Benefits and costs of using HPV testing to screen for cervical cancer Mandelblatt J S, Lawrence W F, Womack S M, Jacobsen D, Yo B, Hwang Y, Gold K, Barter J, Shah K Record Status This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn. Health technology Three screening strategies for cervical cancer were examined. The strategies were Papanicolaou (Pap) test, human papillomavirus (HPV) testing, and their combination. As the three screening strategies could be performed at two different intervals (every 2 or 3 years) and could be stopped at different moments (65 years, 75 years or at death, such as 100 years), eighteen possible screening strategies were examined. These were: Pap every 3 years with screening stopped at 65, 75 and 100 years; HPV every 3 years with screening stopped at 65, 75 and 100 years; Pap every 2 years with screening stopped at 65, 75 and 100 years; HPV every 2 years with screening stopped at 65, 75 and 100 years; Pap plus HPV every 3 years with screening stopped at 65, 75 and 100 years; Pap plus HPV every 2 years with screening stopped at 65, 75 and 100 years. Type of intervention Screening. Economic study type Cost-utility analysis. Study population The study population comprised the average US population of women, aged 20 years and over. Women with human immunodeficiency virus infection were excluded from the analysis. Setting The setting appears to have been the community. The economic study was carried out in the USA. Dates to which data relate The effectiveness and resource use data were derived from studies published between 1986 and 2002. The price year was 2000. Source of effectiveness data The effectiveness evidence was derived from published studies, and was supported by the authors' assumptions. Page: 1 / 9
Modelling A decision analytic model based on a Markov process was constructed in order to simulate the natural history of cervical cancer. Also, to estimate the lifetime costs and benefits of the women included in the analysis. There were seven model states and a hypothetical cohort of 1 million women was considered moving though one-year cycles. Women could remain in the same health state or move between states, as a result of being screened, developing symptoms, having a hysterectomy for noncancer reasons, or dying from cervical cancer of other causes. Outcomes assessed in the review The health outcomes assessed in the review and used as inputs for the model were the prevalence of HPV/low-grade squamous intraepithelial lesion (LSIL); the probabilities of progression from healthy to HPV/LSIL; the probabilities of regression from HPV/LSIL to healthy, persistence of HPV/LSIL; the probabilities of progression from HPV/LSIL to high-grade squamous intraepithelial lesion (HSIL); the probabilities of regression from HSIL to HPV/LSIL; the probabilities of persistence of HSIL; the probabilities of progression from HSIL to invasive cancer; the sensitivity and specificity of the Pap smear, HPV testing, and colposcopy and biopsy; the cure rates for LSIL and HSIL; the compliance rates with screening, diagnosis and treatment; the 5-year survival for local invasive, regional invasive, distant invasive cancer and average annual all-cause mortality; the relative risk of death with local and regional and distant disease; and the utilities for quality-adjusted survival in case of false-positive, LSIL, HSIL, local, regional and distant disease. Study designs and other criteria for inclusion in the review Not stated. Sources searched to identify primary studies Not stated. Criteria used to ensure the validity of primary studies Not stated. Methods used to judge relevance and validity, and for extracting data The authors stated that the data were abstracted from the best-quality published studies. Number of primary studies included The effectiveness evidence were derived from 60 primary studies. Page: 2 / 9
Methods of combining primary studies This was not stated in all cases, but for transition probabilities for oncogenenic HPV and Pap smear performance, data were stated to have been pooled. Investigation of differences between primary studies Not stated. Results of the review The prevalence of HPV/LSIL was 0.245 in the age class 20 to 24 years and 0.009 for women older than 74 years. The probability values were: 0.225 to 0.284 (age class 20 to 65 years) for progression from healthy to HPV/LSIL; 0.284 for regression from HPV/LSIL to healthy; 0.010 to 0.079 (age class 20 to 65 years) for progression from HPV/LSIL to HSIL; 0.250 for regression from HSIL to HPV/LSIL; and 0.011 to 0.057 (age class 20 to 65 years) for progression from HSIL to invasive cancer. For LSIL, the sensitivity of Pap in the case of atypical squamous cell of undetermined significance (ASCUS) negative was 75% for women under 55 years and 47% for women 55 years or older. The corresponding values for HSIL were 84% for women under 55 years and 61% for women 55 years or older. The specificity for LSIL was 97% for