Currently Viewing:
The American Journal of Managed Care December 2017
Chronic Disease Outcomes From Primary Care Population Health Program Implementation
Jeffrey M. Ashburner, PhD, MPH; Daniel M. Horn, MD; Sandra M. O’Keefe, MPH; Adrian H. Zai, MD, PhD; Yuchiao Chang, PhD; Neil W. Wagle, MD, MBA; and Steven J. Atlas, MD, MPH
Expanding the "Safe Harbor" in High-Deductible Health Plans: Better Coverage and Lower Healthcare Costs
A. Mark Fendrick, MD, and Rashna Soonavala
Currently Reading
Impact of Consumer-Directed Health Plans on Low-Value Healthcare
Rachel O. Reid, MD, MS; Brendan Rabideau, BA; and Neeraj Sood, PhD
ED-Based Care Coordination Reduces Costs for Frequent ED Users
Michelle P. Lin, MD, MPH; Bonnie B. Blanchfield, ScD, CPA; Rose M. Kakoza, MD, MPH; Vineeta Vaidya, MS; Christin Price, MD; Joshua S. Goldner, MD; Michelle Higgins, PA-C; Elisabeth Lessenich, MD, MPH; Karl Laskowski, MD, MBA; & Jeremiah D. Schuur, MD, MHS
Evaluation of the Quality Blue Primary Care Program on Health Outcomes
Qian Shi, PhD, MPH; Thomas J. Yan, MS; Peter Lee, BS; Paul Murphree, MD, MHA; Xiaojing Yuan, MPH; Hui Shao, PhD, MHA; William H. Bestermann, MD; Selina Loupe, BS; Dawn Cantrell, BA; David Carmouche, MD; John Strapp, BA; and Lizheng Shi, PhD, MSPharm

Impact of Consumer-Directed Health Plans on Low-Value Healthcare

Rachel O. Reid, MD, MS; Brendan Rabideau, BA; and Neeraj Sood, PhD
Switching to a consumer-directed health plan is associated with reduced overall outpatient spending, but not with reduced spending on low-value healthcare services.
ABSTRACT

Objectives: To assess the impact of consumer-directed health plan (CDHP) enrollment on low-value healthcare spending.

Study Design: We performed a quasi-experimental analysis using insurance claims data from 376,091 patients aged 18 to 63 years continuously enrolled in a plan from a large national commercial insurer from 2011 to 2013. We measured spending on 26 low-value healthcare services that offer unclear or no clinical benefit.

Methods: Employing a difference-in-differences approach, we compared the change in spending on low-value services for patients switching from a traditional health plan to a CDHP with the change in spending on low-value services for matched patients remaining in a traditional plan.

Results: Switching to a CDHP was associated with a $231.60 reduction in annual outpatient spending (95% CI, –$341.65 to –$121.53); however, no significant reductions were observed in annual spending on the 26 low-value services (­–$3.64; 95% CI, –$9.60 to $2.31) or on these low-value services relative to overall outpatient spending (–$7.86 per $10,000 in outpatient spending; 95% CI, –$18.43 to $2.72). Similarly, a small reduction was noted for low-value spending on imaging (–$1.76; 95% CI, –$3.39 to –$0.14), but not relative to overall imaging spending, and no significant reductions were noted in low-value laboratory spending.

Conclusions: CDHPs in their current form may represent too blunt an instrument to specifically curtail low-value healthcare spending.

Am J Manag Care. 2017;23(12):741-748
Takeaway Points
  • Consistent with prior study findings, switching from a traditional plan to a consumer-directed health plan (CDHP) was associated with reduced overall outpatient spending. 
  • However, switching to a CDHP did not reduce spending on low-value healthcare services that offer unclear or no clinical benefit and represent a significant source of waste. 
  • This pattern was consistent for laboratory services, imaging services, and services both more and less sensitive to patient preferences. 
  • CDHPs may encourage patients to curb spending indiscriminately rather than specifically reducing low-value services; more targeted consumer incentives in CDHPs may be necessary to reduce this source of waste.
Low-value healthcare services are medical tests and procedures that provide unclear or no clinical benefit to patients, but still expose them to both risk and expense. Despite evidence of their lack of clinical benefit to patients, these unnecessary services remain frequently ordered and contribute substantially toward wasteful spending within the US healthcare system.1-4 Reducing the use of low-value services offers an opportunity to decrease wasteful spending while improving access and quality. One influential effort to reduce low-value services is the American Board of Internal Medicine Foundation’s Choosing Wisely campaign. This initiative, which assembled recommendations from 75 physician and professional societies on low-value services to avoid in their specialty, has garnered support and partnership from patient and payer organizations alike.5-7 

