatic Lyme disease is the single most important factor in determining the cost-effectiveness of vaccination. At low incidences of Lyme disease, such as the average national incidence, vaccinating for Lyme disease is prohibitively expensive. However, at incidence levels observed in high-risk groups and highly endemic areas, the Lyme disease vaccine becomes a potentially cost-effective tool. Vaccinating for Lyme disease becomes not only more effective, but also cost saving at a Lyme disease incidence of 10% when yearly boosters are given, and at an incidence of 5% when boosters are given every 3 years (Figure 4). Connecticut has the highest reported statewide incidence of Lyme disease (100 cases per 100,000). The highest reported county-specific incidence is in Nantucket, MA, at 1.51% (3). The highest annual incidence of Lyme disease ever reported was 10%, and this incidence was found during an epidemic outbreak in a small community next to an open nature preserve (21). The 5 states and counties with the highest incidence of Lyme disease in the US are listed in Table 1 (12,22). Nantucket, with its population of 3,000 persons, is representative of the implication of our results in a highly endemic area. To vaccinate the entire population of Nantucket would cost $2,500,000 with yearly boosters and $1,300,000 with boosters at 3-year intervals. Two hundred eighty-five cases of Lyme disease would be averted over 10 years at a discounted incremental cost per case averted of $7,000, with 11 people required to be vaccinated to prevent only 1 case of Lyme disease over 10 years. The existing passive reporting system is thought to significantly underestimate the true number of cases of symptomatic Lyme disease. To the extent that symptomatic Lyme disease is underreported, vaccination will be more cost-effective and possibly even cost-saving in highly endemic areas. Similarly, to the extent that a person s vocational and recreational activities increase his or her exposure to ticks, that person s risk will increase beyond the average for a given geographical area. Occupational studies in states where Lyme disease is endemic have found that outdoor workers were 5 times more likely than indoor workers to have been exposed to ticks, to be seropositive for antibody to B burgdorferi, and to have had Lyme disease (23,24). Since Lyme disease is not directly transmitted between human beings, no herd immunity is provided by vaccination. In addition, the incidence of serious sequelae of late Lyme disease is low, especially if people are treated early with antibiotic therapy, and long-term outcomes are generally good for most (25,26). However, there is evidence that the incidence of Lyme disease is increasing as the population density of disease-bearing Ixodes ticks increases and their territory spreads (27 29), and as reforestation and suburbanization increase human exposure to B burgdorferi infection (2,6,30). Even based on the present reported incidence of Lyme disease, the economic burden is substantial. A decision analysis model estimated that a national expenditure of $2.5 billion dollars over 5 years would be required to prevent 55,000 cases of Lyme disease sequelae, using an annual mean incidence of 4.73 cases of Lyme disease per 100,000 (10). The implications of our results are consistent with the recommendations of the American Committee on Immunization Practices (ACIP) of the CDC. The ACIP recommends Lyme disease vaccination for persons ages years who reside, work, or recreate in areas of high or moderate risk and who engage in activities that result in frequent or prolonged exposure to tick-infected habitats (14). Lyme disease vaccination is not recommended for persons who have minimal or no exposure to ticks even if they live in an area of high or moderate risk. Vaccinated persons should continue to practice preventive measures, such as wearing protective, light-colored clothing, using tick repellent, and inspecting the entire body daily for ticks for prompt removal to prevent disease transmission. Furthermore, persons should be reminded that the Lyme disease vaccine does not protect against infection from other tick-borne illnesses, such as ehrlichiosis and babesiosis. Whether vaccination against Lyme disease would reduce precautionary behavior to prevent tick bites and thereby increase the incidence of other tick-transmitted infections is unknown. Other preventive treatment options for Lyme disease include environmental interventions to reduce tick density, education and promotion of precautionary

