Consumer-Directed Health Plans: Do Doctors and Nurses Buy In?

March 16, 2017
Lucinda B. Leung, MD, MPH
Lucinda B. Leung, MD, MPH

,
Jos&eacute
Jos&eacute

,
J. Escarce, MD, PhD
J. Escarce, MD, PhD

Volume 23, Issue 3

Although clinical knowledge positions physicians and nurses well as consumer-directed health plan enrollees, they appeared less likely to choose these plans than nonmedical faculty and staff.

ABSTRACTObjectives: Aiming to increase healthcare value, consumer-directed health plans (CDHPs)—high-deductible health insurance plus a personal spending account—equip enrollees with decision-support tools and expose them to the financial implications of their medical decisions. This study examines whether medically knowledgeable consumers are more or less likely to select a CDHP than individuals without medical knowledge.

Study Design: Using University of California Los Angeles (UCLA) human resources data, our observational cross-sectional study analyzed the health plan enrollment choices of 3552 faculty and 8429 staff employees.

Methods: We compared CDHP selection in 2 cohorts: 1) physicians and nonphysician faculty and 2) nurses and nonmedical staff. We used probit regression models to predict CDHP selection, adjusted for job title, demographics (ie, age, gender, race/ethnicity, education, employee income), and coverage type (eg, single).

Results: Approximately 5% of UCLA employees chose the CDHP. After adjusting for sociodemographic characteristics and coverage type, physicians were less likely to choose these plans than nonmedical faculty, when all other covariates were fixed at their means (predicted probability change [ΔP], —1.6%; standard error [SE], 0.8%; P = .05). Nurses also appeared less inclined to choose these plans than nonmedical staff, which approached statistical significance (ΔP, —1.9%; SE, 1.0%; P = .07).

Conclusions: Overall low rates of CDHP selection were observed in consumers with and without medical knowledge. Although physicians and nurses seem to be better positioned as CDHP consumers, they appeared less likely to select these health plans compared with nonmedical faculty and staff in our study.

Am J Manag Care. 2017;23(3):e89-e94Takeaway Points

This study examines whether medical knowledge impacts consumer-directed health plan (CDHP) selection.

  • CDHPs, or high-deductible health insurance paired with a personal spending account, equip enrollees with decision-support tools and then expose them to the financial implications of their medical decisions.
  • Although clinical knowledge well positions physicians and nurses as CDHP enrollees, they appeared less likely to choose these plans than nonmedical faculty and staff.
  • Physicians’ and nurses’ relative aversion to CDHPs may call into question the whole premise underlying these plans, suggesting that even medically knowledgeably consumers believe they may not be able to make good healthcare decisions with CDHPs.

Often heralded as the key to containing healthcare costs, consumer-directed health plans (CDHPs) place greater financial responsibility for medical decision making in the hands of consumers. These plans are a combination of: 1) high-deductible health insurance; 2) a personal spending account, such as a health reimbursement account or a health savings account; and 3) decision-support tools for enrollees. Although very few CDHP enrollees have sophisticated tools with healthcare price-transparency and quality information, these are increasingly available over time.1 The available tools are generally Web-based and provide basic health information to consumers who are inclined or able to base their healthcare decisions on them.

In essence, CDHPs expose “informed” enrollees to the financial implications of their medical decisions in hopes of reducing the use of low-value services and overall medical spending. There is no cost sharing for preventive services, such as annual physical exams2; however, other medical services, such as referrals or medications, are not exempt from the deductible. Compared with other plans, CDHPs appear to reduce overall utilization by 5% to 14%3 and by even greater proportions in healthy enrollees.4 Although CDHPs have only recently emerged, they have gained steady traction in the health insurance market, where they currently cover nearly a quarter of insured workers.5

The rapid growth of CDHPs, accompanied by few research studies, raises 2 concerns. Fueled by findings from the RAND Health Insurance Experiment,6 the first concern is that CDHP enrollees may avoid medically necessary care because of high cost sharing. Early studies on CDHPs’ effect on utilization of services have reached similar conclusions.3,7 According to employee surveys, CDHP enrollees were more likely than other health plan enrollees to forgo medical care to save money.8 Indiscriminately curbing utilization due to higher cost sharing may put consumers at risk for detrimental health consequences. The second concern is whether this can be fixed by improving consumer decision-support tools. Previous research shows that CDHP enrollees tend to be younger and white, in better health, more educated, and more likely to seek information on cost or treatment alternatives and to have a higher income.9-15 These demographics suggest that these consumers are well positioned to use information tools to make sound decisions regarding their own healthcare.

