Tuesday, October 20, 2009: 1:00 PM
Grand Ballroom, Salon 5 (Renaissance Hollywood Hotel)

* Candidate for the Lee B. Lusted Student Prize Competition

Session Chairs:
Larry Kessler, ScD and Uwe Siebert, MD, MPH, MSc, SD
1:00 PM
Heather Taffet Gold, PhD and Mary Katherine A. Hayes, MD, Weill Cornell Medical College, New York, NY

Purpose: To assess the cost effectiveness of new accelerated partial breast radiotherapy (RT) modalities compared to standard whole breast RT (WBRT), which may be delayed in the community setting. Accelerated strategies (three-dimensional conformal external-beam RT (3D-CRT) and brachytherapy with balloon catheter) ease time and toxicity burdens of WBRT, requiring treatment 2 times/day for 5 days instead of daily for 4-7 weeks, and may deliver RT more efficiently to the tumor bed, where most breast recurrences occur; however, long-term trial evidence is unavailable.

Method: Decision analytic Markov models from the payer and payer+patient perspectives were created to evaluate 4 RT strategies following breast-conserving surgery for Stage I breast cancer: standard WBRT, WBRT delayed by 8+ weeks, 3D-CRT, and brachytherapy. The microsimulation included 10,000 hypothetical 60yo women for 15 years follow-up. Data were from published studies, US life tables, SEER data, and Medicare reimbursement rates (2008US$). Costs and QALYs were discounted at 3%. In the base case we assumed equivalent utility for all RT options (0.92) and equivalent efficacy between accelerated strategies. Assumptions were varied in sensitivity analyses. The model is validated for the WBRT strategy based on pooled trial data on overall survival for the same population (via AdjuvantOnline!).

Results: From the payer perspective, standard WBRT is the dominant strategy (costing $11K and yielding 11.04 QALYs). When 3D-CRT’s utility is 1% higher than standard WBRT, the CE ratio for 3D-CRT is $53,900/QALY gained. When patient time/transportation costs are included, 3D-CRT is the preferred strategy at a willingness-to-pay threshold of $100K/QALY; traditional WBRT then costs $185,700/QALY gained. Brachytherapy-based RT and delayed WBRT are dominated by WBRT and 3D-CRT under all reasonable assumptions.

Conclusion: 3D-CRT appears to be a cost-effective RT modality to reduce the burden of standard WBRT with its high patient costs and lower utility. This is particularly important for individuals that receive delayed RT, who are more likely to be individuals of lower socioeconomic status and non-White race.

1:15 PM
Ba' Pham, MSc, PhD, (c)1, Laura Teague, RN, MN, NP-Adult2, James Mahoney, MD2, Laurie Goodman, RN, BA3, Anita Stern, PhD4, Jeff Poss, PhD5, Jianli Li, PhD2, Nancy Sikich, PhD6, Rosemarie Lourenco, CHIM, IMS6, Luciano Ieraci, MSc4, Steven Carcone, MSc1, William Witteman, MISt4 and Murray D. Krahn, MD, MSc4, (1)Toronto Health Economics and Technology Assessment Collaborative, Toronto, ON, Canada, (2)St. Michael's Hospital, University of Toronto, Canada, Toronto, ON, Canada, (3)The Credit Valley Hospital, Mississauga, ON, Canada, (4)University of Toronto, Toronto, ON, Canada, (5)University of Waterloo, Waterloo, ON, Canada, (6)Ontario Ministry of Health and Long-Term Care, Toronto, ON, Canada

Purpose: New pressure ulcers (PU) occur in many patients undergoing elective surgery (7%-66%). We evaluated the cost-effectiveness of GFOs in support of guideline recommendations by the Ontario Health Technology Assessment Committee.