women under 55 years and 69% for women 55 or older. The sensitivity of Pap in the case of ASCUS positive cut points was 67% for LSIL and 80% for HSIL, while the specificity was 87%. The sensitivity of Pap in the case of thin prep was 70% for LSIL and 88% for HSIL, while the specificity was 89%. For the HPV/DNA testing with hybrid capture II/polymerase chain reaction, the sensitivity was 43% for women under 55 years and 70% for women 55 years or older for LSIL. The corresponding values for HSIL were 63% for women under 55 years and 95% for women 55 years or older. The specificity for LSIL was 80% for women under 55 years and 95% for women 55 or older. The sensitivity and specificity of colposcopy and biopsy (with endocervical curettage as needed) for LSIL were both 100%. The cure rates were 95% (range: 85-98) for LSIL and 98% (range: 90-99) for HSIL. The compliance rates for screening were 80% (range: 10-100) for women under 70 years and 58.7% (range: 10-100%) for women 70 years or older. The compliance rates for diagnosis and treatment were both 100% (range: 10-100). The average (all ages) 5-year survival was 0.875 for local invasive disease, 0.436 for regional invasive disease, and 0.079 for distant invasive cancer. For annual all-cause mortality, the average 5-year survival was 0.0005 for women in the age class 20 to 24 years and 0.147 for women aged 85 years or older. The relative risk of death was 0.54 (95% confidence interval, CI: 0.34-0.86) with local disease and 0.58 (95% confidence interval: 0.40-0.81) with regional and distant disease. The utilities for quality-adjusted survival were 0.97 for false-positive and LSIL, 0.93 for HSIL, 0.90 for local disease, Page: 3 / 9
0.70 for regional disease, and 0.50 for distant disease. Methods used to derive estimates of effectiveness The authors made some assumptions that were used in the decision model. Estimates of effectiveness and key assumptions The following assumptions were made: cervical cancer developed as a result of the progression of unclear HPV infection to high grade and eventually invasive disease; cervical neoplasia represented the natural history of HPV infection and rarely (about 5%) occurred in the absence of this infection; the state of newly acquired HPV infection was combined with LSIL; ASCUS was considered as a negative test result; women would return to screening in the next interval and would not have colposcopy until the development of HPV/LSIL; women treated for HPV/LSIL and cured, returned healthy and acquired new HPV infection at similar rates to women without prior HPV/LSIL, while those women treated but not cured remained in the HPV/LSIL state; the HPV and PAP smear results were conditionally independent; the transition rates were age-dependent; all women with abnormal screening were referred for colposcopy and biopsy; all women complied with diagnostic evaluation and treatment after a positive screening result. Measure of benefits used in the economic analysis The main benefit measure was the quality-adjusted life-years (QALYs). These were derived from modelling and a 3% discount rate was used. The numbers of cervical cancer cases and deaths were also reported as model outcomes. Direct costs A 3% discount rate was used as the costs were incurred over a timeframe of longer than 2 years. The unit costs were given and a complete breakdown of the costs was reported. The cost items included in the analysis were consumable supplies, personnel, laboratory, and procedural costs relative to the two screening tests (Pap and HPV), diagnosis, initial treatment and terminal care. The cost/resource boundary adopted was that of the hospital. When possible, the cost data were based on resource utilisation or a microcosting approach. Otherwise, gross cost accounting methods were used, such as Medicare reimbursement rates. All the costs were inflated to 2000 values using the medical care component of the Consumer Price Index. Statistical analysis of costs Statistical analyses of the costs were not carried out. Indirect Costs The indirect costs were included in the analysis. These included the time spent by the patient for screening, diagnosis, Page: 4 / 9
treatment, travel and waiting. A 3% discount rate was used as the time horizon of the analysis was longer than 2 years. The unit costs were reported. The cost data were obtained from published studies and the National Health Interview Survey. The costs of lost productivity were excluded, as it was assumed that they were accounted for by decrements in utilities. The price year was 2000. Currency US dollars ($). Sensitivity analysis One-way sensitivity analyses were carried out to test the robustness of the decision model to variations in several variables. The variables investigated included the screening interval, HSIL progression, LSIL incidence, HPV test cost, HPV sensitivity and HPV specificity. The probability that strategies were cost-effective was tested using confidence