Gerald Amundson

Population-Based Incidence of Pertussis among Adolescents and Adults, Minnesota, 1995–1996

To estimate the incidence of pertussis, a prospective study was done among members of a managed care organization in Minneapolis/St. Paul, Minnesota. Of 212 patients 10-49 years old enrolled from January 1995 through December 1996, 8 were found to be culture positive, 10 were found to be positive by polymerase chain reaction assay, 13 had a > or =2-fold increase in IgG or IgA to pertussis toxin (PT), and 18 had IgG to PT in a single serum specimen > or =3 SD above the mean of an age-matched control group. At least 1 positive laboratory test result for pertussis infection was found in 27 (13%) patients, among whom the duration of cough illness was a median of 42 days (range, 27-66 days). On the basis of any positive laboratory result, the estimated annual incidence of pertussis was 507 cases per 100,000 person-years (95% confidence interval, 307-706 cases). Bordetella pertussis infection may be a more common cause of cough illness among adolescents and adults than was recognized previously.

Using Simulation Technology to Teach Diabetes Care Management Skills to Resident Physicians

Background: Simulation is widely used to teach medical procedures. Our goal was to develop and implement an innovative virtual model to teach resident physicians the cognitive skills of type 1 and type 2 diabetes management. Methods: A diabetes educational activity was developed consisting of (a) a curriculum using 18 explicit virtual cases, (b) a web-based interactive interface, (c) a simulation model to calculate physiologic outcomes of resident actions, and (d) a library of programmed feedback to critique and guide resident actions between virtual encounters. Primary care residents in 10 U.S. residency programs received the educational activity. Satisfaction and changes in knowledge and confidence in managing diabetes were analyzed with mixed quantitative and qualitative methods. Results: Pre- and post-education surveys were completed by 92/142 (65%) of residents. Likert scale (five-point) responses were favorably higher than neutral for general satisfaction (94%), recommending to colleagues (91%), training adequacy (91%), and navigation ease (92%). Finding time to complete cases was difficult for 50% of residents. Mean ratings of knowledge (on a five-point scale) posteducational activity improved by +0.5 (p < .01) for use of all available drug classes, +0.9 (p < .01) for how to start and adjust insulin, +0.8 (p < .01) for interpreting blood glucose values, +0.8 (p < .01) for individualizing treatment goals, and +0.7 (p < .01) for confidence in managing diabetes patients. Conclusions: A virtual diabetes educational activity to teach cognitive skills to manage diabetes to primary care residents was successfully developed, implemented, and well liked. It significantly improved self-assessed knowledge and confidence in diabetes management.

CC1-01: A Simulated Diabetes Learning Intervention Improves Provider Knowledge and Confidence in Managing Diabetes

Background/Aims Provider performance on diabetes measures is variable, and is at least partially due to differences in provider knowledge and confidence in managing patients with diabetes and multiple co-morbidities. Objective To evaluate whether a simulated learning program can improve provider knowledge and self-confidence in diabetes management. Methods 19 primary care residency programs and 341 consented residents were randomized to (a) intervention (177 residents) or (b) control (164 residents) conditions. Intervention subjects were assigned 18 learning cases using SimCare Diabetes, a web-based immersive simulation program that challenges providers to achieve blood sugar, blood pressure, and lipid goals in 6 months of simulated time, as well as to address other issues such as severe obesity and insulin resistance, hypoglycemia, depression, obstructive sleep apnea, and non-adherence. Between simulated encounters with a patient, providers receive personalized feedback on progress to goals and treatment actions taken or omitted. 92 intervention and 128 control subjects completed a post-intervention follow-up online survey with 10 multiple choice knowledge and 5 self-confidence assessment questions using a 5-point Likert scale (1=not at all confident, 5=very confident). Mean (95% CI) knowledge test and self- confidence measures, adjusting for residency program clustering, were compared by group. Results On knowledge testing, 46% of the intervention group answered more than half the answers correctly compared to 16% of the control group. The mean knowledge score (95% CI) was 5.31 (4.87–5.75) for intervention and 4.1 (3.69–4.50) for control subjects (p<.001). Self-confidence measures were higher for intervention compared to control subjects for: use of all available drug classes to manage diabetes (3.64 vs. 3.09, p<.001), insulin use (4.12 vs. 3.36, p<.001), interpretation of blood sugars (4.21 vs. 3.58, p< .001), setting individualized treatment goals (4.06 vs. 3.42, p< .001), and overall confidence in managing diabetes (3.97 vs. 3.28, p< .001). Discussion An immersive online simulated diabetes learning program was effective at improving knowledge and self-confidence for diabetes management in primary care residents. Support and partnerships to maintain and spread the technology are desirable.

Clinical decision support directed to primary care patients and providers reduces cardiovascular risk: a randomized trial

Objective To test the hypothesis that use of a clinical decision support (CDS) system in a primary care setting can reduce cardiovascular (CV) risk in patients. Materials and Methods Twenty primary care clinics were randomly assigned to usual care (UC) or CDS. For CDS clinic patients identified algorithmically with high CV risk, rooming staff were prompted by the electronic health record (EHR) to print CDS that identified evidence-based treatment options for lipid, blood pressure, weight, tobacco, or aspirin management and prioritized them based on potential benefit to the patient. The intention-to-treat analysis included 7914 adults who met high CV risk criteria at an index clinic visit and had at least one post-index visit, accounted for clustering, and assessed impact on predicted annual rate of change in 10-year CV risk over a 14-month period. Results The CDS was printed at 75% of targeted visits, and providers reported 85% to 98% satisfaction with various aspects of the intervention. Predicted annual rate of change in absolute 10-year CV risk was significantly better in CDS clinics than in UC clinics (-0.59% vs. +1.66%, -2.24%; P < .001), with difference in 10-year CV risk at 12 months post-index favoring the CDS group (UC 24.4%, CDS 22.5%, P < .03). Discussion Deploying to both patients and providers within primary care visit workflow and limiting CDS display and print burden to two mouse clicks by rooming staff contributed to high CDS use rates and high provider satisfaction. Conclusion This EHR-integrated, web-based outpatient CDS system significantly improved 10-year CV risk trajectory in targeted adults.

