Currently Viewing:
Currently Reading
The Effect of EHR-Integrated Patient-Reported Outcomes on Satisfaction With Chronic Pain Care
Christopher A. Harle, PhD; Nicole M. Marlow, PhD; Siegfried O.F. Schmidt, MD, PhD; Jonathan J. Shuster, PhD; Alyson Listhaus, MPH; Roger B. Fillingim, PhD; and Robert W. Hurley, MD, PhD

The Effect of EHR-Integrated Patient-Reported Outcomes on Satisfaction With Chronic Pain Care

Christopher A. Harle, PhD; Nicole M. Marlow, PhD; Siegfried O.F. Schmidt, MD, PhD; Jonathan J. Shuster, PhD; Alyson Listhaus, MPH; Roger B. Fillingim, PhD; and Robert W. Hurley, MD, PhD
Integrating patient-reported outcomes in an electronic health record had no effect on patient or primary care provider satisfaction with chronic noncancer pain care.

Objectives: Given its complexity, chronic noncancer pain presents an opportunity to use health information technology (IT) to improve care experiences. The objective of this study was to assess whether integrating patient-reported outcomes (PROs) data in an electronic health record (EHR) affects provider and patient satisfaction with chronic noncancer pain care.

Study Design: We conducted a pragmatic cluster randomized trial involving 4 family medicine clinics.

Methods: We enrolled primary care providers (PCPs) and their patients with chronic noncancer pain. In the first 7 months (education phase), PCPs in intervention practices received education on how to use PROs for pain care. In the second 7 months (PRO phase), patients in intervention practices reported pain-related outcomes on arrival at their visits. PROs were immediately reported to PCPs through the EHR. Control group PCPs provided usual care. We compared intervention and control practices in terms of provider and patient satisfaction with care.

Results: During the education phase, patients’ mean ratings of their visits did not differ between control and intervention (9.33 vs 9.08; P = .20). During the PRO phase, patients’ mean ratings did not differ between control and intervention (9.28 vs 9.01; P = .20). Similarly, there were no differences between the intervention and control groups in terms of provider satisfaction.

Conclusions: Delivering EHR-integrated PROs did not consistently improve patient or provider satisfaction. Positively, we found no evidence that the PRO tools negatively affected satisfaction. Future studies and technological innovations are needed to translate point-of-care health IT tools into improvements in patient and provider experiences.

Am J Manag Care. 2016;22(12):e403-e408
Take-Away Points

Integrating patient-reported outcomes (PROs) in an electronic health record (EHR) had no effect on patient or primary care provider satisfaction with chronic noncancer pain care. 
  • Integrating clinically relevant information in EHRs may be insufficient to meaningfully improve patient and provider satisfaction with managing chronic noncancer pain. 
  • This study adds to the literature on the value, or potential lack thereof, of current information technology–based clinical decision support tools. 
  • Before acquiring and widely implementing them, administrators should evaluate the ease of using and usefulness of electronic PRO data and related clinical decision support tools.
More than a decade ago, the US federal government began significantly investing in health information technology (IT). Today, 95% of hospitals and 56% of office-based providers use certified electronic health records (EHRs).1,2 Thus, the US healthcare system has more infrastructure than ever upon which health IT tools can be implemented to support better information access, information sharing, and decision making by healthcare providers and patients. Still, we lack sufficient research to understand if and when health IT tools affect patient and provider experiences.3-7 Without these studies, we lack the evidence needed to overcome the challenges that healthcare providers face in embracing health IT to meaningfully improve care for complex and costly patients.

Affecting more than 100 million Americans at a cost of more than $600 billion annually,8 chronic noncancer pain presents an opportunity to utilize health IT and other interventions to improve care. Chronic noncancer pain conditions, such as low back pain and diabetic neuropathy, have complex biopsychosocial roots, symptoms, and comorbidities.8 Moreover, chronic noncancer pain care is directly linked to the ongoing public health epidemic of prescription opioid pain medication use, misuse, and abuse that results in more than 350,000 emergency department visits9 and more than 14,000 deaths each year.10 Given the complexity of pain conditions and poor outcomes associated with chronic opioid therapy, many patients with chronic noncancer pain, and the providers who care for them, are dissatisfied with their respective experiences in receiving and providing care.11-15

Point-of-care health IT tools may improve pain care satisfaction by collecting and aggregating valuable health information. For example, clinical decision support tools could give providers easier and more reliable access to information on patients’ physical and psychological symptoms. In the case of symptoms, patients must report them directly, such as through discussions with clinicians, written questionnaires, or electronic questionnaires. Today, systems developers, researchers, and healthcare providers are engaged in learning how to best collect and integrate patient-reported outcomes (PROs) into care using health IT.16-22 Researchers have also called for studies that focus on the role of PROs in patient–provider relationships.23 On one hand, IT-integrated PROs allow providers to collect unique and clinically relevant information on patients’ symptoms and needs, which may improve care planning, patient–provider communication, and satisfaction. On the other hand, traditional clinical routines and norms in patient–provider communication may inhibit providers’ ability to process and respond to PRO information in ways that improve satisfaction with care.24 Concerns about poor health IT usability and usefulness7,25 could also inhibit IT-integrated PROs from positively affecting patient and provider satisfaction, especially if PRO information is collected but not discussed or acted on during visits. Still, several prior studies have found positive effects of PROs on patient–provider communication and patient satisfaction.26,27

In this study, we describe a pragmatic cluster randomized trial involving primary care providers (PCPs) and their patients with chronic noncancer pain. We examined whether providing PCPs with education on multidisciplinary pain care and then systematically collecting pain-related PROs in the EHR would affect PCP and patient satisfaction with care. Given prior research evidence on low satisfaction with pain care and expressed need within our health system for improvements in pain care satisfaction, we hypothesized that the education and EHR-integrated PROs would both increase provider and patient satisfaction with pain care. The results of this study will inform the literature on how health IT affects patient and provider experiences with care.