8 1658 HSIA ET AL behavior, and antibiotic prophylaxis of tick bites. Although environmental methods such as vegetation removal and chemical treatments have been shown to reduce tick density, none of these methods has been proven to decrease the risk of acquiring Lyme disease (31). The same is true for preventive behavioral interventions (31). A recent randomized controlled trial of 482 subjects demonstrated the efficacy of a single 200-mg dose of doxycycline in preventing Lyme disease after an Ixodes scapularis tick bite, which was contrary to previous data (32). However, many patients who develop Lyme disease are unaware of the initial tick bite and would not present in time for antibiotic prophylaxis to be effective. Cost-effectiveness analysis of these other preventive treatment options requires more data and an approach different from the one used in our analysis, and is beyond the scope of this study. Investigators in 3 other recently reported studies have examined the cost-effectiveness of vaccinating against Lyme disease. Shadick et al determined costeffectiveness in terms of dollars per quality-adjusted life year (QALY), using calculated utilities from rating scale scores obtained from surveying residents of Nantucket Island (33). Their model was very sensitive to the utility values, with a wide variation in their results ($38,500/ QALY to $122,300/QALY). Their base-case utility values for various Lyme disease health states were low and differed from those in a recent Institute of Medicine (IOM) report that estimated the incremental costeffectiveness of Lyme disease vaccination to be $100,000/QALY if given to residents of a highly endemic area (33,34). Shadick et al did report $5,300 per case of Lyme disease averted at a 1% incidence of Lyme disease; however, this result did not include yearly boosters and reflected a one-time cost of vaccination of $150. Our results are consistent with those of Shadick et al if booster costs are taken into account. The IOM report was a general analysis to help policymakers prioritize the development of 26 vaccination programs on the basis of cost-effectiveness (34). The benefits of vaccination targeted to all residents of endemic areas were expressed in QALYs as measured by the Health Utilities Index Mark II for different disease states. The IOM report was not a dedicated analysis of the cost-effectiveness of the present Lyme disease vaccine, and therefore, limited sensitivity analyses were performed. Because the IOM report used an effectiveness measure of QALYs, a set Lyme disease incidence of 4.56 per 100,000, and other contrasting assumptions, it is difficult to compare our results with those in the IOM report. Meltzer et al determined the cost-effectiveness of vaccinating against Lyme disease in terms of the cost per case averted and included yearly boosters, but their analysis differed from ours in a number of respects, including the incorporation of lost productivity costs in the base-case analysis and distinct assumptions such as a higher probability of disseminated disease (35). Nevertheless, the results of our present analysis are in accordance with the similar conclusions of Shadick et al and Meltzer et al from their studies, that vaccination against Lyme disease may be cost-effective only when individual risk exceeds 1%. We chose to measure effectiveness in terms of the number of cases averted, and not in QALYs. Measuring QALYs requires the determination of utilities for various health states of Lyme disease complications that have not been well studied and are not reported in the literature except in the Shadick et al article (33). Utility measurement depends on the methods used, incorporates its own