Pilot Project: Creating a Decision-Support Computer Program to Help Cancer Patients Understand and Manage Their Symptoms

PCORI funded the Pilot Projects to explore how to conduct and use patient-centered outcomes research in ways that can better serve patients and the healthcare community. Learn more.

Background

Patients who are being treated for cancer may have symptoms that affect their quality of life. Many of the clinics that treat cancer do not have a way to help patients decide how to manage treatment-related symptoms such as pain, nausea and vomiting, or constipation. Often, patients and their caregivers are not sure when to contact their doctors about symptoms or what to tell them.

Project Purpose

This study developed and tested a way to help patients with cancer manage and communicate about the symptoms from their treatment. The researchers wanted to know if they could create a decision-support computer program that gives patients suggestions about

• How to manage their symptoms
• When to talk to their doctors about their symptoms
• What to say when they call their doctors

The suggestions provided by the program would be based on symptom and quality-of-life information that the patient provided through a computer survey.

Methods

The project had two parts.

In the first part, the researchers talked with 64 cancer patients and caregivers and 51 clinicians (doctors, physician assistants, nurse practitioners, and nurses) in groups and individual interviews. All of the patients, caregivers, doctors, physician assistants, nurse practitioners, nurses, and experts were from the Dana-Farber Cancer Institute network.

The researchers used group and individual interviews to learn how patients managed information about symptoms during cancer treatment. The researchers also asked the groups about what kind of information would be useful when making decisions, and in what form they wanted to receive the information.

The researchers shared this information with a group of experts, including patients who had survived cancer and their caregivers. This group helped to decide what would be included in the computer program.

In the second part of the project, researchers worked with health communication experts at the Dana-Farber Harvard Cancer Center to create the decision-support computer program. The program asked patients for information about their pain, nausea, vomiting, and constipation. Using the information patients provided, the computer program suggested ways to manage their symptoms, when to call their doctor, and what to tell the doctor.

The researchers tested the decision-support computer program with patients and caregivers to see if it was easy to understand and useful. The researchers asked whether the language in the program was clear and made sense. They then made improvements based on the feedback they received. The researchers also asked patients, caregivers, doctors, physician assistants, nurse practitioners, and nurses to fill out a survey so that they could collect feedback about the program.

Findings

After talking to the participants who used the decision-support computer program, the researchers found five things that patients and caregivers thought the program should do. Participants felt it should

• Make sure patients stay safe
• Make health information easy to understand
• Help patients talk with their doctor
• Help patients feel in charge of their health
• Help patients know whom to get help from

The answers to the survey questions showed that patients, caregivers, and doctors thought the decision-support computer program would be useful.

Limitations

Everyone in the study was from the same health system. All patients were adults getting outpatient cancer treatment in different clinics within that system. The results might be different in other health systems, for children, or for patients who are getting treatment for cancer during a hospital stay.

Conclusion

Patients and caregivers said that they needed help knowing when to call a doctor and what to say about symptoms. The researchers made a decision-support computer program to meet those needs. Patients, caregivers, and clinicians found the program to be useful.

Sharing the Results

The researchers presented this work at national meetings and prepared manuscripts for professional journal articles.

PCORI funded the Pilot Projects to explore how to conduct and use patient-centered outcomes research in ways that can better serve patients and the healthcare community. Learn more.

Project Purpose

The use of decision support for symptom and quality of life (SQL) management is an innovative way to enhance patient engagement, facilitate patient–provider communication, and improve outcomes. Most clinical settings have not established efficient methods to collect SQL data and integrate decision support systems into the flow of care. Although multiple studies have assessed the usability and/or efficacy of decision support systems, little research has examined patient and caregiver perspectives about preferences for decision support program components.

Research Objectives

The overall purpose of this study was to design and pilot test a patient-centered decision support program to process and manage SQL information in the cancer care setting. Phase I objectives were to (1) describe patient preferences for providing, processing, and managing SQL data that can enhance communication during the clinical encounter and (2) identify preferences for the format, specific information, and components for decision support that would be most useful to patients and their clinicians.