Given the above concerns for consumer healthcare decision making, one may wonder how enrollees with detailed clinical medical knowledge (ie, physicians, nurses) view CDHPs compared with lay enrollees. On the one hand, physicians and nurses appear poised to use decision-support tools and their own medical knowledge to make sound healthcare decisions, including when to forgo unnecessary care. On the other hand, physicians and nurses may be particularly attuned to the difficulty of making good healthcare choices, especially when faced with illness in an unpredictable, expensive healthcare system and with strong financial incentives to limit care. Examining their plan choices could, therefore, provide some insight into the viability of these plans. If physicians and nurses are more likely than other consumers to choose CDHPs, then at least, under ideal circumstances, these plans may represent a useful model for curbing utilization while preserving necessary care. By contrast, if we find that physicians and nurses are less likely than other consumers to choose CDHPs, it may suggest that medically knowledgeable consumers are concerned that they may not obtain the care they need through these plans.

Using University of California Los Angeles (UCLA) employee health plan enrollment data, we aimed to examine whether consumers with medical knowledge (eg, physicians, nurses) are more or less likely to select the Blue Shield Health Savings Plan (ie, CDHP) compared with similar nonmedical university employees with less medical knowledge.

METHODS

Data Source and Selection

We conducted an observational cross-sectional study with 2015 UCLA employee health plan enrollment data obtained from Human Resources. One year prior, UCLA revised its health plan offerings and introduced a CDHP. Employees were notified regarding the availability of the new plan and its features—most notably, the health savings account to which University of California (UC) contributes. The CDHP network is a Blue Shield preferred provider organization (PPO), which includes many UCLA providers and is also available to UC Care PPO enrollees. Each employee had to choose a new plan or be defaulted to a new plan similar to his or her previous plan choice. We capitalized on this opportunity to analyze employee health plan choices.

Our data source contained health plan choice, coverage type, job title, job location, and selected sociodemographic characteristics (ie, age, gender, ethnicity, education level, employee income) for 32,721 employees. We excluded non-UCLA employees (n = 2954) and UCLA employees who did not select 1 of the following major UCLA health plans (n = 2750): Kaiser Permanente North or South health maintenance organization (HMO), Health Net Blue & Gold HMO, UC Care PPO, Blue Shield Health Savings Plan (CDHP) (eAppendix A [eAppendices available at ajmc.com), and Core Major Medical (catastrophic insurance) (eAppendix B).

To compare employees of similar educational backgrounds, we conducted stratified analyses for: 1) faculty, which included physicians and nonphysician faculty; and 2) staff, which included nurses and other nonmedical campus staff. (We excluded 2 individuals who did not have a listed job title.) UCLA employees also included Medical Center staff who were not nurses (n = 10,605), such as administrators, clerical staff, maintenance staff, radiology and laboratory technicians, and aides, but we excluded these employees from the study because of the wide variation in their medical knowledge. We also excluded all nonmedical campus staff in the lowest income quartile (n = 4429) because none of the nurses earned salaries in this income bracket; we wanted to ensure full overlap in the income distribution of both staff groups. In our analyses, we examined the health plan choices of the remaining 11,981 UCLA employees who selected 1 of the above plans (faculty: n = 3552; staff: n = 8429). The study received Institutional Review Board approval from the UCLA Human Subjects Protection Committee.

Dependent Variable and Primary Regressor

Our primary outcome variable was selection of the CDHP or another health plan. Our main predictor variables were being a physician (vs nonphysician faculty) in the faculty analyses and being a nurse (vs nonmedical campus staff) in the staff analyses. We presumed physicians and nurses had considerable clinical medical knowledge relative to their counterparts.

Control Variables

We adjusted for 2 types of covariates: sociodemographic characteristics and coverage type. Demographic variables included age, gender, race/ethnicity, educational attainment (in the staff analyses), and employee income; nearly all are known predictors of CDHP selection.9-12,14 We only had education information on 93% of faculty and 76% of staff. Because of the minimal variation in educational attainment (ie, 100% of physicians and 98% of nonphysician faculty possessed a graduate degree), we omitted education from faculty regressions. Given the wider variation in educational attainment for the staff cohort, we performed multiple imputations for education in staff regressions. We controlled for coverage type (ie, single, family, single with children) and employee income—both of which are used to calculate premium costs for each health plan.

