Concurrent Oral Presentation C. COMPARATIVE EFFECTIVENESS

Monday, October 19, 2009: 1:30 PM
Grand Ballroom, Salon 6 (Renaissance Hollywood Hotel)

* Candidate for the Lee B. Lusted Student Prize Competition

Session Chairs:
Kevin D. Frick, PhD and Harold Lehmann
1:30 PM
Sabina Alistar, MS and Margaret L. Brandeau, PhD, Stanford University, Stanford, CA

Purpose: Ukraine has one of the fastest growing HIV epidemics in the world, with half of infections due to injection drug use.  Currently, almost no injection drug users (IDUs) have access to methadone substitution therapy, and only 10% of eligible individuals receive highly active antiretroviral therapy (HAART).  Both types of programs are likely to be scaled up in the near future.  However, the appropriate allocation of resources between the two interventions is unknown.  We estimated the effectiveness and cost effectiveness of strategies for expanding methadone programs (up to 25% of IDUs) and HAART (up to 80% of all eligible patients) in Ukraine.

Method: We developed a dynamic compartmental model of the HIV epidemic in a population of non-IDUs, IDUs who inject opiates, and IDUs on methadone. The parameters for the model were based on data from Ukraine. We considered a population of 1,000,000 individuals aged 15-49 stratified according to HIV status and injection drug use. We analyzed the effects of interventions that focus on expanding methadone, increasing access to HAART, or both. We measured health care costs, quality-adjusted life years (QALYs), HIV prevalence, HIV infections averted, and incremental cost effectiveness.  

Result: With no incremental interventions, HIV prevalence reached 71.7% in IDUs and 1.22% in non-IDUs after 20 years.  Increasing methadone access to 25% of IDUs was the most cost-effective strategy and generated the lowest final HIV prevalence in both IDUs (49.8%) and the general population (0.7%). This strategy averted 5300 infections and added 35,000 QALYs for a cost of $740/QALY gained.  Expanding HAART averted fewer infections, even with 80% coverage: 3700 infections were prevented, adding 57,000 QALYs at a cost of $4,400/QALY gained. Increasing both methadone (to 3.1% of IDUs) and HAART access (to 50% of eligible individuals) averted 3150 infections, adding 44,600 QALYs at a cost of $4,100/QALY gained. Treating a limited number of IDUs (10%) with HAART but offering 80% HAART coverage to eligible non-IDUs averted only 1,680 infections, adding 40,200 QALYs at a cost of $4,800/QALY gained.

Conclusion: Methadone substitution therapy is a highly cost-effective option for addressing the growing HIV epidemic in Ukraine. For programs that focus on expanding HAART, provision of minimal access to methadone significantly improves health benefits. Excluding IDUs from HAART reduces infections averted and increases costs.

1:45 PM
April D. Kimmel, PhD, MSc1, Milton C. Weinstein, PhD2, Xavier Anglaret, MD, PhD3, Sue J. Goldie, MD, MPH2, Elena Losina, PhD4, Yazdan Yazdanpanah, MD, PhD5, Eugène Messou, MD, PhD6, Kara L. Cotich, BS2, Rochelle P. Walensky, MD, MPH7 and Kenneth A. Freedberg, MD, MSc8, (1)Weill Cornell Medical College, New York, NY, (2)Harvard School of Public Health, Boston, MA, (3)INSERM Unité 897, Bordeaux, France, (4)Brigham and Women's Hospital, Boston, MA, (5)Centre Hospitalier de Tourcoing, Lille, France, (6)Centre Hospitalier Universitaire de Treichville, Abidjan, Ivory Coast, (7)Harvard Medical School, Boston, MA, (8)Massachusetts General Hospital, Boston, MA

Purpose: As 2nd-line antiretroviral therapy (ART) availability increases in resource-limited settings, questions remain about the value of laboratory monitoring to guide treatment decisions.  Our objective was to assess the incremental benefits and cost-effectiveness of CD4 count and HIV RNA tests to guide the timing of switching ART in HIV-infected patients.

Methods: In a treatment-eligible cohort in Côte d’Ivoire, West Africa, we used a state-transition model (CEPAC-International) to simulate the average life expectancies and total health care costs (2006 US$) associated with different monitoring strategies to guide switching to 2nd-line ART.  Monitoring strategies included clinical assessment, CD4 cell count, and HIV RNA testing, with criteria for 1st-line ART failure a single WHO stage III-IV event, 50% decrease in peak CD4, and return to baseline HIV RNA level, respectively.  Time on failed 1st-line ART resulted in resistance mutations, which reduced 2nd-line ART efficacy.  Data were derived from clinical trials and cohort studies from Côte d’Ivoire and published literature.  Sensitivity analyses were conducted to explore the impact of uncertain parameters and assumptions, as well as to assess variations of the three main monitoring strategies.

