Hospital Performance Dashboards: What a Health System Dashboard Needs to Show

Every hospital already has the numbers. Infection rates are counted. HCAHPS comes back quarterly. The CFO knows the operating margin to the decimal. Length of stay is on somebody’s screen right now.

What a health system rarely has is one view where those numbers sit next to the strategy they were chosen to serve, with a human name attached to each one.

We can put a size on that gap, because we host it. Across 50 health systems on our platform, the median organization tracks 370 KPIs across 12 to 13 separate plans. Forty percent of them carry more than 500. CMS, for its part, star-rates 64. Most board-dashboard guidance in circulation suggests a board should look at fifteen to twenty.

A hospital performance dashboard is the instrument that resolves that spread — 370 tracked, 64 regulated, 20 governable — without deleting the detail the quality team needs. This page covers what that dashboard has to show, what the data says hospitals actually track, and the one band of metrics that reliably goes unowned.

What a hospital performance dashboard needs to show

We classified 24,171 hospital KPIs from the health systems in our research set. Ten thousand four hundred and thirty-one of them map cleanly to a recognizable dashboard domain, and they fall into seven bands. The shares below are of that classified subset.

The eighth band: the regulated measure sets

The KPIs that did not classify are not noise. They are the regulated clinical measure sets, imported into the plan more or less as the regulator publishes them. The most repeated names in our healthcare data are HEDIS and MIPS strings: CIS - MMR, CIS - DTAP, CIS - Hep B, IMA - Tdap/Td, WCC - Child/Adolescent BMI Percentile, Controlling High Blood Pressure: MIPS Measure, Diabetes HgA1c Poor Control: MIPS Measure. Diabetes A1c control and hypertension control each appear 177 times.

This band behaves differently from the other seven, and a dashboard that treats it the same way will fail. Regulated measures arrive with their definition, denominator and reporting calendar already fixed. They do not get renamed to suit a strategic pillar. The dashboard’s job is to carry them intact and roll them up — never to redefine them.

Hospitals keep their numbers alive better than any sector we host

Strategy software is full of abandoned metrics. Healthcare is the exception, and the margin is not small.

Two forces explain it, and both are structural rather than cultural. Hospital measurement is externally inspected — the Hospital Inpatient Quality Reporting Program, Joint Commission survey cycles and HEDIS submission windows all impose a calendar that a city’s strategic plan does not have. And clinical measurement already has an owner by default: a measure belongs to a service line, a unit or a named medical director before anyone opens a dashboard.

This holds all the way down the size curve. Removing the five largest health systems from the calculation moves the median by less than two points, so it describes the ordinary hospital rather than the flagship academic centre.

The goals go dark while the numbers stay lit

Here is the asymmetry that a hospital dashboard has to be designed around.

That same median health system keeps a live performance status on 45.7% of its measures and on 7.0% of its objectives. The measures stay lit. The goals they were chosen to serve go dark — a 6.6× gap.

The mechanism is easy to picture and hard to escape. Quality reporting is a machine: the infection rate is submitted whether or not anyone reviews the strategic plan, because CMS requires it. The strategic objective it rolls up to — “become the region’s safest hospital” — has no external submission deadline, so it quietly stops being scored. Two years later the health system has a live, accurate, well-maintained wall of numbers and no current answer to whether the strategy is working.

And the projects finish least often of any sector

The same pattern shows up at the far end, where strategy turns into work. Across 6,163 strategic projects run by health systems in our research set, 5.1% reach completion. In government the figure is 15.0% — three times higher — and education sits at 11.9%. Health systems have the lowest strategic project completion rate of any sector we host.

This one gets worse, not better, when you remove the largest accounts: drop the three biggest health systems and the completion rate falls to 2.5%. It is not an artifact of a few struggling organizations — it is the sector’s ordinary condition.

Read the three numbers together and the shape of the problem is clear. Health systems are the best in our data at keeping a metric current, among the worst at keeping the goal above it scored, and the worst of all at finishing the projects meant to move either. Measurement discipline is not the constraint in a hospital. Follow-through is. A hospital dashboard that only displays metrics reinforces the half that already works.

The metric your board asks about is the one nobody owns

We looked at which domains hospitals put a name on, organization by organization.

Of the 37 health systems tracking patient-experience metrics, 23 — nearly two in three — have not assigned an owner to a single one. For quality and safety, it is 17 of 38. The median health system owns 0% of its patient-experience KPIs and 9.8% of its quality-and-safety KPIs.

The pattern is legible. Metrics a surveyor inspects get a name. Metrics the board asks about get a chart.

That ordering is expensive, because patient experience is the largest single band of hospital KPIs at 22.7%, it is publicly star-rated by CMS on Care Compare, and it is the number a board chair is most likely to raise without warning. HCAHPS produces twelve star ratings — one for each of the eleven publicly reported measures plus a summary — and a health system needs at least 100 completed surveys in a twelve-month period to receive them. All of that is public. Most of it, in our data, is unowned.

Assigning an owner is the cheapest intervention available on a hospital dashboard, and it moves the number that decides whether anything else on the dashboard gets maintained.

What board-ready hospital reporting looks like

A board does not need 370 KPIs. It needs a defensible route from 370 to the twelve or fifteen it will actually discuss, and the ability to walk back down that route when it asks a question.

Four layers, one dataset.

The test of the design is a single question. When a trustee asks “why is readmissions yellow?”, can somebody click through from the board page to the measure, its owner, its last update and the initiative meant to move it, without leaving the room to build a slide? If yes, the dashboard is doing its job. If no, the board layer is a document that resembles a dashboard.

Where the ceilings come from: most published board-dashboard guidance settles around fifteen to twenty indicators. CMS’s Overall Hospital Quality Star Rating uses 64 of the 100-plus measures on Care Compare, drawn from the IQR and OQR programs. Neither number is a rule. Both are useful evidence that the board layer is meant to be small and the layers beneath it are meant to be complete.

How Carilion Clinic built it

Building the dashboard in ClearPoint

A hospital performance dashboard in ClearPoint is one dataset presented four ways.

• Roll-ups across the hierarchy. A unit measure aggregates to a service line, a service line to the system, and the system measure to a strategic objective — one definition, calculated once. Carilion pulls its data automatically through the API rather than rekeying it.
• Named ownership on every element. Every measure, objective, initiative and milestone carries an owner, and update reminders go to that person on the measure’s own cycle. This is the lever behind the ownership gap on this page.
• Separate calendars on the same measure. A HEDIS measure can report on the submission window while its parent objective is scored on the board’s quarterly cycle, without a second copy of the data.
• Board briefing books, generated. The board layer assembles from live data with status, owner, trend and narrative — the work that replaced 40 hours of monthly data population with 15 to 20 minutes for one health system on our platform.
• Public and internal views from one source. Publish a community health dashboard to the public while the operating detail stays internal, with role-based access deciding who sees what.
• Departmental dashboards. Quality, finance, nursing, population health and service lines each get their own view of the same numbers. See a worked template in our healthcare quality improvement dashboard example, built on the IOM aims — safety, effectiveness, access and patient-centeredness.

Frequently asked questions