Cost-Effectiveness of Chlamydia Vaccination for Young Women
Cost-Effectiveness of Chlamydia Vaccination for Young Women
We used a deterministic heterosexual transmission model that was relatively simple compared with previously published models to explore the potential health and economic outcomes of a hypothetical chlamydia vaccine focusing on vaccination programs for 14-year-old girls and 15–24-year-old women in the United States. We repeated our analyses by using a higher disease burden. Overall, results from our exploratory analyses showed that a chlamydia vaccine could be cost-effective under many plausible scenarios. Interventions that reduce QALYs lost for <1–3 times per capita gross domestic product (≈$50,000 in the United States) are typically considered to be cost-effective. Our sensitivity analyses suggest that a highly efficacious chlamydia vaccine with long duration of immunity might be cost-saving in countries with high prevalence of chlamydia, as demonstrated by results of our additional analysis. Our results are consistent with preliminary, spreadsheet-based calculations, which suggested that a chlamydia vaccine would cost <$10,000/QALY saved.
Our analyses showed that a high-performance vaccine could potentially eliminate chlamydia infection when coverage was high (>75%) among susceptible persons before their sexual debut. These results were consistent with findings from previous studies, and our estimates of cost-effectiveness of chlamydia screening (versus no screening) were consistent with those of previous studies. In addition, the relative cost-effectiveness of targeting different age groups was consistent with results of previous studies on HPV vaccine. In particular, our study showed that catch-up vaccination of 15–24-year-old women, in addition to 14-year-old girls, resulted in an increase in the ICER, implying that additional QALYs are gained at higher costs. Consistent with results of Elbasha et al. the addition of catch-up vaccination of 15–24-year-old women did not change the long-term prevalence of infection, but did shorten the time needed to realize the effects of vaccination.
An additional aspect of vaccination is that it is easier to implement than an intervention of routine screening because it does not need to be repeated annually. Although health services data have shown chlamydia screening rates ≥30% in young women, time-series insurance data have shown that <1% of women ≤25 years of age are consistently screened at least once per year.
Our exploratory study has several limitations. Notable among them is the inherent limitations associated with models in general because models are simplifications of real-world events. Thus, all limitations associated with models are applicable. Another major limitation is the high uncertainty surrounding the parameter values we used (including illness estimates). Because we focused on heterosexual transmission, our model was largely driven by parameters associated with women; prevaccination prevalence was calibrated to approximate reported illnesses for women, and prevaccination prevalence for men was determined by the model. Because a substantially high proportion of high-impact health outcomes of chlamydia infection are in women (i.e., PID and associated complications), it is reasonable to focus on illness in women in such analyses. Nonetheless, as was conducted for HPV, future studies should also assess cost-effectiveness of chlamydia vaccination for men.
The prevaccination prevalence rates for men determined by our model were substantially different from reported prevalence rates for men in the United States. For instance, the reported prevalence for men of a similar age group (15–24 years) in the United States was approximately half that of women (men 1.66%; women 3.21%), and prevalence in men from our main analysis was substantially higher (men 2.79%; women 3.24%).
We excluded numerous possible outcomes of chlamydia vaccination, such as changes in the number of partners or screening practices, which might arise as a result of vaccination, health benefits for persons vaccinated while infected, and costs and loss in quality of life to persons who experience potential adverse vaccination outcomes, such as side effects (e.g., temporary pain at injection site).
We did not explore potential broader properties of an effective chlamydia vaccine, such as degree (i.e., reducing susceptibility but not completely eliminating it) or infectiousness (i.e., breakthrough infections being less infectious than primary infections and shorter in duration). Future studies should consider assessing these 2 characteristics (degree and infectiousness). We assumed that all members of the hypothetical population (with substantially different sexual activity levels) have equal access to screening, treatment, and vaccination. Thus, treatment rate, screening rate, and vaccination coverage were applied equally across all eligible model compartments (subpopulations). However, this simplifying assumption is probably not realistic. Consequently, benefits of screening and vaccination might have been overestimated if women who are highly sexually active were less likely to be screened, treated, or vaccinated. In addition, it is also conceivable that persons vaccinated might be less likely to be screened for chlamydia annually. Further studies are needed to explore the potential health and economic benefits of a chlamydia vaccine that targets specific subpopulations, such as persons infected, those with limited access to health care, and those who have multiple sexual partners.
Because our model does not account for major factors, such as age-based mixing of sexual partners and ongoing sexual partnerships, our model is not of sufficient complexity to inform chlamydia vaccine recommendations. For example, our model assumed sexual debut at 15 years of age and that sex partners were chosen from a pool of 15–24 year-old persons, thereby ignoring heterogeneity in age at sexual debut, which is a simplification. Similarly, models such as ours that do not specifically keep track of ongoing sexual partnerships can overestimate the effect of chlamydia screening because reinfection of treated women by their untreated sex partner is not specifically taken into account. If the effect of chlamydia screening is overestimated, then the marginal effect of adding chlamydia vaccination to an existing chlamydia screening program might be underestimated. Development of more complex models will be needed over time, and these models would be better suited to examine the effect of vaccination over a wide range of assumptions regarding vaccine coverage, efficacy, and duration of protection.
Notwithstanding these limitations, our model provides useful information on the potential cost-effectiveness of a chlamydia vaccine, as well as a useful basis for future chlamydia vaccine cost-effectiveness analyses and other modeling studies. In particular, determination of the hierarchy of influential parameters in our model would be useful for future analyses, and assist in understanding the relative roles played by numerous variables that are used in models to facilitate discussions around simple and complex model inputs. Finally, our study suggests that a successful chlamydia vaccine could have a substantial effect on chlamydia prevalence, thereby reducing the health and economic burden associated with chlamydia.
