The health and economic impact of scaling cervical cancer prevention in 50 low‐ and lower‐middle‐income countries

To estimate the health impact, financial costs, and cost‐effectiveness of scaling‐up coverage of human papillomavirus (HPV) vaccination (young girls) and cervical cancer screening (women of screening age) for women in countries that will likely need donor assistance.


| Analytic overview
We used a model-based approach to synthesize population, demographic, and epidemiological data from 50 low-and lower-middleincome(LIandLMI)countrieswithpopulationsover1millionpersons and gross national income (GNI) per capita less than or equal to US $2585 (Supplementary material Tables S1 and S2); this represents the midpoint GNI per capita of the World Bank LMI country income tier, 14  direct medical costs of screening, diagnosis, and treatment of precancer; and direct medical costs of cervical cancer treatment by stage) from the literature, using previously described methods. 15 We estimatedcountry-specificepidemiologicdatainputsonburdenofHPV, precancer, and cervical cancer by applying previously described methods, 15 using:(1)multivariateregressionmodelstopredictcountry-and age-specificHPVprevalence(SupplementarymaterialTableS3) 15 ; (2) a peer-reviewed individual-based microsimulation model that was previouslycalibratedtofourseparatelow-andlower-middle-income countries (El Salvador, India, Nicaragua, and Uganda) [16][17][18] to predict country-specificprevalenceofprecancer 15 (Supplementary material Tables S4 and S5). 1 To estimate the effectivenessofHPV-16/18vaccination,wereliedonvaccinetrialdata andepidemiologicdataontheproportionofcervicalcancersattributed to HPV-16/18. 5,19-21 CERVIVAC inputs pertaining to screening and treatmenteffectivenesswerederivedfromthemicrosimulationmodel, whichwasusedtoestimatethereductioninage-specificcervicalcancer incidence and mortality, as well as shifts in stage distribution of detected cervical cancer, associated with each screening strategy. We estimated current access to cancer treatment in each country using published literature to project cervical cancer treatment cost savings associatedwitheachvaccinationandscreeningscenario. 22 The analysis was conducted from a payer perspective.The time horizonwasthelifetimeofbirthcohortswhoreceivedeithervaccina-tionorscreeningbasedonageduringthe10-yearinterventionperiod (2017-2026). Because girls aged 10 years would not be subsequently eligibleforscreeningduringtheinterventionperiod,wedidnotexamine the impact of screening in vaccinated women. We present both undiscounted costs and future costs discounted at an annual rate of3%in2013US$.Healthbenefitsarereportedascervicalcancer cases,deaths,anddisability-adjustedlifeyears(DALYs)averted;DALYs have been discounted at an annual rate of 3%. We present incremental cost-effectivenessratios(ICERs)-asthenetcostperDALYavertedto accountforcancertreatmentoffsets-separatelyforvaccinationand screening(relativetonointervention).Whilethereisnouniversalcriterion that defines a threshold cost-effectiveness ratio,we consider theheuristicthataninterventionwithanICERlessthanGDPpercapitaisestimatedtobe"verycost-effective"andlessthanthreetimes GDPpercapitaisestimatedtobe"cost-effective". 23  Prospects. 24 Eachbirthcohortwastrackedoveritslifetimetocapture health service utilization, burden of disease, and long-term health impactofvaccinationandscreeningduringtheinterventionperiod.

| HPV vaccination and cervical cancer screening strategies
Scale-up assumptions for vaccination and screening are displayed in  For each screening test, we used the microsimulation model to estimate percent reductions in age-specific cervical cancer incidence andmortality(in5-yearagegroupsfromage20toage84years)attributable to a screening program with the features described above (e.g. treatmenteffectiveness,eligibilityforcryotherapy,etc).Descriptionsof thismicrosimulationmodelandtheparameterizationprocess(including model calibration to epidemiologic data for the development of four country-specific models reflecting the natural history of cervical cancer in El Salvador, India, Nicaragua, and Uganda) have been previously published. [16][17][18] Inbrief,weestimatedbaseline"prior"inputvaluesfor natural history transitions using longitudinal data. [25][26][27][28] To reflect heterogeneityinage-andtype-specificHPVincidencebetweensettings, aswell as natural immunity following initial infection and uncertainty in progression and regression of precancer, we set plausible ranges around these input values. We then performed repeated model sim- The health impact of cervical cancer treatment was used for DALY calculations. We used country-specific access to radiation therapy infrastructure as a proxy for access to treatment at any stage, given that most women with cervical cancer present with regionalcancerintheabsenceoforganizedscreening. 22  End Results Program: 92% for local, 57% for regional, and 17% for distant cancer. 28 We assumed that the proportion of women who hadnoaccesstocancertreatmenthad5-yearabsolutesurvivalrates based on a linear regression of survival in the IARC SurvCan database and access to radiation therapy 22,29 (65% for local, 37% for regional,and 16% for distant cancers). Ina sensitivity analysis,we assumed all women with cancer had access to treatment based on the FIGO guidelines.

| Costs
All costs were converted to 2013 US $, and we assumed that intervention costs did not vary with coverage level. For vaccination, we assumedthepriceof$4.50perdoseforGavi-eligiblecountriesandthe PAHO Revolving Fund Price of $8.50 per dose for the four remaining countries (Bolivia, Honduras, Republic of Moldova, and Uzbekistan).
We estimated the country-specific HPV vaccine delivery cost per dose as previously described (Supplementary material Tables S10 and   S11). 15 Forscreening-relatedcosts,weincludedthecountry-specific direct medical costs associated with screening, diagnosis (if relevant), and treatment of precancer, as previously described (Supplementary material Tables S12-S14). 15 Cancer treatment costs were similarly derived for each country, assuming stage-specific treatment protocols based on FIGO guidelines (Supplementary material Tables S15 and S16).

| Cost and cost-effectiveness
Total costs for the HPV vaccination program and cervical cancer screening program, by year, are displayed in Figure 3.