The hidden costs of infant vaccination

Vaccine 19 (2001) 33±41

www.elsevier.com/locate/vaccine

The hidden costs of infant vaccination Tracy A. Lieu a,*, Steven B. Black b, G. Thomas Ray a, Kathleen E. Martin a, Henry R. Shine®eld b, Bruce G. Weniger c a

Division of Research, Kaiser Permanente, Oakland, CA, USA Vaccine Study Center, Kaiser Permanente, Oakland, CA, USA c The National Immunization Program, Centers for Disease Control and Prevention, Atlanta, GA, USA b

Received 14 December 1999; received in revised form 11 April 2000; accepted 11 April 2000

Abstract Combination vaccines to minimize injections required for infant vaccination, and new vaccines with improved safety pro®les, will pose increasingly complex choices for vaccine purchasers in the future. How much of a premium to pay for such vaccines might be determined by taking into account (1) the psychological burden of multiple injections during a single clinic visit, and the costs of any additional visits to minimize these, and (2) the medical, work-loss, and incidental costs of common vaccineassociated symptoms. This cross-sectional survey included randomly-selected parents of 1±8-month-old infants who received vaccines in a Northern California health maintenance organization (HMO) in 1997. Interviewers called parents 14 days after the infant's vaccination to administer a 10-minute closed-ended interview in English or Spanish. Parents were asked about infant symptoms after vaccination, their preferences regarding multiple injections and their (theoretical) willingness to pay to reduce the number of injections their infant would receive, or to avoid the adverse symptoms experienced. Among 1769 eligible infants, interviews were completed with parents of 1657 (93%). The psychological cost of multiple injections was estimated by the willingness of parents to pay a median of $25 to reduce injections from 4 to 3, $25 from 3 to 2, and $50 from 2 to 1. Vaccineassociated symptoms caused mean costs of $42 in medical utilization and $192 in work-loss among the families who experienced those events (Ns=62 and 35, respectively). When averaged among all 1657 study infants, vaccine-associated symptoms after the index vaccination visit resulted in $2.91 in medical utilization, $4.05 in work-loss, and $0.74 in direct nonmedical costs, yielding total ®nancial costs of $7.70. Parents of infants who had vaccine-associated symptoms said they would have paid a median of $50 to avoid these symptoms. Fever and fussiness were associated in logistic regression analysis with a two-fold increase in the odds of medical utilization, and fever with more than a three-fold increase in work loss. We conclude that multiple injections during a single clinic visit entail psychological costs. The psychological costs of vaccine-associated symptoms, as measured by willingness-to-pay methods, are higher than those resulting from multiple injections. The ®nancial costs of medical utilization and work-loss resulting from common vaccine-associated symptoms are non-negligible and should be incorporated in economic analyses. 7 2000 Elsevier Science Ltd. All rights reserved. Keywords: Economics; Combination vaccines; Safety; Immunization

1. Introduction Childhood immunization is one of the most success* Corresponding author. Department of Ambulatory Care and Prevention, Harvard Pilgrim Health Care and Harvard Medical School, 126 Brookline Ave., Suite 200, Boston, MA 02215, USA. Tel.: +1617-421-6218; fax: +1-617-859-8112. E-mail address: [email protected] (T.A. Lieu).

ful practices in all of health care [1]. It is highly coste€ective compared to other interventions, and actually cost-saving in many instances. Even further gains in disease prevention are promised by the biotechnology revolution, which is yielding many new and improved vaccines. Yet, as more maladies become preventable through vaccination, it becomes increasingly dicult to add new vaccines to an already crowded schedule of recommended immunizations. In the United States in

0264-410X/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 2 6 4 - 4 1 0 X ( 0 0 ) 0 0 1 5 4 - 7

