Childhood vaccination against chickenpox: An analysis of benefits and costs

Childhood vaccination against chickenpox: An analysis of benefits and costs

Childhood vaccination against chickenpox: An analysis of benefits and costs Daniel M. Huse, MA, H. C o d y Meissner, MD, M i c h a e l J. L a c e y , ...

608KB Sizes 0 Downloads 37 Views

Childhood vaccination against chickenpox: An analysis of benefits and costs Daniel M. Huse, MA, H. C o d y Meissner, MD, M i c h a e l J. L a c e y , BS, a n d Gerry Oster, PhD From Policy AnalysisInc., Brookline, Massachusetts, and the Department of Pediatrics, New England Medical Center and Tufts UniversitySchool of Medicine, Boston, Massachusetts

Objective. To estimate the economic costs and benefits of routine childhood vaccination against varicella infection. Design. Decision-analytic model of the incidence and costs of chickenpox in children assumed to receive varicella vaccine at a g e 15 months in conjunction with the measles-mumps-rubella vaccine, or not to be vaccinated against varicella. Patients. Hypothetical cohort of 100,000 children. Main o u t c o m e measures. Costs of vaccination, cumulative incidence of chickenpox to a g e 25 years, and related disease costs, including medical treatment and work loss. Results. Vaccination of 100,000 children against varicella at a g e t5 months would cost $4,812,000. The expected number of cases of chickenpox to a g e 25 years would be reduced from 95,400 to 4800; costs of medical treatment and work loss would correspondingly decline by $1,678,000 and $9,781,000, respectively. On balance, vaccination is estimated to yield n e t e c o n o m i c benefits of $6,647,000, or $66.47 per vaccinee. Conclusion. Vaccination against varicella infection is cost-effective and should be part of the routine immunization schedule for U,S. children. (J PEDIATR 1994;124:869-74) A vaccine against the varicella-zoster virus (Oka/Merck live attenuated virus vaccine) may soon be approved for use in the United States.~ In a large multicenter, placebo-controlled trial in children 1 to 14 years of age, it reduced the incidence of chickenpox by more than 90% for a 2-year period.2, 3 There is significant interest in the use of this new vaccine for susceptible adults, as well as immunocompromised children, because of the severity of varicella infection in these groups. 4, 5 Interest is also high in vaccinating hospital employees, to reduce nosocomial transmission of varicella-zoster virus, particularly to high-risk patients. 6 Supported by a research grant from Merck Vaccine Division, Merck & Co, Inc., West Point, Pa. Submitted for publication Nov. 15, 1993; accepted Jan. 14, 1994. Reprint requests: Gerry Oster, PhD, Policy Analysis Inc., Four Davis Court, Brookline, MA 02146. Copyright | 1994 by Mosby-Year Book, Inc. 0022-3476/94/$3.00 + 0 9/20/54347

The value of targeted vaccination seems clear, but the cost-effectiveness of routine childhood immunization against chickenpox may be questioned] Among the young, most varicella infections are benign, and serious morbidity is infrequent, as are fatalities. 8, 9 However, the economic burden of chickenpox may be substantial.9 Children with the disease are routinely kept out of school or day care for NHDS NHIS NMES

National Hospital Discharge Survey National Health Interview Survey National Medical Expenditure Survey

as long as 1 week. 1~The economic cost of home care for sick children accordingly may be sizable, and so, too, may the corresponding cost savings resulting from disease prevention. The economic costs and benefits of universal childhood vaccination against varicella have been reported previously, l I but the cost of the vaccine was unknown at the time this 869

870

H u s e et al.

study was conducted. In addition, the analysis may be outdated. The numbers of children being cared for outside the home has risen substantially as the labor force participation of women has increased 12; the average age at which children are first exposed to the virus therefore may have declined, and the number of parents who miss work to care for sick children may have increased. Similarly, more than one million physician office visits for chickenpox were reported in the United States in the 12-month period ending in September 1992 (unpublished data: National Disease and Therapeutic Index), triple the average number reported for the years 1974 through 1983. 8 We examined the cost-effectiveness of universal childhood vaccination against chickenpox, using current estimates of disease incidence and costs. We posed four distinct questions in our study: (1) What would be the impact of universal vaccination at this age on the expected number of cases of chickenpox? (2) What would be its impact on medical care costs? (3) What would be the impact on the costs of work loss related to chickenpox? (4) Would the economic benefits of vaccination outweigh the costs? METHODS