PS1-47: Primary Care Residents Highly Rate Simulated Diabetes Training

Purpose Simulation training is prevalent in aviation and engineering industries, but acceptance by medical providers is unknown. Our objective was to design and evaluate resident physician satisfaction with simulated diabetes training. Methods This web-based learning program integrated these components: 18 unique diabetes learning cases, an interactive care management interface, a physiologic model to simulate outcomes of actions across a series of patient encounters, and a library of feedback messages to critique and guide provider actions. A total of 341 consented primary care residents in 19 U.S. residency programs were randomized to receive (n=177) or not receive (n=164) the learning intervention. A satisfaction survey evaluating program features was completed by 94 (53%) of intervention subjects. Responses to open-ended questions about features considered valuable and areas needing improvement were assessed using qualitative methods. Results Likert-scale responses were favorably higher than neutral for general satisfaction (93%), recommending to colleagues (91%), training adequacy (90%), navigation ease (95%), blood sugar displays (86%), drug info and help links (76%), goal progress graphs (49%), and feedback received (81%). Difficulty finding time to do cases was an issue for (51%) of responders. Open-ended responses (n=87) indicated that the most valuable learning pertained to insulin management (n=35), general management (n=23), and goal- achievement (n=10). Suggested improvements included software enhancements (n=34) and nothing bad to report (n=27). Discussion Learning through case simulations in a web-based dynamic environment is rated highly for satisfaction and ease of use by resident physicians. Most would recommend it to colleagues.

PS2-20: Racial Disparities in A1c Change and Medication Intensification.

Introduction: While it is well documented that disparities exist across racial groups of diabetes patients relative to glucose control, the underlying causative factors are not well understood. The purpose of this study was to examine differences in physician orders for adjustments of glucose control medications in diabetes patients between African Americans and Caucasians. Methods: A cohort of 434 African American and 1,471 Caucasian diabetes patients was followed for a period of 18 months from 6/1/2006 through 11/30/2007. All patients were being treated in one of 6 primary care clinics within a large mid-western medical group. First and last A1c values were collected during the study period and an A1c change score was calculated. Patients were also classified as to whether English was their primary language and whether they were on medical assistance at any time during the study period. The presence of a medication adjustment was tracked for metformin, thiazolidinediones, sulfonylureas (initiation or titration) and insulins (initiation only). Medication adjustment data was from the prescribing data in the medical record and thus represents physician actions. Results: There was a significant difference between African Americans and Caucasians on their initial A1c values (8.2% vs 7.3%; P<0.0001). Adjusted for the initial A1c, mean number of visits was similar between the two groups (16.9 vs. 15.8, P=.50). When change in A1c was calculated while controlling for initial A1c African Americans had less than half the decline found in Caucasians (.20% vs .45%; P=0.0009). English as a primary language (P=.81) and medical assistance (P=.81) status were not related to the disparity in change value. Examination of medication adjustment for African Americans relative to Caucasians found that they were less likely to have adjustments of metformin (OR .69; P=.0021) and TZDs (OR .65; P=.0122) but no differences were found for Sulfonylureas (P=.49) or insulins (P=.30). Conclusions: This study suggests that part of the racial disparity in glucose control is related to physician orders for medication intensification. This indicates a need for better understanding of the reasons for medication adjustment disparities and effective interventions to reduce them.

C-B4-02: Effect of Point-of-Care A1c Testing in Primary Care Clinics on Diabetes Medication Intensification

Background: A1c results are often not available until after the outpatient visit is completed. Despite the potential for rapid point- of- care (POC) A1c testing to improve the process of diabetes care, published results have not conclusively shown a link to improved diabetes care in primary care settings. Methods: All HealthPartners Medical Group primary care clinics use protocols for nurses to remind patients with diabetes to have A1c tested before upcoming medical appointments. In June 2007, one clinic began POC A1c testing for all diabetic patients who did not have an A1c in the previous six months or if the most recent A1c was more than 1 month ago and >7%. Using generalized linear mixed model regression, we compared diabetes medication intensification at encounters with diabetes patients in the pre testing period (PRE, 6/1/06 5/31/07, 22932 encounters) and post testing period (POST, 6/1/07 5/31/08, 27056 encounters) at the intervention clinic and five comparison clinics with no POC A1c testing capability. Results: The analysis included 3261 patients (mean age 57, 29% minority, median encounters 8/year) seen by 42 primary care physicians (PCP). The median A1c PRE was 7.2% at the intervention clinic, 6.9% at comparison clinics. At intervention clinic encounters, mean days since A1c testing fell from 72 to 44 (with 60% of POST encounters preceded by an A1c less than 1 month old), while there was no change PRE to POST at the comparison clinics. Medication was intensified at 16.3% of PRE encounters with the PCP when A1c was >7% at the intervention clinic, compared to 15.6% at the comparison clinics. Medication was intensified at 12.8% of POST encounters with the PCP when A1c was >7% at the intervention clinic, compared to 12.6% at the comparison clinics (P=0.92). Medication intensification also did not differ at other types of encounters (A1c <7%, not with PCP). Conclusions: These results add to previous research by examining a mediating step between POC A1c testing and improved glycemic control. Despite more recent A1c test results, medication intensification was not greater in the primary care clinic using POC A1c testing.