34

T.A. Lieu et al. / Vaccine 19 (2001) 33±41

1989, only eight injections were required to fully immunize a child from birth through 16 years of age according to the nationally recommended immunization schedule [2]. A decade later, to satisfy the 2000 schedule [3], a minimum of 16 injections are required, taking advantage of combination vaccines available as of March 2000. Although the bene®ts of vaccination may outweigh them, adverse side e€ects of vaccines do pose risks, ranging from common and mild fever and injection site pain, to uncommon but serious reactions such as seizures [4], and in very rare cases, paralysis. The incorporation of safer vaccines into immunization practice Ð such as those for diphtheria, tetanus, and pertussis with acellular pertussis antigens (DTPa) and (injected) inactivated polio vaccine (IPV) [5,6] Ð has contributed to some of the recent increase in injections. This ``pincushion syndrome'' [7±9] will likely worsen with the introduction of e€ective new vaccines, such as for pneumococcal disease in infants [10], which would add four injections to the overall schedule and result in four or ®ve simultaneous injections for an infant at the 2-month clinic visit. Combining antigens into the same vial or syringe is an obvious solution which has been promoted by the Children's Vaccine Initiative and pursued vigorously by vaccine manufacturers [11]. However, combination vaccines face challenging biomedical and regulatory obstacles [12], which make their development quite lengthy and expensive. How much more should one pay for a new combination vaccine that reduces the required number of injections, or for a safer product which decreases the adverse e€ects of vaccination? How would one choose among a growing menu of alternative products with di€ering numbers of antigens? To answer such questions, algorithms for vaccine selection have been proposed which use mathematical techniques to compare alternative vaccine formularies and serve as practical tools for decisions that would be too complicated for an intuitive process [13,14]. Their fundamental principle is to recognize the economic value of various distinguishing features of competing products, so as to provide incentives for continuing innovation by industry for new and improved vaccines. These algorithms require, however, reliable economic data on the savings conferred by various vaccine innovations. How much does it save to avoid one injection? What are the direct and indirect costs for treating the adverse side e€ects of vaccines? What is the cost per dose to keep vaccines refrigerated that would be saved by heat-stable products? How often do vaccines expire before use, and how much would be saved if they had longer shelf lives? The current study was designed to address the ®rst two questions above by ®lling gaps in existing knowl-

edge about the direct and indirect medical, work-loss, psychological, and incidental costs of multiple injections and the adverse e€ects of vaccination. 2. Methods 2.1. Overall design This cross-sectional study linked information from vaccination visits of 1±8-month old infants with data collected by telephone survey 2 weeks later of their parents about subsequent patient symptoms, and the resulting actual medical experiences, work loss, and hypothetical preferences of the parents. 2.2. Study population The target population was parents of infants aged 1±8-months who received vaccines at Northern CA Kaiser Permanente (NCKP) facilities. NCKP is a nonpro®t, prepaid group-model health maintenance organization that serves approximately 2.7 million members in northern CA via 33 outpatient clinics, 17 of which are based at hospital medical centers. All nationally recommended childhood vaccines are provided with no out-of-pocket co-payment for vaccination, although some members have a co-payment (most commonly $5±15) for each clinic visit. Eligible families were identi®ed using the Kaiser Immunization Tracking System, a computerized database which captures all immunizations delivered in NCKP with better than 95% accuracy. Each week during the study period (April 1997±June 1997), an age-strati®ed random sample of vaccinated infants was identi®ed for attempted interviews with parents. During the study period, NCKP did not have a standard regional recommendation for the choice between a diphtheria-tetanus-acellular pertussis vaccine (DTPa) plus a separate Haemophilus in¯uenzae type b vaccine (HIB), or a combined diphtheria-tetanus-whole-cell-pertussis-Haemophilus in¯uenzae type b vaccine (DTPw-HIB), nor a recommendation between (injected) inactivated polio vaccine (IPV) or oral polio vaccine (OPV). Each of NCKP's medical centers could de®ne its own guidelines for which of these vaccine types to use, and/or allow individual pediatricians to select from among alternative products. 2.3. Data collection Fourteen days after the infant's vaccination, an interviewer from the NCKP Division of Research telephoned each parent or caregiver to request participation in a 10-minute, structured, closed-ended interview in English or Spanish. Interview calls were

T.A. Lieu et al. / Vaccine 19 (2001) 33±41

attempted until the interview was complete or 35 days had elapsed since the vaccination. Verbal informed consent was obtained under the study protocol, which was approved as ethical research on human subjects by the NCKP Institutional Review Board. 2.4. Questionnaire content The interviewer used a standardized form to ask questions in the following topic sequence: The ®rst part inquired about fever, fussiness, injection site redness, swelling, and pain during the ®rst 48 hours after the ``shot(s)'' and during the subsequent 2 weeks, using questions developed for previous studies. The interviewer asked whether the family had called the advice nurse or made a clinic visit due to such symptoms or other problems related to the vaccination(s). The interviewer then asked if anyone had missed work to bring the infant for the visit at which the vaccination(s) were given, or for a subsequent visit due to symptoms of adverse side e€ects from the vaccine(s). Because the mother, father, or other caregivers of an infant may have di€erent salaries and di€erent propensities to miss work, respondents were also asked which individual(s) had missed work and what their annual incomes were. The interviewer then asked about parent preferences regarding multiple injections. One question asked, ``Imagine that (child's name )'s doctor had recommended four shots and said it was safe to give them all at once. Would you have chosen for (child's name ) to receive the four shots at once, or take two or three at once and come back another time for the rest?'' In addition, the parent was asked whether anyone would have needed to miss work if the infant needed to come back a second time if injections had been deferred. Finally, two interview questions measured how much parents would have been willing to pay to avoid their child experiencing vaccine-associated symptoms or multiple injections. During the ®rst half of the study period an open-ended format was used: ``Health policymakers need to understand how much value to place on preventing symptoms related to shots. A common measure of value is our own money Ð dollars. I am asking you to think how much you would have been willing to pay to avoid any symptoms your baby had after his vaccinations. Don't count any money lost from your salary for time o€, or any co-payment at the clinic. Consider all the ways you spend your money. If you could have paid money to prevent (child's name )'s symptoms, what is the maximum amount of your own money you would be willing to pay?'' The open-ended question on multiple injections was: ``I see that (child's name ) had (X number ) shots at the