Result: Compared with 1st-line ART only, the incremental benefits from the availability of 2nd-line ART ranged from a 23% (clinical monitoring) to a 39% (CD4 count) to a 48% (HIV RNA) increase in undiscounted life expectancy.  The incremental cost-effectiveness ratio of switching to 2nd-line ART based on clinical monitoring was $1,750 per year of life saved (YLS) compared to 1st-line ART only; biannual CD4 monitoring was $2,300 per YLS.  The incremental cost-effectiveness ratio of biannual HIV RNA testing ranged from $3,760 ($87/test) to $2,290/YLS ($25/test), compared to the next best strategy. While continued ART following virologic failure provided life-expectancy gains, the costs associated with never stopping ART were substantial.  If 2nd-line ART costs were reduced, the cost-effectiveness of HIV RNA monitoring became more attractive.  Results were also influenced by the impact of resistance on 2nd-line ART efficacy.

Conclusion: Use of simulation models to examine the costs and consequences of different laboratory monitoring strategies can assist in rational use of HIV treatment resources in settings like Côte d’Ivoire.  Results can inform not only whether CD4 count and HIV RNA monitoring are used, but how they are implemented in clinical care.

2:00 PM
Anjali D. Oza, PhD, (Candidate), University of Chicago, Chicago, IL

Purpose: About half of all pregnancies in the US are unintended. Emergency contraceptive pills are 89 percent effective in preventing pregnancy if taken within 72 hours of unprotected sex. In August 2006, the FDA announced that Plan B, commonly referred to as the “morning after pill”, would be made over-the-counter (OTC) for women 18 and older. Women age 17 and under would still require a prescription from a physician. Prior to the legislation, nine states had pharmacy access legislation allowing a woman of any age to purchase Plan B from a pharmacist. This study assess whether by increasing access to Plan B, we observe a decrease in number of abortions and an increase in sexually transmitted infections (STIs). The latter would occur if increased access to Plan B increases the frequency of unprotected sex or causes women to abandon effective forms of birth control.

Method: Data source is the Thompson MarketScan Database 2005-2007 with sample size of 1,685,669 unique females ages 15-44 in each year. STIs and abortion identified using ICD-9 codes and CPT4 procedure or supplies codes provided in outpatient claims data. Using a pre-post design with an early adopter control group, I identify the effect of the FDA ruling on two outcomes of interest: STIs and abortion using a difference-in-difference estimator allowing for age trends. A log-link model and logit regression is specified and bootstrapped standard errors around the average marginal effect are obtained. Cost-effectiveness analysis is used to compare medical costs across alternatives: physician visit (time and monetary) and prescription cost, retail cost of OTC pill, and abortion in the event of unintended pregnancy.

Result: Making Plan B OTC resulted in 416 additional STIs per 100,000 women age 15-44, 95% CI [258, 575] and 108 fewer abortions per 100,000 women age 15-29, 95% CI [54, 171]. The relative risk of STI among women in early adapter states compared to those in late adapter states is 1.045, 95% CI [10.014, 1077]. Women 15-29 in the early adopter states have a RR of abortion over the laggard states of 0.898, 95% CI [0.827, 0.976].

Conclusion: Physicians should counsel women on safer sex practices and pharmacists could offer information on safer sex practices with purchase of Plan B.

2:15 PM
Jesse D. Ortendahl, BS1, Jane J. Kim, PhD1, Maxi Raymonville, MD2, Renel Saintard, MD2 and Sue J. Goldie, MD, MPH1, (1)Harvard School of Public Health, Boston, MA, (2)Zanmi Lasante, Cange, Haiti

Purpose: Invasive cervical cancer is the most frequent cause of cancer death among women in Haiti. While conventional cytology screening has not been feasible, two promising options could soon be accessible: for women over age 35, secondary prevention using rapid HPV testing and same-day treatment, and for girls ages 9-11, primary prevention with vaccination against HPV-16 and -18. Both will require price reductions to be cost-effective. For country-level decision making on cervical cancer prevention, the relative value of each new technology will depend on effectiveness and achievable coverage.  We used an empirically-calibrated simulation model of HPV infection and cervical cancer to estimate the mean reductions in cancer risk and mortality associated with cervical cancer prevention strategies.

Method: Model calibration was conducted using a likelihood-based approach, fitting model outcomes to country- and region-specific epidemiologic data (e.g., age-specific prevalence of HPV and precancerous lesions, HPV type-distribution, and cervical cancer incidence). Face validity of the model was assessed using primary data collected from women in rural Haiti. Analyses were conducted with a random sample of good-fitting parameter sets. Strategies included screening 3 times per lifetime at ages 35, 40 and 45 with rapid HPV testing followed by same-day treatment if needed, vaccination at age 9 prior to sexual debut, and combined screening and vaccination. Vaccine efficacy, achievable vaccination coverage, and achievable screening coverage were varied in sensitivity analysis.