assumptions, and may differ substantially across individuals, populations, and perspectives. In addition to the problems with utility measurement, we believe that the number of cases averted is the effectiveness measure more relevant to policymakers, clinicians, and patients. The goal of Lyme disease vaccination is to prevent Lyme disease cases, and therefore, the number of cases averted is the unit of effectiveness that is understood by the majority of persons interested in the cost-effectiveness of the Lyme disease vaccine. However, since most economic evaluations provide comparisons in terms of cost per life-year gained or per QALY gained, it becomes more difficult to judge what incremental cost per case averted is acceptable to society and therefore cost-effective. Several studies have used the number of cases prevented as the effectiveness measure. One study which examined the cost-effectiveness of misoprostol in preventing serious gastrointestinal events associated with the use of nonsteroidal antiinflammatory drugs found an incremental cost-effectiveness ratio of $94,766 per serious gastrointestinal complication averted, while for medium- and high-risk groups, the ratios were $14,943 and $4,101, respectively (36). Investigators in another cost-effectiveness study reported that preventing infective endocarditis by prescribing oral amoxicillin for all patients with mitral valve prolapse cost an incremental $119,000 per case averted (37). Ultimately, determining what incremental cost per Lyme disease case averted is acceptable to society is a value judgment to be made by decision-makers. As is true with all decision analyses, assumptions were necessary in formulating our model and in estimat-

9 COST-EFFECTIVENESS OF LYME DISEASE VACCINE 1659 ing variable values. Since there has been no reported evidence of serious adverse effects unequivocally due to the vaccine, we assumed in our analysis that there were none. Reported followup, however, has only been on the order of 2 years, and adverse events resulting from medication and biologics are severely underreported. The recently found homology between leukocyte function associated antigen 1 and OspA and the consequent potential for autoimmune-induced arthritis provide scientific support for theoretical concern about vaccine-induced side effects (38). Since the approval of the Lyme disease vaccine, 1 million doses were distributed in the first year, and the FDA has received 774 reports of patients who may have had adverse effects related to the vaccine, of which 64 were considered serious events involving death, hospitalization, or disability (39 41). However, as stated by the FDA, these reports do not indicate a confirmed link to the vaccine (39,41). Studies by the FDA of reported adverse events reveal that only hypersensitivity reactions (a total of 22 reports of urticaria and dyspnea, of which 3 were serious) follow a pattern that suggests a connection to the vaccine (41). If the vaccine is found to have serious side effects like arthritis, then its cost-effectiveness will be reduced depending on the frequency of such adverse events and the magnitude of associated morbidity and costs. Finally, this model examined only direct medical costs. If indirect costs had been included, the costeffectiveness of the vaccine would have been better than that estimated by our results. We chose not to model chronic Lyme disease (arthralgias/myalgias associated with post Lyme disease syndrome), since patients with this syndrome have a high prevalence of fibromyalgialike symptoms due to other causes and, currently, most experts do not recognize it as a separate diagnostic entity (16). Our analysis did not incorporate the problem of misdiagnosis and inappropriate treatment of Lyme disease, a problem exacerbated by the tremendous public anxiety and confusion surrounding this issue (42,43). Inappropriate use of health services associated with the overdiagnosis and overtreatment of Lyme disease has been estimated to be substantial (44 46). Potentially, vaccination may decrease anxiety about acquiring Lyme disease and thereby reduce the societal costs