35

last visit. Think about how much you would have been willing to pay to avoid one of those shots. (Child's name ) would get the same protection against disease but the number of shots would go from (X ) to (Xÿ1 ). What is the maximum amount of your own money you would be willing to pay?'' A total of 592 parents were queried once, without repeating the question, only about a reduction from the actual number of injections their infant received. For example, a parent whose infant received three injections was asked only the amount he or she would have been willing to pay Ð assuming the same protection from disease Ð to reduce the number of injections from three to two, but NOT from four to three or from two to one. During the second half of the study, interviewers asked similar questions about willingness-to-pay using dichotomous-choice response formats. Because these questions yielded similar results as the open-ended formats, for simplicity, only the results of the open-ended questions are presented here. 2.5. Analysis of vaccine symptoms Symptoms associated with vaccination were analyzed in three ways. First, they were classi®ed by types of symptoms, regardless of their degree of severity. Because a patient may have more than one symptom, with each of varying degree, a second analysis classi®ed the patient's worst symptom according to mild, moderate, or severe degree. If fever was reported, it was classi®ed as mild if the temperature was R100.48F (388C), moderate if 100.5±102.78F (38.1±39.38C), and severe if R102.88F (R39.338C). Fussiness was classi®ed by the parent as mild (with normal activity), moderate (prolonged crying and refusal to play), or severe (persistent crying and inability to be comforted). Injection site pain was classi®ed by the parent as mild (cried or reacted a little when the leg was touched), moderate (cried a lot when the leg was touched), or severe (cried when the leg was moved). The third method of classifying vaccine symptoms was adapted from a Dutch system [15]. Each vaccine reaction was classi®ed as ``simple'' (a single vaccine injection and a single reaction), ``compound'' (a single vaccine injection and more than one reaction), ``multiple'' (more than one vaccine injection but only one reaction), or ``compound-multiple'' (more than one vaccine injection and more than one reaction). As published, this system applied only to major reactions, but it was adapted here for the minor reactions of fever, fussiness, and injection site pain. 2.6. Medical and work-loss costs Most NCKP medical centers administer infant vaccinations in specialized injection clinics with dedicated

36

T.A. Lieu et al. / Vaccine 19 (2001) 33±41

sta€. The direct medical costs of such an injection clinic visit, as well as telephone consultation with a nurse, and a return clinic visit for vaccine-associated symptoms, were estimated with the assistance of NCKP administrators. The injection clinic cost estimate reviewed internal cost accounting data for personnel, disposable supplies, and administrative costs for a one-year period at a large, typical injection clinic with 25,000 visits annually. To calculate work-loss for both clinic visits and vaccine-associated symptoms, we multiplied the number of hours of work the parent or caregiver was reported to have missed by that person's hourly wage rate, derived from their reported hourly or annual salary. For families who missed work but who did not provide information on either wages or hours lost, work-loss costs were imputed at the average of those families that did provide these data. The projected cost of parent work-loss due to a follow-up visit was estimated as the proportion of families who would need to miss work for the visit times the mean cost of work-loss among families who missed work for a clinic visit. The cost of work-loss for a follow-up clinic visit for adverse e€ects was assumed to be similar to that of work-loss for the initial clinic visit. To determine a ``cost per vaccination visit'', dollar amounts were divided by the total number of index vaccination visits studied. 2.7. Statistical methods The sample size of approximately 1600 was chosen, based on the assumption that 5% of parents would miss work due to vaccine-associated symptoms. This sample size has power of 80% to identify a 50% reduction in the proportion of parents with work-loss between two equally divided groups at a one-tailed alpha of 0.05. Bivariate analyses were conducted using the chi-square test for categorical variables and the Kruskal-Wallis test for ordinal and non-normally distributed continuous variables. Logistic regression was used to analyze the relationship between demographic characteristics and work-loss, and the association of vaccine symptoms with medical utilization and workloss. Iterative forced-entry logistic regression models were used to evaluate predictors of whether or not work was lost for the initial clinic visit. All analyses were conducted with SAS software (SAS Institute, Cary, NC). 3. Results 3.1. Enrollment During the study period, 1838 families of 1±8-