intervals, which were determined using bootstrap replications of the cohort sample. In correspondence with the authors, they stated that,selected two way analysis' was also performed. Estimated benefits used in the economic analysis There were: 2,382 cancers, 1,822 deaths and 26.8666 QALYs with no screening; 1,020 cancers, 434 deaths and 27.0175 QALYs with Pap every 3 years and screening stopped at 65 years; 817 cancers, 305 deaths and 27.02 QALYs with Pap every 3 years and screening stopped at 75 years; 750 cancers, 253 deaths and 27.0204 QALYs with Pap every 3 years and screening stopped at 100 years; 1,009 cancers, 418 deaths and 27.0183 QALYs with HPV every 3 years and screening stopped at 65 years; 800 cancers, 284 deaths and 27.0209 QALYs with HPV every 3 years and screening stopped at 75 years; 729 cancers, 229 deaths and 27.0213 QALYs with HPV every 3 years and screening stopped at 100 years; 796 cancers, 352 deaths and 27.0315 QALYs with Pap every 2 years and screening stopped at 65 years; 523 cancers, 185 deaths and 27.035 QALYs with Pap every 2 years and screening stopped at 75 years; 437 cancers, 124 deaths and 27.0355 QALYs with Pap every 2 years and screening stopped at 100 years; 749 cancers, 319 deaths and 27.0321 QALYs with Pap plus HPV every 3 years and screening stopped at 65 years; 792 cancers, 348 deaths and 27.0314 QALYs with HPV every 2 years and screening stopped at 65 years; 525 cancers, 182 deaths and 27.0347 QALYs with Pap plus HPV every 3 years and screening stopped at 75 years; 450 cancers, 127 deaths and 27.0352 QALYs with Pap plus HPV every 3 years and screening stopped at 100 years; 515 cancers, 177 deaths and 27.035 QALYs with HPV every 2 years and screening stopped at 75 years; 425 cancers, 113 deaths and 27.0356 QALYs with HPV every 2 years and screening stopped at 100 years; 607 cancers, 285 deaths and 27.0408 QALYs with Pap plus HPV every 2 years and screening stopped at 65 years; 317 cancers, 113 deaths and 27.0444 QALYs with Pap plus HPV every 2 years and screening stopped at 75 years; Page: 5 / 9
225 cancers, 51 deaths and 27.045 QALYs with Pap plus HPV every 2 years and screening stopped at 100 years. Cost results The total costs were: $6,804 with Pap every 3 years and screening stopped at 65 years; $6,833 with Pap every 3 years and screening stopped at 75 years; $6,851 with Pap every 3 years and screening stopped at 100 years; $6,904 with HPV every 3 years and screening stopped at 65 years; $6,941 with HPV every 3 years and screening stopped at 75 years; $6,964 with HPV every 3 years and screening stopped at 100 years; $7,230 with Pap every 2 years and screening stopped at 65 years; $7,280 with Pap every 2 years and screening stopped at 75 years; $7,308 with Pap every 2 years and screening stopped at 100 years; $7,348 with Pap plus HPV every 3 years and screening stopped at 65 years; $7,388 with HPV every 2 years and screening stopped at 65 years; $7,393 with Pap plus HPV every 3 years and screening stopped at 75 years; $7,422 with Pap plus HPV every 3 years and screening stopped at 100 years; $7,452 with HPV every 2 years and screening stopped at 75 years; $7,489 with HPV every 2 years and screening stopped at 100 years; $7,857 with Pap plus HPV every 2 years and screening stopped at 65 years; $7,934 with Pap plus HPV every 2 years and screening stopped at 75 years; and $7,980 with Pap plus HPV every 2 years and screening stopped at 100 years. Synthesis of costs and benefits An incremental cost-utility analysis was carried out to combine the costs and QALYs. Each alternative was compared with the next most effective non-dominated option. The ranking was as follows: Pap every 3 years and screening stopped at 65 years, not cost-effective; Pap every 3 years and screening stopped at 75 years, $11,830 per QALY; Pap every 3 years and screening stopped at 100 years, not cost-effective; HPV every 3 years and screening stopped at 65 years, dominated; HPV every 3 years and screening stopped at 75 years, not cost-effective; Page: 6 / 9
HPV every 3 years and screening stopped at 100 years, not cost-effective; Pap every 2 years and screening stopped at 65 years, not cost-effective; Pap every 2 years and screening stopped at 75 years, $29,871 per QALY; Pap every 2 years and screening stopped at 100 years, $56,440 per QALY; Pap plus HPV every 3 years and screening stopped at 65 years, dominated; HPV every 2 years and screening stopped at 65 years, dominated; Pap plus HPV every 3 years and screening stopped at 75 years, dominated; Pap plus HPV every 3 years and screening stopped at 100 years, dominated; HPV every 2 years and screening stopped at 75 years, dominated; HPV every 2 years and screening stopped at 100 years, not cost-effective; Pap plus HPV every 2 years and screening stopped at 65 years, not cost-effective; Pap plus HPV every 2 years and screening stopped at 75 years, $70,347 per QALY; and Pap plus HPV every 2 years and screening stopped at 100 years, $76,183 per