of overdiagnosis and overtreatment of Lyme disease; however, there are no data to support such a claim. In summary, our analysis shows that an individual s risk for Lyme disease is the most important parameter that determines the cost-effectiveness of vaccination. This risk is a function of both the geographic incidence of Lyme disease and an individual s exposure to ticks. For the vast majority of persons in the US, Lyme disease vaccination is not a cost-effective measure. This conservative analysis, based only on direct costs, shows that the vaccine can be considered cost-effective in certain populations where the risk of Lyme disease is 1%, and that it is cost saving with yearly boosters when incidence exceeds 10% as well as with boosters every 3 years when incidence exceeds 5%. Thus, the decision to vaccinate should be based on an individual assessment of risk that takes into account both local geographic area and personal history of the amount of tick exposure. Vaccination should only be recommended for those with considerable individual risk. REFERENCES 1. Orloski KA, Hayes EB, Campbell GL, Dennis DT. Surveillance for Lyme disease United States, Mor Mortal Wkly Rep CDC Surveill Summ 2000;49: Evans J. Lyme disease. Curr Opin Rheumatol 1999;11: Centers for Disease Control and Prevention. Lyme disease United States, MMWR Morb Mortal Wkly Rep 1997;46: Gardner P. Lyme disease vaccines [editorial]. Ann Intern Med 1998;129: Wormser GP. Vaccination as a modality to prevent Lyme disease: a status report. Infect Dis Clin North Am 1999;13: Steere AC, Sikand VK, Meurice F, Parenti DL, Fikrig E, Schoen RT, et al, the Lyme Disease Vaccine Study Group. Vaccination against Lyme disease with recombinant Borrelia burgdorferi outer-surface lipoprotein A with adjuvant. N Engl J Med 1998;339: Steere AC. Lyme disease. N Engl J Med 2001;345: Parenti D, Schoen R, Sennewald E, Buscarino C, van Hoecke C. Evaluation of reactogenicity and immunogenicity of booster doses of LYMErixTM, recombinant L-Osp-A vaccine against Lyme disease (LD). Presented at the Interscience Conference on Antimicrobial Agents and Chemotherapy; 1999 Sep; San Francisco (CA). 9. Sigal LH, Zahradnik JM, Lavin P, Patella SJ, Bryant G, Haselby R, et al, the Recombinant Outer-Surface Protein A Lyme Disease Vaccine Study Consortium. A vaccine consisting of recombinant Borrelia burgdorferi outer-surface protein A to prevent Lyme disease [published erratum appears in N Engl J Med 1998;339: 571]. N Engl J Med 1998;339: Maes E, Lecomte P, Ray N. A cost-of-illness study of Lyme disease in the United States. Clin Ther 1998;20: Magid D, Schwartz B, Craft J, Schwartz J. Prevention of Lyme disease after tick bites: a cost-effectiveness analysis. N Engl J Med 1992;327: Centers for Disease Control and Prevention. Summary of notifiable diseases, United States, MMWR Morb Mortal Wkly Rep 1998;47:x, Nichol G, Dennis DT, Steere AC, Lightfoot RW Jr, Wells G, Shea B, et al. Test-treatment strategies for patients suspected of having Lyme disease: a cost-effectiveness analysis. Ann Intern Med 1998;128: Centers for Disease Control and Prevention. Recommendations for the use of Lyme disease vaccine: recommendations of the

10 1660 HSIA ET AL Advisory Committee on Immunization Practices (ACIP). MMWR Morb Mortal Wkly Rep 1999;48(RR-7):1 17, Thanassi WT, Schoen RT. The Lyme disease vaccine: conception, development, and implementation. Ann Intern Med 2000;132: Wormser GP, Nadelman RB, Dattwyler RJ, Dennis DT, Shapiro ED, Steere AC, et al, the Infectious Diseases Society of America. Practice guidelines for the treatment of Lyme disease. Clin Infect Dis 2000;31 Suppl 1:S Anonymous. Lyme disease vaccine. Med Lett Drugs Ther 1999; 41: Cardinale V, editor. Drug topics red book. Montvale (NJ): Medical Economics Company; Lightfoot RW Jr, Luft BJ, Rahn DW, Steere AC, Sigal LH, Zoschke DC, et al. Empiric parenteral antibiotic treatment of patients with fibromyalgia and fatigue and a positive serologic result for Lyme disease: a cost-effectiveness analysis. Ann Intern Med 1993;119: US Bureau of Labor Statistics Web site. Consumer price indexes. Washington (DC): US