As part of a Quality Improvement initiative you are tasked with improving ED length of stay for patients presenting with ankle pain. You remember learning about the Ottawa Ankle Rules to help screen which patients require ankle x-rays.

How will you implement that knowledge into creating a clinical pathway?
Will you utilize quality improvement (QI) or traditional clinical epidemiology methodology?

In this short review, we will discuss differences between quality improvement (QI) methodology and traditional clinical epidemiology, and show how they can co-exist.

Quality Improvement versus Traditional Clinical Epidemiology

QI and traditional clinical epidemiology are both valuable tools to help improve patient care. Both are concerned with knowledge generation and have robust methodology to generate questions to then collect data, test solutions, and critically evaluate data to portray results.

However, there are also key differences. Traditional clinical epidemiology aims to provide generalizable knowledge, while QI assesses local improvements in process and outcomes. Whereas traditional clinical epidemiology attempts to eliminate context (e.g., aiming to have two groups in a trial be similar besides the measured variable), QI embraces context when making improvement changes (e.g., variations in ED wait-times/volumes based on day/time). Finally, traditional clinical epidemiology often requires prolonged data collection for definitive results. In contrast, QI methodology uses rapid Plan-Do-Study-Act (PDSA) cycles to formulate rapid tests of change. Investigators also use a family of measures, which is discussed below.

Quality Improvement

Family of Measures

In QI projects, we acknowledge there will be variation in a real-life uncontrolled environment. Unlike in traditional clinical epidemiology, where variability is undesirable, a family of measures in QI lets you better picture what is happening in real-time. We utilize outcome, process, and balancing measures which are defined as follows:

Outcome measures: What you are trying to improve

Process measures: Intermediate steps that you measure to get to that outcome measure

Balancing measures: Negative impact/consequences of your intervention

Now, let’s return to our example scenario looking at patients presenting with ankle pain. If we were to implement triage nurse-initiated radiography based on Ottawa Ankle Rules, the measures could be the following:

Outcome measure: ED length of stay for ankle sprain

Process measure: Time from ED triage to x-ray

Balancing measure: Number of patients sent back for repeat x-rays (rationale: are patients getting inappropriate/inaccurate radiographs and requiring repeat x-rays?)


Measuring Change

In traditional clinical epidemiology, you measure pre and post data, compare one group to another and look for significant change (e.g., p < 0.05). In QI, we utilize run charts/statistical process control charts. The purpose is to:

  • Make process performance visible
  • Determine if a change is an improvement
  • Ensure we are sustaining improvement

Afterwards, you utilize various rules to analyze whether a change is due to “common cause variation” or “special cause variation.”

Common cause variation: What you expect to happen in a usual system functioning at baseline. There will be some day-to-day variability.

Special cause variation: When something occurs that is more than common cause, resulting in a change in the system.

When you see special cause variation, this indicates that something out of the ordinary has happened, resulting in a change to the system. However, this does not show you what has occurred. Fortunately, as the QI expert, you can go back and analyze the events leading to that change.


Putting it all together

Lee et al. saw that triage nurse-initiated radiography using the Ottawa ankle rules resulted in a decrease in ED length of stay (LOS) (median = 20 minutes, mean = 28 minutes)(1). This randomized control trial looked at 146 patients over 11 months and utilized traditional clinical epidemiology. Harries et al. performed a comparative evaluation utilizing QI methodology(2). They took the outcome of interest (ED LOS) and first created a baseline. Afterwards, they used run charts and were able to see special cause variation after 6 months. Therefore, the authors concluded that they could identify sustained improvement quicker by utilizing QI methodology.

Ultimately, this study highlights that in situations where the aim is to accelerate the implementation of practice improvement, there should be increased consideration of using QI analytic methodology.

Quality Improvement

Bottom Line

Both QI and traditional clinical epidemiology are important for providing and advancing patient care. Importantly, they are not mutually exclusive and can be used to supplement each other. When your research question is geared towards implementing a protocol/practice, QI methodology should be used, especially when it is supplementing knowledge generated from traditional clinical epidemiology.



  1. Lee WW, Filiatrault L, Abu-Laban RB, Rashidi A, Yau L, Liu N. Effect of triage nurse initiated radiography using the Ottawa ankle rules on emergency department length of stay at a tertiary centre. Can J Emerg Med. 2016;18(2):90–7.
  2. Harries B, Filiatrault L, Abu-Laban RB. Application of quality improvement analytic methodology in emergency medicine research: A comparative evaluation. Can J Emerg Med. 2019;21(2):253–60.


  • Dr. Mok is an emergency medicine at the University of Ottawa, with special interests in evidence-based medicine and resuscitation.

  • Dr. Samara Adler is a junior editor for the EMOttawa Blog, and is an FRCPC resident in the Department of Emergency Medicine at the University of Ottawa.

  • Dr. Josée Malette is an Emergency Medicine Resident in the Department of Emergency Medicine, University of Ottawa. She is a Junior Editor with the Digital Scholarship and Knowledge Dissemination team for the EMOttawaBlog.