We conducted a pragmatic cluster randomized trial in 4 family medicine clinics affiliated with an academic health center (see eAppendix A [eAppendices available at] for a diagram of the trial flow).28-30 The 4 clinics were matched as 2 pairs based on practice and patient demographics and randomized to intervention or control groups. For PCPs in intervention clinics, we implemented pain-related education modules and an EHR-integrated PRO system. The EHR system (Epic) served as a conduit for collecting and communicating the PROs to providers at the point of care. We registered the study as a randomized controlled trial with (identifier: NCT02188667). The University of Florida Institutional Review Board approved the study.


We recruited PCPs (physicians and physician assistants) from the 4 clinics. We also recruited patients, aged 18 to 89 years, whose health record history indicated chronic musculoskeletal, headache, or neuropathic pain, and no cancer (see eAppendix B). Upon arriving for a visit with a participating provider, eligible patients received a written notice about the study and a follow-up phone survey about their visit satisfaction. Patients could refuse to participate in the survey by signing an opt-out form or declining to complete the phone survey when called.


The intervention involved 2 phases. First, we provided a 6-module education curriculum that was available to providers for 7 months. The education emphasized evidence-based, multimodal, and comprehensive pain care and was based on published evidence-based guidelines and systematic reviews.31-33 The education also included training on PRO instruments that providers could use to assess risks and benefits of different treatment modalities34 and patients’ levels of pain, function, sleep, and mental health.

Second, we implemented the EHR-integrated PRO system. We evaluated the effect of this system for 7 months. When patients arrived for a visit with a participating provider, a clinic staff member asked them to complete a series of PRO surveys using a tablet computer and the Collaborative Health Outcome Information Registry (CHOIR) software.35 We used CHOIR to collect 13 measures and send results to the EHR for providers to view during patient visits. Nine measures (pain interference, pain behavior, fatigue, anger, depression, sleep disturbance, sleep-related impairment, anxiety, and physical function) were assessed via computer-adaptive instruments from the National Institutes of Health Patient-Reported Outcomes Measurement Information System (PROMIS) system.36 We also measured pain catastrophizing,37 risk of opioid-related aberrant behavior,34 pain intensity,38 and location of pain.

Immediately following completion, CHOIR sent numeric PRO assessment results to the EHR, where they could be accessed by providers. Each PROMIS measure was recorded in the EHR’s results module both as a T score and a population percentile. In addition, the system provided narrative clinical interpretation for any scores with established interpretations. Providers could insert the numeric results and clinical interpretations in their notes and graph PRO results over time alongside medication start and stop dates.


The 2 primary outcomes were patient satisfaction and provider satisfaction with receiving and providing pain care, respectively (see eAppendices C and D for survey questions). We collected patient satisfaction after each study visit. First, we asked patients to rate their overall healthcare on a scale of 0 (worst visit possible) to 10 (best visit possible). Second, patients who reported discussing pain with their provider during the visit were asked 2 questions adapted from the Hospital Consumer Assessment of Healthcare Providers and Systems survey.39 The first item addresses patient perception that their provider did everything they could to help with pain. The second item addresses patient confidence that their pain will be well controlled in the future based on their visit experience.

We measured provider satisfaction monthly, starting just prior to implementing the education component of the intervention and then monthly for 14 months. This survey included a 14-item questionnaire assessing general experiences and satisfaction with providing care to patients with chronic pain. The questionnaire was modified from a visit-specific physician satisfaction questionnaire,40 such that the item wordings referred to general satisfaction with visits by patients who have chronic pain. The 14 items included 1 global satisfaction measure and multi-item subscales assessing the quality of the patient–provider relationship, appropriate use of time during visits, adequacy of data collection during visits, and patients’ cooperation during visits. Each item contained a 5-point agree/disagree response scale anchored by “strongly disagree” and “strongly agree.”

Sample Size

Our a priori sample size estimations were based on the 0-to-10 patient satisfaction measure, accounting for clustering within providers. We expected unbalanced enrollment based on PCP staffing levels at the clinics during study planning (11 control PCPs and 9 intervention PCPs) and a uniformly distributed random number of patients per PCP between 20 and 40 (606 patients per phase). The following simulated powers with 20,000 replications were obtained to detect a true mean difference in scores of 1.5 on a 0-to-10 point scale at 2-sided type I error of 5%, respectively:

(σBetween Providers [1.0, 1.5, 85%], σWithin Providers [1.0, 2.0, 83%], Power [1.0, 2.5, 81%]).


For patient satisfaction outcomes, we first computed a personal mean for each physician within phases and compared the intervention and control groups using a weighted least squares analysis, with weights proportional to the physician total sample size. The independent variable was the intervention group (ie, intervention vs control). Next, we assessed the differences between the education phase and the PRO phase for intervention versus control physicians. We used the paired difference in physician mean between the phases using weighted least squares, as above. The model for assessing the PRO phase and education phase difference in mean satisfaction for physician j (Yj) was: Yj = β0 + β1Tj + εj, where Tj is 0 (control) or 1 (intervention) and εj is random error for physician j. The slope β1 represents the mean treatment difference, asking if the treatment difference depends on the phase.

Copyright AJMC 2006-2019 Clinical Care Targeted Communications Group, LLC. All Rights Reserved.
Welcome the the new and improved, the premier managed market network. Tell us about yourself so that we can serve you better.
Sign Up