month-old infants who had received vaccinations were identi®ed for telephone interview attempts. Of these, 16 (1%) were ineligible due to disconnected or wrong telephone numbers, and 53 (3%) were ineligible because the parents did not speak English or Spanish. Of the 1769 eligible families, 94 (5%) were unreachable after repeated attempts, and 18 (1%) declined to participate, resulting in 1657 (93%) who completed interviews and were analyzed in this study. 3.2. Demographics and index clinic visit The study population had diverse race/ethnicity, with more than half of the respondents from minority groups (Table 1). One-third of mothers had an educational level of a high school degree or lower. In 403 families (24%), at least one person had missed work to bring the infant to the clinic for the studied index visit at which the initial vaccinations were given. The mean value of work lost due to the index visit was $16 per vaccination visit (averaged over all infants), or $69 for a€ected families (averaged over the 403 families who reported work-loss). In the logistic regression model, three predictors were associated with whether or not work was lost for the index clinic visit. Families were more likely to report work-loss for the initial visit if they had higher income (odds ratio [OR] 1.11 for each $10,000 increment, 95% con®dence interval [CI] 1.07±1.15, P = 0.0001); non-white mothers (OR 1.44, 95% CI 1.14± 1.83, P = 0.0025); and older infants. Compared to families with infants aged 1±3 months, the OR for missing work was 1.97 for those with infants aged 4±5 months (95% CI 1.51±2.97, p = 0.0001) and 1.63 for those with infants aged 6±8 months (95% CI 1.20± 2.20, p = 0.0016). The direct medical cost (including labor, supplies, and administrative overhead) of using the injection clinic during a pediatric oce visit was estimated as $11.40 per visit, regardless of the number of injections received. Similarly, the cost of a telephone consultation with a nurse by a parent regarding an adverse vaccination symptoms was estimated at $10.00. 3.3. Multiple injections Most infants received multiple injections at their index visit for this study. Two percent (31) received ®ve injections on that day, 10% (164) received four injections, 35% (579) received three, 41% (688) received two, and 12% (195) received only one. A total of 4119 injections were received by all the infants during the 1657 index vaccination visits studied. These included 388 DTPa, 1269 DTPw-HIB, 383 HIB, 1172 HBV, 266 IPV, and 639 investigational vaccine (either pneumococcal conjugate vaccine or control), and two

T.A. Lieu et al. / Vaccine 19 (2001) 33±41

measles mumps rubella (plus 1336 OPV which were not counted as injections). 3.3.1. Psychological costs The amounts parents reported they were willing to pay to reduce by one the number of simultaneous injections their infant had received, while still receiving the same protection against disease, was a median of $25 to reduce from 4±3 injections, $25 from 3±2, and $50 from 2±1 (means and ranges provided in Table 2). 3.3.2. Projected costs of deferred vaccinations If four injections had been due at a single clinic visit, 708 (43%) of the 1657 parents said they would have preferred their infant receive all four at once, while 898 (55%) said they would have preferred fewer than four injections and to make a follow-up visit for the rest. Among these 898, 381 (43%) said they or the other parent would need to miss work for the followup visit. The projected cost of work-loss for such follow-up visits was $30 per follow-up visit (proportion of families missing work [0.43] times the mean cost of

37

each work-loss episode for a clinic visit [$69]). If 55% of families actually chose to defer some injections, this would translate to an injection clinic cost of $6.27 (0.55  $11.40), plus a work loss cost of $16.50 (0.55  $30), for a total of $22.77 per index vaccination visit. However, in contrast to this hypothetical question, none of 1657 respondents reported actually missing work to make a follow-up visit for injections deferred from the index visit. 3.4. Adverse e€ects of vaccination 3.4.1. Symptoms Parents reported that 1207 (73%) of the 1657 infants experienced one or more adverse symptoms following vaccination (Table 3). Fussiness was most common (46%), followed by injection site pain (35%), fever (31%), and injection site swelling (27%). In infants with any symptoms, all were mild in 40%, while symptom(s) were moderate in 21% and severe in 9%. The Dutch compound-multiple classi®cation [15] was the

Table 1 Family demographics and work-loss associated with a routine vaccination visit among 1±8-month olds, Northern CA Kaiser Permanente, 1997 No. (%) among families. . . Demographic characteristic Mother's race/ethnicity White Latino Asian Black Mixed race, including white Other Mother's education Less than high school High school graduate Some college or more Mother's annual income $20,000 or less $20,001±40,000 $40,001 or more Father's annual income $20,000 or less $20,001±40,000 $40,001 or more Both parents employed Index infant age 1±3 months 4±5 months 6±8 months Total children in family 1 2 3 or more Index infant in day care a