Model overview. To examine the economic consequences of universal childhood vaccination against varicella, we set forth a model of the incidence and costs of chickenpox in a hypothetical cohort of 100,000 children aged 15 months. These children were assumed either to have received the varicella vaccine or not to have been vaccinated. For each scenario, we then calculated the expected number of cases of chickenpox and associated expected costs from ages 15 months to 25 years, when more than 90% of all reported cases of chickenpox occur, s The costs of chickenpox were assumed to include expenses for medical treatment as well as the value of illnessrelated work loss. The former was assumed to include outpatient visits, prescription drugs, and hospitalizations. The latter included work loss for the care of sick children and of employed persons who acquire the disease. Consistent with previous analyses of pediatric vaccine programs, 13"16 we focused attention on the net economic cost of this intervention, which we defined as the cost of vaccination (including treatment of side effects) minus the resulting savings in the costs of treating chickenpox and related work loss. We evaluated all such costs from a societal perspective, 17 expressed at 1991 average price levels, and discounted all future costs to the date of vaccination (i.e., to age 15 months), using an annual rate of 5%. Data sources. To estimate the safety and efficacy of varicelia vaccine, we used data from a large, randomized trial of the Oka/Merck live attenuated virus vaccine.2, 3 To estimate the current incidence and costs of chickenpox, we used data from the National Health Interview Survey, the National Hospital Discharge Survey, and the National

The Journal o f Pediatrics June 1994

Medical Expenditure Survey. The NHIS is an annual survey of the health status of the U.S. civilian, noninstitutionalized population, based on interviews with 40,000 to 50,000 sample households comprising more than t00,000 persons. 18 The NHDS is an annual survey of patients discharged from U.S. nonfederal, acute-care hospitals, based on discharge abstracts for approximately 200,000 patients treated in 400 sample hospitals. 19 The NMES was a one-time survey of expenditures on personal health-care services during calendar year 1987, based on repeated interviews with approximately 14,000 households during that year. 2~ Each of these surveys is based on a national probability sample (i.e., of households or hospitals), and each individual respondent or hospital discharge is assigned a sampling weight accordingly; we employed these weights in all analyses of survey data. To obtain samples sufficiently large to generate age-specific estimates of chickenpox incidence and related utilization of health-care services, we pooled 11 consecutive years (1980 through 1990) of the NHIS and NHDS. These series yielded data on approximately 1.2 million persons and 2.4 million hospital discharges, respectively. All medical-care costs were adjusted to 1991 price levels by means of the Medical Care Component of the Consumer Price Index for All Urban Consumers (Bureau of Labor Statistics); the value of work loss was estimated with 1991 data and hence required no such adjustment. All other sources of data are as noted in the text below. Incidence of chickenpox. To estimate the expected numbers of cases of chickenpox from ages 15 months to 25 years in an unvaccinated cohort, we multiplied age-specific estimates of disease incidence by the estimated number of cohort members remaining alive at each age. Age-specific disease incidence was estimated with the use of NHIS data. We divided the age-specific average annual number of reported cases of chickenpox by the total number of persons of similar age surveyed. We obtained singleyear incidence rates for ages 1 to 15 years and an average rate for ages 16 to 25 years. Incidence from ages 15 to 23 months was assumed to be equal to three fourths of that for all 1-year-old children. The number of cohort members remaining alive at each age was estimated by multiplying the initial size of the cohort (i.e., 100,000) by the proportion surviving to each future age, based on 1987 U.S. life tables. 21 We estimated the expected survival of 15-month-old children by dividing the proportions of a U.S. birth cohort surviving to single-year ages by the proportion surviving to age 15 months. Survivorship data are not reported for age 15 months; we interpolated this value by adjusting the proportion alive at 1 year downward by one fourth of the decline in the survival rate between ages 1 and 2 years. Vaccine safety and efficacy. In the previously described trial of the Oka/Merck varicella vaccine, 27% of children