An emerging body of research has begun to measure low-value services in the US healthcare system. Some study results have demonstrated that the volume of low-value services delivered to Medicare patients varies across regions and physician organizations.8-11 Another recent study's results demonstrated regional variation among commercially insured patients and that patients from more advantaged groups (ie, white, higher-income) receive more low-value services.12

In a related trend, consumer-directed health plans (CDHPs) are growing in popularity. These plans combine high deductibles with tax-sheltered health savings accounts (HSAs) that allow patients to pay out-of-pocket costs using pretax dollars. This benefit structure results in greater cost sharing for patients, which is intended to spur value-conscious care choices and reduce wasteful spending. In the employer-sponsored insurance market, CDHP enrollment increased from 4% to 29% over the last decade.13 In the individual market, nearly 90% of Affordable Care Act Marketplace enrollees are in CDHPs.14 Prior research has shown that CDHPs do reduce overall healthcare spending.15-17 If CDHPs encourage more value-conscious choices, then these spending reductions should be concentrated among low-value services that offer unclear or no clinical benefit. However, the effects of CDHPs on low-value services have not been studied. In this study, we assessed the impact of enrolling in a CDHP on low-value healthcare service spending.

METHODS

Study Design

In this quasi-experimental analysis, we used a difference-in-differences (DID) approach to compare the change in patients’ spending after switching to a CDHP from a traditional plan with that of matched patients who remained in a traditional plan.

Data

We used a 25% random sample of 2011 to 2013 Optum Clinformatics Datamart insurance claims for UnitedHealthcare-affiliated commercial plan members across all 50 states. To enable comparisons across patients and geographic areas, Optum standardizes allowed payments in their data as follows: facility outpatient charges are priced at a percent of the submitted charge, professional services are priced at approximately 130% of Medicare fee-for-service pricing for the relative value units (RVUs) assigned to the service, and ancillary services are priced at approximately 120% of the Medicare pricing for the RVUs assigned to the service.

Patient demographic data included age, sex, race, household income, and geographic region via census divisions. Race and household income were estimated by Optum via proprietary algorithms using residential address and other personal information. Health plan information included plan type and whether the plan included CDHP features. We measured comorbidity as the count of diagnoses contributing to the Charlson Comorbidity Index using 2011 claims.18

Inclusion Criteria

We included patients aged 18 to 63 years in 2012 who were continuously enrolled from 2011 to 2013. We excluded patients without complete sociodemographic information, those who were enrolled in a CDHP before 2013, and those enrolled in health maintenance organization and exclusive provider organization plans, as these plan types only rarely offered CDHP options.

Matching

We compared 2 groups of patients. The first group comprised patients who switched from a traditional plan to a CDHP between 2012 and 2013; the second included patients who remained in a traditional plan. To reduce the impact of selection bias, we matched the traditional-plan patients to the CDHP patients on observable characteristics (ie, age, sex, race, household income, census division, comorbidity, and 2012 health plan type). To do so, we employed exact matching, which is more stringent and robust than propensity score methods.19 First, we identified patients in the traditional-plan group who exactly matched patients in the CDHP group based on the observable patient characteristics described above. We allowed more than 1 patient in the traditional-plan group to match each patient in the CDHP group. Then, we excluded patients within each group who did not have at least 1 patient who was an exact match in the other group. Finally, to account for one-to-many matching, we weighted the patients within the traditional-plan group so that their distribution of characteristics was the same as the CDHP group.

Measuring of Low-Value Service Spending

We employed 26 previously published measures of low-value services, focusing on services delivered in the outpatient setting, where the impact of CDHPs on consumer behavior is greatest (Table 1).8,9,12,20,21 These measures are based on Choosing Wisely recommendations, expert consensus, or literature evidence. Detailed specifications are provided in eAppendix Table 1 (eAppendices available at ajmc.com).

