Metrics as Leverage Points

Metrics earn their power to change systems when they are simultaneously measurable, actionable, and meaningful. When a metric achieves that status, it tends to cascade across multiple levels simultaneously - clinical practice, reimbursement policy, funding priorities, public health infrastructure, and even cultural norms around disease.

Studies have found that the key ingredients are transparency (publishing results publicly), benchmarking (comparison over time and across contexts), and tight linkage to accountability structures. Remove any one of those, and the metric remains academic.

Examples of Metrics That Moved Systems

1. Infant Mortality Rate - The Original Accountability Metric

The infant mortality rate (IMR) is one of the oldest examples of a single metric restructuring a health system from the ground up. In the late 19th century, IMR emerged as a preferred index of societal health over crude mortality rates. That reframing was enormously consequential.

During the early 20^th century, unsanitary urban conditions and poverty led to up to 30% of infants in some cities dying before their first birthday. Between 1915 and 1950, the U.S. IMR fell from 100 to under 30 per 1,000 live births - before vaccines and modern medicine played much of a role. The metric drove massive structural investments: water chlorination and filtration, milk pasteurization, sewage systems, birth and death registration infrastructure, and new public health institutions. The number itself became the political and social target that mobilized these and continues to spur investments today. In 2025,  IMR for the U.S. dropped to a historic low of  5.4 per 1,000 live births. Recent improvements in IMR are attributed to advances in monoclonal antibody treatments and recommended use of the respiratory syncytial virus (RSV) vaccine in pregnant women, reducing serious infections from influenza and respiratory syncytial virus (RSV) as well as increased education about safe sleeping for infants reducing deaths due to sudden infant death syndrome (SIDS).

System-level changes triggered: Infrastructure investment, public health institution creation, maternal education programs, vaccine and drug development, civil registration systems, legislative action.

2. Blood Pressure - From "Essential" Malady to Treatable Target

Perhaps the most dramatic example of a metric catalyzing systemic change is blood pressure (BP). Until the 1960s, hypertension was considered a benign, unavoidable, and even necessary feature of aging. For instance, in 1944 President Roosevelt was given a clean bill of health with a BP of 220/120 (normal is less than 120/80). He died of a hypertensive hemorrhagic stroke a year later with an estimated BP of 350/195 mm Hg.

The mercury sphygmomanometer gave clinicians a number, and that number turned out to be the foot in the door. Once the VA Cooperative Studies (starting in 1967) showed in the first-ever hypertension RCT that treating diastolic BP >90 slashed stroke and heart failure, the metric became a treatment target. Each subsequent decade added nuance:

• 1979: The HDFP trial introduced goal-directed stepped care - now there was not just a metric but a protocol for hitting it.
• 1991: The SHEP trial showed isolated systolic hypertension was treatable too, expanding the treated population by an estimated 50 million in the U.S. alone.
• 2015: The SPRINT trial moved the systolic target from 140 to 120, reducing all-cause mortality by 27%.

The AHA/ACC regularly update guidelines in response to the evolving evidence around this metric, each update reshaping how tens of millions of patients are treated globally.

System-level change triggered: New drug classes developed, treatment protocols codified, primary care workflows restructured, insurance coverage requirements adjusted, population-wide screening normalized.

3. HIV: A Cascade of Metrics Reshaping a Global Epidemic

HIV is arguably the richest modern example of metric-driven systemic change - and it happened in stages, each new metric unlocking a new level of response.

CD4 count was the first actionable biomarker, establishing thresholds for when to initiate antiretroviral therapy (ART). But as NIH HIV guidelines document, the metric that truly transformed HIV care was viral load. Measuring HIV RNA copies per milliliter gave clinicians a real-time window into treatment efficacy - and more importantly, defined the state of "viral suppression."

That viral suppression concept then spawned two enormous systemic movements:

• The 90-90-90 targets: UNAIDS translated viral suppression into a cascade of population-level metrics - 90% of people with HIV knowing their status, 90% of those on treatment, 90% of those virally suppressed. These three numbers became the organizing framework for global HIV policy, funding allocation, country accountability, and resource flows. In countries that achieved 90% treatment coverage by 2024, new infections have declined by 72%.
• Undetectable = Untransmittable (U=U): Once the metric of "undetectable viral load" was established and validated, it became the evidentiary backbone for the U=U message - endorsed by WHO, CDC, and the NIAID. U=U did not just change treatment guidelines - it changed the social and legal landscape around HIV, reducing stigma, reshaping criminalization debates, and transforming how prevention is framed globally.

System-level change triggered: Global funding mechanisms restructured, country-level health strategies rewritten, clinical monitoring protocols redesigned, legal frameworks around criminalization challenged, prevention messaging overhauled.

4. HbA1c and Diabetes - A Metric That Standardized an Entire Disease Field

Hemoglobin A1c was not always the universal diabetes management yardstick. Its adoption as the standard metric of glycemic control over the past few decades has had enormous systemic effects. It became the first HEDIS quality measure for diabetes, as part of the original Comprehensive Diabetes Care measure set in 1997, meaning health plans across the U.S. track and report it, and performance is tied to reimbursement.

The ADA Standards of Care - updated annually - use HbA1c as a central organizing metric. Targets have been continuously refined with the current 2026 guidelines beginning to deemphasize HbA1c as a stand-alone metric in favor of continuous glucose monitoring data - itself a systemic shift driven by the recognition that HbA1c is an imperfect proxy.

