Quality Measures and Doctor Behaviors

by Vadim Shteyler

Efforts to improve health care quality under the ACA have been directed towards expanding EHR use and health IT, improving care delivery by promoting care coordination and population health, and laying incentives for providers to meet quality measures. The 33 ACO quality measures include 8 measures to evaluate preventive care, 12 measures to address goals of managing 4 common diseases, and 7 to assess patient satisfaction. Though quality improvements have not been consistently shown, studies have found modest Medicare spending reductions. In fiscal year 2013, CMS began reducing health care reimbursement rates to hospitals with excessive 30-day readmission rates, as generalized by their readmission rates for heart attacks, heart failure, and pneumonias. These were extended to include readmission rates for hip and knee surgeries in fiscal year 2015. And, as readmissions were estimated to account for $17.5 billion of Medicare costs in 2012 (in part attributable to insufficient discharge services, access to outpatient care, and follow-up), efforts to curb them are expected to continue.

The ACO quality measures have been criticized for being too process oriented (as opposed to outcomes oriented). And, undoubtedly, so few quality measures can’t encapsulate all of health care. Noted shortcomings of readmission rates as a valid indicator of quality include that they do not differentiate between planned and unplanned readmissions and they don’t adequately control for different case-mixes between hospitals. As psychiatric illness is often poorly recorded in medical records, it is a major confounder that may impact different hospitals differently. In this blog post, I add the speculative concerns of a medical student from limited experiences on the wards.

Regarding 30-day readmission rates, I wonder whether efforts to prevent readmissions may, in some instances, unnecessarily prolong the initial hospital stay. When considering discharge for a hypothetical ailment, instead of being 70% confident that a patient will recover without additional medical intervention, clinicians may wait to become 85% confident, thus prolonging the initial hospital stay. Fear of added readmissions from nosocomial infections might mitigate the process, however, for many conditions, this may not be enough. Though readmission rates may fall, the average number of hospital days for the condition may increase.

Besides prolonging the initial admission, readmission rate indicators may not always incentivize good follow-up care, access to outpatient services and coordinated services. In fact, from one aspect, improving coordination of care may (appropriately) increase readmission rates. For example, a hospitalist may want to keep patients with poor follow-up long enough to ensure that they are beyond the window for some short-term complication. If patients had excellent follow-up, the hospitalist might feel more comfortable discharging them earlier knowing that if their condition were to decline, they would be promptly attended to (and, possibly, readmitted). Furthermore, as access to home services improves, more patients would be able to get hospital care at home (e.g., finishing a course of IV antibiotics) and qualify for discharge even earlier. Since more outpatient access to services and well coordinated care may allow for earlier discharges and assurance that patients who need further care would be readmitted, they might result in appropriately higher readmission rates for some conditions. Insofar as readmission rates are currently disincentivized, coordinated care for certain conditions may not be properly incentivized.

One minor concern about the ACO quality incentives is that providers, in an effort to reach the indicated ranges for blood pressure and sugar control as surely as possible, may feel even more pressure to underemphasize lifestyle interventions than they do now. I also wonder how these incentives may influence the manner in which providers present antihypertensive and hypoglycemic drugs to patients.

Along similar lines, I worry that of cancer screening may be more frequently offered but their risks may not be adequately discussed. In a previous blog post, I discussed data suggesting that patients are largely unaware of the risks of overdetection with cancer screening. When those risks are explained, patients become less willing to get screened. As mammography and colorectal cancer screening are directly incentivized, the growing overdetection issue might become even bigger. As providers aim to meet quality standards, regardless of what they are, they may (wittingly or not) increasingly nudge patients during informed consent discussions.

These concerns provide a very limited (and speculative) view of how the ACA may impact providers’ behaviors. Though tying quality metrics to health care reimbursements holds much promise, how to do it best still requires more attention.