Using Health Information Technology to Support Behavior Change

Patient-centered care promises to deliver effective care consistent with a patient’s preferences and reduce costs. The Office of the National Coordinator for Health Information Technology has suggested that health information technology (health IT) can facilitate shared decision making. The Agency for Healthcare Research and Quality issued a program announcement requesting proposals focused on using health IT to improve health care quality and outcomes with one focus on using clinical data to improve care delivery and support shared decision making. These announcements suggest these questions:

  1. What decisions should we be helping patients make?
  2. What information can health IT provide to help patients make these decisions?
  3. How might we engage other members of the health care team in the decision making process?

I’ll focus this post on those clinical scenarios where the diagnosis is clear, but the best treatment option is not.

What decisions should we be helping patients make?

Current health care spending estimates may help us determine what health conditions might benefit from shared decision making. Dieleman et al. found that three conditions (diabetes [$101.4B], ischemic heart disease [$88.1B] and low back/neck pain [$87.6B]) accounted for nearly 13.2% of Between 1996 and 2013, the four conditions with the highest increases in health care spending were diabeteslow back/neck painhypertension and hyperlipidemia.

The National Institute of Diabetes and Digestive and Kidney Diseases believes diabetes can be prevented with weight loss, physical activity and healthy eating. The Mayo Clinic suggests back pain can be avoided through exercise, core muscle strengthening exercises and maintaining a healthy weight. The Centers for Disease Control state hypertension can be prevented by adopting a healthy lifestyle, including eating a healthy diet, maintaining a healthy weight, getting enough physical activitynot smoking and limiting alcohol intake. The American Heart Association describes four health behaviors to reduce the risk of developing hyperlipidemia: eating a heart healthy dietregular exercise, avoiding tobacco smoke and losing weight if you are overweight or obese.

The Global Burden of Disease 2015 Mortality and Causes of Death Collaborators (funded by the Bill and Melinda Gates Foundation) identified smoking, body-mass index, fasting plasma glucose, blood pressure and total cholesterol as the five risk factors with the largest contributors to reductions in disability-adjusted life-years for both sexes in America in 2015 (drug use and alcohol use were numbers five and six, respectively).

Two Behavioral Change Theories to Design Health IT to Help Patients Make Better Decisions

If the American health care system is being overwhelmed by non-communicable disease with strong behavioral components (smoking cessation, weight loss with dietary changes, physical activity and medication adherence), then a focus on empowering patients to change their behavior might help us create relevant health IT interventions to address these challenges.

Ralf Schwarzer developed the Health Action Process Approach model (HAPA) to include motivational and volitional elements to predict health behavior change. The motivational components refer people considering a behavior change, but has not taken any steps to initiate that behavior change. Drivers include risk perception, outcome expectancies and self-efficacy. The volitional elements focus on people initiating a behavior change. Drivers include different types of self-efficacy, barriers and resources to initiating (or maintaining) a behavior change.

BJ Fogg developed his own behavior model that suggests behavior is the product of motivation, ability and triggers that occur within a specific moment. Fogg stratifies behavior change into fifteen categories based on duration of behavior change (one-time, finite period, indefinite), stopping or starting a behavior, increasing or decreasing a behavior’s intensity or duration and if the behavior is familiar or unfamiliar. He believes that long-term behavior change can occur as the result of

  1. An epiphany (e.g., a family member dies unexpectedly),
  2. Changing the environment (e.g., removing unhealthy foods from the home), or
  3. Taking baby steps (e.g., walking 10 minutes a day with increases each month to a goal of 45 minutes a day)

Fogg makes two claims that might challenge HAPA:

  1. When influencing people, simplicity is more important than motivation.
  2. To change behavior, the most powerful element is hot triggers, not information.

What information can health IT provide to help patients make these decisions?

Published literature

Ackley et al. proposed seven levels of evidence to help determine the strength of any findings. The highest level of evidence is based on meta-analyses of randomized, controlled trials and the lowest level of evidence is based on consensus recommendations of clinical experts. This information could inform risk perception and outcome expectancy (HAPA) or motivation and ability (Fogg).

Electronic health records

Electronic health records could provide information about disease prevelance and incidence, as well as what types of interventions were suggested (behavioral, pharmaceutical, other). For conditions where meta-analyses or randomized, controlled trials are not available, data from electronic health records could provide insights into local disease and intervention patterns to help patients make “educated guesses” about what they might expect if trying specific interventions. Those health systems with larger patient populations could potentially produce more robust estimates of specific intervention frequency and effectiveness.

Patient-generated health data

As patients’ comfort with technology in all domains of their lives increases, the likelihood that they will provide health information outside traditional healthcare encounters also increases. In addition to surveys about overall health, they could be asked about medication adherence, quality of life and their ability to complete activities of daily living. This information could:

  1. Provide the patient and health care team information about earlier health states, self-efficacy, barriers and resources around other health behavior changes, AND
  2. Supply information to help patients gauge the likelihood of success when considering a new health behavior change (e.g., for patients with your gender, age group and constellation of medical conditions, about 75% lowered their cholesterol 25 points after starting a new statin prescription. Only 10% of those patients reported any new-onset fatigue after starting the medication).

For those health IT interventions that include a social component, patients could learn new skills from similar patients through “reviews” of specific healthcare interventions or interacting with other users through a discussion forum.

Combining data sources

In some cases, patients may feel overwhelmed by the possible behavior changes to improve their health. A “mash-up” of the expected benefit of each suggested behavioral intervention along with the patient’s self-efficacy to initiate and maintain that behavior change over time could provide patients with the appropriate information about what behavior change to pursue.

How might we engage other members of the health care team in the decision making process?

This paradigm of health IT places the patient at the center of the health care interactions. Physicians and other health care team members could benefit from understanding:

  1. The strength of the patient’s risk perception (HAPA) or motivation (Fogg)
  2. The validity of the outcome expectancy (HAPA) information presented to patients
  3. The patient’s perceived self-efficacy (HAPA) or ability (Fogg) to make a behavior change
  4. For those patients engaged in changing a behavior or maintaining that behavior change, understanding the barriers and resources available (HAPA)
  5. Triggers (Fogg)

Health IT that can elecit these data elements from patients and present them to the healthcare team in a workflow-friendly way may be more likely to support meaningful conversations between patients and healthcare team members about changing health behaviors in those ways that most benefit patients.


Health IT could help patients initiate and maintain new health behaviors using published literature, electronic health record information and patient-generated health data. Data collection based on the Health Action Process Approach and BJ Fogg’s theories of behavior change could help identify those health behavior changes that are most likely to be adopted and sustained over time. Each cycle of behavior change could then increase a patient’s confidence to increase the intensity of an existing health behavior or adopt a new health behavior to further improve their health.