- ABSM —
- area-based socioeconomic measure
- BCBSMA —
- Blue Cross Blue Shield of Massachusetts
In their article in this issue of Pediatrics, Chien et al1 appear to raise an interesting policy dilemma. By using data from Blue Cross Blue Shield of Massachusetts (BCBSMA) during the period from 2008 to 2012, the authors linked annual plan payment claims to geocoded socioeconomic background information on the plan’s pediatric participants aged 0 to 20 years. They found that children living in census tracks with higher socioeconomic characteristics used more services at higher unit prices than children living in census tracts with lower socioeconomic characteristics. If payments to plans are risk-adjusted on the basis of the socioeconomic status of enrollees, then these findings suggest that resources will be directed to plans that enroll children from better-off families relative to plans that enroll poorer children, which is an impact at odds with the expectations of those who advocate for adding socioeconomic status to risk stratification algorithms.
The study’s principal finding should not be all that surprising. Health care services are “normal” goods, meaning that as income rises, parental health investments in children rise as well.2 Once age, sex, and health status are held constant, it would have been counterintuitive were those with more disposable income not found to have used some of this income to acquire greater amounts of discretionary care, such as outpatient services and prescription drugs, without necessarily acquiring more emergency care. The authors’ finding that children from census tracts with higher area-based socioeconomic measures (ABSMs) had more outpatient encounters but fewer emergency visits than those from low ABSMs locations is consistent with these basic economic predictions.
Whether these findings have significant policy implications, however, depends on both the magnitude of the effect and the extent to which the findings are generalizable. A careful reading of the study would suggest that the authors have overemphasized the importance of risk adjustment for socioeconomic status on expected expenditures within their sample and underemphasized its importance beyond their sample.
The authors’ estimated relationship between the continuous measure of ABSM and plan payments (Table 2) is fairly modest relative to the impact of other factors. The difference in plan payments between a child living in a census tract with a median household income of only $28 000 and one coming from a census tract with a median household income of $100 000 is 15.4%. If we compare the effect of these large differences in socioeconomic status with the impact of chronic conditions, we find that a marginal increase from 0 to 1 chronic condition (a characteristic of 29% of the sample) raises plan payments by 10 times as much (156%) as the large increase in ABSM noted above.
The question of generalizability rests on how much confidence we have that the children enrolled in BCBSMA are representative of children elsewhere. Risk adjustment regimes are designed not to reallocate resources within health plans but to reallocate resources among them. Should it turn out, for example, that poor children in this plan, compared with their counterparts in other plans, are generally low users of care, then risk-adjusting for socioeconomic status among plans might still achieve the objective of redistributing the differential cost burden in favor of less advantaged populations.
National health expenditure data from the Centers for Medicare and Medicaid Services for the year 2012 reveal that mean per-enrollee expenditures for children 0 to 18 enrolled in Medicaid was $3453, whereas the same figure for children enrolled in private health insurance plans was $2166.3 By contrast, the children in Chien et al’s1 sample from the census tracks with the lowest socioeconomic status had mean annual expenditures of only $1924, suggesting some favorable selection into BCBSMA compared with national averages, even for poor children enrolled in other commercial plans. The application of risk adjustment for socioeconomic status in this context would tend to result in directing resources away from a plan with enrollees similar to those in BCBSMA and toward commercial plans in which poor children are more costly.
On the basis of Chien et al’s1 analysis, we conclude that risk adjustment by using socioeconomic status is unlikely to have profound effects on resource distribution among commercial health insurance plans, one way or the other. This is not to say that it would be inconsequential for comparing quality scores among providers or in adjusting shared savings payments for accountable care organizations in which Medicaid-covered patients are admixed with those who are commercially insured. There is much we have yet to understand regarding the dynamics of adverse or favorable selection into health insurance plans, but as with most everything else in debates over health care, the devil is resting comfortably nestled somewhere amid the details.
- Accepted August 16, 2017.
- Address correspondence to Andrew D. Racine, MD, PhD, Executive Offices, Montefiore Health System, 111 East 210th St, Bronx, NY 10467. E-mail:
Opinions expressed in these commentaries are those of the authors and not necessarily those of the American Academy of Pediatrics or its Committees.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2017-1640.
- Chien AT,
- Newhouse JP,
- Iezzoni LI,
- Petty CR,
- Normand ST,
- Schuster MA
- Centers for Medicare and Medicaid Services
- Copyright © 2017 by the American Academy of Pediatrics