• The longest ever study of consumer-directed health plans

    It’s by Paul Fronstin, Martín Sepúlveda, and M. Christopher Roebuck in the latest Health Affairs and spanned four years, 2007-2010. The study also exploits the natural experiment of one, large, midwestern manufacturer that fully replaced its PPO with a CDHP + HSA option (consumer directed health plan + health savings account). Except for voluntary switches to plans outside the employer (e.g., spousal coverage or by leaving the employer altogether), this eliminates selection bias among plan enrollees. If an employee wanted coverage from this employer, she had to switch to the CDHP.

    Still, this is not a randomized trial. The authors selected a matched comparison group from another, similar employer and obtained a very close match on age and other demographics, comborbidities, and prior use of health care services. There were 10,509 subjects in each of the CDHP and comparison groups.

    Two CDHP + HSA options were offered at the plan-switching employer. They differed in level of annual deductible ($1,250 vs. $2,150 for an individual and $2,150 vs. $4,300 for a family), employee premium contribution ($271 vs. $0 for an individual and $2,184 vs. $1,417 for a family). Employer HSA contributions were the same for both plans ($700 for an individual and $1,300 for a family). Almost 90% of employees chose the higher deductible/lower premium option.

    The results:

    Relative to a matched comparison group whose members did not face the same change in health insurance, active workers in the CDHP significantly reduced their physician office visits [by 0.22 to 0.47 visits per person per year, depending on year and relative to the pre-study (2006) level] and prescription drug use [by 0.80 to 1.27 fills per person per year] after the plan was implemented, and these reductions were sustained throughout the study period. This is consistent with the expectation that CDHPs will tend to dampen demand for some medical services.We did not study the effects on costs, outcomes, or health status, but these should be foci of future studies to further understand the effects of reduced utilization that result from these types of insurance arrangements.

    Hospitalization The absence of utilization changes for hospitalizations runs counter to an expectation of reduced utilization. However, it is not surprising, since hospitalizations involve clinical conditions of greater severity, acuteness, and high patient concern, which means that patients may be less sensitive to the costs of these services. Inpatient services are also significantly more costly than outpatient care, which serves to blunt the incentive to avoid care created by the initial high deductibles in CDHPs.

    Emergency Department Use We did find that ED visits increased slightly in year 3 [by 0.011 per  person and 0.018 per person in year 4] of the CDHP relative to the matched comparison group. This is contrary to expectations, and it is more difficult to understand these changes,especially because we did not find that they were associated with higher inpatient use. The increase in ED use might stem from the long-term implications of reductions in physician office visits and prescription drug use after the CDHP was implemented. Fewer physician office visits may lead to the writing of fewer prescriptions, which could in turn mean that individuals with chronic conditions are less adherent to recommended medication therapy. The intent is for CDHPs to reduce the use of discretionary rather than essential care. However, more research is needed to better understand what may have caused the longer-term increase in ED use.

    Prescription Drugs When it comes to the initial decline in prescription drug use, this may be the result of fewer outpatient office visits or simply the introduction of the CDHP. However, it is unknown whether people only reduced unnecessary prescriptions or reduced the use of necessary pharmaceutical services. If the latter occurred, it may explain the longer-term increase in ED use.

    Cancer Screening Contrary to one of the goals of CDHPs, which is to encourage preventive care, recommended cancer screening declined in the first year of the CDHP introduction [by about 0.04 screenings per person for breast, cervical, and colorectal cancer]. With the exception of cervical cancer screening, which fell sharply following a change in screening guidelines, the rates for the CDHP group did rebound and steadied over the remaining years of observation. However, it is worth noting that after the first year of the CDHP, there was no difference in cancer screening rates between the CDHP group and the matched comparison group.

    Declines in cancer screening rates during the first year of the CDHP may reflect reductions in the use of outpatient visits in that first year. This is notwithstanding the fact that expenses associated with the cancer screenings examined were not subject to patient cost sharing but were covered at 100 percent. Given that this employer designed its health plan so that participants would not skimp on preventive cancer screenings, the reduction in cancer screenings was a surprising finding and may suggest that enrollees did not sufficiently understand their health insurance coverage.

    It is also possible that the observed reductions in preventive cancer screenings occurred because plan participants reduced physician office visits since they were subject to the deductible and not considered preventive services. If this was the case, allowing CDHPs with a health savings account to exempt primary care services from the deductible could mitigate the negative impact on preventive services such as cancer screening.