Result: Assuming complete protection against HPV-16,-18 in girls previously unexposed, the mean reduction in lifetime cancer risk was 30% (range 26.4%-39%) with a school-based vaccination program reaching 90% of primary school enrollees (~56% enrollment at age 9). If this strategy included rapid HPV testing in adulthood, reductions would increase to 43% with 25% screening coverage, and 52% with 50% screening coverage. For older girls, vaccine efficacy is likely to be reduced as girls increasingly will have been exposed to HPV infection and school enrollment rates rapidly decline. For example, with a 25% decline in efficacy and a 25% drop-off in school enrollment, vaccinating in the mid-teens would provide a mean reduction of 13.8%.

Conclusion: Even if attainable vaccination and screening coverage rates are low in Haiti, a program of school-aged vaccination and rapid HPV testing 3 times per lifetime in adulthood could cut cervical cancer deaths in half.

2:30 PM
Sean P. Elliott, M.D., Timothy J. Wilt, M.D., M.P.H. and Karen M. Kuntz, ScD, University of Minnesota, Minneapolis, MN

Purpose: To project and compare clinical outcomes of adjuvant radiation soon after radical prostatectomy for locally advanced prostate cancer vs. initial observation with salvage radiation delivered at early prostate specific antigen recurrence. 

Methods: A state-transition Markov model was constructed to modify the published results of randomized studies of observation vs. adjuvant radiation such that the observation patients were given salvage radiation at early prostate specific antigen recurrence.  Transition probabilities and utility inputs were drawn from randomized controlled trials of adjuvant radiation and cohort studies of salvage radiation and calibrated to validate the model against the reference articles.  Outcomes of prostate specific antigen recurrence-free probability, metastasis-free survival, overall survival and projected quality-adjusted life expectancy were modeled through 10 years.  Sensitivity analysis and probabilistic sensitivity analysis were performed.

Results: At 10 years, observation with early salvage radiation yields similar results compared to adjuvant radiation for the outcomes of freedom from prostate specific antigen recurrence (53% and 52%, respectively), metastasis-free survival (72% and 70%, respectively) and overall survival (74% and 74%, respectively).  Findings were robust to sensitivity analyses and probabilistic sensitivity analyses.  After adjusting for the disutility of radiation therapy, observation with early salvage radiation was slightly superior to adjuvant radiation through 10 years (7.0 and 6.2 quality-adjusted life years, respectively).

Conclusion: This decision analysis model suggests that for patients with locally advanced prostate cancer after radical prostatectomy, a strategy of initial observation with salvage radiation given at early prostate specific antigen recurrence yields cancer control and survival outcomes similar to adjuvant radiation while minimizing the morbidity of radiation. 

2:45 PM
Bartholomew K. Abban, MS and Karen M. Kuntz, ScD, University of Minnesota, Minneapolis, MN

Purpose: The use of colonoscopy as a primary screening test for colorectal cancer (CRC) has increased substantially since 2000.  At the same time there is a growing concern of early post-polypectomy surveillance (i.e., surveillance colonoscopy done at shorter intervals than recommended) and inadequate capacity to meet colonoscopy demand in the next decade.  We used recent screening uptake data to forecast the demand for colonoscopy under various screening and surveillance scenarios and examined the impact of early surveillance on the demand for colonoscopy.

Method: We used a population-based state-transition Markov model to simulate the natural history of CRC and superimposed a screening mechanism than could either identify and remove adenomas or detect and treat CRC.  We used data from the US census to simulate the population 50 years of age and older and projected the demand for colonoscopies and the colonoscopy-related deaths each year. We modeled current screening uptake from the National Health Interview Survey (NHIS), as well as the type of screening (e.g., fecal occult blood testing, colonoscopy). Recommended surveillance intervals are 3 years following polypectomy of an adenoma of 10 mm or larger or more than 3 adenomas, and 5 years for all other findings.  We modeled early surveillance by specifying that a proportion of patients with history of adenomas undergo surveillance before the recommended time interval.   We varied the timing of surveillance colonoscopies to assess its impact on colonoscopy demand.

Result: We estimated the demand for screening and surveillance colonoscopies in 2009, based on the current guidelines and current screening uptake, to be 6.56 million.  Early surveillance resulted in an additional 1.03 million colonoscopies and a 14% increase in colonoscopy-related mortality.  We projected the demand for colonoscopy under current screening practices to reach 12.02 million by 2015 and 13.26 million by 2020 assuming a gradual increase to a 70% screening uptake.  The current estimated capacity in the US would have to increase by 40% to meet the demand.

Conclusion: The projected demand for colonoscopy is expected to surpass the capacity if screening uptake increases.  Early post-polypectomy surveillance represents a relatively large part of the increase in colonoscopy demand, with a substantial impact on colonoscopy-related deaths.  Efforts need to be made to increase colonoscopy capacity and ensure adherence to surveillance guidelines.