Use of PROs in Primary Care: PROMIS or Disappointment?
Elizabeth R. Pfoh, PhD, MPH and Michael B. Rothberg, MD, MPH
Center for Value-Based Care Research, Cleveland Clinic, Cleveland, OH, USA.
J Gen Intern Med
© Society of General Internal Medicine 2018
n 2004, the National Institutes of Health (NIH) launched its
Roadmap as a way to transform healthcare research by
making it more efficient.
PROMIS (the Patient-Reported
Outcomes Measurement System) was developed as part of
the Roadmap to fill the need for an evidenced-based, state-
of-the-art assessment system for measuring self-reported
PROMIS standardizes collection of patient-reported
outcomes (PROs) (such as depression and anxiety) across
research studies and enables efficient assessment of various
aspects of self-reported health. Further, PROMIS provides a
way to capture the patient’s full experience of their disease and
In the clinical setting, physicians and researchers
can assess symptoms, inform treatment decisions, and track
outcomes using PROs.
SPADE symptoms (sleep problems, pain, anxiety, depres-
sion, and low energy/fatigue) are prevalent and can be mea-
sured using PROMIS. They are the types of outcomes the NIH
was hoping to improve. In this month’s Journal of General
Internal Medicine, Kroenke and colleagues’ report results of a
randomized trial in which they offered primary care clinicians
a visual display of patients’ SPADE symptoms, then measured
the change in 3-month composite SPADE scores.
no difference in SPADE symptom improvement or clinician
action between patients whose clinicians saw the visual dis-
plays and the control group. Documented clinician action
included medication prescriptions (66% of study participants)
followed by other treatments such as education (36%) or
referrals (28%). Patient-reported clinician action was not as-
sociated with patient characteristics nor satisfaction. Impor-
tantly, two thirds of threshold-level symptoms were unre-
solved at 3-month follow-up.
While disappointing, the findings are unsurprising. The
study’s randomized clinical trial design provides strong evi-
dence that measuring symptoms alone does not change prac-
tice. Displaying patients’ symptom scores did not improve
documentation, symptom-specific clinician action, or patient
satisfaction. There are three potential explanations: (1) physi-
cians overlooked the displays; (2) clinicians prioritized other
diseases over SPADE symptoms; or (3) SPADE symptoms are
difficult to treat. While interviewing the clinicians could an-
swer these questions definitively, some insight can be gleaned
from the clinicians’ actions and interviews with the patients.
The first explanation seems unlikely because according to the
patients, the vast majority of threshold symptoms were
Patients also reported that clinicians often priori-
tized other diseases. In fact, the most common reason for not
discussing the SPADE symptom was a more pressing medical
issue (49%). Finally, many patients reported not needing
(47%) or wanting (29%) treatment. Yet, treatment was desired
the majority of the time when symptoms were threshold level
for pain, sleep, and fatigue.
Perhaps the most likely reason that presenting the informa-
tion did not result in symptom improvement is that SPADE
symptoms are difficult to treat. Until recently, opioids were the
most common treatment for pain. Opioids are addictive and
increasingly regulated. There are medications for sleep, anxi-
ety, and depression, too, but their effectiveness is low and side
effects are common. Cognitive behavioral therapy (CBT) is
effective but patients may face barriers to accessing an expert
in CBT, including insurance, time, or availability. As the
authors acknowledge, the lack of systematic supports to ad-
dress patients’ symptoms likely blunted the impact of the
Prior randomized trials of depression
screening found screening improved morbidity and mortality
only when substantial supports are in place.
are clearly needed.
If we cannot readily improve SPADE symptoms, does it still
make sense to measure them? Historically, patients have
shared their medical history and symptoms for their own
benefit, for the purposes of diagnosis and treatment. Routine
measurement can improve outcomes for patients with chronic
disease (e.g., monitoring blood pressure or arthritis symp-
toms), but this assumes we have effective treatments available.
Monitoring SPADE symptoms may be more akin to monitor-
ing body mass index, an often futile and frustrating activity for
physician and patient alike. Yet, to improve overall healthcare
quality, learning health systems need data, especially the kind
provided by patients completing PROMIS measures. Learning
health systems can identify ways to improve care delivery by
leveraging these data using real-time analysis.
can link PROs and electronic health record data to understand
the effectiveness of the treatments they provide. They can
study positive deviants—physicians whose patients improve
more than the average—to spread their practices. A defined set
of measures for SPADE symptoms, such as one used in this