Statistical Analysis

In descriptive analyses, we compared health plan choices, coverage types, and sociodemographic characteristics among employees of different job titles using χ2 tests. Stratified by faculty and staff cohorts, we used multivariate probit models to estimate the likelihood to choose a CDHP by employee job title after adjusting for sociodemographic characteristics and coverage type. Then, we computed predicted probabilities and calculated standard errors with the delta method using estimates from adjusted regression models. Postestimation analysis also included likelihood ratio χ2 tests and Hosmer-Lemeshow tests. Finally, we performed sensitivity analyses on faculty and staff cohorts by excluding the catastrophic health plan (ie, retaining the CDHP and other noncatastrophic health plans) in regression models. We determined significance using a 2-tailed alpha of 0.05 and conducted all analyses using Stata version 14.0 (StataCorp LLC, College Station, Texas).

RESULTS

Descriptive Analyses

Employee demographic characteristics and coverage selection patterns were all significantly different (χ2 test P <.001) (Table 1). UCLA physicians and nurses were generally younger, more likely to be female, and more likely to be Asian than their nonmedical counterparts. Income distributions were fairly even, except physicians and nurses more often had mid-range salaries ($101,000-$152,000) compared with their counterparts. Although physician and nonphysician faculty coverage-type choices were nearly identical, nurses were more likely to select family coverage than other staff.

Overall, only 4.8% of UCLA employees chose the CDHP over other health plans. Unadjusted analyses did not show a significant difference in likelihood of choosing a CDHP between physicians and nonphysician faculty (5.5% vs 6.1%; predicted probability change [ΔP], —0.6%; odds ratio [OR], 0.89; P = .42), nor between nurses and nonmedical campus staff (4.1% vs 5.0%; ΔP, —0.9%; OR, 0.82; P = .06) (Figure). Consistent with previous research, there was some evidence that employees who chose the CDHP trended toward being younger, male, and white, with a higher income than employees who chose non-CDHPs [data not shown].

Multivariate Analyses

The difference between members of the faculty and staff cohorts grew when we used probit regression to adjust for sociodemographic characteristics and coverage type. Among faculty, physicians were less likely to choose these plans than nonmedical faculty when all other covariates were fixed at their means (ΔP, —1.6%; standard error [SE], 0.8%; P = .05). Among staff, nurses also appeared less inclined to choose these plans than nonmedical staff when all other covariates were fixed at their means, but this finding did not reach conventional levels of statistical significance (ΔP, —1.9%; SE, 1.0%; P = .07) (Table 2).

Sensitivity Analyses

Because the catastrophic health plan also had a high deductible and may target similar consumers, we excluded it from sensitivity analyses and examined only the CDHP versus other noncatastrophic health plans. We estimated regression models for both faculty and staff cohorts without a Core Major Medical health plan and found that the point estimates were similar but slightly farther from statistical significance at the 0.05 level. Among faculty (n = 3340), physicians appeared less likely to choose these plans than nonmedical faculty, when all other covariates were fixed at their means, but the result was borderline statistically significant (ΔP, —1.6%; SE, 0.9%; P = .07). Among staff (n = 8230), nurses also appeared less likely to choose these plans than nonmedical staff, when all other covariates were fixed at their means, but the result was borderline statistically significant (ΔP, —1.8%; SE, 1.1%; P = .08).

DISCUSSION

Overall, we observed low rates of CDHP selection in consumers both with and without medical knowledge in the study. Previous research has found an association between educational attainment and CDHP selection,14 but has not further investigated the content of expertise conferred by education. To our knowledge, this is the first study to specifically examine the role that medical knowledge plays in CDHP selection. After adjusting for sociodemographic characteristics and coverage type, physicians in our study were less likely to choose the CDHP over other health plans compared with nonphysician faculty. Nurses similarly appeared less likely to choose the CDHP compared with nonmedical campus staff, although this difference only approached statistical significance.

As medically knowledgeable consumers, physicians and nurses appear to be better CDHP enrollees than lay consumers. In addition to directly using their own medical knowledge, they can apply their knowledge to decision-support tools when making healthcare choices. On the other hand, they may be unwilling CDHP consumers because they are more cognizant of unforeseeable and high healthcare costs or they have unique insight into how cost sharing may discourage appropriate medical use. Our study findings showing their low rates of CDHP selection lead us to consider why this occurs.

