Strategies to Prevent Varicella among Newly Arrived Adult Immigrants and Refugees: A Cost-Effectiveness Analysis

In temperate, industrialized countries, such as Canada, varicella is a common disease in childhood [1]. In tropical countries, for reasons that are not entirely clear, varicella occurs among young adults. Seroprevalence data from tropical regions suggest that up to 30% of individuals are susceptible to varicella at 20 years of age, and 5%–10% remain susceptible at 30 years of age [2, 3]. There have been many reported outbreaks of varicella in immigrant populations in industrialized countries, suggesting that foreign-born adults are disproportionately susceptible [4–6]. Recent immigrants also have a high risk of exposure, because many are the parents of young children, who have a high annual incidence of varicella unless they are vaccinated. Varicella is more severe in adults than in children and is associated with a greater risk of complications and death [7]. The fetuses and neonates of pregnant women who develop varicella may develop congenital or neonatal varicella, with high case-fatality rates [8, 9]. Preventing varicella in young adult immigrants, therefore, is important because of the susceptibility of this population and the potential for adverse outcomes. A live-attenuated vaccine was licensed in Canada in 1998 and is now universally available for children <13 years of age and for at-risk adults [10]. Several studies have documented the cost-effectiveness of routine childhood immunization for varicella [11–14]. There is also growing evidence supporting the vaccination of selected adults [15–18]. Figueira et al. [19] documented the cost-effectiveness of immunization for varicella in refugee children. We conducted a cost-effectiveness analysis to identify the optimal vaccination strategy for adult immigrants and refugees arriving in industrialized countries. The model. We constructed a decision-analysis model to compare the cost-effectiveness of 4 potential vaccination strategies for new adult immigrants and refugees (using TreeAgePro 2005; TreeAge Software). Markov processes were incorporated to address the recurrent annual risk of varicella, changing immunity over time since arrival, and waning protection after successful vaccination. The model was analyzed from a societal perspective and included the direct costs of health care and indirect costs to patients and families (notably, time lost from work). Estimates of relative cost-effectiveness reflected cost per case of varicella prevented and cost per quality-adjusted life-year (QALY) gained. All future outcomes and costs were discounted at an annual rate of 3%, as recommended by the Panel on Cost-Effectiveness in Health and Medicine [20–22]. We also examined cases of permanent disability from varicella encephalitis and death as secondary outcomes. We constructed an additional model to specifically investigate the impact of these vaccination strategies on congenital and neonatal outcomes. Vaccination strategies. Four vaccination strategies were compared with a “no intervention” strategy. These strategies were (1) vaccination of all individuals, (2) serological testing of all individuals and vaccination of those with results indicating susceptibility to varicella, (3) vaccination of individuals with a negative or uncertain history of varicella, and (4) serological testing of those individuals with a negative or uncertain history of varicella and vaccination of those with results indicating susceptibility to varicella. Figure 1 illustrates the Markov process used to estimate varicella-related outcomes. Illustration of the Markov process used to determine the expected number of varicella cases, the number of cases of varicella-related permanent disability from encephalitis, and the number of varicella-related deaths. It includes 4 end states (death, permanent disability, immune, and susceptible). Seroprevalence estimates. This analysis builds on a previous seroprevalence study of vaccine-preventable diseases among immigrants and refugees in Montreal, Quebec [23]. A cohort of 1480 adult immigrants and refugees from 6 major geographic regions of the world [24] were recruited from 5 Montreal clinics and hospitals. The mean age (±SD) of the cohort was 32.3 ± 9 years, and individuals had lived in Canada for a mean (±SD) of 21 ± 16 months. Subjects were administered a questionnaire to gather information on age, country of origin, and past history of vaccine-preventable disease. Serological testing was performed for varicella and other vaccine-preventable diseases. Table 1 summarizes the observed seroprevalence of varicella antibodies by age and region of origin. Seroprevalence of varicella IgG antibodies by region of origin and age. In addition, we estimated varicella seroprevalence within a hypothetical cohort of new immigrants arriving in the United States: we standardized the population structure to reflect the regions of origin of immigrants who arrived in the United States between 2003 and 2005. In our primary analysis, the population of interest is a hypothetical cohort of 100,000 