Diabetes Population Management by Telephone Visits *

Introduction

There is significant room for improvement in diabetes outcomes, both at our institution and nationwide. ^1,2 Finding novel, effective ways to deliver care to patients with chronic diseases such as diabetes is essential, given the substantial associated morbidity, mortality, and healthcare costs ^{3 –6} as well as patient and provider dissatisfaction with traditional modes of delivering primary care. ^{7 –9} Some experts envision chronic disease management by physician-supervised nurses ¹⁰ and other medically trained staff through various extraclinic communication modalities, including asynchronous two-way telemonitoring equipment. ^7,8 Given a patient population with limited resources and significant barriers to rapid redesign of chronic care delivery across thousands of patients, we explored an intermediate model for enhanced chronic disease management: dedicated primary care provider time for panel management and telephone visits.

Research Design and Methods

The study was conducted at Westside Family Health Center (WS), one of Denver Health Medical Center's eight federally qualified community health centers. WS serves over 1,700 adult diabetics who are primarily Latino (81%), largely uninsured (41%), or on Medicaid or Medicare (56%).

Intervention patients were defined as those who had an established primary care relationship with the resident (author A.V.) or with the supervising staff physician (author H.H.F.) and were active in our diabetes registry (at least one primary care clinic visit and ICD-9 code for diabetes in the previous 18 months). There were no exclusions and patients included in the postintervention analysis were only those who remained in the diabetes registry at study end.

Control patients were all other established patients in the WS diabetes registry and received “usual care” from their providers, consisting of (1) provider clinic visits with laboratory checks; (2) follow-up on laboratory results with a telephone call and adjustment of medications if necessary; and (3) recommendation for a follow-up appointment, generally within 2–3 months if the patients were not at target on outcome measures.

Telephone contacts with the intervention patients were prioritized in the following order: (1) low-density lipoprotein (LDL) overdue, (2) HbA1c overdue, (3) last LDL >99 mg/dL, (4) last HbA1c >9%, and (5) last blood pressure >130/80 mm Hg. The diabetes registry was queried on a monthly basis. A.V. dedicated an average of 2 h weekly to telephone outreach. H.H.F. conducted five half-day clinic sessions per week, where two face-to-face visits were replaced by three telephone visits dedicated to diabetes panel management throughout the 8-month intervention. Patients not available by phone were mailed letters.

We utilized existing patient resources, such as home glucometers and blood pressure cuffs, and also asked patients to present to clinic for overdue laboratory (no HbA1c in 6 months and/or no LDL in >12 months) or blood pressure checks, which required no copay or provider visit. Patients not at glycemic or blood pressure goals provided home glycemic measurements and blood pressure checks every 1–2 weeks. Provider's problem-solved barriers to care and titrated medications during phone contact. Lifestyle council and medication titration for lipid management occurred approximately every 6 weeks for those not at lipid goal.

We also tracked the number of phone calls, letters, and “visits avoided,” defined as the telephonic facilitation of overdue laboratory tests and/or the initiation or titration of a hyperglycemic, lipid, or blood pressure medication.

Differences in lipid, glycemic, and blood pressure outcomes were compared pre- and postintervention, utilizing the last value in the study interval for both the control and treatment groups using an intention-to-treat principle. Analyses were adjusted for differences in age, race/ethnicity, and gender when t-tests and contingency tables suggested a significant difference between the intervention and control groups at baseline (p-value <0.05). Baseline levels for each outcome variable were also adjusted for in multivariate analyses, which included generalized estimated equations to account for the within-subject correlation of repeated measures by individual patients.

These analyses were adequately powered (0.80) to detect 11% improvements in rates for HbA1c <9%, LDL <100 mg/dL, and blood pressure <130/80 mm Hg. Statistical analyses were performed using SAS (version 9.1; SAS Institute, Inc., Cary, NC) software and power analyses were performed using PASS 2008 software.