System-level change triggered: EHR alert systems built around HbA1c targets, pay-for-performance contracts structured around it, specialist referral thresholds defined, diabetes registries organized around it, pharmacological innovation focused on it.

5. LDL-C / Cholesterol - From Biomarker to Treatment Architecture

The establishment of LDL cholesterol as a key cardiovascular risk metric has arguably been one of the most commercially and clinically consequential events in modern medicine. Established as an independent risk factor for cardiovascular disease in 1977 following landmark findings from the Framingham Heart Study, it turned cholesterol from a broad lab value to an actionable target for preventing heart attacks and stroke. The statin era was built on it, which shifted the pharmaceutical industries focus from drugs that cured acute symptoms to preventative maintenance. It also reshaped healthcare delivery by pushing health systems, payers, and clinicians toward routine lipid measurement, treatment intensification, and follow-up testing.

The ACC/AHA guidelines - just updated in 2026 - now recommend LDL-C <70 mg/dL for intermediate-risk patients and even tighter targets for high-risk patients. These new 2026 guidelines demonstrate a major systemic shift: the focus has moved toward earlier prevention, starting statin conversations with people in their 30s and 40s. That is the metric driving the system to act earlier and more broadly than before.

System-level changes triggered: Statin prescribing at population scale, lipid panel ordering as routine primary care, insurance coverage for preventive statins codified (USPSTF Grade B recommendation), new drug classes developed targeting LDL (PCSK9 inhibitors and emerging one-time gene-editing therapies).

6. Hospital Readmission Rates - A Metric That Changed the Payment System

The Hospital Readmissions Reduction Program (HRRP) by CMS is the clearest example of a metric directly rewiring financial incentives across an entire national health system. By tracking 30-day readmission rates for conditions like heart failure, pneumonia, and hip/knee replacement, CMS began penalizing hospitals financially for poor performance.

As a result, readmission rates declined from 21.5% to 17.8% for targeted conditions. The metric drove hospitals to invest in discharge planning, care coordination, post-acute follow-up, and patient education programs they'd never previously prioritized - not because it was always the right clinical move, but because the metric made it the financially imperative one.

System-level changes triggered: New care coordination roles created, post-discharge phone call programs launched, partnerships with SNFs and home health expanded, EHR tracking systems built around 30-day windows, financial penalties codified in law.

Cross-Cutting Themes: What Makes a Metric Systemically Powerful?

Looking across these cases, several common features emerge that allow a metric to "punch up" to the system level:

Feature	Explanation
Construct validity	The metric reflects a real, underlying physiological reality (viral load = actual HIV burden)
Actionability	Clinicians or policymakers can actually do something when the number changes
Standardization	The metric is measured the same way across sites/contexts (e.g., HEDIS technical specs) and over time
Public accountability	Results are reported, compared, and tied to consequences (financial, reputational, regulatory)
Cascade potential	The metric organizes multiple downstream behaviors into a coherent "pipeline" (like the HIV care cascade)

Goodhart's Law and the Limits of Metrics

We also cannot ignore the failure modes. Economist Charles Goodhart captured the core problem: "When a measure becomes a target, it ceases to be a good measure." At its core, Goodhart’s Law denotes the structural breakdown in inference about a system that occurs when rules change, and people and organizations start optimizing for the score instead of the goal, especially when tied to rewards or control.

The healthcare world has generated some striking examples of this. CMS's 4-hour pneumonia antibiotic rule is a textbook case. Based on a 1997 JAMA article stating that elderly people who received antibiotics  within 8 hours of arrival to the hospital had  improved survival rates,  CMS created a quality metric mandating antibiotics administered to people with pneumonia within 4 hours of hospital arrival—time to first antibiotics dose (TFAD). The result was that harried emergency room physicians began over diagnosing pneumonia in anyone with respiratory symptoms to meet the timeline – often prescribing unnecessary antibiotics to patients with heart failure and viral infections. The goal of TFAD was to reduce inpatient mortality from pneumonia, and the measure did improve differences between hospitals in time to antibiotic delivery. However, the end result was often misdiagnosis, flawed triage, and inappropriate delivery of antibiotics, and most importantly no association between TFAD and mortality. The metric was shifted to 6 hours in 2008 and in 2014 the entire set of Pneumonia Core Measures was withdrawn.

Similarly, aggressive HbA1c targets can incentivize physicians to push diabetic patients into hypoglycemia risk rather than address the social and economic determinants of poor control. Readmission penalties have been criticized for inadvertently penalizing safety-net hospitals serving sicker, lower-income populations.

The pattern is consistent: metrics designed for populations applied rigidly to individuals produce unintended harm.

Implications/Key points to keep in mind:

1. Metrics are leverage points, not endpoints. The most powerful disease metrics (viral load, HbA1c, blood pressure, DALY) do not just measure - they organize the entire system around a concept of what "better" looks like. Designing a new metric is, in effect, designing a policy.
2. The journey from biomarker to system lever takes time and validation. CD4 count preceded viral load, which preceded U=U, which preceded policy change. The pipeline is long, but the eventual impact is enormous.
3. Composite and cascade metrics outperform single-number approaches for driving systemic change. The 90-90-90 framework is more powerful than any single viral load cutoff because it maps the entire system's performance across the care continuum.
4. Metrics require governance. Without clear definitions, public reporting, accountability structures, and built-in review mechanisms, the best-designed metric will either drift into irrelevance or get gamed.
5. Watch for Goodhart effects. Any metric used for accountability purposes needs ongoing surveillance for unintended behavioral consequences - especially when financial penalties or rewards are attached.