. . .of all infants (N = 1657)

. . .with any work-loss (N = 403)

Unadjusted odds of work-loss

p

756(46) 410(25) 171(10) 112(7) 69(4) 128(8)

168(42) 102(25) 58(14) 29(7) 13(3) 33(8)

Referent 1.2 1.8 1.2 0.8 1.2

0.30 0.001 0.39 0.51 0.38

172(10) 387(24) 1092(66)

30(7) 88(22) 285(71)

Referent 1.4 1.7

0.16 0.016

769(51) 469(31) 223(15)

150(39) 140(36) 85(22)

Referent 1.8 2.5

< 0.001 < 0.001

281(18) 609(39) 623(40) 785(48)

81(21) 147(38) 143(37) 263(66)

Referent 0.80 0.74 2.7a

0.14 0.06 < 0.001

870(53) 457(28) 327(20)

166(41) 143(35) 94(23)

Referent 1.9 1.7

< 0.001 < 0.001

719(44) 538(33) 365(23) 208(13)

209(53) 127(32) 55(14) 92(23)

Referent 0.75 0.43 2.9a

0.31 < 0.001 < 0.001

For parent employment and day care variables, the odds of any parent (or adult caregiver) of the infant missing work for the index vaccination visit studied are expressed in relation to the referent groups: only one or neither parent employed, and not in day care, respectively.

38

T.A. Lieu et al. / Vaccine 19 (2001) 33±41

Table 2 Dollar amounts parents reported they were willing to pay to avoid multiple injections and adverse symptoms after vaccination of their 1±8month-old infants, Northern CA Kaiser Permanente, 1997 Amount willing to pay (U.S. $)

To reduce injections received by infant from:a 4±3 3±2 2±1 To avoid vaccine-associated symptoms among infants with: Any symptomb Fever Fussiness Injection site pain Injection site swelling

No. responses (% of eligible)

Mean (SD)

Median

Range

37(67) 154(65) 192(64)

$104(238) 149(828) 115(305)

$25 25 50

$0±1000 0±9999 0±3000

50 50 50 50 50

0±9999 0±9999 0±9999 0±5000 0±9999

356(55) 144(22) 200(31) 161(25) 113(17)

261(1129) 321(1427) 266(1218) 149(427) 369(1415)

a

See Methods section for how questions were posed and di€ered for parents of infants receiving di€ering numbers of injections. Among those respondents with missing values, 58% were unable to decide on a dollar value, 28% gave no reason for non-response, 13% did not understand the question, and 2% were o€ended by the question. b Of the 764 parents asked this question, 118 (15%) did not respond because their infant had not experienced any symptoms. The remaining 646 parents were considered eligible to answer this question. Of these parents, 290 did not respond: 84% were unable to decide on a dollar value, 11% did not understand the question, 2% were o€ended by the question, and 2% gave no reason for non-response.

most common (40%), followed by multiple (20%), compound (6%), and simple (3%) symptoms. 3.4.2. Medical costs Of the 1657 respondents, 107 (6.5%) said they telephoned the advice nurse due to symptoms resulting

from the vaccination. The rate of such calls was 0.09 per vaccination visit because some families made more than one call. Based on the estimated cost of $10 per advice call, the cost of such calls was $0.90 per infant vaccination visit. Only 25 (2.8%) of the 893 families asked if they had

Table 3 Medical, work-loss, and other nonmedical costs, and rates of selected events from symptoms associated with vaccination visits of 1±8-month-old infants (N = 1657), Northern CA Kaiser Permanente, 1997

Costs (U.S. 1997 dollars) Medical utilizationa Work-loss Direct nonmedical expensesb Total costs Rates of selected events due to subsequent vaccine-associated symptoms Per index vaccination visit, among all infants Days of work lost by mother Days of work lost by father No. of telephone calls to advice nurse No. of outpatient clinic visits Among only families with any medical utilizationc No. of telephone calls to advice nurse No. of outpatient clinic visits Total cost of medical utilization Among only families with work-loss due to vaccine symptoms (N = 35)d Days of work lost by mother Days of work lost by father Total days of work lost Total cost of work-loss a

Mean (SD)

Median

$2.91(15) 4.05(34) 0.74(11) 7.70(46)

$0 0 0 0

$0±205 0±865 0±350 0±970

0 0 0 0

0±4.5 0±5 0±4 0±3

1.0 0 10

0±4 0±3 10±205

0.02(0.2) 0.01(0.14) 0.09(0.38) 0.03(0.2) 1.26(0.77) 0.45(0.62) $41.94(41) 1.04(1.04) 0.42(0.98) 1.5(1.14) $192(166)