The Journal o f Pediatrics Volume 124, Number 6

who received the vaccine had adverse reactions within 48 hours, and 4% had "varicella-like" rashes within 8 weeks of vaccination. 2 Seroconversion was reported in 94% of vaccine recipients overall, and in 96% of those 12 to 23 months of age. In the first year of follow-up, there were no reported cases of chickenpox among vaccinated children, in comparison with 39 cases in the placebo group overall and four cases among those exposed to siblings with chickenpox (i.e., 100% effficacy).2 In the second year, there was one mild case in the vaccine group, in a child exposed to an infected sibling; protective efficacy was 96% overall and 92% among those exposed. 3 Long-term follow-up studies have reported persistence of antibody in all vaccinees tested after 6 years and in 97% of those tested after 7 to 10 years), 22 In our analysis, we assumed that vaccination would reduce the expected incidence of chickenpox at each future age, to age 25 years, by 95%. Costs of vaccination. We assumed that the Oka/Merck varicella vaccine would be administered as a separate injection in conjunction with the measles-mumps-rubella vaccine at age 15 months 1, 11 and therefore would not require an additional office visit. We assumed that the cost of the vaccine would be $35 per dose (personal communication: Merck & Co., Oct. 15, 1993) and that physicians would charge an additional $13 for handling and administration. The latter estimate was based on the difference between the reported median fee charged by pediatricians for administration of the measles-mumps-rubella vaccine ($37) and the acquisition cost of this vaccine ($24 per dose), consisting of the direct manufacturer's price plus the federal excise tax. 23, 24 One percent of vaccinated children with adverse reactions, however, or 0.31% of all vaccinees, were assumed to require an additional office visit at a cost of $38, based on the average fee charged by pediatricians for visits with established patients. 25 Costs of treating ehickenpox. We calculated an expected cost per case of treating chickenpox, focusing attention on outpatient visits, prescription drugs, and hospitalizations. For each type of medical service, we multiplied the estimated rate of utilization per case of chickenpox by its estimated average cost. Outpatient visits. We estimated that there would be 32 physician visits per 100 cases of chickenpox among children aged 1 to 15 years, and 95 visits per 100 cases among persons aged 16 to 25 years; our estimate was based on the average annual number of physician visits for chickenpox reported in the NHIS, divided by the corresponding average annual numbers of cases. The cost per outpatient visit for treatment of chickenpox was estimated, on the basis of the average total expense for such visits reported in the NMES, to be $43.80. Prescription drugs. Our estimate that there would be 11 prescriptions for antiinfective and antipruritic medications

H u s e et al.

871

per 100 cases of chickenpox was based on the number of such prescriptions for treatment of chickenpox among persons aged 1 to 25 years reported in the NMES, divided by the corresponding number of cases of chickenpox reported for the same year (i.e., 1987) in the NHIS. The cost per prescription was estimated to be $15.50 on the basis of the mean total expenses reported for the selected prescriptions in the NMES. Hospitalization. To estimate a rate of hospitalization for treatment of varicella complications,8,26 we first used N H D S data to calculate the average annual number of discharges with a principal diagnosis of chickenpox (International Classification of Diseases [ninth edition]: Clinical Modification [ICD-9-CM] 052), or a secondary diagnosis of chickenpox with a principal diagnosis of encephalitis (323), Reye syndrome (331.81), cerebellar ataxia (334.3), epilepsy (345), pneumonia (480-486), viral complication of pregnancy (647.6), or cellulitis (682). We then divided the age-specific numbers of hospitalizations with these diagnoses by the corresponding annual numbers of cases of chickenpox, yielding rates of 16 and 60 per 10,000 cases of chickenpox, respectively, for persons aged 1 to 15 years and 16 to 25 years. The total cost per hospital admission, including hospital and professional services, was calculated as the product of the estimated average duration of stay and average cost per day. Average duration of stay was estimated to be 4.2 days on the basis of data for the selected patients in the NHDS. The average cost per day in hospital was estimated from the N M E S data. Because few admissions for varicella complications were reported in this survey, we selected all hospitalizations of patients 1 to 25 years of age with a similar medical conditions on the basis of a principal diagnosis of infectious disease (ICD-9-CM 001-139), encephalitis (323), pneumonia (480-486), or cellulitis (682). For these selected hospital stays, we then divided the total reported expenses for hospital and professional services by the total number of in patient days, which yielded an average cost per day of $1115. We therefore estimated the cost per hospitalization to be $4683 (i.e., 4.2 • $1115). Costs of work loss. We estimated the expected cost of work loss separately for patients 15 months to 6 years, 7 to 12 years, 13 to 17 years, and 18 to 25 years of age. For each group, we calculated expected costs as the product of the percentage of cases involving work loss, the average number of days of work loss per case, and the value of a day of work. For patients 15 months to 6 years and 7 to 12 years of age, respectively, we estimated that 33% and 46% would involve parental work loss; our estimate was based on the product of the age-specific percent of children whose mothers were in the labor force and the percentage of women in the labor force with children of corresponding age who were employed full time (Bureau of Labor Statistics, Current Pop-

