|A medical home|
It cannot resist.
As readers will recall, the offending JAMA article described how a large three year-long Patient Centered Medical Home (PCMH) multi-payer pilot involving approximately 64,000 patients failed to reduce health care costs or increase quality. The pilot program was called the "Chronic Care Initiative" (CCI), and was the brainchild of then Governor Rendell's reform-minded "Prescription for Pennsylvania."
In the AMJC study, 6940 "intervention" patients with a) at least 3 months of primary care physician follow-up, plus b) at least 6 months of assignment to one of the medical home practices were retrospectively compared to 6940 similar "control" patients from a single non-participating practice. The control patients were matched using "DxCG" risk adjustment software* that was combined with propensity matching.
Pediatric practices were excluded, as were outlier patients with more than $100,000 in medical expenses.
In addition to looking at those patients, the top 10% of risk DxCG patients from the medical home (654 patients) were compared to matched high-risk non-medical home practices (734 patients).
The analysis was complicated by the later attainment of NCQA medical home recognition among some clinics that were taking some of the control patients. This limited the pool of patients in the 3rd year to just over a thousand in both arms, and just over 100 patients in the high risk groups.
There was no difference in the evolution of health care costs among all patients included in the analysis. This confirmed the JAMA drama.
For the top 10% high-risk patients, there were reductions of 61, 48 and 94 hospitalizations per thousand over each of the three years study. This was accompanied by a difference of the per member per month (PMPM) inpatient costs of $115 and $62 in years 1 and 2. While there was also an increase in outpatient specialist visits, the downward change in inpatient utilization drove the difference in combined overall costs in years 1 and 2 of $107 and $75 PMPM.
All these differences were statistically significant. The 3rd year was not because there were too few patients to achieve statistical significance.
While the study was retrospective, the matching methodology is credible enough for the peer reviewers of AJMC and for the PHB. Using control patients from just one clinic is problematic, but no study is perfect.
Which brings us to the punchlines:
1. Two years ago, the prescient Population Health Blog described how modern Ver. 2.0 "disease" (better described as "population") health management can financially succeed. It said that one key ingredient is risk segmenting the population and targeting services at the highest risk patients. This AJMC article says it was right. Most patients won't benefit, but vulnerable patients will. They are the PCMH's customer.
2. The AMJC article also comports with an accompanying JAMA editorial that is discussed here. As the PHB quoted, the JAMA drama....
".... has done a great service for the advocates of the Patient Centered Medical Home by effectively ending promotion of this care model as a generic, low-level, unselective approach to health care delivery for all. The next critical phase of PCMH development should focus on its strategic deployment for the care of high-utilization patients...."
* This uses "linear additive formulas obtained from ordinary least squares regression to combine expenses associated with clinical groups and demographic factors to generate predictions." Wasn't that easy?