QALY. When the cost of the HPV test fell to below a threshold of $5, the screening option of HPV test every 2 years and screening stopped at 100 years became cost-effective ($50,100 per QALY). The final ranking was also sensitive to changes in the HPV test sensitivity and HPV/LSIL prevalence. Other changes may affect the overall costs, but not the ranking. Authors' conclusions The maximum savings in life could be gained with the screening option of Papanicolaou (Pap) plus human papillomavirus (HPV) tests every 2 years, from 20 years to death. The cessation of screening at 65 or 75 years resulted in less costs and captured 86.6 and 97.8%, respectively, of the benefits of lifetime biennial screening. CRD COMMENTARY - Selection of comparators The rationale for the choice of the comparators was clear. All eighteen strategies represented possible screening options for cervical cancer. The option of no screening was not considered practical since standard care employs some form of screening. You should assess which screening option is currently implemented in your own setting. Validity of estimate of measure of effectiveness The effectiveness analysis used estimates derived from published studies. Although it was not stated whether a formal review of the literature was undertaken, the authors stated that data from the best-quality published studies were used. The methods used to combine the data from the primary studies were not specified for all parameter estimates. In addition, it was unclear whether the authors considered the impact of differences in the populations of the studies, although prevalence was varied in the sensitivity analysis. However, the use of several assumptions and short-term utility data may have limited the validity of the analysis, although many assumptions were tested in the sensitivity analyses. Also, in correspondence with the authors, they state that a formal literature review was undertaken. Validity of estimate of measure of benefit The benefit measure used in the economic analysis consisted of QALYs. These were derived from a decision model, Page: 7 / 9
which appears to have been appropriate to simulate the natural history of the disease. A panel of experts validated the model. Appropriate discounting was also carried out. The use of QALYs ensures the comparability with benefits of other interventions carried out in the health care system. Validity of estimate of costs The analysis of the costs was carried out from a societal perspective, and the direct medical and non-medical costs were included in the study. The unit costs and ranges used in the sensitivity analyses were reported. A complete breakdown of the costs was given, thus enhancing generalisability. The incremental analysis was correctly performed. The price year was appropriately indicated. Other issues The authors made some comparisons of their findings with those from other studies. The issue of the generalisability of the study results was not explicitly addressed. However, the results of the sensitivity analyses were reported clearly, and the unit costs and the price year were given. These enhanced the external validity of the analysis. All the results were reported clearly and in full. The study results appear to be generalisable to the general population of women older than 20 years of age in the USA. The authors pointed out some strengths and limitations of their analysis. Implications of the study The authors suggest that "regardless of modality chosen, the greatest gains from screening will depend on reaching all women end ensuring access to diagnosis after an abnormal screening result (and treatment, if malignant)". This was generally a very well conducted and reported study. Source of funding Supported by the National Institute on Aging, grant number RO1-AG15340, and the Department of the Army, contract DAMD17-9444-J-4212. Bibliographic details Mandelblatt J S, Lawrence W F, Womack S M, Jacobsen D, Yo B, Hwang Y, Gold K, Barter J, Shah K. Benefits and costs of using HPV testing to screen for cervical cancer. JAMA 2002; 287(18): 2372-2381 PubMedID 11988058 Other publications of related interest Comment: JAMA 2002;287:2428-9. Indexing Status Subject indexing assigned by NLM MeSH Adult; Aged; Cervical Intraepithelial Neoplasia /diagnosis /economics /prevention & control /virology; Cost-Benefit Analysis; DNA, Viral /analysis; Female; Humans; Mass Screening /economics /methods; Middle Aged; Models, Theoretical; Papanicolaou Test; Papillomaviridae /isolation & purification; Papillomavirus Infections /diagnosis; Polymerase Chain Reaction /economics; Quality-Adjusted Life Years; Tumor Virus Infections /diagnosis; United States; Uterine Cervical Neoplasms /diagnosis /economics /prevention & control /virology; Vaginal Smears /economics AccessionNumber 22002008122 Page: 8 / 9
Powered by TCPDF (www.tcpdf.org) Date bibliographic record published 31/01/2003 Date abstract record published 31/01/2003 Page: 9 / 9