872

H u s e et al.

The Journal o f Pediatrics June 1994

T a b l e I. Expected costs of chickenpox to age 25 years* among 100,000 children aged 15 months, with and without vaccination. Costs

Medical care Vaccination Treatment of side effects Treatment of chickenpox TOTALMEDICAL-CARECOSTS Work loss TOTALCOSTS *Discountedat 5% annually.

Not vaccinated (A)

Vaccinated (B)

Net cost of vaccination (B minus A)

--$ 1,766,000 1,766,000 10,296,000 $12,062,000

$4,800,000 12,000 88,000 4,900,000 515,000 $5,415,000

$4,800,000 12,000 ( 1,678,000) 3,134,000 (9,781,000) (6,647,000)

ulation Survey). We estimated the average number of days of home care per caseof chickenpox to be 3.7 days; our estimate was based on the average annual number of days lost from school because of chickenpox among children aged 5 to 12 years reported in the NHIS, divided by the corresponding average annual number of cases. We assumed that children less than 5 years of age would require the same number of days of home care, on average. Finally, our estimate that the value of a day lost from work would be $103 was based on one fifth of the 1991 average weekly earnings for U.S. women (Bureau of Labor Statistics, Current Population Survey), plus average fringe benefits equal to 40% of earnings (Bureau of Labor Statistics, Employment Cost Index). For patients with chickenpox who were aged 13 to 17 years, we assumed that no home care would be required and that there would be no associated work loss. For patients aged 18 to 25 years, we estimated, on the basis of the reported work loss related to chickenpox in the NHIS, that 40% of these persons would be employed and would miss work for an average of 5.5 days. Our estimate that the value of a day of work for these persons would be $78 was based on one fifth of the average weekly earnings for persons 18 to 25 years of age in 1991 (Bureau of Labor Statistics, Current Population Survey), plus fringe benefits equal to 40% of earnings (Bureau of Labor Statistics, Employment Cost I n d e x ) . Sensitivity analyses. We examined the sensitivity Of our results to changes in a number of model variables, including the assumed efficacy of the vaccine, the number of doses required to ensure lifelong immunity, the expected costs of medical treatment for chickenpox (including hospitalization and outpatient care), the expected costs of work loss (to reflect a less stringent school exclusion policy for infected children),27 and the discount rate.

