Wednesday, October 23, 2013: 10:00 AM - 11:30 AM
Key Ballroom 7,9,10 (Hilton Baltimore)
Category Reference for Presentations
AHEApplied Health Economics DECDecision Psychology and Shared Decision Making
HSPHealth Services, and Policy Research METQuantitative Methods and Theoretical Developments

* Candidate for the Lee B. Lusted Student Prize Competition

Session Chairs:
Tara A. Lavelle, MS, PhD and Scott B. Cantor, PhD
10:00 AM
Ganesh Sivarajan, MD1, Glen Taksler, Ph.D.2, Dawn Walter, MPH3, Marc Bjurlin, MD1, Cary P. Gross, MD4, R. Ernest Eosa, MD1 and Danil V. Makarov, MD, MHS2, (1)New York University Langone Medical Center, New York, NY, (2)New York University School of Medicine, New York, NY, (3)Cancer Institute, New York, NY, (4)Yale University School of Medicine, New Haven, CT


The rapid diffusion of the surgical robot has been fraught with controversy because of the technology's high costs and disputed marginal benefit. Some, however, have suggested that adoption of the surgical robot has facilitated partial nephrectomy for renal malignancy, an underutilized procedure considered to be more challenging yet less morbid than radical nephrectomy. We sought to determine whether institutional acquisition of the surgical robot was associated with a higher local rate of partial nephrectomy.


We used the 2001, 2005 and 2008 Health-care Cost and Utilization Project State In-patient Databases from 7 states to identify 21,569 surgical procedures for renal tumors. These patient-level records were aggregated to the hospital-level then merged with the American Hospital Association Survey and with publicly available data on timing of surgical robot acquisition. We used a difference-in-difference model to assess at the hospital-level whether robot acquisition was associated with an increase in the rate of partial nephrectomy, adjusting for total nephrectomy rate, year of surgery, year of robot acquisition, geographic state and several hospital-level factors. We also performed two sensitivity analyses to determine whether there was a time lag between robot acquisition and changes in the utilization of partial nephrectomy (because of a presumed necessity to develop requisite surgical skill) and to ensure no association existed between robot acquisition and performance of an unrelated procedure (the presence of which might suggest than an unmeasured confounding characteristic, rather than robot acquisition, was the cause for increased surgical volume).


In the multivariable adjusted differences-in-differences model (Figure 1), hospitals acquiring a robot between 2001-2005 performed more partial nephrectomies in both 2005 (31% increase) and 2008 (36% total increase) (p<0.01 for both). Hospitals acquiring a surgical robot in the time period between 2005-2008 also demonstrated higher rates of partial nephrectomy 2008 (16% increase) (p=0.02). Results of the secondary lag time analysis were not substantially different. As expected, there was no association between robot acquisition and utilization of an unrelated surgical procedure.


Hospital acquisition of the surgical robot is associated with increased utilization of partial nephrectomy, an underutilized, guideline-encouraged procedure. This is one of the few studies to suggest acquisition of the surgical robot was associated with improvement in quality of patient care.

Figure 1

10:15 AM
Janelle Z. Seah, MSc, Paula K. Lorgelly, PhD and Anthony Harris, M.A., M.Sc., Monash University, Melbourne, Australia

Chronic diseases often occur in combination, their interaction effects possibly leading to a non-linearity in healthcare costs that is often ignored in economic evaluations. This paper aims to quantify the comparative effect of single and multiple chronic diseases on hospital resource use.


Using records of all admissions to public hospitals in the state of Victoria, Australia in 2010-2011 we estimate multiple regression models of hospital length of stay (and total annual discharges) for combinations of 11 chronic diseases. For length of stay we run separate models for patients with a 1 day stay and those with multiple day stays and adjust for observed and unobserved characteristics of patients.


Having a higher chronic disease count decreases the odds of having a same-day hospitalization (day case) exponentially while some disease combinations increased these odds. Having ischemic heart disease (IHD) & dementia doubled the odds of a day case compared to a patient with dementia only. Among overnight stays, having a mental disease had the highest single disease effect on length of stay (LOS) – increasing LOS by 3-4 days. Some diseases when combined had non-additive effects (i.e. their combined effect was greater or less than the sum of their single disease effects) on LOS while others were additive. The interaction effect in a depression-renal failure combination added 3 more days to the sum of its single disease effects, while in cancer-osteoporosis it was -2 days. We observed that disease combinations that produced a positive interaction effect were usually unrelated diseases. The number of chronic diseases an individual had was found to be positively correlated with their number of admissions. Among single diseases, cancer had the highest effect on admissions – increasing admissions by 5. Having a combination of diseases was generally found to have a less-than-additive effect on the number of admissions.


Patients with chronic diseases have a resource use pattern that includes longer lengths of stays and more admissions. Combinations of unrelated diseases are particularly correlated with longer lengths of stay therefore it is the disease and combination type that is associated with higher lengths of stay and admitted episodes. Economic evaluations which assume additivity when estimating joint-disease costs could either overestimate or underestimate costs depending on the type of disease combination.

10:30 AM
Lawrence J. Cheskin, MD1, Vincent Wu, BS2, Sunil Karnawat, PhD3 and Weiyu W. Liu, MHA3, (1)Johns Hopkins Weight Management Center, Baltimore, MD, (2)Independent Contractor, San Francisco, CA, (3)VIVUS, Inc., Mountain View, CA

Purpose: A substantial portion of the economic burden of obesity is the cost of medications to manage obesity-related comorbidities, including hypertension, dyslipidemia, and type 2 diabetes mellitus (T2DM). This post hoc analysis evaluated the effects of magnitude of weight loss (WL) on annual antihypertensive, lipid-lowering, and antidiabetic medication costs in obese/overweight subjects.


Methods: The CONQUER trial was a double-blind, Phase 3 study of 2487 obese/overweight subjects (body-mass index [BMI] ≥27 and ≤45 kg/m2) with ≥2 weight-related comorbidities randomly assigned to placebo (n=994), phentermine (PHEN) 7.5mg/topiramate extended-release (TPM ER) 46mg (7.5/46; n=498), or PHEN 15mg/TPM ER 92mg (15/92; n=995) plus lifestyle modifications for 56 weeks. Subjects included in this post hoc analysis completed ≥12 weeks of therapy and received medications at baseline or endpoint for the treatment of ≥1 of the following comorbidities: hypertension (n=830), dyslipidemia (n=340), or T2DM (n=207). Annual antihypertensive, lipid-lowering, and antidiabetic medication costs were calculated at baseline and end of treatment by multiplying the unit cost (Medi-Span's PriceRx database) by number of doses per day and by 365. Subjects were stratified by magnitude of WL (<5%, ≥5%-<10%, ≥10%-<15%, and ≥15%), and changes in annual medication costs were evaluated from baseline to end of treatment.


Results: Most subjects were female (70%) and Caucasian (86%); mean weight was 103.117.9kg and mean BMI was 36.64.5kg/m2. At treatment end, the majority of subjects with <5% WL were from the placebo group, while the majority of subjects achieving ≥5% WL were from the PHEN/TPM ER groups (Table). Compared with subjects losing <5% body weight, changes in annual medication cost for ≥5%-<10% WL were -$107.27, -$167.86, and -$76.18 for the treatment of hypertension, dyslipidemia, and T2DM, respectively. Similarly, changes in annual medication cost in subjects with ≥10%-<15% WL were
-$99.54, -$242.83, and -$271.17, respectively, and for subjects with ≥15% WL were -$229.26, -$594.48, and -$244.94, respectively. Common treatment-emergent adverse events were constipation, dry mouth, and paraesthesia.


Conclusions: In this obese/overweight population, increasing magnitudes of WL were associated with greater reductions in annual medication costs for the treatment of hypertension, dyslipidemia, and T2DM. This suggests that ≥5% WL may have a meaningful and beneficial impact on the economic burden of obesity.


Dr. Cheskin is a stockholder and member of the National Advisory Board of VIVUS, Inc.



10:45 AM
Ankur Pandya, PhD, Ajay Gupta, MD, Hooman Kamel, MD, Pina C. Sanelli, MD, MPH and Bruce R. Schackman, PhD, Weill Cornell Medical College, New York, NY

Purpose: There are an estimated 400,000 patients over the age of 65 years in the United States with extracranial internal carotid artery stenosis, which accounts for 15-20% of ischemic strokes. Revascularization has been shown to decrease stroke risk by 45%, but this procedure carries substantial costs and risks. Carotid stenosis patients with image-determined cerebrovascular reserve (CVR) impairments have been shown to be at a four-fold increased risk of stroke. We projected health benefits, risks, and costs of three competing strategies (treat, test, or neither) for carotid stenosis patients.

Methods: We developed a decision analytic model to compare the following interventions for asymptomatic carotid stenosis patients (50-99% extracranial internal carotid artery blockage): 1) treat all with revascularization (carotid endarterectomy); 2) only perform revascularizations for those with image-determined (transcranial Doppler ultrasound [TCD]) CVR impairments, and use medical therapy alone for all others; and 3) medical therapy alone for all patients. Model input parameters were estimated from published sources. Healthcare costs for revascularization, TCD, acute stroke, and annual long-term stroke care were $14,130, $265, $65,800, and $31,000, respectively. Estimates for baseline annual risk of stroke and probability of revascularization complications (death, stroke, or myocardial infarction) were 1.5% and 2%, respectively. Discounted lifetime costs and health benefits (quality-adjusted life years [QALYs]) were projected for each strategy.


Results: The medical therapy alone strategy resulted in higher total costs ($47,000 per person, driven by stroke-related costs) and lower lifetime QALYs (11.379) compared to image-based CVR testing ($40,800, 11.535 QALYs) or revascularization for all ($44,800, 11.555 QALYs). The incremental cost-effectiveness ratio for the revascularization for all versus CVR testing strategy was $210,000/QALY. Cost-effectiveness results were most sensitive to plausible variations in revascularization risks, costs and benefits; strength of association between CVR impairment and future stroke; and baseline stroke risk. Figure 1 shows the two-way sensitivity analysis results for revascularization complication rates and baseline stroke risk. Model results were robust to variations in TCD costs; acute and long-term stroke costs; and stroke utility values.

Conclusions: CVR testing can be a cost-effective tool to identify asymptomatic carotid stenosis patients most likely to benefit from revascularization. The economic value of this tool depends on accurate assessments of baseline stroke risk and the likelihood of procedure complications, factors that vary by patient and provider.



11:00 AM
Julia Thornton Snider, Ph.D.1, Timothy Juday, PhD2, Yuri Sanchez, Ph.D.1, Daniel Seekins, MD2, John Romley, Ph.D.3, Neeraj Sood, Ph.D.3 and Dana P. Goldman3, (1)Precision Health Economics, Los Angeles, CA, (2)Bristol-Myers Squibb, Plainsboro, NJ, (3)University of Southern California, Los Angeles, CA
Purpose: To understand the likely impact on clinical and economic outcomes of changes to state policies governing access to combination antiretroviral therapy (cART) among low-income individuals living with HIV/AIDS

Method: Retrospective analyses of cART access were conducted on state Medicaid and AIDS Drug Assistance Programs (ADAP) policies, using data from ADAP Monitoring Reports, the Kaiser Family Foundation, and the Centers for Medicare and Medicaid Services over 2001-2010.  The survival effects and lifetime health care expenditures for policies were quantified based on the clinical and cost-effectiveness literature.  The effect of cART access on employment, disability and earnings were analyzed with probit and Poisson regression analyses of data from the HIV Cost and Service Utilization Study.

Result: As expected, stricter eligibility requirements based on income, assets or medical status, as well as the lack of a medically needy program, reduce the number of individuals with access to cART through ADAP and Medicaid.  In turn, decreased cART use results in higher mortality by 1.4 quality-adjusted life years (QALYs) per beneficiary.  For example, in a scenario in which the ADAP income eligibility cutoff is decreased by 50 percentage points nationally, there would be 4,876 fewer individuals with cART access.  Based on a $34,900 cost per QALY from the literature, this policy would save $325 million in health care expenditures (2012 dollars), but result in 6,729 QALYs lost.  Assuming a $100,000 value of a statistical life-year, the cost of reduced cART access in terms of decreased QALYs would be 2.9 times as large as the benefits in terms of decreased health care spending.  In addition, reduced access would lead to a 13% increase in work-limiting disabilities, an 11% decrease in employment, and a 3% decrease in individual earnings.

Conclusion: In the face of increasing budget pressures, state legislators may be considering cuts to programs that increase access to cART for low-income HIV/AIDS patients.  While such an approach may be financially advantageous in the short term, it overlooks the longer term advantages not only to patients in terms of improved clinical outcomes, but the economic benefits that derive from increased employment and earnings and decreased disability.

11:15 AM
William W. L. Wong, Ph.D.1, Hong Anh Tu, PhD1, Wendong Chen, MD, PhD1, Jordan J. Feld, MD, MPH1, Kristen Chelak, BSc, (Pharm), MSc, RPh2, Karen M. Lee, MA2 and Murray D. Krahn, MD, MSc3, (1)University of Toronto, Toronto, ON, Canada, (2)Canadian Agency for Drugs and Technologies in Health (CADTH), Ottawa, ON, Canada, (3)Toronto Health Economics and Technology Assessment (THETA) Collaborative, Toronto, ON, Canada
Purpose: Prior to 2011, pegylated interferon plus ribavirin (PR) was the standard therapy for chronic hepatitis C (CHC).  In 2011, the first direct-acting antiviral agents (DAA) (boceprevir, telaprevir) were approved.  More DAA currently under development and expected to receive market approval in the next few years.  While these treatments appear to be more effective at achieving sustained virologic response in CHC patients; they are more expensive than PR and have different adverse event profile. These aspects need to be considered collectively when making reimbursement decisions.  The objective of this study is to build a flexible model platform that can evaluate the cost-effectiveness of the pipeline of hepatitis C drugs for various treatment and patient scenarios once new treatments are available.

Method: We developed a flexible state transition model of CHC.  Health states relate to treatment and adverse events, fibrosis stages (F0–F4), and CHC complications states.  Our model was divided into a treatment and a natural history modules.  The treatment module can be easily changed to reflect different treatment algorithms.  The natural history module is a robust model that reflects the natural history of CHC and was validated against other published models. Simulated population can be stratified by CHC genotype, Interleukin-28B genotype, treatment status (naïve, experienced), age, and fibrosis stages.  Disease progression parameters were from our published systematic review, which estimated the annual transition probabilities between fibrosis stages.  The CHC-related costs were from our large published costing study using administrative data.  Utility data were from our published study of CHC patients across different health states.

Result: As a preliminary test, we evaluated our model using data from the telaprevir clinical trials (ADVANCE, ILLUMINATE).  Table 1 summarizes the outcomes associated with our base-case analysis (50-years, genotype 1, treatment naïve) stratified by fibrosis stages.    Validation was done against QALY gained from other published models.  Based on the current structure, the generated ICERs are congruent with other studies.

Conclusion: Our model platform provides a robust CHC natural history model while allowing for flexibility in changing treatment algorithms.  When new agents are brought forward for approval, the model will be used to compare the relative cost-effectiveness of available CHC treatment strategies.

 Table1:     Fibrosis Strategy Cost QALYs ICER
Mild PR $112,534 11.36
Telaprevir $137,995 11.73 $68,228
Severe PR $111,530 10.18
Telaprevir $137,465 11.07 $29,089