This investigation was not placed before the Colorado IRB as it was a quality improvement initiative with minimal risk to patients and obtaining consent from the 1,522 control patients and 167 intervention patients would not have been feasible.

Results

There was no significant difference in ethnicity between the two groups. The intervention group was younger (mean age: 54.7 vs. 56.8; p<0.01) and had more predominantly male (49% vs. 39%; p=0.02) than the control group.

The intervention group achieved greater absolute percentage of increases than the control group for HbA1c checked in the past 6 months, LDL checked in the past year, and blood pressure checked in the past year (17.2% vs. −1.2%, 18.3% vs. 1.0%, and 10.3% vs. −2.6%, respectively; p<0.01 for each measure; Table 1). Absolute clinical outcome performance increased in the intervention and control patients by 20.4% and 5.5% (p<0.01) for HbA1c <9%, 20.1% versus 6.2% (p<0.01) for LDL <100 mg/dL, and 14.1% and 5.8% (p=0.01) for blood pressure <130/80 mm Hg, respectively (Table 1).

The intervention resulted in 201 phone contacts and 49 mailed letters to 82 unique patients, as well as 119 avoided visits. Medication adjustment took place in 29 patients with a mean number of 3.6 calls per patient. Medication adjustments were made 51 times, accounting for 43% (51/119) of the “visits avoided.” The remaining 57% of the “visits avoided” were due to patients coming in for overdue laboratory tests. We were unable to contact 35 patients by phone and thereby mailed letters asking these patients to present for laboratory tests. We mailed 14 additional letters to patients who fell out of contact (moved or phone disconnected). The control and intervention groups had the same number of primary care visits per year at baseline (3.0 for each; p=0.98) as well as during the intervention period (3.0 and 2.8, respectively; p=0.37).

Conclusions

This intervention explores the use of telephone visits for diabetic patients. Although the telephone visits did not cut down the number of face-to-face visits per patient, clinical outcomes improved. Also, the two participating providers reported an enhanced ability to monitor and impact important diabetes outcomes. This pilot study suggests that provider-driven telephone visits may be a means for healthcare systems to improve chronic disease outcomes as we transition to new paradigms of chronic care delivery. Ultimately, one could envision a more efficient use of staff in this intervention: (1) clerks could be trained to algorithmically query the diabetes registry, facilitate laboratories, and schedule patients for telephone visits and (2) physician-supervised nurses could conduct the telephone visits, a model explored in recent expert panel discussions. ⁷

Optimal chronic disease self-management requires daily attention on the part of the patient, and providing adequate support for this endeavor calls for significant changes in chronic disease management. ⁸ Engaging patients on a frequent basis, as often as every 1–2 weeks, diverges from the traditional model of chronic disease management in which patients interact with their provider in a face-to-face visit every 3 months at best. It allows for frequent assessment of barriers and titration of medications to achieve diabetes outcome goals. From a resource utilization standpoint, incorporation of phone management of diabetes avoids the use of ancillary staff required for face-to-face visits as well as the cost of transport and time to the patient. As we ponder new modalities of chronic disease management, further studies of telephone outreach as well as other communication modalities, such as text messaging, e-mail, and the use of IVR technology, will help inform future directions.

Important limitations include lack of blinding and randomization and the difficulty of distinguishing the impact of telephone visits from the “interventionist effect,” whereby patients may have responded to the increased commitment of their physicians. We were also unable to distinguish between the impact of telephone visits scheduled into a clinic template (H.H.F.) and dedicated time for panel management (A.V.). Further, a significant portion of our patients fill medications at pharmacies outside of Denver Health and we were therefore not able to perform a comparison of baseline and postintervention medication use. There were baseline differences in gender and age between the two groups, which were adjusted for in the multivariate analysis.

Author Contributions

H.H.F. and A.V. performed this intervention. M.J.D. and K.M. contributed to the analysis of the intervention. T.D.M. contributed to the design and analysis of the intervention.