Adult Asthma Surveillance and Intervention in a Managed Care Setting

1. To determine gender differences, a questionnaire was mailed to a random sample of 800 members with asthma in Kaiser Permanente Northwest (KPNW) and 800 in KP Colorado (KPC) Regions, which are large, group-model health maintenance organizations. Results were combined with information from administrative databases on health care utilization..
2. The telephone intervention study was conducted within KPNW. 6,948 eligible HMO members were identified from the electronic medical record (EMR). Individuals 18 or older were eligible for the study if they were either (1) on the KPNW high risk asthma registry or (2) had at least 180 days of anti-asthma medication dispensings and at least one medical contact for asthma during a specified two-year period. Excluded were 850 individuals who had COPD listed in the electronic medical record.

Eligible individuals were randomly assigned to either usual care (n=3,367) or telephone outreach (n=3,581); 192 members of the telephone outreach group were randomly selected to receive calls from a real person, while the remaining 3,389 received automated calls.

The intervention consisted of three rounds of calling, spaced ~five months apart. Calls consisted of a brief series of questions to assess (1) recent emergency department or hospital care without a follow-up visit, (2) current level of asthma control, (3) current patterns of asthma medication use, and (4) whether the member could identify a primary provider for asthma care. (Sel-identified COPD patients received an abbreviated intervention but were kept on the call list unless a diagnosis of COPD was subsequently added in thier EMR.)

Based on the responses to these initial questions, members were offered tailored feedback regarding their overall level of asthma control and use of asthma medications. Feedback was designed to convey a positive message and not be prescriptive. The call concluded with an offer to receive information about additional Kaiser Permanente (KP) resources (a toll-free health line and KP's online website for members), followed by an offer to call again in five months.

The calls were brief and lasted less than 10 minutes. Participants could refuse the tailored feedback. To maximize participation, calls were made using speech recognition (or speech enabled) technology, rather than requiring the respondent to respond by pushing buttons.

Automated and live person calls used identical scripts. Responses to the questions were computerized and used to generate different text messages that were entered into the electronic medical record and logged as telephone encounters. Members identified as being at high risk of a future acute exacerbation were brought to the attention of the system, ensuring that a clinician would have to physically review the encounter and initiate a follow-up contact with the patient.

Consistent with a standard intention to treat analysis, all randomized individuals were included in the analysis, even if they opted not to participate in the calls.

Primary outcomes of the study related to health care utilization, medication use, and quality of life. A secondary outcome was the extent to which the intervention could be successfully implemented, and the reaction to the intervention by patients and providers. Extensive process data was collected, including a survey of providers whose patients were enrolled in the intervention, mailed surveys to members, feedback from members obtained immediately following the third intervention call, and an audit of those charts where the encounter was left open to determine what, if any, follow-up occurred. Participation at each round of calling was also tracked.

Follow-up data was obtained from a survey of 1,583 randomly selected KPNW members one month following the final calling period and from an analysis of health care utilization data from the start of the initial calling until one month following the last round of calling. The response rate for the follow-up survey was 65% for both intervention participants and control participants.