Two recent reports extoll the virtues of the Patient-Centered Medical Home (PCMH) and direct primary/concierge care. One is the annual report of the Patient-Centered Primary Care Collaborative. The other is a study of a network called MD-Value in Prevention (MDVIP).
Before I continue, let me be clear that there are positives in these reports and these models. Something in here is working. I just take issue with the purported mechanisms, especially in regards to costs and utilization.
The PCMH report is long and detailed and I won’t try to cover all the information contained within it. I’ll focus on the table that summarizes peer-reviewed studies. On the cost/utilization side, many of the studies reported some degree of less utilization, usually in the form of fewer ER visits and hospital admissions, though the results weren’t consistent. The “additional outcomes” information in the table is inconsistent. One study reported more primary care visits, another less. Some reported increases in cervical cancer screening, hemoglobin A1cs, LDL testing, and colon cancer screening, others reported decreases. The percentage change in these measures was usually much less than 10%, which means none are likely to influence total cost of care much.
The MDVIP study followed the outcomes of patients who paid $125-170 per month for “executive-style” primary care than included an annual 60-90 minute “physical” with lots of screening tests both verbal (a depression screening instrument, e.g.) and blood work. The majority of patients in the MDVIP group did not achieve any cost savings in the first 2 years – only about 25% did. 63% of the patients were reported to achieve some degree of savings in year 3. The absolute cost reduction was not reported, nor was an overall tally of costs/savings. The authors concluded that prevention explained the year 3 cost savings.
The two major problems with these reports are that they assume that simplistic quality metrics such as LDL cholesterol testing explain savings and that these preventive expenses result in savings later. No study that I’ve ever seen about the PCMH or any other model has drilled down the details to support these assumptions. In other words, no study says we spent $X more on doctor visits, cholesterol tests and cholesterol medicine and saved $XX in fewer heart attacks, cardiac bypasses, and strokes. Or $X more dollars on Pap smears and $XX less on cervical cancer treatment. All PMCH studies to date have been superficial in their reported cost metrics.
The reality is that none of them will ever say this. Prevention using medical tests and treatments almost never saves money even in the long run. An ounce of prevention costs a ton of money (read more about this on my blog section on the topic). I think there is a smattering of studies that show that extra attention and effort given to very high-risk patients pays off, but not across broad populations. A diabetic who is prone to go into DKA or severe dehydration/hyperosmolar state, feel horrible, and show up at the local ER vs. the diabetic who is just on metformin and whose hemoglobin A1C is 8.5 on no meds. Reaching out with care managers, home visits, pharmacy vouchers, etc. saves money for the first patient, but not for the second.
If prevention doesn’t explain the modest cost savings, what does? It’s the gray zone of primary care: the problem solving of unique situations, understanding the social influences and patient preferences, respecting patient wishes even if they don’t follow the text book, ignoring some published guidelines for a variety of reasons, balancing co-morbid competing interests, being comfortable with uncertainty and ambiguity, being comfortable with complexity, and even being comfortable with death. None of these processes can be measured with electronic health records so they remain a black box hidden to researchers who don’t understand the true value of family medicine.
Just because simplistic inputs (LDL levels, depression screenings, etc.) can be measured simultaneously with simplistic outputs (ED visits, hospital admissions), does not mean they are connected. The PCMH study authors correctly pointed out a limitation of the studies they reviewed. “More … ‘total cost of care’ research is needed that assesses the costs associated with PCMH transformation (or ‘upstream’ spending) that results in ‘downstream’ savings.” Unless they get this right, there is a high likelihood that value-based payment models being proposed will actually reward interventions with little value and leave silent the factors that family physicians use to produce real value.