and 171 hospitalizations, at a total (discounted) cost of $1,766,000. Expected work loss related to chickenpox would amount to 133,000 days, at a total (discounted) cost of $10,296,000. Vaccination of 100,000 children against varicella at age 15 months would be expected to cost $4,812,000:$4,800,000 for vaccine purchase and administration, and an estimated $12,000 for 310 outpatient visits for treatment of side effects. The expected number of cases of chickenpox by age 25 years would decline to 4800, and the numbers of visits, prescriptions, and hospitalizations for treatment of chickenpox would decline accordingly to 2000, 500, and 9, respectively, at a total (discounted) cost of $88,000. Work loss would be expected to decline to 7000 days, valued at $515,000 (disCounted). Vaccination of 100,000 children aged 15 months against chickenpox would increase total expected medical-care costs by $3,134,000 ($31.34 per vaccinee). Expected costs of work loss, however, would decline by $9,781,000. On balance, therefore, vaccination would yield expected net economic benefits of $6,647,000, or $66.47 per vaccinee (Table I). Sensitivity analyses. We conducted one-way sensitivity analyses on a number of key model variables (Table II): vaccine efficacy was varied from 80% to 100%; the addition of a second dose of vaccine at age 12 years was considered, with an assumed 95% percent efficacy for the two-dose regimen; the costs of treating chickenpox, including rates of hospitalization, numbers of doctor visits, and numbers of prescriptions, were alternatively halved and doubled; the expected cost of work loss was halved; and the discount rate was varied from 2% to 8%. In each instance, vaccination increased total medical care costs but yielded overall net economic benefits. DISCUSSION

RESULTS Among 100,000 unvaccinated children, we estimated that there would be 95,400 cases of chickenpox by age 25 years, resulting in 33,200 doctor visits, 10,500 prescriptions,

Our estimate of the net economic benefits of vaccination against varicella is not substantially different from that reported by Preblud et al. 11 in 1985, but this similarity conceals important differences in assumptions and findings. For

The Journal o f Pediatrics Volume 124, Number 6

one, the actual cost of the vaccine is higher than Preblud and colleagues assumed. Our estimate of the expected savings in costs of treating chickenpox is also twice as great, which refleets a significantly higher frequency of physician visits (i.e., 32 vs 10 per 100 cases). Preblud et al. assumed that 54% of cases in children would involve parental work loss; we estimated, on the basis of reported patterns of employment among mothers of young children, that such work loss would occur in only 33% to 46% of cases. They assumed that parental work loss would average 5 days; we estimated that only 3.7 days are lost. We employed a higher estimate of the value of a day lost from work, which accounted for fringe benefits as well as wage-and-salary income. In estimating the benefits of vaccination, we assumed that immunity acquired from a single dose of varicella vaccine would persist. In long-term follow-up studies, antibody titers among vaccinees have been similar to those after natural varicella infection.3, 25 However, even if a second dose of the vaccine were required--for example, in conjunction with the measles booster at age 12 years--vaccination would nonetheless yield net economic benefits. We estimated the impact of vaccination on chickenpox incidence and costs to age 25 years, by which age approximately 95% of unvaccinated persons in the United States become infected. Although a disproportionate number of varicella complications occur in older persons, the expected costs of these events would be modest if discounted to age 15 months; costs incurred at age 40 years, for instance, would be discounted by 86%, assuming a 5% annual rate. It has been suggested that universal childhood vaccination may increase the incidence of varicella among adults, because reduced childhood incidence permits unvaccinated individuals and those with waning immunity to enter adulthood susceptible to infection,v, 27 The experience of universal childhood vaccination against measles, which formerly was as ubiquitous in childhood as varicella, suggests that these concerns may be overstated. The share of measles cases in adults has increased since the vaccine's introduction, but the incidence has declined from 25 to 2 per 100,000 annually (unpublished data, Centers for Disease Control and Prevention).28 If the experience with measles is any guide, therefore, there is no reason to expect universal childhood vaccination against varicella to affect adult incidence of the disease adversely. We believe that our analysis was conservative and may have understated the benefits of vaccination. We considered in our analysis only the benefits of preventing cases of chickenpox among vaccinees, and did not take into account any possible benefits associated with reduced transmission of varicella to immunocompromised or other susceptible persons. Vaccination also may reduce the risk of congenital or neonatal varicella resulting from infection of pregnant women. The risk of nosocomial varicella may be diminished

Huse et al.

