3 minute read

Concentration is a scarce commodity in healthcare delivery. Although our systems are flush with data, we can only process a finite amount of information to make clinical decisions for engaging, treating, and monitoring patients in our catchment areas. We screen and triage patients as stewards of scarce resources using measures that we may not know a lot about or have validated on our own. We trust that someone else has done the testing for us.

In many industries, quality engineers use a process called Measurement System Analysis to determine whether a measurement is capable of providing information that is reliable enough to base decisions on its output. The assumption is that an unreliable measurement system is simply guessing and not reliable. If the data is poor quality, then problems can develop downstream.   

The importance of measurement system analysis is not lost on our colleagues in pharmaceuticals, medical equipment manufacturing, and clinic laboratories. They predominantly use objective measures and have access to vast resources to refine and validate their measures. In collaborative care, we use self-reported data out of necessity and also convenience. One example, the Patient Health Questionnaire (PHQ), is ubiquitous in our field, often used in case identification, diagnostic support, clinical decision making, treatment outcome monitoring, and more.

All this to say that the following article from Carlo et al in Medical Care recently caught the attention of the CFHA community. Recognizing the need for more measurement validation, the authors analyzed PHQ and patient-centered outcome (PCO) data to determine an association between depression response and remission metrics with concomitant clinical improvement in PCO outcomes (all self-reported) like general health, quality of life, and disability. The authors wanted to know if changes in PHQ scores were related to changes in other patient variables.

In short, the authors found the likelihood of PCO improvement was most improved by absolute PHQ-9 score decreases of 7–9 and 14–16 points. These findings suggest that clinicians can use a 7-9 point deduction as a standard or benchmark, knowing that there is evidence that such a reduction is associated with improvement in other areas (PCO outcomes).

The results of this study are certainly important for managed care systems that use the PHQ for treatment response monitoring. Other systems can decide if they will change their measurement practices. The authors of the article are connected with the AIMS center and Collaborative Care model which relies on careful screening and outcome monitoring.

As you consider how you may maintain or change your use of the PHQ, here are a few parting thoughts. First, in healthcare, like other industries, not everything that can be counted counts, and not everything that counts can be counted. Most of us rely on self-reported measures because there is no lab test for depression or poor quality of life. Only the patient can tell us how they are feeling today or how their life is going. The PHQ is a measure of depression symptoms. Not every patient we see has depression.

Second, there is a difference between case identification and treatment monitoring. Some measures were designed and tested for both activities, while others were not. In my opinion, the PHQ-2 (pre-screening) is useful for identifying patients with depression symptoms; a quality of life measure is more appropriate for measuring treatment outcomes. Third, we work in a data-centric society. Data is the lifeblood of healthcare, the coin of the kingdom. If you are not already monitoring the performance of your integrated care service, there are many indicators and metrics you can choose. See this recent article for additional ideas.