Module 4
In Section 1.4, you were introduced to two types of data every QI team relies on: quantitative data (the numbers) and qualitative data (the stories). Below, we'll apply both to a real measurement scenario — then go one level deeper into how QI teams categorize the measures themselves.
Quick Refresher: Quantitative vs. Qualitative Data
As a reminder:
- Quantitative data — the numbers: volumes, rates, percentages, and time.
- Qualitative data — the stories: experiences, perceptions, and context that numbers alone can't capture.
Note: For the full list of quantitative and qualitative data examples, revisit Section 1.4: Measure and Learn.
Going Deeper: The Four Types of QI Measures
Let's apply both types of data to a familiar scenario: a PMHCA program measuring how well its telepsychiatry consult line is working for an ED partner site.
Quantitative and Qualitative Data in Action: Telepsychiatry Consult Line
Quantitative data in action:
- Average time from consult request to connection with a psychiatric provider (in minutes)
- Number of completed telepsychiatry consults per week
- Percentage of consult requests connected within the target response window
Qualitative data in action:
- ED nursing staff feedback on how easy the consult line is to activate during a busy shift
- Notes from a debrief with the on-call psychiatric provider about workflow friction points
- Family reflections on what the telepsychiatry consult experience was like for their child
Together: The numbers might show average connection time creeping up during evening shifts — the staff feedback might reveal why: night-shift nurses aren't sure which extension to call. Neither data type tells the whole story alone.
The rest of this section goes a level deeper into how QI teams categorize their measures. It's most useful once you're ready to design a full QI project measurement plan — you don't need to master it just to understand the data types above.
Knowing what kind of data you're collecting is only part of the picture. To build a well-rounded measurement plan, QI teams also consider what the measure is telling them. Is it about how the system is set up, what's happening day to day, or what results you're getting? That's where the four core types of QI measures come in: process, outcome, structural, and balancing measures.
Note: You don't need to track all four types for every project. Most QI projects rely most heavily on process and outcome measures, with structural and balancing measures used where they add useful context.
The Four Types of Measures
Each type answers a different question about your improvement work:
| Measure Type | What It Tells You | Example |
|---|---|---|
Structural |
Whether the systems, staffing, equipment, or infrastructure needed to do the work are in place. These are often "yes/no" or count-based measures set early in a project. |
Does the ED have a documented protocol and a phone line for behavioral health consults? (Yes/No) |
Process |
Whether the steps you intended to happen are actually happening, as intended, and how reliably. |
Percentage of eligible ED patients who received a trauma screening during their visit. |
Outcome |
Whether the result you ultimately care about (tied to your AIM Statement) is improving. |
Percentage of screened patients who were connected to appropriate behavioral health follow-up within 7 days. |
Balancing |
Whether a change aimed at improving one thing is causing unintended problems somewhere else in the system. |
Average ED length of stay for behavioral health patients, tracked to confirm screening isn't adding a delay somewhere else. |
Adapted from Institute for Healthcare Improvement (2024) and Donabedian (1966).
A note on structural measures: these are optional to track formally, but many QI collaboratives use them as a starting checklist before a project begins. Confirm that the basic building blocks (staffing, protocols, technology) are in place before layering on process and outcome measures.
A PMHCA team is supporting an ED partner site in improving trauma screening for pediatric behavioral health patients. Rather than tracking a single number, the team defines one measure of each type so they can see the full picture:
- Structural: Is a trauma screening tool built into the ED's EMR, with a designated staff role responsible for administering it? (Yes/No — confirmed before the project launched)
- Process: Percentage of eligible pediatric ED patients who receive a completed trauma screening during their visit, tracked weekly.
- Outcome: Percentage of patients who screen positive and are connected to a behavioral health referral or consult within 48 hours.
- Balancing: Average time added to ED throughput for patients who complete a screening, to confirm the new step isn't creating unintended delays elsewhere in the visit.
Result: Together, these four measures tell the team not just whether screening rates are going up, but whether the system was ready to support the change, whether the change is producing the outcome they care about, and whether it's creating new problems along the way.
Once you know which type of measure you're building, the next step is defining it precisely enough that anyone on your team would count it the same way. For rate-based measures (most process and outcome measures), this means deciding on two things:
- Numerator: the specific group or event you're counting — the "top number."
- Denominator: the total population the numerator is measured against — the "bottom number," or everyone who was eligible.
Being specific here matters. A vague measure like "screening rate" can be counted in several ways, depending on who's doing the counting. A well-defined measure removes the guesswork:
Numerator: Number of pediatric ED patients ages 8–17 who completed a trauma screening during their visit.
Denominator: Number of pediatric ED patients ages 8–17 who presented to the ED during the same time period and met eligibility criteria for screening.
Measure: Numerator ÷ Denominator = Weekly trauma screening completion rate.
Result: With the numerator and denominator both clearly defined, every staff member counting the data — this week or six months from now — will count it the same way, making the measure trustworthy over time.
- QI measures generally fall into four types: structural, process, outcome, and balancing.
- Structural measures confirm the systems and infrastructure are in place; process measures track whether the steps are happening; outcome measures track the result you care about; balancing measures watch for unintended consequences.
- Most projects lean on process and outcome measures, adding structural and balancing measures where they add useful context.
- A well-defined measure has a clear numerator and denominator that anyone on the team would count the same way.