1 0 1 $180

Range

0±4.5 0±5 0.4±5 $17±865

Post-index visit medical utilization included making a call to the advice nurse, making a clinic visit, or both. Other post-index visit expenses incurred by families included transportation, parking, and extra babysitting. c The rates and costs reported are for 62 families who experienced any medical utilization. The denominator for the N of 62 was 893. All 1657 respondents were asked the question about calls to the advice nurse, but only 893 respondents were also asked the question about clinic visits. d The days and costs reported are for the 35 families who experienced work-loss. For four of these families, data on income were missing and were inferred as the mean income of families with work-loss with data on income. b

T.A. Lieu et al. / Vaccine 19 (2001) 33±41

39

made a follow-up clinic visit due to problems from the vaccination had done so; the rate of follow-up visits was 0.0315 per index vaccination visit. The calculated cost of these follow-up visits was $2.04 per (index) vaccination visit.

overall symptom severity than those who received DTPw-HIB (n = 1269) ( p < 0.0001). However, the odds of medical utilization and work-loss were not signi®cantly di€erent between recipients of these two alternative vaccine types.

3.4.3. Work-loss and incidental costs In 35 families (2.1%), at least one person missed work during the 2 weeks after the index vaccination visit because of the adverse e€ects. The mean estimated cost of work-loss due to vaccine symptoms was $192 per family experiencing work-loss (n = 35), or $4.05 per index vaccination visit among all infants (Table 3).

3.4.6. Psychological costs Parents of infants with adverse e€ects of vaccination reported willingness to pay a median of $50 (range: $0± > 9,999, mean: $261) to have avoided such sequelae (Table 2). This median amount did not vary according to the types of symptoms actually experienced.

3.4.4. Associations between symptoms, medical utilization and work-loss Vaccination-induced fever was the symptom most strongly associated with medical utilization (calling the nurse or returning to the clinic, OR 2.2, p < 0.001) and with parental work loss (OR 3.6, p < 0.001) (Table 4). Fussiness was also associated with medical utilization (OR 1.9, p < 0.001), but not with work loss. 3.3.5. Associations between DTP vaccine type and outcomes Infants who received DTPa (n = 388) had lower

4. Discussion This study found that the hidden costs of infant immunization can be substantial. Both multiple injections and adverse e€ects incur medical and psychological costs that might be averted by combination vaccines or less reactogenic vaccines. For multiple injections, we estimated a psychological cost, via willingness-topay questions, of at least $25 per injection. We also calculated potential additional costs of $22 per index visit for the medical and work-loss costs incurred by parents who said they would decline four or ®ve simul-

Table 4 Vaccination symptoms within 48 hours, and associated medical utilization and work-loss within 14 days after a vaccination visit of 1±8-monthold infants (N = 1657), Northern CA Kaiser Permanente, 1997 Any medical utilizationa

Any work-loss

No. (% of all infants) No. (% of those in subgroup) Odds ratiob No. (% of those in subgroup) Odds ratiob Symptom presence None 450(27) Any 1207(73) Symptom type:c Fever 571(31) Fussiness 766(46) Injection site swelling 443(27) Injection site pain 584(35) Degree of worst symptom Mild 666(40) Moderate 346(21) Severe 148(9) Modi®ed Dutch classi®cation Simple 57(3) Compound 102(6) Multiple 332(20) Compound-multiple 667(40)

15(3) 96(8)

Referent 2.5f

5 (1) 30 (2)

Referent 2.3

58(11) 73(10) 31(7) 48(8)

2.2f 1.9f 0.8 1.1

22 20 14 16

(4) (3) (3) (3)

3.6f 1.0 1.4 1.1

33(5) 34(10) 26(18)

1.4 2.9f 5.7f

7 (1) 14 (4) 8 (5)

0.9 3.5d 4.7e

0(0) 12(12) 23(7) 58(9)

± 3.6f 2.0 2.5f

1 (2) 6 (6) 4.1 18 (3)

1.5 5.1e 1.0 1.2

a Medical utilization included making a call to the advice nurse, making a clinic visit, or both. Three percent of infants with no symptoms within 48 hours after vaccination developed delayed symptoms within 14 days prompting medical utilization. b Odds ratios for parent work-loss are expressed in relation to the referent group of infants with no symptoms. c Symptom percentages that do not add to 100 are because of multiple symptoms in some infants. Odds ratios are from a logistic regression model that included all symptoms as predictors. d P R 0.05. eP R 0.01. fP R 0.001.

