By the time I finished reading “Standardizing Patient Outcomes Measurement,” the New England Journal of Medicine article by Michael Porter, Tom Lee, and Stefan Larsson, I was cheering. The authors point out how a “let a thousand flowers bloom” approach to metrics development has resulted in a glut of process measures in health care and slowed progress toward metrics that reflect real outcomes. They decry the artisanal process by which organizations reinvent the metrics wheel over and over but with enough small differences to render broader comparisons of performance useless.
Our team at Weill Cornell Medicine recently compiled an inventory of the metrics by which we are measured by our four largest commercial payers and for our Medicare ACO. We counted 60 individual measures, with a grand total of 2 that apply across the entire group. All track processes rather than outcomes. There are similarities among the other 58 measures but also nuances that introduce a friction into our operations, to which we must devote significant resources to overcoming.
For example, 18% of the measures relate to management of patients who are diabetic. Every payer agrees on measuring the rate at which our attributed patients have an annual retinal eye exam. But while one payer simply measures whether an annual HbA1C level has been documented, another wants the level to be less than 9%. For one payer it is enough that we document prescribing a 180-day supply of a medication, where for another we need to demonstrate that the patient is taking that medication.
Is there is a business case for this complexity? Possibly, if you take the cynical viewpoint that the entities who administer payment for health care services want to set up providers to fail. I don’t believe that’s true — but the quantity of process measures is a real impediment in the quest for value in health care. Payers with a stated goal of improving quality are actively inhibiting progress toward more meaningful outcome measures.
If this patchwork quilt of process measures is the result of a focus on perfection — by specialty societies that choose measures they can control — then let me beg for imperfect uniformity instead. Variability has been linked to poorer outcomes in health care; surely we should apply the same lens to the metrics we use to measure those outcomes.
We’ve heard much recently about moonshots in medicine. This is a call for a metrics moonshot. The International Consortium for Health Outcomes Measurement, which seeks to identify and propagate well validated measures, is a great launching pad. But the planet on which health care needs to land is one where providers, payers and, most importantly, patients, define shared goals for health and measure efforts on how close we all are to achieving those goals. If all parties can devote the time and space to agree on the measures that matter, then health care’s complexity and cost will be reduced. Process measures are the pre-launch equipment checks that ensure our ship will function. Outcomes measures are what will make that giant step forward.
Meaningful outcomes measures will be out of reach as long as we expend our collective energies on reporting processes. Reducing the complexity of that reporting might just be the catalyst that practicing physicians need in order to focus on patient health.
Your email address will not be published. Required fields are marked *
Note: This is a moderated forum and all comments are reviewed before posting. By clicking on the "Post Comment" button below, you agree to abide by the NEJM Catalyst Terms of Use. We reserve the right to not post every comment; including those that are submitted anonymously or that are potentially illegal, vulgar, libelous, or commercial in nature.
Joe Humphry, MD, FACP
At the Lana’i Community Health Center (a small FQHC), our board and leadership focus on quality care and not on quality metrics as defined by health plans and governmental agencies. Our two metrics are risk reduction for individuals and populations and quality of life particularly for those with disabilities and limited life expectancy in spite of medical interventions. Risk reduction is based on disease modeling with a weight composite metrics that influence 10 year outcomes. A current example is the ACC/AHA CVD risk calculator used in the Million Heart Model. (https://innovation.cms.gov/initiatives/Million-Hearts-CVDRRM/) Risk reduction models exist for most common diseases and usually the greatest determinate of risk is age encouraging us to focus resources on the younger patient with more years of productive life to gain. There are also a number of validated measures of quality of life. For the elderly with disability and social isolation, chronic pain patients, behavioral health patients including substance abuse and those with social disparities, quality of life becomes an important metric that likely also improves long term outcomes. We do report metrics to the appropriate agencies, but ours is an organizational culture of quality. As we provide increasing high quality care for our population based on outcomes and quality of life metrics, we wait for the external organizations driven by financial incentives to recognizes that system change is driven from the internal mission of the organization, and not by health plans and governmental institutions that have little incentive to truly change to delivery of care.
April 24, 2016 at 10:10 pm
Geraldine McGinty
Joe, it sounds as if you are doing double work. You say you're reporting metrics to the various agencies but you are also tracking your performance based on measures that you believe, and I agree, are more meaningful to your patients. Wouldn't it be ideal if you only had to report one set and it incorporated only measures you felt mattered!
May 04, 2016 at 2:08 pm
michael posner
In the pediatric practice arena, the major issue in standardization is not counting process events, but outcomes. Still even when we know how many immunizations against measles are given, we still have parents who worry about autism with the vaccine. We still have parents who will not allow their children to be treated with stimulants for diagnoses of ADHD, which meet appropriate criteria and for which there are good outcome data. Perhaps that's because we are not earning our objectively lowish incomes by unnecessary procedures at objectively highish costs. Insurers quickly discovered that pediatricians nearly always demonstrate the presence of streptococci before treating a child with pharyngitis, or document bacteriuria before continuing treatment for a prospective urinary tract infection. How many abdominal CT scans, or now ultrasounds, are ordered by pediatricians for the diagnosis of appendicitis? Probably fewer than most surgeons. Outcomes are easily measured. If procedures are considered processes, I would think that it is the procedurally driven specialties that need review rather than primary care to explain the necessity for so much emphasis on process, but I wouldn't make any bets on the insurers checking on outcomes.
April 20, 2016 at 11:57 am