    Clearly there’s a lot more one could learn from this study (e.g., costs and health outcomes), and this may be just the first publication off it. Still, given its relatively strong design (as work in this area goes) and long follow-up, it warrants attention. Based on results to date, it’s fair to say that the consequences of CDHPs are uncertain. One could read them and be encouraged (lower outpatient visits and prescription drug use without long-term reduction in preventative screening). Or, one can read them for signs of trouble (no change in hospitalization and down-stream increases in ED use).

    The accompanying appendix includes a CDHP literature review that generally paints a mixed picture. There appears to be a genuine trade-off between cost savings and use of health-inducing or maintaining services.


    • This is still a fairly short term study. Will be interesting if they follow it to see if drug use goes back up. My first thought is that this is what you might expect to happen. It cuts spending in the low cost areas, office visits, but does little on big ticket spending.


    • It seems to me that utilization is not a useful metric. What we tend to care about is health and cost. Perhaps this study will be seen through the prism of prior assumptions: if utilization went down, costs went down or the other hypothesis: utilization is a proxy for health and more leads to more health.

      The only thing that’s interesting is changes in measurable health and/or cost.

      I suspect to understand what’s going on, you have to look at individual behavior. Since new patient visits to GPs tend to be expensive, it wouldn’t surprise me if young, healthy new employees never hook up with a doctor, then, when they need one, they have to go to the nearest emergency room. Does this result in higher or lower overall costs?

      Or is it that people who are quite sick skip the doctor’s office and on the weekend or in the middle of the night decide they need the ER, when the GP might have made them better for less?

      Or is all this true, but the overall cost of GPs makes the occasional trip to the ER the more cost effective option?

    • By year 3 people who had gone light on health care in the prior years would have had balances in their HSA accounts. If a health situation occured they may have felt that since they had the HSA cash and needed to be seen why not skip the wait for an office visit and be seen ASAP?

      My health plan has sent me several notices about places in my area I can go for immediate care that aren’t hospital emergency rooms. An HSA plan probably needs to reach out to not-frequent users and keep them informed about places where they can get immediate care that are more cost effective. Many hospitals advertize their emergency rooms, it’s natural someone looking for urgent but occasional care would go there.

    • Doctors in our area often have weak after-hours response capacity, and will advise patients to “go to the emergency room” even when the conditions would warrant a lesser level of care.

      Also, several large employers in our area have first self-insured, then moved to CDHP plans with increased co-pays and deductibles — to reduce employer medical costs.

      These employers found that medical costs increased despite — or perhaps bedause of — the reductions in benefits. Now those same employers are implementing employer health clinics with well-advertised free immunization and preventive services as well as free health checkups. They have realized that making basic benefits unaaffordable, or creating a fear on the part of employees that they will be unaffordable, had the effect of driving up overall medical costs rather than reducing them as hoped.

      Now they hope that the free services — despite the employer costs to provide them — will bring overall medical costs back down by earlier interventions in potentially chronic or serious health conditions.

    • 1) So if these are consumer directed plans why does the employer need incentives (100% reimbursement, no less) to get people to use preventative services? If people would stint on cancer screenings, what makes us think they won’t stint on diet and exercise or outpatient mental health? How about office visits that catch a potentially serious illness before it becomes acute? The evidence that people are not fully forward looking agents maximizing their health is really pretty overwhelming. If they are not, why would anyone expect a CDHP to save large amounts of money? A victory for ideology over evidence.

      2) Most health care dollars spent by the non-elderly are spent on behalf of the minority of the population that is seriously ill, especially those with chronic illnesses. Healthy people are not a problem for insurance markets. Finding out the healthy use less health care when they pay for their care is really not interesting or useful. Finding out they are just as healthy is really not interesting or useful. It is the costs and outcomes of the sickly minority that is interesting, and study designs like this will not have enough of them to tell us much. For a person with lupus who is going to hit their out of pocket max every year a CDHP is basically a plan with a zero marginal cost and a lump sum payment of the out of pocket max. So why would it have a behavioral effect? Sure their are sick people who may not spend the max, hypertensives may not. The same goes for diabetics. But do we really want them to stint on doctors? Aren’t regular checkups essentially preventative for the chronically ill?

      3) The null for CDHPs should be that they do not save money and do not improve outcomes. Publish only when you have something that says they save financially significant sums of money and produce better clinical outcomes among sick people than a standard HMO or PPO.