Thursday, April 23, 2015

Block Granting Medicaid is a Terrible Idea


Today’s Managing Health Care Costs Number is $2 trillion


Source
The New England Journal published an op ed yesterday on Republican proposals to convert Medicaid to become a block grant to each state. This block grant would not increase at the rate of medical inflation –but rather at a much smaller rate.   The legislation (The Patient CARE Act) sponsored by chairs of the Senate Finance Committee (Orin Hatch) and the House Energy and Commerce Committee (Fred Upton), would exempt state Medicaid programs for the disabled and the elderly, and federal matching funds would continue for these programs.

Thus, the total weight of the $2 trillion cut over a decade would fall on programs for children and moms (and other poor adults under age 65 in states which expanded Medicaid).

The authors note that:

Such a law would be bad news for beneficiaries and for providers, especially those that serve low-income communities, since under such financing terms few, if any, states could maintain existing coverage for affected populations. With projected federal budget cuts from Medicaid of nearly $2 trillion over 10 years, these changes would force most states to put eligibility, benefit, and cost-sharing protections on the line as they attempted to cope with the brunt of future cost growth. Financial risks associated with health care inflation, changes in technologies and services, and the long-term care needs of an aging population would be largely shifted to the states, as they reached the limits of their capped federal allotments. Moreover, over time, the federal government could ratchet down the caps in an effort to avert further outlays….

The likely adverse effects on the poor and state economies of block-granting Medicaid have been well documented. Indeed, experts attribute the size of Medicaid's budget to the sheer number of people it serves (more than 66 million by 2014) and their health status. On a per-capita basis, spending is actually modest. Put another way, there is little to cut in Medicaid other than people, health care, and (already heavily discounted) provider payments.

The elderly and disabled represent 24% of Medicaid beneficiaries, but are responsible for almost 2/3 of the total program cost.  21% of Medicare beneficiaries are “dual eligible” who have Medicaid for coverage of services not covered by Medicare, such as chronic care in a nursing home. Source: KFF Medicare covers 45% of newborn deliveries in the US – 70% in Louisiana.

Medicaid has been shown to reduce financial strain for beneficiaries, and it’s even been shown to increase the likelihood that eligible children go to college.   Medicaid pays for the delivery of our babies, and for the care of the frailest of our elderly.

States have shown variable willingness to adequately fund Medicaid.  Remember that a family of 4 in Texas would be eligible for Medicaid if they had income of 19% of the federal poverty limit – or under $6000.   Further, the need for Medicaid is countercyclical – and when there is an economic downturn states are in the worst position to increase funding for this program.

This proposed change to Medicaid is a terrible public policy idea.

Wednesday, April 22, 2015

Small study shows statistically insignificant cost savings from Medicaid case management


Today’s Managing Health Care Costs Number is $318



The third of three articles about primary care interventions and payment reform in this month’s Health Affairs is a report on cost savings from severely ill Medicaid beneficiaries in Washington State, who were offered an intensive case management program based on their calculated risk score. There weren’t enough resources to offer these services to all those eligible – so the Medicaid agency recruited members to participate in three waves.  

The researchers did not compare the costs and quality results of the different waves, which would have been difficult given short eligibility duration of many Medicaid beneficiaries.  They rather did  propensity matching where they matched the 907 beneficiaries who had these services with 907 “control” Medicaid patients who were of similar demographics and had similar severity of past illness including drug abuse.  The findings:

·         Hospital costs down $318 pmpm (more than 100% of savings). Note that this was the only statistically significant difference
·          Total medical costs down $248 pmpm (not statistically significant)
·         Inpatient admissions down by 9.6 (not statistically significant)
·         Emergency department visits up by 10.8 (not statistically significant)
·         12 month mortality down 18% (not statistically significant)
·         Cost of intervention $1847 per person ($1.7 million total)

This study is small, shows statistically equivocal results, and is likely to be subject to substantial selection bias despite the propensity matching.    These results are intriguing –but need to be tested in a far larger population before we can draw any public policy conclusions.

Monday, April 20, 2015

BCBS Michigan study touts PCMH savings which are small, but doesn’t include all costs.


Today’s Managing Health Care Costs Number is 1.1%



This is a review of the second of three articles about payment reform that are in the April issue of Health Affairs.   Each proclaims success at lowering cost and improving value, and I’m evaluating them using criteria I laid out last week.