873

T a b l e II. Sensitivity analyses Model parameter

Net* costs of medical care

Vaccine efficacy 80% $3,399,000 95% (base case) 3,134,000 100% 3,045,000 Number of doses One (base c a s e ) $3,134,000 Two 5,973,000 Costs of treating cbickenpoxt 0.5 • base c a s e $3;973,000 Base case 3,134,000 1.5 x base case 2,295,000 Costs of work loss 0.5 x base c a s e $3,134,000 Base case 3,134,000 Discount rate 2% $2,805,000 5% (base case) 3,134,000 8% 3,375,000

Net* total costs

($4,838,000) (6,674,000) (7,250,000) ($6,674,000) (3,808,000) ($5,808,000) (6,674,000) (7,486,000) ($1,757,000) (6,674,000) ($8,627,000) (6,674,000) (5,116,000)

*Vaccinated minus unvaccinated. ?Simultaneous variation of the rate of hospitalization, number of office visits, and number of prescriptions.

because pediatric patients and susceptible health care workers with children of their own are less likely to serve as vectors for transmitting varicella in a hospital setting. We did not take account of rates of death caused either by natural infection with varicella or by vaccination. An estimated 100 deaths annually in the United States are attributed to chickenpox9; no deaths have been associated with the Oka/Merck varicella vaccine.2, 29, 30 The possibility of a fatal adverse reaction cannot be ruled out, but it seems likely that fewer deaths would occur among vaccinated than unvaccinated persons. We also did not consider any possible effect of vaccination on the severity of chickenpox cases that occur among vaccinees. These cases have been reported to be mild, involving relatively few lesions and a low incidence of fever. 30 We did not consider any possible effect of vaccination on the incidence of herpes zoster. Although there have been reports of herpes zoster among vaccinees, the incidence of herpes zoster after vaccination appears lower than that after natural varicella infection.31 However, most cases of herpes zoster occur among elderly persons, 32so the expected benefits of prevention would be small when discounted to the age of vaccination. Our estimates of prescription drug use do not reflect the recent approval by the U.S. Food and Drug Administration of acyclovir for treatment of uncomplicated chickenpox, which may increase future costs of this disease. We also did not consider the costs of nonprescription medications.

874

Huse et al.

Finally, we did not include a n y estimate of work loss a m o n g p a r t - t i m e workers, a n d we used data on the average earnings of women to value work loss for care of children with chickenpox. Some h o m e care m a y be provided by men, for w h o m average earnings are higher (Bureau of L a b o r Statistics, C u r r e n t Population Survey). O u r analysis indicates t h a t childhood vaccination against chickenpox would yield substantial economic benefits, principally by reducing the b u r d e n of p a r e n t a l work loss associated with this c o m m o n childhood disease. O u r findings suggest t h a t vaccination against varicella is cost-effective a n d should be part of the routine i m m u n i z a t i o n schedule for U.S. children. REFERENCES

1. Marwick C. Varicella vaccine expected to be ready by 1993. JAMA 1992;268:851-2. 2. Weibel RE, Neff BJ, Kuter BJ, et al. Live attenuated varicella virus vaccine. N Engl J Med 1984;310:1409-15. 3. Kuter BJ, Weibel RE, Guess HA, et al. Oka/Merck varicella vaccine in healthy children: final report of a 2-year efficacy study and 7-year follow-up studies. Vaccine 1991;9:643-7. 4. Gershon AA, Steinberg SP, Varicella Vaccine Collaborative Study Group (National Institute of Allergy and Infectious Diseases). Persistence of immunity to varicella in children with leukemia immunized with live attenuated varicella vaccine. N Engl J Med 1989;320:892-7. 5. Gershon AA, Steinberg SP, LaRussa P, et al. Immunization of healthy adults with live attenuated varicella vaccine. J Infect Dis 1988;158:132-7. 6. Weber D J, Rutala WA, Parham C. Impact and costs of varicella prevention in a university hospital. Am J Public Health 1988;78:19-23. 7. Mclntosh K. Varicella vaccine: decisions a little nearer. N Engl J Med 1984;310:1456-7. 8. Guess HA, Broughton DD, Melton L J, Kurland LT. Populationzbased studies of varicella complications. Pediatrics 1986; 78(suppl):723-7. 9. Preblud SR, Orenstein WA, Bart KJ. Varicella: clinical manifestations, epidemiology and health impact in children. Pediatr Infect Dis 1984;3:505-9. 10. Moore DA, Hopkins RS. Assessment of a school exclusion policy during a chickenpox outbreak. Am J Epidemiol 1991; 133:1161-7. 11. Preblud SR, Orenstein WA, Koplan JP, Bart K J, Hinman AR. A benefit-cost analysis of a childhood varicella vaccination programme. Postgrad Med J 1985;61:17-22. 12. Bureau of Labor Statistics. Handbook of labor statistics. Bulletin 2340. Washington, D.C.: U.S. Department of Labor, Bureau of Labor Statistics, 1989. 13. Cochi SL, Broome CV, Hightower AW. Immunization of U.S. children with Hemophilus influenzae type b polysaccharide vaccine. JAMA 1985;253:521-9. 14. Koplan JP, Preblud SR. A benefit-cost analysis of mumps vaccine. Am J Dis Child 1982;136:362-4. 15. Koplan JP, Schoenbaum SC, Weinstein MC, Fraser DW.