Discussion
We used a deterministic heterosexual transmission model that was relatively simple compared with previously published models to explore the potential health and economic outcomes of a hypothetical chlamydia vaccine focusing on vaccination programs for 14-year-old girls and 15–24-year-old women in the United States. We repeated our analyses by using a higher disease burden. Overall, results from our exploratory analyses showed that a chlamydia vaccine could be cost-effective under many plausible scenarios. Interventions that reduce QALYs lost for <1–3 times per capita gross domestic product (≈$50,000 in the United States) are typically considered to be cost-effective. Our sensitivity analyses suggest that a highly efficacious chlamydia vaccine with long duration of immunity might be cost-saving in countries with high prevalence of chlamydia, as demonstrated by results of our additional analysis. Our results are consistent with preliminary, spreadsheet-based calculations, which suggested that a chlamydia vaccine would cost <$10,000/QALY saved.
Our analyses showed that a high-performance vaccine could potentially eliminate chlamydia infection when coverage was high (>75%) among susceptible persons before their sexual debut. These results were consistent with findings from previous studies, and our estimates of cost-effectiveness of chlamydia screening (versus no screening) were consistent with those of previous studies. In addition, the relative cost-effectiveness of targeting different age groups was consistent with results of previous studies on HPV vaccine. In particular, our study showed that catch-up vaccination of 15–24-year-old women, in addition to 14-year-old girls, resulted in an increase in the ICER, implying that additional QALYs are gained at higher costs. Consistent with results of Elbasha et al. the addition of catch-up vaccination of 15–24-year-old women did not change the long-term prevalence of infection, but did shorten the time needed to realize the effects of vaccination.
An additional aspect of vaccination is that it is easier to implement than an intervention of routine screening because it does not need to be repeated annually. Although health services data have shown chlamydia screening rates ≥30% in young women, time-series insurance data have shown that <1% of women ≤25 years of age are consistently screened at least once per year.
Our exploratory study has several limitations. Notable among them is the inherent limitations associated with models in general because models are simplifications of real-world events. Thus, all limitations associated with models are applicable. Another major limitation is the high uncertainty surrounding the parameter values we used (including illness estimates). Because we focused on heterosexual transmission, our model was largely driven by parameters associated with women; prevaccination prevalence was calibrated to approximate reported illnesses for women, and prevaccination prevalence for men was determined by the model. Because a substantially high proportion of high-impact health outcomes of chlamydia infection are in women (i.e., PID and associated complications), it is reasonable to focus on illness in women in such analyses. Nonetheless, as was conducted for HPV, future studies should also assess cost-effectiveness of chlamydia vaccination for men.
The prevaccination prevalence rates for men determined by our model were substantially different from reported prevalence rates for men in the United States. For instance, the reported prevalence for men of a similar age group (15–24 years) in the United States was approximately half that of women (men 1.66%; women 3.21%), and prevalence in men from our main analysis was substantially higher (men 2.79%; women 3.24%).
We excluded numerous possible outcomes of chlamydia vaccination, such as changes in the number of partners or screening practices, which might arise as a result of vaccination, health benefits for persons vaccinated while infected, and costs and loss in quality of life to persons who experience potential adverse vaccination outcomes, such as side effects (e.g., temporary pain at injection site).
We did not explore potential broader properties of an effective chlamydia vaccine, such as degree (i.e., reducing susceptibility but not completely eliminating it) or infectiousness (i.e., breakthrough infections being less infectious than primary infections and shorter in duration). Future studies should consider assessing these 2 characteristics (degree and infectiousness). We assumed that all members of the hypothetical population (with substantially different sexual activity levels) have equal access to screening, treatment, and vaccination. Thus, treatment rate, screening rate, and vaccination coverage were applied equally across all eligible model compartments (subpopulations). However, this simplifying assumption is probably not realistic. Consequently, benefits of screening and vaccination might have been overestimated if women who are highly sexually active were less likely to be screened, treated, or vaccinated. In addition, it is also conceivable that persons vaccinated might be less likely to be screened for chlamydia annually. Further studies are needed to explore the potential health and economic benefits of a chlamydia vaccine that targets specific subpopulations, such as persons infected, those with limited access to health care, and those who have multiple sexual partners.
Because our model does not account for major factors, such as age-based mixing of sexual partners and ongoing sexual partnerships, our model is not of sufficient complexity to inform chlamydia vaccine recommendations. For example, our model assumed sexual debut at 15 years of age and that sex partners were chosen from a pool of 15–24 year-old persons, thereby ignoring heterogeneity in age at sexual debut, which is a simplification. Similarly, models such as ours that do not specifically keep track of ongoing sexual partnerships can overestimate the effect of chlamydia screening because reinfection of treated women by their untreated sex partner is not specifically taken into account. If the effect of chlamydia screening is overestimated, then the marginal effect of adding chlamydia vaccination to an existing chlamydia screening program might be underestimated. Development of more complex models will be needed over time, and these models would be better suited to examine the effect of vaccination over a wide range of assumptions regarding vaccine coverage, efficacy, and duration of protection.
Notwithstanding these limitations, our model provides useful information on the potential cost-effectiveness of a chlamydia vaccine, as well as a useful basis for future chlamydia vaccine cost-effectiveness analyses and other modeling studies. In particular, determination of the hierarchy of influential parameters in our model would be useful for future analyses, and assist in understanding the relative roles played by numerous variables that are used in models to facilitate discussions around simple and complex model inputs. Finally, our study suggests that a successful chlamydia vaccine could have a substantial effect on chlamydia prevalence, thereby reducing the health and economic burden associated with chlamydia.