Regression discontinuity designs in healthcare research

Venkataramani, Atheendar S.; Bor, Jacob; Jena, Anupam B.

doi:10.1136/bmj.i1216

Cited by 123 publications

(132 citation statements)

References 25 publications

Supporting

Mentioning

126

Contrasting

Order By: Relevance

“…To isolate the effects of the program, we used a regression-discontinuity design 12,13 that took advantage of the eligibility rules of the 340B Program for general acute care hospitals, which establish eligibility above a threshold of 11.75% in the Disproportionate Share Hospital adjustment percentage (DSH percentage) of each hospital. The DSH percentage, a federally defined measure that determines additional payments for uncompensated care, is largely based on the percentage of admissions at a hospital that are for Medicaid patients and low-income Medicare patients (see the Supplementary Appendix, available with the full text of this article at NEJM.org).…”

Section: Methodsmentioning

confidence: 99%

Consequences of the 340B Drug Pricing Program

Desai

McWilliams

2018

N Engl J Med

View full text Add to dashboard Cite

BACKGROUND The 340B Drug Pricing Program entitles qualifying hospitals to discounts on outpatient drugs, increasing the profitability of drug administration. By tying the program eligibility of hospitals to their Disproportionate Share Hospital (DSH) adjustment percentage, which reflects the proportion of hospitalized patients who are low-income, the program is intended to expand resources for underserved populations but provides no direct incentives for hospitals to use financial gains to enhance care for low-income patients. METHODS We used Medicare claims and a regression-discontinuity design, taking advantage of the threshold for program eligibility among general acute care hospitals (DSH percentage, >11.75%), to isolate the effects of the program on hospital–physician consolidation (i.e., acquisition of physician practices or employment of physicians by hospitals) and on the outpatient administration of parenteral drugs by hospitalowned facilities in three specialties in which parenteral drugs are frequently used. For low-income patients, we also assessed the effects of the program on the provision of care by hospitals and on mortality. RESULTS Hospital eligibility for the 340B Program was associated with 2.3 more hematologist–oncologists practicing in facilities owned by the hospital, or 230% more hematologist–oncologists than expected in the absence of the program (P = 0.02), and with 0.9 (or 900%) more ophthalmologists per hospital (P = 0.08) and 0.1 (or 33%) more rheumatologists per hospital (P = 0.84). Program eligibility was associated with significantly higher numbers of parenteral drug claims billed by hospitals for Medicare patients in hematology–oncology (90% higher, P = 0.001) and ophthalmology (177% higher, P = 0.03) but not rheumatology (77% higher, P = 0.12). Program eligibility was associated with lower proportions of low-income patients in hematology–oncology and ophthalmology and with no significant differences in hospital provision of safety-net or inpatient care for low-income groups or in mortality among low-income residents of the hospitals’ local service areas. CONCLUSIONS The 340B Program has been associated with hospital–physician consolidation in hematology–oncology and with more hospital-based administration of parenteral drugs in hematology–oncology and ophthalmology. Financial gains for hospitals have not been associated with clear evidence of expanded care or lower mortality among low-income patients. (Funded by the Agency for Healthcare Research and Quality and others.)

show abstract

Section: Methodsmentioning

confidence: 99%

Consequences of the 340B Drug Pricing Program

Desai

McWilliams

2018

N Engl J Med

View full text Add to dashboard Cite

show abstract

“…16,17 This method leverages the quasi-random nature of the change in program design at the 1939 birth year to derive causal estimates of "real world" impacts. 18 Because individuals on either side of this threshold are likely similar in terms of other characteristics, any abrupt differences in their drug and health services utilization can be attributed to the effects of the plan design. Observations in years further from the threshold aid in estimating the overall age-related trend.…”

Section: Discussionmentioning

confidence: 99%

Impact of income-based deductibles on drug use and health care utilization among older adults

Law

Cheng

Worthington

et al. 2017

CMAJ

View full text Add to dashboard Cite

“…Regression discontinuity is an established method for testing hypotheses when random assignment to experimental and control groups is not possible . Although not yet widely used in health care research, it is suitable for many studies in this setting . In a regression discontinuity design, the probability of assignment to a treatment increases discontinuously when a continuous “rating variable” exceeds a threshold value (known as a cut point).…”

Section: Evaluation Of Rtcimentioning

confidence: 99%

“…In the absence of a valid control group, we employed a regression discontinuity design to approximate experimental conditions while leveraging the existing program structure. Regression discontinuity is a powerful, but under‐utilized method in health care research . This paper describes our regression discontinuity design and our findings about RTCI's impact on community discharge rates.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Assessing the impact of Minnesota's return to community initiative for newly admitted nursing home residents

Hass

Woodhouse

Grabowski

2019

Health Services Research

View full text Add to dashboard Cite

Objective To evaluate Minnesota's Return to Community Initiative's (RTCI) impact on community discharges from nursing homes. Data Sources Secondary data were from the Minimum Data Set and RTCI staff (April 2014 – December 2016). The sample consisted of 18 444 non‐Medicaid nursing home admissions in Minnesota remaining for at least 45 days, with high predicted probability of community discharge. Study Design The RTCI facilitates community discharge for non‐Medicaid nursing home residents by assisting with discharge planning, transitioning to the community, and postdischarge follow‐up. A key evaluation question is how many of those transitions were directly attributable to the program. Return to Community Initiative was implemented statewide without a control group. Program impact was measured using regression discontinuity, a quasi‐experimental design approach that leverages the programs targeting model. Principal Findings Return to Community Initiative increased community discharge rates by an estimated 11 percent (P < 0.05) for the targeted population. The program effect was robust to time and increased with level of facility participation in RTCI. Conclusions The RTCI had a modest yet significant impact on the community discharge rates for its targeted population. Findings have been applied in strengthening the RTCI's targeting approach and transitioning process.

show abstract

Regression discontinuity designs in healthcare research

Cited by 123 publications

References 25 publications

Consequences of the 340B Drug Pricing Program

Consequences of the 340B Drug Pricing Program

Impact of income-based deductibles on drug use and health care utilization among older adults

Assessing the impact of Minnesota's return to community initiative for newly admitted nursing home residents

Contact Info

Product

Resources

About