When Better Healthcare Metrics Can Produce Worse Care

How Goodhart's law explains the growing distance between performance dashboards and the lived experience of medicine

Modern healthcare has never measured more. We track door-to-balloon times, length of stay, readmission rates, patient-satisfaction scores, documentation compliance, relative value units, discharge-before-noon rates, and dozens of other performance indicators.

By many of these measures, healthcare is becoming more efficient. Yet physicians are burning out. Patients feel rushed. Clinical encounters are increasingly shaped by computer screens. Experienced clinicians can find themselves spending more time satisfying documentation requirements and less time doing the reflective work medicine requires. Hospitals may meet their targets while both patients and professionals sense that something important is being lost.

This apparent contradiction can be explained partly by what is commonly called Goodhart's law, often summarized this way:

When a measure becomes a target, it ceases to be a good measure.

A metric initially reflects something we genuinely value. Door-to-balloon time, for example, was built to measure how quickly certain heart attack patients receive reperfusion. I have spent my career in cath labs where those minutes are measured in heart muscle, so I have deep respect for what that metric was built to protect. Shortening unnecessary delays saves myocardium, and it saves lives.

But once the number becomes a rigid target, the system begins reorganizing itself around the metric rather than the underlying purpose. The question subtly changes from how do we provide excellent care to how do we make the number look right.

That distinction matters.

Metrics can help healthcare teams see patterns, but the numbers should begin the conversation, not replace clinical judgment.

A hospital can reduce length of stay by accelerating discharge, but the patient may leave confused, inadequately supported, or medically vulnerable. It can improve patient-satisfaction scores by prioritizing immediate expectations, even when good medicine requires difficult conversations or declining an unnecessary treatment. It can increase physician productivity by demanding more RVUs, while decreasing the time available for diagnostic reasoning, counseling, coordination, and compassion.

Documentation metrics may produce more complete charts without producing better communication. Quality dashboards may become greener while the clinical environment becomes more fragmented.

Why reasonable people produce unreasonable systems

This does not usually happen because healthcare workers are dishonest or indifferent. It happens because people adapt rationally to the incentives placed before them.

If compensation, promotion, staffing, and professional standing depend on a number, that number will influence behavior. Exceptions become inconvenient. Complexity becomes penalized. The things that cannot be readily counted receive less institutional attention: judgment, trust, restraint, continuity, mentorship, and moral responsibility.

The result is a healthcare system that can demonstrate improvement without always delivering an improved experience.

Metrics also tend to measure what is easiest to quantify, not necessarily what matters most. We can precisely count how many patients a physician sees in a day. It is much harder to measure whether the physician recognized that one patient's vague fatigue represented an early malignancy, prevented an unnecessary procedure, helped a family understand a grave prognosis, or earned enough trust for a frightened patient to follow a lifesaving treatment plan.

Those contributions are difficult to place on a dashboard, but they are central to medicine.

Measurement should serve judgment

The answer is not to abandon measurement. Many quality measures have exposed unacceptable variation, reduced delays, standardized effective practices, and improved outcomes. A complex healthcare system cannot function on intuition alone.

The strongest healthcare systems use data to support the people making decisions at the bedside, not to substitute for their judgment.

But metrics should remain instruments of judgment, not substitutes for it.

First, pair every target with balancing measures. If we shorten length of stay, we must examine what happens after discharge. If we increase productivity, we must monitor errors, turnover, access, continuity, and clinician well-being.

Second, use metrics to identify questions rather than dictate conclusions. An outlying number should prompt investigation, not immediate punishment.

Third, involve frontline clinicians. They should help design and interpret the measures governing their work, because they understand the exceptions and unintended consequences invisible in administrative data.

Finally, preserve room for professional judgment. Not every patient can move through healthcare at the same speed, and not every patient should.

Medicine is not a factory, and patients are not standardized units of production. Efficiency matters because resources are finite. But efficiency is valuable only when it serves the patient.

I see the same principle from another vantage point in my medical-legal work. A chart assembled to satisfy documentation metrics can look complete while showing very little of the judgment behind the decisions. The record is supposed to show who decided, with what information, and on what reasoning. When documentation becomes a target instead of a tool, that visibility is what quietly disappears, and everyone who relies on the record later feels the loss.

The deepest danger of Goodhart's law is not merely that people game the system. It is that institutions gradually forget the difference between the metric and the mission.

Healthcare may continue producing better numbers.

We must make sure it is also producing better care.