40

T.A. Lieu et al. / Vaccine 19 (2001) 33±41

taneous injections and make a follow-up visit for vaccines remaining due. Although 55% of parents said they would defer some injections to a follow-up visit, this is likely a high-end estimate because of the hypothetical nature of the question. In a 1993 study in St. Paul, MN, 71% of parents indicated that three shots were too many for one visit [9]. Yet, most parents at the end of the 1990s routinely accept three injections at a visit. Further, a 1992 study in Rochester, NY, found that 58% of parents would accept four injections at one visit if due and recommended by their physician [7]. All these studies, including the one reported here, are limited by their hypothetical nature. More de®nitive determination of the tolerable number of multiple injections will need to be studied in rigorous, empirical, observational studies of what parents and physicians actually do as more injections are added to the schedule and new combination vaccines become available. This study revealed that the hidden cost of vaccine adverse reactions are also not negligible, totaling more than $7 per infant vaccination visit in the 1±8-month age range. Even minor symptoms, such as fever and fussiness, should be considered in economic algorithms for vaccine selection. However, cost savings from less reactogenic vaccines must be weighed against any hidden costs if they result in additional injections, such as the use of newer DTPa vaccine plus a separate HIB, instead of the older DTPw-HIB combination. In addition, routine prophylactic antipyretics may have value as a means of reducing fever and thus costs (although the cost of such medications would need to be factored in). We used the willingness-to-pay method to quantify psychological costs, i.e., parent preferences for avoiding multiple injections and vaccine adverse reactions. This method, also known as contingent valuation, is a standard approach for valuing outcomes in public-policy decision analyses [16,17], and has been used in other studies to assign economic values for dicult-tomeasure health outcomes. For example, it has been used to measure patient preferences in chronic arthritis [18], lung disease [19], and angina [20]. Compared with other methods of assessing patient preferences, the willingness-to-pay technique has the advantage of using a well-understood measure of value: out-ofpocket money to be spent. Respondents usually must decide Ð in real-life scenarios Ð how much of their own funds they are actually willing to spend for speci®c services or bene®ts in real-life situations [18]. A common alternative approach to valuing health outcomes is utility elicitation, in which patient preferences are translated into quality-adjusted life-years saved [21]. However, utility elicitation tends to be better-suited for valuing chronic conditions than for

minor, temporary conditions such as vaccine injection pain or adverse reactions [22]. It also has the disadvantage of using more abstract measures of preference, such as the risk of death or the loss of time to be lived, rather than money. Legitimate skepticism has been voiced about the willingness-to-pay method. One criticism is that any dollar amounts speci®ed are arti®cial because most families in the USA do not pay directly for health care. Responses may correlate poorly with what the subject would really pay: It is hard to believe half the parents would pay more than $25 to avoid the fourth injection. Another problem is that the responses tend to vary widely [19], as we found with ranges from $0± > $9999 to avoid some shots or symptoms. Despite these limitations of the willingness-to-pay methodology, the results credibly indicate the relative strength of parent preferences, such as the greater value of avoiding vaccine-associated symptoms (willing to pay a median of $50) compared to a third or fourth injection ($25). One concern about this study is the uncertainty about whether telephone interviewing produces reliable and valid results using willingness-to-pay questions. Another limitation is that the results for infants may not apply to older children. The psychological costs for toddlers and school-age children could be higher because of needle phobia. In this study, we did not ask participants to place a value on avoiding vaccine-preventable diseases such as pertussis, meningitis, sepsis, hepatitis, varicella, pneumonia, or otitis media. These disease outcomes would also need to be evaluated using willingness-to-pay and/ or utility assessment to enable calculating tradeo€s relevant to vaccine selection. For example, it would be useful to quantify whether vaccines with greater e€ectiveness might justify accepting additional injections. The NCKP population studied was racially and ethnically diverse, although low-income families may still have been under-represented [23]. Most families served by NCKP receive bene®ts through employment. In half of the families we studied, both parents worked, and these families may have been more likely to experience work loss than other families. The work-loss attributed to index vaccination visits should be prorated to re¯ect the fact that these infant visits include time for preventive services other than vaccination, such as laboratory testing and preventive health counseling [24]. These ®ndings provide heretofore scarce primary data on the various costs associated with vaccine injections and adverse vaccine reactions, as well as the parental work loss associated with infant immunization. Although the methods entail many limitations, this information should be useful in vaccine economic modeling of various aspects of immunization practice. As a

T.A. Lieu et al. / Vaccine 19 (2001) 33±41

multitude of combination vaccines with ever-higher valencies are introduced [25], one such application is to expand and re®ne economic algorithms for achieving the overall ``best value'' when HMOs and other immunization providers assemble their vaccine formularies [13,14]. In selecting vaccines, it should be recognized that purchase price alone is but a small portion of the total costs in protecting children from disease through immunization, and that the broader economic impact of such decisions can now be weighed.

