Counterpoint: How Quality Reporting Made Me a Worse Doctor

PATIENT SATISFACTION

My partners complained about the conflict between good medical practice versus giving patients what they demanded (such as unneeded antibiotics and/or opioids) to increase patient satisfaction scores. System factors beyond the direct control of the clinician may also demoralize clinicians who feel they are being unfairly judged.³ In one study, whether patients chose (higher satisfaction) or were assigned (lower satisfaction) their doctor was 10 times more influential than clinician behavior.³ Might one also expect an inverse association between patient satisfaction scores and open access scheduling? I asked myself that question as I continued to keep my practice open to “work-ins,” “walk-ins,” and new patients.

MEASURES BASED ON OPINION NOT EVIDENCE

A measure based on expert opinion increased the death rate of the unlucky patient group that achieved the “quality” benchmark.⁴ It took 4 years from the time of publication of a randomized controlled trial—(Action to Control Cardiovascular Risk in Diabetes [ACCORD])⁴ that provided Level 1 evidence that aggressive treatment of Type 2 diabetes to an A_1c below 7 increased mortality—before this measure was eliminated from my practice.⁵ During those 4 years we who were aware of ACCORD had to choose between evidence-based practice and looking good on the “quality” measure. Here’s another example, though not as deadly. There is no strong evidence for mass screening of depression because all existing trials are of high-risk groups (eg, high utilizers of medical care⁶ or groups containing patients with previous depression diagnoses).^7,8 Depression measures should encourage case finding.⁹ In my experience, mass screening wastes resources.

PATIENT-ORIENTED MEASURES

Advocates for disease-oriented measures (eg, HgbA_1c) argue that surrogates (like laboratory values) are easier to measure than patient-centered outcomes (such as morbidity and mortality). Is it better to do the inadequate thing systematically than back off and do the right thing whenever possible? A growing number of validated patient-reported outcome measures are available (eg, PHQ-9 for depression, the Asthma Control Test for asthma). Process measures for depression (visit frequency) and asthma (medication use) are of limited utility¹⁰ or valueless.¹¹ Only surrogate measures validated in practice-based research effectiveness randomized controlled trials should be considered for use. Directly measuring patient-important outcomes is better.

ARBITRARY BENCHMARKS

Current quality measures reward achieving a high prevalence of performance and rank performance based on arbitrary benchmarks (eg, the benchmark for systolic blood pressure [SBP] control is <140). Is achieving a SBP of 138 better than a SBP of 142? This approach inevitably leads to gaming.¹² Performance measures should draw attention to clinical conditions that most warrant attention (eg, treating a SBP of 220 to 150 is clinically important but does not meet the current benchmark). My partners and I knew this and felt helpless to do anything about it. The original intent of quality measurement was to inform valid quality improvement activities. Benchmarks need to be reconfigured to fulfill this aim.

SHARED DECISION MAKING

Shared decision making is a process in which the clinician offers options to her patient who is encouraged to apply his own values to making the choice that is best suited to him.^13,14 Shared decision making is appropriate for clinical preventive services and management of chronic conditions that form the bulk of current primary care practice. Quality assessment should focus on the shared decision making process, not on the prevalence of the choices made by the patient.² A clear conflict of interest exists for clinicians practicing in settings that link achievement of arbitrary benchmarks to clinician pay or other incentives/disincentives. This may be the most disturbing unanticipated consequence of the “quality” movement. I was an early advocate of clinical preventive service delivery in primary care^15–17 and knew I could manufacture high numbers if I wanted to.¹⁸ I refused to play the game, however, because I had learned that shared decision making was more personally rewarding. This inevitably meant that “quality” reporting made me look like a worse doctor.

My partners complained that we were “not making widgets.” I wonder to what extent clinician burnout may be attributable to knowing that one is being judged unfairly by metrics that undermine effective practice.¹² Measures must be improved. They should provide actionable information. They should align with good clinical practices and promote patient-centered care, especially shared decision making. They should encourage reflection and valid continuous quality improvement. They should undergo regular evaluation and should allow for changes in response to data and provider input.¹ Measures should not be used to arbitrarily and spuriously reward or punish clinicians. Current “quality” measures do not address many things that stakeholders (patients, clinicians, payers) feel are important.¹⁹ The Institute of Medicine has recently outlined a radically different set of core quality measures.²⁰ How many years (or decades) before we see better measures?