Diabetes is on the rise in California, as it is in many other parts of the U.S. As of 2003, nearly 1.7 million California adults age 18 and over have been diagnosed with diabetes, up from 1.5 million in 2001, according UCLA researchers. The risk for developing diabetes is significantly higher for minority populations and individuals with lower levels of educational attainment. Many of these patients have limited English proficiency or a poor understanding of health issues, making communication with their clinicians difficult and creating roadblocks to proper treatment. To promote better access to diabetes care and improve patient outcomes, Dean Schillinger, M.D., professor of medicine at the University of California, San Francisco, worked with his team at the UCSF Center for Vulnerable Populations to develop interventions to reach this population. Two forms of support—automated telephone diabetes management and group medical visits—were implemented with a group of patients who had poor control of type 2 diabetes. Dr. Schillinger talked to us about the results of these interventions and next steps.
How does limited English proficiency or low health literacy affect patients' ability to manage chronic conditions like diabetes or asthma?
Having limited literacy presents not only a barrier to written materials but, more importantly, a barrier to verbal communication. Individuals with limited literacy skills are more likely to report that their doctor—or nurse or health educator—uses words they don't know or speaks too fast, and that they just don't understand their treatment plan. Having an interpreter can be helpful, but communication can remain impaired even in the presence of the interpreter.
In our setting, we have a lot of Spanish speakers with less than an eighth-grade education—the double whammy of language and literacy. And diabetes itself is a very communication-sensitive disease. There have been a lot of self-management support studies and interventions tested over the years that show you can improve people's diabetes health care. The problem is that none have been designed for the more vulnerable population—the underserved, the uninsured, and those with limited literacy.
Are there other reasons that existing disease management strategies don't work with low-health-literacy populations?
Schillinger: Well, they're costly, and someone's got to pay for them. A typical health plan will contract with a company to implement a range of interventions, from home visits to telephone calls to Web-based tools. And for a significant proportion of vulnerable populations, there is no payer for that. In addition, many of the self-management support interventions require people to jump through hoops—drive to a hospital many miles away, find child care, et cetera. And then, lastly, I think most disease management interventions tend to be reactive rather than proactive—in other words, they wait for the patient to show up with a problem. Often, the patients with the greatest clinical and communication needs are not the ones coming in with lists of questions. They mostly seek care when things have really fallen apart.
In the intervention that you designed, you worked with two groups: patients who received automated telephone calls and follow-up phone calls from a nurse, and patients who were part of group visits. What were your expectations for each?
Schillinger: When I talked to colleagues about literacy and language and chronic disease support, I find that they're divided into two groups: the techie doctors who say, 'Give patients a beeper and have it beep when it's time to take their insulin.' The other half—the other extreme of the interpersonal support—say, 'We just need more time with patients to talk about diabetes and teach them how to care for it.' What was exciting was the ability to actually compare those two approaches from a public health standpoint. I went into it pretty open-minded.
So what proved to be successful?
Schillinger: The telephone arm was associated with a much higher proportion of people taking part in the intervention at least once and then doing it on an ongoing basis, compared with the group medical visits. But when we looked at literacy and language patterns, those with limited literacy and non-English speakers were more likely to use it on an ongoing basis and to make behavioral action plans—like walking 30 minutes a day, three times a week—when compared with those with those with high literacy and English speakers. With group medical visit models—which, in a certain way, reproduce the clinical encounter: the reach was lower, with a smaller proportion of people attending and continuing to attend.
In contrast, people with higher literacy and English speakers were more likely to use and engage in the group visit model, compared with lower literacy and non-English speakers. This was despite the fact that we did extensive training of facilitators on patient-centered education and health literacy sensitivity. There was just something about the group model that led to lower levels of engagement. The difference in reach was fivefold.
Were there any other surprising findings?
Schillinger: In the automated phone arm, I was surprised to find the degree to which we were picking up unanticipated problems in patient safety. When we followed up with people, we would find out, for example, that they thought their Metformin (a diabetes medicine) was Tylenol. Or that they had symptoms of a urinary tract infection and hadn't told anyone.
Were you able to see physicians' responses to the changes that they saw in patients?
Schillinger: Yes, and we are just completing this analysis. Briefly, patients exposed to either program were perceived by their physicians as more likely to create and achieve goals for chronic care. Physicians rated quality of care as higher among patients exposed to the automated phone model compared to usual care, and compared to group visits.
Are there any plans for broader implementation?
Schillinger: Our local health plan is going to take on the automated phone model, which is really exciting. Medicaid is increasingly being pulled into the managed care model and increasingly dealing with middle-aged, poor people with diabetes, heart failure, and other chronic conditions. Our health plan was looking for cost-effective, homegrown solutions to help their populations manage their own conditions. They recognize that most of their members have lower literacy skills and don't speak English, and so were excited that we had a tool that they could implement on their own, with only modest training of their existing staff. In my new capacity as chief of the Diabetes Prevention and Control Program, we are working on developing modules to tackle pre-diabetes.
What are some of the challenges involved with implementing this kind of project?
Schillinger: For vulnerable populations, the numbers of barriers to self-management are just exponentially greater than for non-vulnerable populations. If you're poor and you can't buy healthy food, it's hard to self-manage. If it's not safe in your neighborhood, it's hard to go out walking. All these things make the nature of self-management support so much more complex when compared with self-management support for a more affluent population, which might focus on issues like using insulin correctly.
There are also IT challenges. How do you integrate with the other IT programs within the health plan? How do you maximize the impact? There may be ways to find out if a certain patient picked up his or her insulin that week. That's an opportunity as well as a challenge. Another challenge is identifying and training the labor force—one that is not so costly and can get the job done. I do not think it has to be a nurse, and it certainly doesn't have to be a doctor. It could be done by medical assistants and health educators who can identify the 10 percent of the time that you need a nurse and the 1 percent of the time you need a doctor.