Physicians and nurses may differ from their university counterparts for several reasons. First, consumers with medical knowledge, due to the aforementioned difficulties in making sound medical decisions, may be more risk aversive with their own healthcare needs than lay consumers. Second, physicians and nurses may place a higher value on healthcare than other consumers, making them wary of any supply-side restrictions (ie, high deductible) as CDHP enrollees. Third, UCLA physicians and nurses may not have taken the time to choose among new health plan offerings and were then assigned a default plan compared with other university employees. Of these explanations, we are unable to discern if any of these behaviors were more frequent among medically knowledgeable or lay employees.

Strengths and Limitations

A major strength of our study is the sample size of university-affiliated employees, which allows us to compare a large cohort of physicians and nurses against nonmedical faculty and staff of similar education level. Our sample is also uniquely diverse in age, gender, and race/ethnicity. However, our data is limited in external validity given that it is only from 1 large public university; for example, our UCLA employee cohort was able to choose a CDHP among many other health plans, unlike employees who have a limited array of health plan choices. Additionally, we did not have information on whether an employee chose a new plan or was defaulted to a new one similar to his or her previous plan choice. Future research may expand this study to several large universities or diversify the setting to large employers of health professionals.

Another limitation is that we lacked data on several factors that may influence CDHP selection, including health status, household income, and family structure. Nevertheless, we believe that not having data on health status probably biased our results toward the null hypothesis—that is, toward finding no difference in CDHP choice between physicians and nonphysician faculty and between nurses and nonmedical campus staff. Previous research indicates that better health is associated with a higher likelihood of choosing CDHPs.9 UCLA physicians and nurses were, on average, younger than their lay counterparts and, therefore, likely to be in better health. Consequently, it is likely that our findings would have been even stronger had we been able to control for health status. Finally, we used employee income as a proxy for household income and coverage type to approximate family structure. We lacked information on outside spousal coverage, however, which may have affected health plan selection.

CONCLUSIONS

To date, existing studies have done little to explore how cognitive factors affect CDHP selection. In our literature review, we were only able to identify 1 doctoral dissertation examining the relationship between employee health consumerism facilitated by a workplace health program (conceptually similar though not an actual CDHP) and social cognitive theory constructs, such as self-efficacy.16 Otherwise, we do not know how personality traits (eg, risk aversion, patient activation, self-efficacy, health locus of control) may influence consumer health plan choice. More importantly, we do not know what consumers are actually thinking (ie, the cognitive processes they used) when they chose their health plans. In understanding the thought processes of consumers, we believe we will be able to learn whether the conceptual underpinnings behind these plans are valid and, consequently, likely to accomplish their goals.

Although physicians and nurses may be better positioned to choose CDHPs and use them wisely, our study results suggest that they may not. Given their understanding of the unpredictability of healthcare needs and expenditures, as well as the difficulty of making good decisions in the face of anxiety or illness, physicians and nurses may reasonably be concerned about making the wrong choice. Under these conditions, they naturally do not want to partake in strong financial disincentives to use care. It is unclear if improving decision-support tools will necessarily ameliorate these concerns, because our study results insinuate that even highly sophisticated consumers with medical knowledge believe they may not be able to make good healthcare decisions with CDHPs.

Physicians’ and nurses’ relative aversion to CDHPs may call to question the whole premise underlying these plans. Our data and analyses, however, limit the implications we can draw about the conceptual underpinnings of CDHPs in that we do not have data on cognitive factors that may affect decision making. No existing study offers insight into what employees are actually thinking, including their beliefs, concerns, fears, and the tradeoffs they make when they chose CDHPs. Future research should address major gaps in understanding the cognitive processes that surround choosing these plans, especially regarding a consumer’s ability to make good decisions and his or her worries about potentially forgoing needed care.

Acknowledgments

The authors would like to thank Lydia Oller (UCLA Benefits Services director), University of California Office of the President Human Resources and the University Committee on Faculty Welfare’s Health Care Task Force for their support of this research.Author Affiliations: Veterans Affairs Greater Los Angeles Healthcare System, Los Angeles, CA (LBL); Division of General Internal Medicine and Health Services Research, University of California, Los Angeles (LBL, JJE), Los Angeles, CA.

Source of Funding: Support for this article was provided by the Robert Wood Johnson Foundation Clinical Scholars program and the US Department of Veterans Affairs. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the US government.