We measured spending for instances of low-value services using 3 approaches. First, for most low-value services, we simply used the cost from the service’s claim as the spending for that service. Second, for low-value services for which there are predictable related services that co-occur (eg, venipuncture for a blood test), we also included the cost for any claims for a narrow set of related services that occurred on the same day in the spending for that low-value service. We applied this approach to the following measures: homocysteine testing in cardiovascular disease, parathyroid hormone testing for stage I-III chronic kidney disease, hypercoagulability testing for venous thromboembolism, preoperative chest radiography, preoperative pulmonary function testing, stress testing in stable coronary artery disease, and inferior vena cava filters to prevent pulmonary embolism. (Specifications for the co-occurring services are provided in eAppendix Table 1.) Finally, for complex services where the true cost of the service included a wider array of co-occurring related services, we summed outpatient costs for the entire day of the low-value service. We applied this approach to the following low-value services: renal artery angioplasty or stent, arthroscopic surgery for knee osteoarthritis, spinal injection for lower back pain, and vertebroplasty or kyphoplasty for osteoporotic vertebral fractures.

After measuring spending for each instance of a low-value service, we summed each patient’s annual spending for each low-value service. Then, we summed each patient’s annual spending across all low-value and all outpatient services. To reduce the impact of spending outliers on our analyses, we winsorized annual spending for each low-value service and for overall outpatient spending by imputing the spending amounts at the 5th and 95th percentiles for any patients whose spending fell outside these percentiles.

We used these spending calculations to assess 3 spending outcomes: 1) annual outpatient spending overall, 2) annual low-value spending (ie, spending on the 26 low-value service measures), and 3) annual low-value spending per $10,000 in overall outpatient spending. In essence, this proportional outcome allowed us to analyze low-value spending controlling for overall spending.

Regression Analyses

Employing a DID approach to estimate spending, our regression models included a variable identifying patients in the CDHP group, a variable identifying the year after the switch, and an interaction term between these variables that assessed the association between CDHP enrollment and spending. This approach accounts for both spending trends over time and any observed or unobserved differences between the CDHP and traditional-plan groups that were stable over time. We used 2-part models because of the frequency of patients with zero spending. In these models, the first part (a probit model) estimated the probability of any spending and the second part (a generalized linear model with a γ-distribution and a log link function) estimated the amount of spending for those patients who had any spending.22 Our models adjusted for patient and plan characteristics, including age, sex, race, household income, census division, comorbidity, and plan type. We present our results as average marginal effects, or the average change in spending attributable to switching from a traditional plan to a CDHP.

To address whether CDHP effects differed by service type, we repeated these analyses limited to laboratory (Current Procedural Technology [CPT] codes 80000-89999) or imaging (CPT codes 70000-79999) spending. Although a physician or provider is the one ultimately ordering the low-value services, some services are more likely to be subject to patient demand or preferences than others. Therefore, we repeated these analyses for 8 services deemed more sensitive to patient preferences (sinus CT for uncomplicated acute rhinosinusitis, head imaging for syncope, head imaging for uncomplicated headache, back imaging for patients with nonspecific low back pain, imaging for diagnosis of plantar fasciitis, stress testing for stable coronary artery disease, arthroscopic surgery for knee osteoarthritis, and spinal injections for lower back pain) versus the remaining 18 services.

The University of Southern California Institutional Review Board exempted this study. We used SAS version 9.2 (SAS Institute; Cary, North Carolina) for descriptive analyses and STATA (StataCorp LP; College Station, Texas) for regression analyses.

RESULTS

Study Cohort and Matching

A total of 11,149 CDHP patients and 408,019 traditional-plan patients met inclusion criteria. Of these, 11,075 (99.3%) CDHP patients and 365,016 (89.5%) traditional-plan patients had at least 1 exact match in the other group. After weighting, the groups were exactly matched on patient characteristics and had similar 2012 spending (Table 2).



 
Copyright AJMC 2006-2017 Clinical Care Targeted Communications Group, LLC. All Rights Reserved.
x
Welcome the the new and improved AJMC.com, the premier managed market network. Tell us about yourself so that we can serve you better.
Sign Up
×

Sign In

Not a member? Sign up now!