Phase II objectives of this study were to (1) develop three computable algorithms for patient self-assessment and management of pain, nausea and vomiting, and constipation; (2) develop a simulated model of a symptom assessment and management intervention for self-care (SAMI-SC); and (3) evaluate the usability and acceptability of a simulated model of SAMI-SC with adults recently treated for cancer, their caregivers, and clinicians.

Study Design

Focus groups elicited information from patients, caregivers, and clinicians about their experiences with managing SQL data during cancer treatment and their preferences for desired components for decision support. An expert panel (cancer survivor, family caregiver, clinician, administrator, quality improvement specialist, and health equity officer) assisted in interpreting data, identifying potential interventions, and co-designing the intervention. Using a mixed methods sequential exploratory design, researchers built and iteratively tested usability and acceptability. Researchers evaluated the simulated SAMI-SC using focus groups, individual interviews, and questionnaires with patients, caregivers, and clinicians.

Participants, Interventions, Settings, and Outcomes

Sixty-four patient and caregiver participants (with a median age of 60, 61 percent female, 64 percent white, and 67 percent with college education) participated in Phase I focus groups. Fifty-one clinician participants (with a median age of 42 years, 76 percent female, 88 percent white, and an average of nine years of experience in cancer care) participated in Phase I interviews. Patients included individuals >18 years of age, English or Spanish speaking, treated for cancer within the previous six months. Enrolled patients were invited to identify caregivers (>18; English or Spanish speaking) to participate in the focus group. All participants were from the Boston-based Dana-Farber Cancer Institute (DFCI) network. Clinicians included medical doctors, physician assistants, nurse practitioners, or registered nurses in ambulatory settings within DFCI.

Researchers developed a simulated model of an algorithm-based decision support program and tested it in collaboration with the health communication core at Dana-Farber Harvard Cancer Center for self-management of pain, nausea and vomiting, and constipation. After patients responded to the questions in the program, a report provided specific suggestions for self-management of the symptom, guidance about when to call the patients’ clinicians, and a script for what to tell the clinicians.

In Phase I, researchers used a topic guide to prompt participants to discuss their experiences collecting, processing, and managing SQL during treatment for cancer and to identify preferred formats for decision support programs. In Phase II, researchers used a topic guide to conduct cognitive interviews with patients and caregivers to ensure that the wording of questions was understandable. A Likert-type response was presented to elicit feedback from patient, caregiver, and clinician participants. The a priori target for acceptability of SAMI-SC was a mean score of four or greater on each item.

Data Analysis

In Phase I, transcripts were transcribed, de-identified and entered into NVivo. Three research team members conducted inductive content analysis of transcripts. In Phase II, patient, caregiver, and clinician demographic data were summarized using descriptive statistics. Acceptability E-scale results were summarized as means and standard deviations. Three study team members coded and grouped the data into themes from the audio-recorded session content. The research team combined themes (reconciled through discussion); created a list of suggested revisions to improve SAMI-SC algorithms; created the simulated model; and then ranked the list of suggested revisions to the simulated model, which were based on audio-recorded sessions and notes taken during those sessions. Researchers implemented critical revisions immediately and monitored less critical revisions for repetition, repeating this process until saturation (no new themes or changes).

Findings

Cognitive testing of all three algorithms showed that questions were easy to read and understand and identified barriers to use of decision support from patient and caregiver, clinician, or key stakeholders. Based on the results of the qualitative data, researchers identified five design objectives—ensure patient safety, communicate clinical concepts effectively, promote communication with clinicians, support patient activation, and facilitate navigation—that were relevant to the development of the algorithm-based CDS program. Patient and clinician acceptability surveys indicated strong support for the CDS program.

Limitations

Data were from one health system but across affiliated centers (a comprehensive cancer center, a community-based center, and a federally qualified health center). Also, the sample consisted of adults with cancer and their caregivers who received treatment in an outpatient setting and findings cannot be generalized beyond this setting.

Conclusions

Patients and caregivers identified that an element of decision support missing in current systems was a mechanism to let them know when they should contact their clinicians about distressing symptoms and what to tell their clinicians. Researchers developed a program that addressed this need and that was acceptable to patients, caregivers, and clinicians.