Author Disclosures: The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (LBL, JJE); acquisition of data (LBL); analysis and interpretation of data (LBL, JJE); drafting of the manuscript (LBL, JJE); critical revision of the manuscript for important intellectual content (LBL, JJE); statistical analysis (LBL, JJE); provision of patients or study materials (LBL); obtaining funding (LBL); administrative, technical, or logistic support (LBL); and supervision (JJE).

Address Correspondence to: Lucinda B. Leung, MD, MPH, UCLA Robert Wood Johnson Foundation VA Clinical Scholars Program, 10940 Wilshire Blvd, Ste 710, Los Angeles, CA 90024. E-mail: lleung@mednet.ucla.edu. REFERENCES

1. Whaley C, Schneider Chafen J, Pinkard S, et al. Association between availability of health service prices and payments for these services. JAMA. 2014;312(16):1670-1676. doi: 10.1001/jama.2014.13373.

2. Internal Revenue bulletin: 2004-15. Notice 2004—23, health savings accounts—preventive care. IRS website. https://www.irs.gov/irb/2004-15_IRB/ar10.html. Published April 12, 2004. Accessed January 20, 2017.

3. Beeuwkes Buntin M, Haviland AM, McDevitt R, Sood N. Healthcare spending and preventive care in high-deductible and consumer-directed health plans. Am J Manag Care. 2011;17(3):222-230.

4. Lo Sasso AT, Helmchen LA, Kaestner R. The effects of consumer-directed health plans on health care spending. J Risk Insur. 2010;77(1):85-103. doi: 10.3386/w15106.

5. With the excise tax in their sights, employers hold health benefits cost growth to 3.8% in 2015. Mercer Global website. https://www.mercer.com/newsroom/national-survey-of-employer-sponsored-health-plans-2015.html. Published November 19, 2015. Accessed January 20, 2017.

6. Newhouse JP, Manning WG, Morris CN, et al. Some interim results from a controlled trial of cost sharing in health insurance. N Engl J Med. 1981;305(25):1501-1507.

7. Rowe JW, Brown-Stevenson T, Downey RL, Newhouse JP. The effect of consumer-directed health plans on the use of preventive and chronic illness services. Health Aff (Millwood). 2008;27(1):113-120. doi: 10.1377/hlthaff.27.1.113.

8. Dixon A, Greene J, Hibbard J. Do consumer-directed health plans drive change in enrollees’ health care behavior? Health Aff (Millwood). 2008;27(4):1120-1131. doi: 10.1377/hlthaff.27.4.1120.

9. Tollen LA, Ross MN, Poor S. Risk segmentation related to the offering of a consumer-directed health plan: a case study of Humana Inc. Health Serv Res. 2004;39(4, pt 2):1167-1188.

10. Buntin MB, Damberg C, Haviland A, et al. Consumer-directed health care: early evidence about effects on cost and quality. Health Aff (Millwood). 2006;25(6):w516-w530.

11. Parente ST, Feldman R, Christianson JB. Evaluation of the effect of a consumer-driven health plan on medical care expenditures and utilization. Health Serv Res. 2004;39(4, pt 2):1189-1210.

12. Fowles JB, Kind EA, Braun BL, Bertko J. Early experience with employee choice of consumer-directed health plans and satisfaction with enrollment. Health Serv Res. 2004;39(4, pt 2):1141-1158.

13. Barry CL, Cullen MR, Galusha D, Slade MD, Busch SH. Who chooses a consumer-directed health plan? Health Aff (Millwood). 2008;27(6):1671-1679. doi: 10.1377/hlthaff.27.6.1671.

14. Christianson JB, Parente ST, Feldman R. Consumer experiences in a consumer-driven health plan. Health Serv Res. 2004;39(4, pt 2):1123-1140.

15. Agrawal V, Ehrbeck T, O’Neill K, Mango P. Consumer-directed health plan report—early evidence is promising. McKinsey & Company website. http://www.mckinsey.com/client_service/public_sector/people/~/media/A4A25CE45A1344C8BF8B0C1EDDBCD1C2.ashx. Published June 2005. Accessed January 20, 2017.

16. Cedergren A. Personal, behavioral, and environmental influences on employer facilitated health consumerism among employees of a large health system: a mixed methods study [dissertation]. Cincinnati, OH: University of Cincinnati; 2013.