Methods: Study design: Cost-utility analysis, third-party payer perspective, 1-year time horizon, 2009 Canadian $. Population: patients undergoing a planned major surgical procedure (e.g. > 90 minutes). Comparators: GFO versus standard OR surfaces (STD). Decision model: A 21-state, 1-day-cycle Markov model calibrated to reproduce stage-specific hospital-acquired PU (H-A PU) prevalence reported by the Toronto Tri-Hospital Survey (n=1,118 patients). Input data: Effectiveness: Recent systematic reviews of preventive interventions identified three RCTs evaluating overlays for OR tables. In surgical patients, gel-filled overlays (GFO) significantly decrease early and mostly intact skin PUs one-day post-operation (RR: 0.53; 95% CI: 0.33-0.85), according to a well-designed 1998-RCT (n=446). Cost: Each GFO costs $878. A case-control design was used to obtain stage-specific PU-attributable inpatient [post discharge] costs, n=3,780 PU cases [n=21,578 home care clients] from the Ontario Case Costing Initiative [interRAI -Home Care] database. Controls were matched to PU cases by age groups, gender, admission diagnosis, and Charlson co-morbidity [Activity of Daily Living] score. PU-attributable inpatient [home care] costs were cost differences between cases and controls: $11,967 [$57/week], $12,951 [$81/week] and $21,797 [$105/week] for stages 2-4, respectively. Health utility: Similarly, a utility decrement estimate attributable to PU stage 2-4 versus 0-1 of 6.1% (3.9%, 8.3%) was obtained using data from the Minimum Data Set – Health Status Index (n=18,321 nursing home residents). This was applied to a mean utility of 0.44 for hospitalized patients from a published study (n=1,207). 

Results: The prevalence of hospital acquired PU was projected to be 13.3% (12.3, 13.6) among GFO patients, and 16.6% (15.6, 17.0) among STD patients, with an estimated NNT=30 (29, 35). A GFO could be used for approximately 1,500 procedures over 2 years. Estimated QALY gained per person with GFOs was 0.00152 (0.00054, 0.00360) and cost saving $225 ($76, $647; $169 in-patient and $56 post-discharge costs). Probabilistic sensitivity analysis showed a 99% chance of higher net benefit with GFOs.

Conclusions: The use of GFOs in surgical patients with a major surgical procedure is highly economically attractive.

1:30 PM
Anirban Basu, PhD and Jing Guo, MS, University of Chicago, Chicago, IL
Background: Recent research demonstrates a significant growth in the prevalence of diabetes in patients with schizophrenia that is attributable to the growth in the use of atypical antipsychotics (A-APDs). Less is known about how comparative effectiveness research (CER) identifying the effect of individual A-APDs on diabetes risk helped curb this growth in diabetes risk, presumably by allowing the clinicians to avoid riskier treatments .

Purpose: To estimate the net change in the risk of diabetes in the population of patients with schizophrenia that may be attributable to the CER in this field, after adjusting for community trends in diabetes prevalence.

Method: Using data from the National Hospital Discharge Survey, we perform trend analyses for prevalence of diabetes and insulin resistance-unrelated general illnesses in patients with schizophrenia, as well as in individuals without known mental illness (control group), during three time periods: 1) CER-naïve era (1979–1994); 2) the hypothesis generation era (1995–2001), and 3) CER-rich era (2002-2006).

Result: Identical trends in the prevalence of diabetes and general illnesses are observed between the schizophrenia and the control groups during the CER naïve era. These allow us to use the trends in the control group in the later years as counterfactuals for corresponding trends in the schizophrenia group.  Risk of diabetes, attributable to the use of A-APDs, grew at an increasing rate (0.7% per year, p<0.001) during 1995-2001, and by 2001, represented 33% (p=0.016) of the 10% diabetes prevalence in this population. During the CER-era, we find a break in the trend of diabetes prevalence in this population, but no such break in either the community trend in diabetes prevalence or the general illness trend among the schizophrenia population. This break translates to a potential reduction in the prevalence of diabetes in the schizophrenia population from 16% to 11% (p=0.038), which saves $1.25 billion per year.    

Conclusion: CER identifying individual A-PD’s risk for diabetes appears to have curbed the A-APD attributable epidemic of diabetes in schizophrenia patients. Longer-term follow-up are necessary to confirm this positive role of CER.

1:45 PM
Yoko Ibuka, PhD1, Alison Galvani, PhD1 and A. David Paltiel, PhD2, (1)Yale University, New Haven, CT, (2)Yale University School of Medicine, New Haven, CT
Background: Vaccination provides an indirect benefit to unvaccinated individuals, however existent cost-effectiveness analyses on vaccination programs often exclude the indirect benefits of vaccination.

Purpose: We incorporated indirect benefits of vaccination into a cost-effectiveness analysis for vaccination against seasonal influenza by using a mathematical model of disease transmission dynamics. Specifically, we evaluated costs and effectiveness at different vaccination coverage levels to examine how the indirect benefits affect the costs and effectiveness.

Methods: A dynamic epidemiological model of influenza transmission was constructed and parameterized based on the estimates from published literature. Our model assumed a homogeneous random-mixing population of100,000 individuals. Using the model, we evaluated the costs and health outcomes of vaccination program for influenza with various levels of vaccination coverage, ranging 0% to 90% with 10% increments. Effectiveness was measured in terms of the number of influenza cases averted in the population, and the costs include vaccination costs as well as direct medical costs for treatment of influenza. Incremental cost effectiveness ratios (ICERs) were calculated as a comparison to the next best strategy in order to examine additional costs required to avert one infection.

Results: The number of influenza cases averted increased in non-linear fashion as vaccination coverage increased (Figure). 38,211 cases were additionally averted when vaccination coverage increased from 10% to 20%; on the other hand, the additional gain was only 804 cases when the coverage increased from 80% to 90%. In contrast, the costs monotonically increased once the vaccination coverage exceeded 50%. As a result, the ICERs were $1, $84 and $380 per influenza case averted, at 20%, 50%, and 90% vaccination coverage levels, respectively.

Conclusions: The additional gain in the effectiveness of vaccination program could decrease with wider vaccination coverage due to reduced additional indirect benefits from vaccination. Therefore, it is critical to incorporate the indirect benefit of vaccination program into its cost-effectiveness analysis using a dynamic model of disease transmission.

2:00 PM
Kenneth J. Smith, MD, MS, Mahlon Raymund, PhD, Mary Patricia Nowalk, PhD, Mark S. Roberts, MD, MPP and Richard K. Zimmerman, MD, MPH, University of Pittsburgh School of Medicine, Pittsburgh, PA

Purpose: In prior influenza pandemics, pneumococcal complications of influenza have caused substantial morbidity and mortality.  The usefulness and cost-effectiveness of pneumococcal polysaccharide vaccine (PPV) for healthcare workers (HCW) during an influenza pandemic is unknown.

Method: We used a Markov model to estimate the cost-effectiveness of PPV use in HCW compared to nonuse during an influenza pandemic.  Age-specific incidence data on invasive pneumococcal disease (IPD) from the CDC were incorporated into the model, assuming that IPD occurred at twice the usual rate during the year of pandemic influenza in HCW (average age 45 years); both the IPD rate multiplier and average HCW age were varied widely in sensitivity analyses.  Model time horizon was 15 years, taking a societal perspective and discounting costs and benefits at 3%/year.  We obtained IPD costs from the National Inpatient Sample and utilities from the medical literature; an expert panel estimated PPV effectiveness.  We assumed that: pneumococcal disease transmission from HCW to patient did not occur, heightened IPD risk occurred for only 1 year, and PPV did not prevent noninvasive pneumonia, all of which potentially bias against PPV.

Result: In the base case analysis, PPV prevented 39.1% of IPD cases and deaths in HCW, saving 0.00083 QALY (about 0.3 days) at a cost of $5.33, or $6207 per QALY gained compared to no vaccination.  If heightened IPD risk related to pandemic influenza lasts 2 years, PPV costs $3302/QALY.  Results are sensitive to HCW age, with incremental cost-effectiveness ratios rising to >$100,000/QALY at ages <29 under base case assumptions; however costs/QALY gained remain <$100,000 for HCW < 30 years old if IPD rates during the pandemic year are ≥ 4.2 times those seen during non-pandemic years.  Pandemic-based PPV is cost saving in HCW aged ≥50 under base case assumptions.  In a probabilistic sensitivity analysis, vaccinating all HCW is favored >74% of the time if the acceptability threshold is ≥$20,000/QALY and in >94% with thresholds ≥$50,000.

Conclusion: Vaccinating all healthcare workers to protect against pneumococcal disease during a pandemic influenza outbreak is likely to be economically reasonable in an analysis biased against vaccination.  Vaccinating healthcare workers aged ≤30 may be more expensive, while vaccinating those ≥50 appears to be cost saving.