[8] [9]

[10]

[11] [12]

Acknowledgements This research was supported through the Vaccine Safety Datalink Project, a collaboration with the Vaccine Safety and Development Branch, National Immunization Program, Centers for Disease Control and Prevention, Atlanta, GA; and the National Vaccine Program Oce. We thank Maureen Sinclair and Soccoro Ramirez for excellent work conducting interviews, and acknowledge Robert Deuson, PhD, Robert Chen, MD, MA, Frank DeStefano, MD, MPH, and Mark Messonnier, PhD, for their ideas in developing this work, and comments on the manuscript. References [1] Centers for Disease Control, Prevention. Achievements in public health, 1900±1999: Impact of vaccines universally recommended for children Ð United States, 1900±1998. Morb Mort Weekly Rep 1999;48:243±8. [2] Centers for Disease Control, Prevention. Recommendations of the Immunization Practices Advisory Committee (ACIP): General recommendations on immunization. Morb Mort Weekly Rep 1989;38:205±27. [3] Centers for Disease Control, Prevention. Notice to readers: Recommended childhood immunization schedule Ð United States, 2000. Morb Mort Weekly Rep 2000;49:35±47. [4] Howson CP, Howe CJ, Fineberg HV, et al., editors. Adverse e€ects of pertussis and rubella vaccines. Washington, DC: Institute of Medicine, 1991. [5] Committee on Infectious Diseases. Recommended childhood immunization schedule Ð United States, January±December 1999. Pediatrics 1999;103:182±5. [6] Committee on Infectious Diseases. Poliomyelitis prevention: Revised recommendations for use of inactivated and live oral poliovirus vaccines. Pediatrics 1999;103:171±2. [7] Woodin KA, Rodewald LE, Humiston SG, Carges MS, Scha€er SJ, Szilagyi PG. Physician and parent opinions: Are

[13]

[14]

[15] [16] [17] [18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

41

children becoming pincushions from immunizations? Arch Pediatr Adolesc Med 1995;149:845±9. Melman ST, Chawla T, Kaplan JM, Anbar RD. Multiple immunizations. Archives of Family Medicine 1994;3:615±8. Madlon-Kay DJ, Harper PG. Too many shots? Parent, nurse, and physician attitudes toward multiple simultaneous childhood immunizations. Archives of Family Medicine 1994;3:610±3. Black SB, Shine®eld HR, Fireman BH et al., Ecacy, safety, and immunogenicity of heptavalent pneumococcal conjugate vaccine (Wyeth) in children, Pediatr. Infect. Dis. J. 2000;19:187± 95. Robbins lA. The children's vaccine initiative. Am J Dis Child 1993;147:152±3. Plotkin SA, Fletcher MA. Combination vaccines and immunization visits. Pediatr Infect Dis J 1996;15:103±5. Weniger BG, Chen RT, Jacobson SH, et al. Addressing the challenges to immunization practice with an economic algorithm for vaccine selection. Vaccine 1998;16:1885±97. Jacobson SH, Sewell ED, Deuson R, Weniger BG. An integer programming model for vaccine procurement and delivery for childhood immunization: a pilot study. Health Care Management Science 1999;2:1±9. Chen RT. Safety of vaccines. In: Plotkin SA, Orenstein WA, editors. Vaccines. Philadelphia: WB Saunders Company, 1999. Gafni A. Willingness-to-pay as a measure of bene®ts. Med Care 1991;29:1246±52. Zeckhauser R. Procedures for valuing lives. Public Policy 1975;23:419. Thompson MS, Read JL, Liang M. Feasibility of willingnessto-pay measurement in chronic arthritis. Med Decis Making 1984;4:195±215. O'Brien B, Viramontes JL. Willingness to pay: A valid and reliable measure of health state preference? Med Decis Making 1994;14:289±97. Chestnut LG, Keller LR, Lambert WE, Rowe RD. Measuring heart patients' willingness to pay for changes in angina symptoms. Med Decis Making 1996;16:65±77. Gold MR, Siegel JE, Russell LB, Weinstein MC, editors. Coste€ectiveness in health and medicine. New York: Oxford University Press, 1996. Wakker P, Stiggelbout A. Explaining distortions in utility elicitation through the rank-dependent model for risky choices. Med Decis Making 1995;15:180±6. Krieger N. Overcoming the absence of socioeconomic data in medical records: validation and application of a census-based methodology. American Journal of Public Health 1992;82:703± 10. LeBaron CW, Rodewald L, Humiston S. How much time is spent on well-child care and vaccinations? Arch Pediatr Adolesc Med 1999;153:1154±9. Centers for Disease Control, Prevention. Combination vaccines for childhood immunization. Recommendations of the Advisory Committee on Immunization Practices (ACIP), the American Academy of Pediatrics (AAP), and the American Academy of Family Physicians (AAFP). Morb Mort Weekly Rep 1999;48(No.RR-5):1±15.