Blue Cross Blue Shield of Michigan established a patient centered medical home program (PCMH) for its primary care physicians – over time about 2/3 participated.   The plan measured costs and quality changes of the early and the later adopters of PCMH compared to the results of the population that never adopted PCMH.   

The study is big (over 3 million members with 12 months of continuous enrollment), though it is not long-term.   The study included all medical costs, so there was no cherry-picking of certain claims lines that went down.  The researchers applied a risk adjustment to account for known demographic or severity differences between assessed groups.  

Note this was a study performed on a PPO style health plan, so members were attributed retrospectively to each primary care group.  

The researchers found that costs grew more slowly ($4.00 pmpm) in the early participant group compared to the nonparticipant group. However, groups’ costs increased by more than this ($5.95 pmpm) in the year they transitioned into the PCMH program.   Higher initial costs for PCMH programs are consistent with other published accounts.   The study showed no inpatient savings – only savings in outpatient and professional costs.  This differs from other published accounts.   It’s good the researchers shared this subanalysis, although it would better sense for PCMH to prevent hospitalizations than to prevent ambulatory care expenses.  The researchers do not report on pharmaceutical costs, which is unfortunate as we’d expect this line item to go up with increased adherence.

The researchers did not account for the cost of administering the program –and did not account for the cost of bonuses given to participating groups that met performance standards. This is a huge deal – the BCBS Massachusetts “Alternative Quality Contract” studies initially reported only on claims costs and showed savings, but only had “all in” annual savings in the fourth year after accounting for payments to providers that were not based on fee for service.   We simply have no idea whether Blue Cross Blue Shield of Michigan actually saved money in this program – it’s not likely, as bonuses were up to 20% of PCP fees.  

On the quality side, the researchers have shown an impressive statistical difference between participating groups and nonparticipating groups – but that difference seems pretty stable pre and post intervention.  In fact, preventive screening rates went down in 5 of the 14 measures reported.  It’s hard for me to conclude that this program led to higher quality – it appears to me that it merely segmented the practices that had processes in place at the outset to lead to higher rates of various preventive services.  

Thursday, April 16, 2015

Cataract Routine PreOp Testing Won't Go Away


Today’s Managing Health Care Costs Number is 53%



Today’s New England Journal has a review of Medicare claims data from 440,000 Medicare beneficiaries who had cataract surgery in 20 that shows that 53% of patients getting cataract surgery get preoperative testing. The authors state that the likelihood that patients have preoperative testing hasn’t decreased since the American Academy of Ophthalmology and other specialty societies recommended against such testing for healthy patients in 2002. Patients got tested based on their surgeon – not their level of illness burden, their age, or the site of the procedure.

It’s easy to disseminate guidelines for a new high margin service – it’s hard to effectively implement guidelines that tell us to do less.  The authors conclude:   These data underscore the fact that publishing evidence-based guidelines alone does not necessarily change individual physician behavior. 

I’ll  be back to the Health Affairs articles on provider payment reform with my next post.

Wednesday, April 15, 2015

Geisinger Study of PCMH Savings - Can't Tell ROI and Can't Apply to non-Medicare Populations


Today’s Managing Health Care Costs Number is $53



This is a review of the first of three articles in this  month’s Health Affairs on savings from provider payment initiatives.

Geisinger Health reports on experience of its patient centered medical home since 2006.    The PCMH program has been rolled out in waves, so researchers compared the regression-adjusted costs of practices that had already implemented PCMH with those that had not.

The research was limited to those who were Medicare eligible and were enrolled in a Geisinger health plan which required designating a primary care physician.   This meant that about 1/3 of each practice’s Medicare patients were included in the evaluation.   The evaluation was done by practice, which represented the “team” nature of the PCMH, rather than by individual patient.

The study was big – 3 million “life months” and over 6400 site months. 

The results showed   a cost savings of 7.9%, including a 19% drop in the cost of inpatient care.  The inpatient calculated savings represented about 2/3 of the total cost savings, and practices which had been in PCMH longer had lower costs.

The researchers recognized that the sites were fundamentally different For instance, the practices that were in PCMH were bigger and had more patients who had Medicare Part D through the Geisinger Health Plan.  They also had more asthma – the other illnesses appear close to equal, although no “p” values are given.  The researchers say that they’ve accounted for these differences using multivariate regression analysis – and they have used the same analysis to set the exposure to PCMH to zero to calculate the cost of care if PCMH had not been implemented.