The Journal of Pediatrics June 1994

Pertussis vaccine: an analysis of benefits, risks and costs. N Engl J Med 1979;301:906-11. 16. Schoenbaum SC, Hyde JN, Bartoshesky L, Crampton K. Benefit-cost analysis of rubella vaccination policy. N Engl J Med 1976;294:306-10. 17. Eisenberg JM. Clinical economics: a guide to the economic analysis of clinical practices. JAMA 1989;262:2879-86. 18. Massey JT, Moore TF, Parsons VL, Tadros W. Design and estimation for the National Health Interview Survey, 198594. Hyattsville, Maryland: National Center for Health Statistics, 1989. (Vital and health statistics; series 2; No. 110.) 19. Graves EJ. Detailed diagnoses and procedures, National Hospital Discharge Survey, 1990. Hyattsville, Maryland: National Center for Health Statistics, 1992. (Vital and health statistics; series 13; No. 113.) 20. Edwards W, Berlin M. Questionnaires and data collection methods for the Household Survey and the Survey of American Indians and Alaska Natives. (DHHS publication No. [PHS] 89-3450; National Medical Expenditure Survey Methods 2, National Center for Health Services Research and Health Care Technology Assessment.) Rockville, Maryland: U.S. Public Health Service, 1989. 21. National Center for Health Statistics. Vital statistics of the United States, 1987; vol II: Mortality (part A) (DHHS publication No. [PHS] 90-1101). Washington, D.C.: U.S. Public Health Service, 1990. 22. Asano Y, Nagai T, Miyata T, et al. Long-term protective immunity of recipients of the OKA strain of live varicella vaccine. Pediatrics 1985;75:667-71. 23. Medical Economics Research Group. Pediatricians: facts about their practices. Montvale, New Jersey: Medical Economics Publishing, 1992. 24. Medical Economics Data. 1991 Red Book: annual pharmacists' reference. Oradell, New Jersey: Medical Economics Company, 1991. 25. AMA Center for Health Policy Research. Physician marketplace statistics, 1989. Chicago: American Medical Association, 1989. 26. Jackson MA, Burry VF, Olson LC. Complications of varicella requiring hospitalization in previously healthy children. Pediatr Infect Dis J 1992;11:441-3. 27. Brunell PA. Chickenpox--examining our options. N Engl J Med 1991;325:1577-9. 28. Langmuir AD. Medical importance of measles. Am J Dis Child 1961;103:54-6. 29. Arbeter AM, Starr SE, Preblud SR, et al. Varicella vaccine trials in healthy children. Am J Dis Child 1984;138:434-8. 30. White C J, Kuter BJ, Hildebrand CS, et al. Varicella vaccine (Varivax) in healthy children and adolescents: results from clinical trials, 1987 to 1989. Pediatrics 1991;87:604-10. 31. Hardy I, Gershon AA, Steinberg SP, LaRussa P, Varicella Vaccine Collaborative Study Group. The incidence of zoster after immunization with live attenuated varicella vaccine: a study in children with leukemia. N Engl J Med 1991 ;325:154550. 32. Weller TH. Varicella and herpes zoster: changing concepts of the natural history, control, and importance of a not-so-benign virus. N Engl J Med 1983;309:1362-8, 1434-40.