# Short-Term Analysis to Support Mental Health and Substance Use Disorder Parity Implementation. Implications of Limited Data for Understanding Effects on Medical Costs

Typically the best way to resolve the question of whether higher utilization of intermediate services generates higher overall total plan costs given variability in plan benefits and utilization would be to estimate a statistical model that accounts for the other factors plausibly related to total costs. We tried such an approach with these data, but the results of models tested yielded estimates of the effect of residential treatment on total PMPM costs that seemed implausible in light of residential treatment utilization being a rare event and a tiny proportion of average total costs.

TABLE 6. Comparing Cost in Plans with Residential Treatment Claims to Cost in all Plans

Final Analytic Sample
(n=290 plans)
Number of Residential Treatment Claims
>0 (n=52)
Mean   Std. Dev.   Min Max Mean   Std. Dev.   Min Max
Charlson-Deyo Index of Chronic Conditions   0.02 0.02 0.00 0.17 0.02 0.01 0.00 0.04
Total Average PMPM \$268.49 \$125.70   \$50.33     \$734.70     \$271.93   \$82.06   \$118.13     \$431.74
Average Medical PMPM   \$200.90   \$84.27 \$34.45 \$518.98 \$207.07 \$60.92 \$90.73 \$341.92
Average MH/SUD PMPM \$12.22 \$10.54 \$0.08 \$113.84 \$12.65 \$5.75 \$3.39 \$37.00
Average MH/SUD Prescription PMPM \$7.46 \$6.51 \$0.08 \$49.52 \$7.18 \$3.10 \$2.70 \$17.22
PANEL B
Final Analytic Sample
(n=290 plans)
Number of Residential Treatment Claims
>10 (n=13)
Mean Std. Dev. Min Max Mean Std. Dev. Min Max
Charlson-Deyo Index of Chronic Conditions 0.02 0.02 0.00 0.17 0.02 0.01 0.01 0.03
Total Average PMPM \$268.49 \$125.70 \$50.33 \$734.70 \$329.19 \$77.72 \$158.69 \$431.74
Average Medical PMPM \$200.90 \$84.27 \$34.45 \$518.98 \$244.64 \$48.78 \$129.54 \$312.71
Average MH/SUD PMPM \$12.22 \$10.54 \$0.08 \$113.84 \$17.15 \$7.40 \$4.84 \$37.00
Average MH/SUD Prescription PMPM \$7.46 \$6.51 \$0.08 \$49.52 \$9.93 \$3.98 \$2.71 \$17.22

To gauge the potential impact of increased utilization of residential treatment services on total plan costs, we compared average PMPM costs for our small subsample of plans that experienced a residential claim with the overall sample. These findings are reported in Table 6. In Panel A we compare plans for which there is a claim for residential treatment and in Panel B we subset this sample further to plans with more than 10 residential treatment claims. It appears in Panel A that by selecting on plans that had a residential treatment claim in 2008, average behavioral health care spending across plans increases by about \$0.43 (from \$12.22 for all plans to \$12.65 for plans with a residential treatment claim). Importantly, the mean difference in average medical spending and total PMPM costs across these groups rises by more than what is observed for behavioral health care spending. Although the plans likely differ on many dimensions, it is difficult to imagine how a \$0.43 difference in residential treatment could influence a \$6.17 difference in average medical costs and a \$3.44 difference in average total plan costs. (Note that behavioral health prescription costs fall a bit on average as we move to this sample, which may be part of the reason why total plan costs rise by less than medical costs alone. Omitted from the table is non-MH/SUD prescription drug costs, which is the other factor causing total plan costs to rise by less than medical costs). The Charlson-DeyoIndex, which we presumed would capture the general health of the plan population by indicating presence of expensive chronic illnesses, does not suggest any differential severity in health across these two groups and indeed the variance in this value is reduced in the plans that cover residential treatment.

When we make the comparison more selective and consider only those plans that had more than ten residential treatment claims (Panel B), we still do not see differences in average severity of illness among enrollees (using the Charlson-Deyo Index), and yet we see even larger differences in average medical costs and total costs than those observed for behavioral health (MH/SUD). Plans providing more generous coverage for these intermediate services appear to provide more generous coverage for medical services as well. However, we cannot rule out other potential explanations for the positive association, including unobserved case mix differences in plan populations (that are not adequately accounted for by the Charlson-Deyo Index).

Table 7 shows that the results presented for partial hospitalization are similar to those found for residential treatment, even though these visits are more common across plans. Panel A shows that health plans that paid claims for partial hospitalization visits in 2008 exhibit a far greater rise in average medical costs (\$200.90 to \$207.11) than average behavioral health costs (\$12.22-\$12.90). The differences between all plans and plans covering partial hospitalization visits get even more pronounced when we focus on plans with more than 20 claims for partial hospitalization (Panel B).

TABLE 7. Comparing Cost in Plans with Partial Hospitalization Claims to Cost in all Plans

Final Analytic Sample
(n=290 plans)
Number of Partial Hospitalization Claims
>0 (n=165)
Mean   Std. Dev.   Min Max Mean   Std. Dev.   Min Max
Charlson-Deyo Index of Chronic Conditions   0.02 0.02 0.00 0.17 0.02 0.02 0.00 0.17
Total Average PMPM   \$268.49   \$125.70   \$50.33     \$734.70     \$277.01   \$110.87   \$92.97     \$734.70
Average Medical PMPM \$200.90 \$84.27 \$34.45 \$518.98 \$207.11 \$73.37 \$73.68 \$420.40
Average MH/SUD PMPM \$12.22 \$10.54 \$0.08 \$113.84 \$12.90 \$8.43 \$2.47 \$54.08
Average MH/SUD Prescription PMPM \$7.46 \$6.51 \$0.08 \$49.52 \$7.72 \$5.66 \$0.75 \$41.71
PANEL B
Final Analytic Sample
(n=290 plans)
Number of Partial Hospitalization Claims
>20 (n=63)
Mean Std. Dev. Min Max Mean Std. Dev. Min Max
Charlson-Deyo Index of Chronic Conditions 0.02 0.02 0.00 0.17 0.02 0.02 0.01 0.16
Total Average PMPM \$268.49 \$125.70 \$50.33 \$734.70 \$292.17 \$90.78 \$106.20 \$713.09
Average Medical PMPM \$200.90 \$84.27 \$34.45 \$518.98 \$220.93 \$58.85 \$87.57 \$407.61
Average MH/SUD PMPM \$12.22 \$10.54 \$0.08 \$113.84 \$13.54 \$7.54 \$3.46 \$54.08
Average MH/SUD Prescription PMPM \$7.46 \$6.51 \$0.08 \$49.52 \$8.15 \$5.13 \$1.51 \$41.71

This evidence reinforces the interpretation that plans providing more generous coverage for these intermediate services provide more generous coverage for medical services as well.12 However, such an interpretation can only be verified through the construction of a statistical model using multiple years of data so that unobserved plan characteristics, including case mix of enrollees, are fully accounted for.

12. Similar analyses are not presented for IOV because the vast majority of plans have claims, and hence there is no statistical difference in means for plans with positive claims. Because the average number of visits across plans are generally below 20, we also do not find significant difference in means for plans with episode lengths within the 75th percentile.

#### View full report

"mhsud.pdf" (pdf, 669Kb)