The researchers have not accounted for the incremental cost of the PCMH intervention itself,    They say that they had no access to this information, although the study was performed by Geisinger researchers who presumably could have gained access to estimates of the cost of the intervention.

The unadjusted cost of care was substantially higher ($869 vs. $735 pmpm) for the sites with PCMH, but after adjustment by the regression analysis the expected cost was $670 and the adjusted observed cost was $617 – a $53 savings.


The researchers say that there was no “cost shifting, noting approvingly that costs went down in all areas,   I expect costs to go up in pharmacy or ambulatory professional services, but perhaps the increased costs are simply not captured because the PCMH intervention itself is not associated with any cost. 

The graphic above evaluates the article based on  the standards I noted in my last post:

Monday, April 13, 2015

Evaluating Reports of Savings from New Provider Payment Models


Today’s Managing Health Care Costs Number is 3


This month’s Health Affairs has three articles trumpeting the effectiveness of provider payment interventions in improving the quality and cost of medical care.  This includes a report from Geisinger on cost savings from its patient-centered medical home program for Medicare beneficiaries, a report on care management done in the Washington state on high risk Medicaid beneficiaries, and a report from BCBS of Michigan on impact of its physician pay for performance program.

I’ll review these over the next few days.  First, here’s a framework for evaluation of provider payment interventions.  

1.     Have the researchers explicitly stated the cost of the intervention?  It’s great to save $5 million; less great if the cost of the program is $10 million! 

2.     What is the comparison group?  Provider practices which voluntarily enroll early in alternative payment arrangements tend to be systematically different than the provider groups which continue in legacy fee for service arrangements. 

3.     How big is the study?   Small studies tend to have large confidence intervals – and should always be confirmed by larger studies.   Smaller studies are also more subject to the influence of a single charismatic effective leader, and thus less likely to be scalable.

4.     Does the researcher report on all expenses, or just a subset? You’d expect that certain costs (primary care office visits, pharmacy) would go up with better management; it’s suspicious when only cherry-picked categories are reported, or when all actuarial categories show lower costs.

5.     What’s the population?   Be careful about extrapolating savings opportunities in the Medicare or dual eligible Medicare/Medicaid population to the commercial (under 65 employer-insured) population.   There are a lot of inpatient and skilled nursing costs and end-of-life costs in the Medicare population; not so in the commercial population.

Saving medical costs is hard – and the triple aim of saving costs while improving population health 
and patient experience is harder still.   It’s great to see so much effort at both implementing AND measuring changes in provider payment.

We should also recognize that publication bias is rife in the health policy world.  Health care organizations rarely publish reports that don’t show the success of their efforts.

ADDENDUM: I'll add one more criterion (4-15-15):

6. Are the researchers independent?  Best that at least some of the researchers are academics not affiliated with the organization reporting the results.   Publication bias is a huge problem, and involving outside independent researchers can make it more likely that negative results would be reported. 

Friday, April 10, 2015

The High Cost of False Positive Mammograms


Today’s Managing Health Care Costs Number is $4 billion





This month’s Health Affairs has an article demonstrating the costs of false positive mammograms in women from 40-59.   The researchers used claims data to show the cost of each false positive, and to estimate the likelihood of a false positive. They calculated national costs at $4 billion by multiplying their findings by national mammography rates.

This is a study based on insurance claims– so it’s not precise.   Some women might have had their final diagnostic test more than 12 months after the initial positive mammogram – and thus been incorrectly labeled as false positives.  The rate of ductal in situ positives is lower than other studies, which could have incorrectly reduced the national cost estimate.

Mammograms save lives – but false positive mammograms are common – and women getting annual screening for a decade have as much as a 61% chance of a false positive during that time period.

Estimates of financial cost also miss the enormous human cost of false positives.  This ranges from worry before a followup mammogram to months of chemotherapy that wasn't really needed.

This study reinforces the importance of moving toward the US Preventive Services Task Force (USPSTF) recommendations of mammography starting at 50 (not 40) for low risk women, and mammography performed every other year rather than annually.   Reducing mammography screening doesn’t apply to women with positive family histories,  certain genetic abnormalities, personal history of breast cancer,  or worrisome physical findings.

We eventually need to migrate to individualized screening recommendations. Women with a high pre-test probability should have more frequent screening; women with low pre-test probability should have less frequent screening.  This would dramatically lower the false positive rate for mammography, and reduce the financial and human costs of false positive screening mammograms.