individuals aged 30 years with the same overall seroprevalence (92%) as in the Montreal study [23] who present for routine care during the year following their arrival in Canada. The simulation covers a 20-year period. Probability and cost estimates. Probabilities and costs were derived from published studies and cost data from Public Health Departments in the province of Quebec [27]. Table 2 lists base-case estimates and ranges for key probabilities and costs. Probabilities and costs for varicella prevention strategies. Force of infection. In the base-case analysis, we estimated an annual attack rate of 7% for varicella among susceptible adults. Brisson et al. [29] reported attack rates of 0.04–0.09 infections per susceptible-year for individuals aged ≥19 years in Canada that were based on physician billing data from the province of Manitoba. Similar rates were documented in the United Kingdom [29], Luxembourg [35], and Australia [36]. Vaccine efficacy. We assumed seroconversion rates of 78% and 99% after the first and second doses of vaccine, respectively [11, 12, 15, 16, 31, 38]. Because the varicella vaccine does not provide life-long immunity, we assumed that 3% of successfully vaccinated individuals would revert to susceptibility each year [31]. Acceptance of and adherence to vaccination. In a large varicella screening and vaccination program conducted among health care workers in Singapore, 26%–57% refused vaccination despite counseling [28]. We estimated that 30% of individuals would refuse vaccination despite screening according to history or serological test results, similar to previous analyses [17, 18]. Among individuals accepting vaccination, we estimated that 90% would return for the second dose of vaccine, as reported in previous studies [15]. Value of history and serological test results. A self-reported history of varicella is highly predictive of immunity in most populations, although it is somewhat less so in 1 study involving immigrants [18, 41]. In contrast, up to 87% of individuals with a negative or uncertain history of varicella will have serological evidence of immunity. In our previous seroprevalence study, we found that a self-reported history of prior varicella had a sensitivity of 43% and a specificity of 89% with respect to the presence of varicella antibodies [23]. These values were used in the base-case analysis. We assumed a sensitivity of 99% and a specificity of 99% for serological testing [16, 17, 37]. Pregnancy-related events. For women aged 30 years, we estimated an annual live birth rate of 9.1% [42]. This rate decreased to 3.6% by age 35 years and to 0.6% by age 40 years. Congenital varicella was estimated to occur in 2.2% of fetuses exposed to maternal varicella during the first 20 weeks of gestation [15]. If maternal varicella occurred within 5 days of delivery, an estimated 20% of neonates were expected to contract varicella [15]. Costs. All costs were calculated in Canadian dollars for the year 2005. Published costs from previous years were converted into 2005 dollars using the Consumer Price Index for health care products from Statistics Canada [42]. Direct costs for hospitalizations were derived from a published report by the Montreal Public Health Department [27], which describes 909 hospitalizations for varicella in Quebec over the period 1994–1996. Additional costs, including physician charges for treating mild disease, costs of prescription medications, and household expenditures, were derived from 2 separate Canadian studies [33, 43]. We estimated an annual direct medical cost of Can$45,000 for individuals who developed permanent disability from varicella encephalitis [15]. Indirect costs were calculated on the basis of previously published data from Quebec [34], as well as from Germany [13] and France [14]. To estimate the cost of lost productivity associated with varicella infection, we attributed lost hours of work to all adults with varicella. Although only 60% of individuals participate in the labor force [42], we also intended to capture the value of lost productivity by individuals not involved in paid employment, including the value of housework and child care. Adults who contracted varicella were expected to miss 10.7 days of work and/or other duties [13–15]. We derived hourly wage rates specific to the age and sex distribution of our population from national data [42]. We then halved these values, because immigrants to Canada earn an average of 50% of the earnings of Canadian-born persons of the same age and sex during the first 5 years after arrival [44]. We assumed that death or permanent disability caused by varicella resulted in forgone income of Can$20,000 per person per year, a figure based on the average household income of participants in the Montreal seroprevalence study [23]. The varicella vaccine was assumed to cost Can$35.01, including administration fees and medical costs associated with minor adverse reactions (which are expected to occur in 2% of vaccinated individuals) [12, 15, 32]. Materials, processing, and administrative fees for serological testing were assumed to cost Can$20.00 [12, 16]. We did not assign an additional cost for the initial health care visit, because we assumed that screening for varicella would occur during a routine visit. Utilities. Varicella infection in adults is usually associated with transient symptoms that have a consistent impact on an individual's quality of life. We used previously published utility estimates for varicella-related health states [16, 39, 40]. We estimated a mean duration of illness of 7 days for varicella infection. Patients requiring hospitalization were estimated to have lengths of stay of 4 days for supportive care and 7 days for treatment of complications. Patients were assigned a health utility score of 0.4 for hospitalized days and 0.7 for nonhospitalized days. Sensitivity analyses. One-way sensitivity analyses were performed for all variables, and a 2-way sensitivity analysis was performed for the sensitivity and specificity of a self-reported history of varicella. Wherever possible, the lowest and highest values derived from the literature were used to estimate the range across which parameters varied. Threshold analyses were performed when variation in any parameter resulted in a change of optimal strategy. In the past decade, varicella exposure may have become less frequent and attack rates may have decreased because of the introduction of childhood vaccination programs. Therefore, in sensitivity analysis, we considered annual attack rates that were as low as 0.033% among individuals susceptible to varicella [16]. Base-case analysis. Results for total costs, effectiveness, and incremental cost-effectiveness from the base-case analysis are summarized in table 3. With no intervention offered, we predicted that 5020 cases of varicella would occur over a 20-year period in our cohort of 100,000 individuals, costing Can$1.2 million in direct medical costs and Can$4.3 million in lost productivity. Base-case analysis of cost, effectiveness and incremental cost-effectiveness. In the base-case analysis, selective serological testing of individuals with no self-reported history of varicella was the least costly method, saving Can$585,000 relative to no intervention. This strategy was expected to prevent 37% of cases, relative to no intervention. Serological testing of all individuals was also cost-saving relative to no intervention and was expected to prevent 42% of cases. This strategy was more costly than the selective serological testing strategy and was associated with an incremental cost-effectiveness ratio of Can$2618 per additional case avoided and Can$384,600 per QALY gained, relative to the selective strategy. Figure 2 illustrates how variation in seroprevalence yields a range of net savings for each strategy relative to no intervention. At a seroprevalence of 84% (which is the seroprevalence found among South Asian immigrants <35 years old), all strategies were cost-saving relative to no intervention, although selective serological testing remained the cheapest strategy. At a seroprevalence of 97%, corresponding to the overall seroprevalence among individuals aged ≥35 years, no intervention was the cheapest strategy. At a seroprevalence of ≤70%, the expected costs of the strategies changed, such that selective serological testing was no longer the cheapest. However, the probability of varicella immunity will exceed 70% among most adult immigrants. The threshold seroprevalences below which each of the strategies were cost-saving, compared with no intervention, were as follows: selective serological testing, 95%; serological testing of all individuals, 92%; selective vaccination, 90%; and vaccination of all individuals, 85%. Variation in net savings per 100,000 immigrants by seroprevalence for each vaccination strategy, relative to no intervention. Net savings are in millions of Canadian dollars. Positive values refer to net savings, and negative values indicate net costs. Secondary outcomes. In the absence of any intervention, we predicted 14 deaths and 12 cases of permanent disability per 10 million immigrants. The selective serological testing strategy was expected to prevent 38% of deaths and cases of permanent disability. Congenital and neonatal sequelae from varicella were also rare outcomes (table 4). The selective serological testing strategy was expected to prevent one-half of the cases of congenital and neonatal varicella and, therefore, one-half of the ensuing outcomes of permanent disability and death. Comparison of expected fetal and neonatal varicella outcomes with no intervention versus with selective serological testing. Sensitivity analyses. One-way sensitivity analyses suggested that our results were robust across a wide range of assumed values for most key parameters, including utility estimates. However, an important finding was that it became cheaper not to intervene when the annual varicella attack rate was <3.8% among susceptible persons. Other thresholds at which it became cheaper not to intervene were a vaccine-refusal rate>54%, a serological testing cost>Can$31 per test, or productivity losses

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.