Bureau of Labor Statistics; Accessed Sep. 14, URL: Lastavica CC, Wilson ML, Berardi VP, Spielman A, Deblinger RD. Rapid emergence of a focal epidemic of Lyme disease in coastal Massachusetts. N Engl J Med 1989;320: Lyme disease statistics [computer file]. Centers for Disease Control and Prevention; Smith PF, Benach JL, White DJ, Stroup DF, Morse DL. Occupation risk of Lyme disease in endemic areas of New York state. Ann N Y Acad Sci 1988;539: Schwartz BS, Goldstein MD. Lyme disease in outdoor workers: risk factors, preventive measures, and tick removal methods. Am J Epidemiol 1990;131: Shadick N, Phillips C, Sangha O, Logigian E, Kaplan RF, Wright EA, et al. Musculoskeletal and neurologic outcomes in patients with previously treated Lyme disease. Ann Intern Med 1999;131: Seltzer E, Gerber MA, Cartter M, Freudigman K, Shapiro E. Long-term outcomes of persons with Lyme disease. JAMA 2000; 283: Herrington JE Jr, Campbell GL, Bailey RE, Cartter ML, Adams M, Frazier E, et al. Predisposing factors for individuals Lyme disease prevention practices: Connecticut, Maine, and Montana. Am J Public Health 1997;87: White DJ, Chang H-G, Benach JL, Bosler EM, Meldrum SC, Means RG, et al. The geographic spread and temporal increase of the Lyme disease epidemic. JAMA 1991;226: Stafford KC III, Cartter ML, Magnarelli LA, Ertel S, Mshar PA. Temporal correlations between tick abundance and prevalence of ticks infected with Borrelia burgdorferi and increasing incidence of Lyme disease. J Clin Microbiol 1998;36: Campbell G, Fritz C, Fish D, Nowakowski J, Nadelman R, Wormser G. Estimation of the incidence of Lyme disease. Am J Epidemiol 1998;148: Poland GA. Prevention of Lyme disease: a review of the evidence. Mayo Clin Proc 2001;76: Nadelman RB, Nowakowski J, Fish D, Falco RC, Freeman K, McKenna D, et al. Prophylaxis with single-dose doxycycline for the prevention of Lyme disease after an Ixodes scapularis tick bite. N Engl J Med 2001;345: Shadick NA, Liang MH, Phillips CB, Fossel K, Kuntz KM. The cost-effectiveness of vaccination against Lyme disease. Arch Intern Med 2001;161: Stratton KR, Durch JS, Lawrence RS. Vaccines for the 21st century: a tool for decision making. Washington (DC): National Academy Press; Meltzer MI, Dennis DT, Orloski KA. The cost effectiveness of vaccinating against Lyme disease. Emerg Infect Dis 1999;5: Maetzel A, Ferraz MB, Bombardier C. The cost-effectiveness of misoprostol in preventing serious gastrointestinal events associated with the use of nonsteroidal antiinflammatory drugs. Arthritis Rheum 1998;41: Devereux RB, Frary CJ, Kramer-Fox R, Roberts RB, Ruchlin HS. Cost-effectiveness of infective endocarditis prophylaxis for mitral valve prolapse with or without a mitral regurgitant murmur. Am J Cardiol 1994;74: Gross DM, Forsthuber T, Tary-Lehmann M, Etling C, Ito K, Nagy ZA, et al. Identification of LFA-1 as a candidate autoantigen in treatment-resistant Lyme arthritis. Science 1998;281: Warner S. Patients sue over effects of vaccine. Philadelphia Inquirer Jun 13;1, Food and Drug Administration. Vaccine adverse event reporting system: report for Lymerix manufactured by SmithKline Beecham. Rockville (MD): Food and Drug Administration; p Ball R, Braun M, Lathrop S, Haber P, Mootrey GT, Hayes N. Lyme vaccine safety data reported to the vaccine adverse event reporting system (VAERS) of the USA from the first year of use [abstract]. Pharmacoepidemiol Drug Safety 2000;9 Suppl 1:S Sigal LH. Anxiety and persistence of Lyme disease. Am J Med 1995;98 Suppl 4A:74S 78S. 43. Sigal LH. The Lyme disease controversy: social and financial costs of misdiagnosis and mismanagement. Arch Intern Med 1996;156: Reid MC, Schoen RT, Evans J, Rosenberg JC, Horwitz RI. The consequences of overdiagnosis and overtreatment of Lyme disease: an observational study. Ann Intern Med 1998;128: Coyle BS, Strickland GT, Liang YY, Pena C, McCarter R, Israel E. The public health impact of Lyme disease in Maryland. J Infect Dis 1996;173: Fix AD, Strickland T, Grant J. Tick bites and Lyme disease in an endemic setting: problematic use of serologic testing and prophylactic antibiotic therapy. JAMA 1998;279: