Sunday, January 10, 2016: 9:00-10:30
Shaw Auditorium, 1/F (Jockey Club School of Public Health and Primary Care Building at Prince of Wales Hospital)

Valeria E. Rac, MD, PhD, Yeva Sahakyan, MD, MPH, Nida Shahid, HBSc., CCRP, Aleksandra Stanimirovic, MSc, PhD (candidate), Petros Pechlivanoglou, MSc, PhD, Welson Ryan, Lusine Abrahamyan, MD MPH PhD and Murray D Krahn, MD, MSc, FRCPC, Toronto Health Economics and Technology Assessment (THETA) Collaborative, University of Toronto, Toronto, ON, Canada

   The purpose of the quantitative descriptive study included evaluating overall patterns of change observed in the remotely monitored clinical parameters of chronic obstructive pulmonary disease (COPD) and heart failure (HF) patients across Ontario.


   Monitoring parameters such as blood pressure (BP), oxygen levels and weight were analyzed for patients enrolled in the program from July 2012 to March 2015 in the Central West, North East and Toronto Central Local Health Integration Networks (LHINs). Clinical data was extracted from a database managed by the Ontario Telemedicine Network (OTN) and analyzed using repeated measures with generalized linear mixed model procedures in SAS. The outcome measures were estimated for change during participation in Telehomecare.


   Findings show, overall 2470 patients enrolled in Telehomecare and completed at least one month. During the first month of participation, one third of patients (n=810) had an elevated BP, with a monthly average systolic BP (SBP) of 150.2+9.5 mmHg and diastolic BP (DBP) of 78.8+12.5 mmHg, compared with adequately controlled (n=1660) patients who had a mean  SBP of 120.6±12.7 mmHg and mean DBP of 68.3±9.7 mmHg. Authors found clinically and statistically significant reduction in systolic and diastolic BP in initially hypertensive patients, over the seven-month program period when adjusted for age, gender and condition.  For SBP levels, reduction was 10.8 mmHg (95% CI= 9.6-12.0) and DBP reduction was 6.5 mmHg (95% CI=5.6-7.3). Although not clinically meaningful, there was a statistically significant reduction in impaired oxygen saturation levels and weight fluctuations in program participants.  


   The observed changes in monitored patient parameters over the course of time leads us to interpret that hypertensive patients may benefit the most from participating in Telehomecare. This is regardless of their age, gender, condition or geographical area. Further analyses are on the way to confirm our results.


Eric A. Finkelstein, PhD, MHA, Chetna Malhotra, MD, MPH, Junxing Chay, MSc and Semra Ozdemir, PhD, Duke-NUS Graduate Medical School Singapore, Singapore, Singapore

Purpose: This study uses a stated preference survey to examine the extent to which financial assistance, in the form of subsidies for life extending treatments (LETs) or cash pay-outs, distorts the demand for end-of-life treatments.

Method(s): A discrete choice experiment, consisting of 10 choice tasks, was administered to 290 patients with breast, lung or colorectal cancer (Stage I-IV) from outpatient clinics in Singapore to elicit their preferences for LETs and palliative care (PC) only. For each task, participants selected their most preferred scenario from two LET choices and PC only (sample task in Figure 1). Each task systematically varied in terms of net cost after cash pay-out (from a hypothetical insurance plan), median length of survival, 5-year survival rate and quality of life. Responses were fitted to a latent class conditional logistic regression model, accounting for potential preference heterogeneity among subgroups of participants. Based on estimated relative preferences, we quantified patients' willingness to accept a less effective LET or PC only. We then simulated the effects of various LET subsidy and cash pay-out policies on treatment choices.

Result(s): We identified 3 main classes of patients according to their end-of-life preferences (see Figure 2). The first class (26.1% of sample) has a strong preference for PC such that they are willing to give up gains in life expectancy and even pay for receiving only PC. The second class (29.8% of sample) has a strong preference for LETs and prefers to extend life regardless of cost or quality of life. The final class (44.1% of sample) prefers LETs to PC, but actively trades off costs, length and quality of life when making end-of-life treatment choices. Marital status and cancer type were found to be predictive of class membership. Policy simulations show that LET subsidies have a greater distortionary effect on treatment choices compared to a cash pay-out, which was not shown to distort demand for PC or LETs.

Conclusion(s): Cancer patients have heterogeneous end-of-life preferences. Unlike cash-pay-outs, LET subsidies distort relative prices of different treatments and encourage greater utilization of LETs. Policymakers should be mindful of these differences when designing healthcare financing schemes for patients with life-limiting illnesses.


Kimberly Babiarz, PhD, Stanford University, Stanford, CA and Jeremy D. Goldhaber-Fiebert, PhD, Stanford Health Policy, Centers for Health Policy and Primary Care and Outcomes Research, Department of Medicine, Stanford University, Stanford, CA

Purpose: School based food transfer programs are a common policy approach to address malnutrition among school-aged children.  Such programs function in >70 low- and middle-income countries. They have generally demonstrated effectiveness in increasing caloric intake at school, but net nutritional benefits depend critically on changes in household purchasing and consumption induced by programmatic food transfers. While accounting for changes in home-based consumption, we assess the net nutritional effects of the largest school feeding program in the world, India's Cooked Mid Day Meals (CMDM) program which serves cooked meals to 120 million primary school children.

Method(s): With a large longitudinal dataset including 730,110 households over 19 years (1994-2012), we estimate CMDM's effects on household consumption (household food purchases and corresponding caloric, macronutrient and micronutrient content). We use difference-in-difference models employing ordinary least squares regressions controlling for household characteristics, and place and time fixed-effects. We compare induced changes in home consumption to CMDM program per-meal nutrient norms, accounting for mean program exposure.

Result(s): CMDM's transfers cause CMDM-eligible households to reduce daily at-home consumption of rice and wheat (124.8 gram reduction [95% CI: 104.7 - 144.9]), legumes (5.5 gram reduction [95% CI: 3.6 - 7.5]), and vegetables (28.3 gram reduction [95% CI: 13.3 - 43.4]). At the same time oil consumption increases (4.5 gram increase [95% CI: 3.4 - 5.5]). Changes in household consumption induced by the program imply a reduction in home-based calorie, protein, iron and zinc intake, and an increase in calcium (primarily driven by households with lower levels of education). Because reductions in home-based legume and vegetable consumption do not fully offset transfers that occur via school meals delivered according to program norms, CMDM induces some dietary diversification if the program meets meal standards (Figure 1). However, overall net calorie consumption is reduced and protein intake is unchanged under the CMDM program. Effects are largest for poorer households.

Conclusion(s): The effect of school based feeding programs depends critically on how households respond to benefits transferred to school children. Our study shows that CMDM's nutritional effects are blunted by reduced household food purchases. Households reduce total caloric consumption, sometimes overcompensating for program food transfers. However, households also respond by diversifying their diets towards nutrient-rich foods. Future policy designs should carefully account for household responses to them.


Lea Prince, MA, PhD, Centers for Health Policy and Primary Care and Outcomes Research, Department of Medicine, Stanford University, Stanford, CA, Kimberly Babiarz, PhD, Stanford University, Stanford, CA and Jeremy D. Goldhaber-Fiebert, PhD, Stanford Health Policy, Centers for Health Policy and Primary Care and Outcomes Research, Department of Medicine, Stanford University, Stanford, CA
Purpose: Nearly half of children under age 6, an estimated 230 million children, are malnourished in India. The national Integrated Child Development Services (ICDS) program aims to reduce childhood malnutrition through supplemental nutrition and food transfers.  We estimated the net impact of ICDS on nutritional intake among program-eligible households, and measured spillover or learning effects associated with ICDS among ineligible households.   

Method(s): We analyzed longitudinal data describing household expenditures on food goods and their nutrient content using the 2004-2011 Indian National Sample Surveys (n=362,701).  We used government reports to construct an annual measure of state-level program intensity (% of children below 6 years receiving food from ICDS). We performed a difference-in-difference ordinary least squares regression analysis of the impact of ICDS on food expenditures and dietary nutrients controlling for household characteristics and time/place fixed effects. Program eligibility was defined as households with children below 6 years with a comparator of otherwise similar households with no children. Spillover and learning effects were measured among households with children recently aging out of eligibility (6-10 years) also compared to households with no children. 

Result(s): At average program intensity (37%), ICDS increased eligible household daily net caloric intake (126 calories [95% CI: 39 - 212] (rural); 206 [95% CI: 123 - 288] (urban)), assuming program transfers accorded with ICDS norms. Eligible households shifted daily food purchases away from dairy (-44 grams [95% CI: -69 - -18] (rural); -21 grams [95% CI: -45 - 3] (urban)) and cereals (-36 grams [95% CI: -47 - -25] (rural); 7 grams [95% CI: -13 --0.2] (urban)) toward rice and wheat (rural households) or meat (urban households). ICDS had similar but smaller effects on nutritional consumption and food expenditure in households with children too old for program eligibility, providing evidence of spillover effects or household learning associated with ICDS.

Conclusion(s): India’s ICDS modestly increased net caloric intake especially in rural areas even as households responded to food transfers by reducing and shifting their food expenditures. As these net effects were smaller than the ICDS transfers and also occurred in households with older children, evaluations of ICDS’s benefits should account for both changes in household consumption that may offset the programs transfers and spillovers to non-eligible households.


Mathilde Francois, MD - MPH, Jonathan Sicsic, PhD and Nathalie Pelletier-Fleury, MD - PhD, Centre de Recherche en Epidémiologie et Santé des Populations - Equipe 1 'Economie de la santé - Recherche sur les services de santé' (CESP, INSERM, UMR 1018 ), Villejuif, France
Purpose: In 2011, the French National Authority for Health decided to downgrade the medical service provided by cholinesterase inhibitors and memantine. The purpose of this study was to assess the impact of this policy decision on drug prescribing patterns for people with dementia.

Method(s): A longitudinal study using data from the French national-health insurance database was performed over a 9-year period. The study population included patients over 65 years old, without exclusion criteria. Rates of drugs prescribing were calculated for each year and the policy’s impact was tested using adjusted segmented regression analysis.

Result(s): During the 2006 – 2014 period, 119,731 individuals were followed. Overall, the annual rates of drug prescribing increased between 2006 and 2011 (from 2.23% (95%CI: 2.13-2.34%) to 2.64% (95%CI: 2.54-2.75%), p < 0.0001) and decreased from 2012 to 2014 (from 2.36% (95%CI: 2.26-2.46%) to 1.92% (95%CI: 1.84-2.01%), p < 0.0001). Drugs for dementia were prescribed at higher rates for females (p < 0.0001), patients older than 85 (p < 0.0001) and patients suffering from at least one other chronic disease (p < 0.0001). The decrease in drug prescribing after 2012 was significantly greater for females (p = 0.0415), patients aged 65-74 (p < 0.0001) and patients suffering from a least one other chronic disease (p < 0.0001).

Conclusion(s): The policy decision implemented in 2011 had a significant impact on the prescribing patterns of drugs used in the treatment of dementia and mainly affected women, younger patients and patients with multimorbidity. This decrease in prescribing could reduce the cost of dementia without diminishing the quality of care.


Shi-Yi Wang, MD, PhD1, Siwan Huang, BS1 and Sylvia Hsu, MD, PhD2, (1)Yale School of Public Health, New Haven, CT, (2)York University, Toronto, ON, Canada
Purpose: Hospice use has increased dramatically in the United States since last decade. This study aims to examine current racial differences in end-of-life care expenditures and intensity among cancer decedents, and to compare the findings in prior literature based on data in an era with low hospice use. 

Method(s): Using the Surveillance, Epidemiology, and End Results–Medicare data, we identified 88,751 beneficiaries who died from breast, prostate, lung, colorectal, pancreas, liver, kidney, melanoma, or hematological cancer within 3 years of cancer diagnosis from 2006 to 2011. We calculated last month of life healthcare expenditure and end-of-life care intensity, measured by 1) chemotherapy received within 14 days of death; 2) >1 emergency department (ED) visit or hospitalization within 30 days of death; 3) ≥1 intensive care unit (ICU) admission within 30 days of death; 4) in-hospital death; or 5) hospice enrollment ≤3 days before death. The magnitudes of racial differences in end-of-life care intensity were compared with the published results based on similar population from 1992 to 2001.

Result(s): The mean expenditure on end-of-life cancer care per decedent in the last month of life was $10,400 for Whites, $12,300 for Blacks, $12,900 for Hispanics, and $14,500 for Asians. Approximately 70.6% of White decedents enrolled in hospice; whereas enrollment was lower for Blacks (62.9%) and Asians (53.5%). Higher proportions of Asian and Black decedents than of White decedents had >1 hospitalization (17.1%, 17.4%, 13.8%, respectively), >1 ED visit (39.5%, 42.2%, and 34.1%, respectively), were admitted to the ICU (27.9%, 23.4%, 17.2%, respectively) in the last month of life and died in the hospital (35.7%, 26.1%, 20.7%, respectively). However, Asian and Black decedents, compared with White decedents, were less likely to receive late chemotherapy (3.2%, 3.3%, 4.2%, respectively) and have late hospice enrollment (5.7%, 6.9%, 8.5%, respectively). These racial differences remained after adjustment. Compared with the findings of 1992-2001, hospice use almost doubled across all groups, and end-of-life care intensity generally increased. The magnitude regarding racial differences in the proportion of decedents who had ICU admission or repeated ED visit increased substantially.

Conclusion(s): Asian and Black cancer decedents were more likely to receive aggressive end-of-life care than White decedents and incurred higher end-of-life expenditures. Recent increase in hospice use did not mitigate racial differences in end-of-life care intensity.