Cooking Well with Diabetes. In 2003, the Texas AgriLife Extension Service dietitians and/or diabetes educators to be involved …
M E D I C A L - S E R V I C E U T I L I Z AT I O N
Diabetes Disease Management in a Community-Based Setting
GREGORY D BERG, PHD1, AND SANDEEP WADHWA, MD, MBA2 1 Senior Research Scientist, McKesson Corp; 2Vice President, Population Care Management, McKesson Corp, and Clinical Assistant Professor of Medicine, University of Colorado Health Sciences Center
ABSTRACT Purpose The medical cost of diabetes in the United States in 1997 was at least 98 billion This study illustrates the behavioral change and medical-care utilization impact that occurs in a community-based setting of a diabetes disease-management program that is applied to program participants in a health insurance plans health maintenance organization and preferred provider organization Design A historical control comparison of diabetes-management participants Methodology One hundred twenty-seven identified diabetes patients are followed from baseline
Author correspondence: Gregory D Berg, PhD McKesson Corp 335 Interlocken Parkway Broomfield, CO 80021 E-mail: GregBerg@McKessoncom Phone: 303-926-6208 Fax: 303-466-5949 Author Contributions: Dr Berg had full access to all the data in this study and takes responsibility for
the integrity of the data and accuracy of the data analysis DrWadhwa provided study supervision and medical expertise This study was funded by McKesson Corp This paper has undergone peer review by appropriate members of MANAGED CARES Editorial Advisory Board
through 1 year Differences in behavior are compared at program intake and at a 6-month reassessment Differences in medical-service utilization are compared in the baseline year and the year subsequent to program enrollment Poisson multivariateregression models are estimated for counts of inpatient, emergency department, physician evaluation and management, and facility visits, while also controlling for potential confounders Principal findings Behaviors improved between program intake and the 6-month reassessment From patient reports, the number of participants having a hemoglobin A1c test increased by 449 percent p001, and there was a 532-percent decrease in symptoms of hyperglycemia p002 From medical claims after program enrollment, a drop occurred during the program year in every dimension of medical-service utilization Regression results show that inpatient admissions decreased by 391 p001 per 1,000 for each group, while
controlling for age, length of membership, and the number of comorbid claims for congestive heart failure In the analysis of costs that were pre- and post-enrollment, which included disease-management program costs, a 434:1 return on investment was calculated Conclusion The diabetes program provides patients with comprehensive information and counseling relative to practicing self-management of diabetes through a number of integrated program components This study strongly suggests that the implementation of such a program is associated with positive behavioral change and, thus, with substantial reduction in medical-service utiliza-
tion In addition, the intervention resulted in a net decrease in direct medical costs INTRODUCTION Approximately 103-million people in the United States have been diagnosed with diabetes mellitus, a serious, life-long disorder that remains, as yet, without a cure An additional 54-million people have diabetes but are unaware that they have the disease1 Individuals with diabetes are at higher risk of heart disease, stroke, high blood pressure, blindness, kidney disease, nervous-system disease, amputations, dental disease, and complications of
pregnancy,1,2 resulting in diabetes mellitus being ranked as the seventh leading cause of death in the United States3 The long-term complications of diabetes, particularly when poorly managed — as with the elderly,4 give rise to grave financial as well as human consequences The number of health care services directed toward diabetes patients is high,5,6 with a large portion of that figure representing inpatient hospital care7 The cost of diabetes was estimated at 982 billion in 19978 Diabetes disease management has been shown to save money and improve outcomes in opt-in programs sponsored by health maintenance organizations,9 resulting in significant short-term improvements in glycemic control in the treatment population10 Furthermore, the integration of individualized goals with educational surveillance reduces risk factors in diabetes patients and thus reduces diabetes complications11 The McKesson Health Solutions MHS Diabetes CareEnhance program specifically targets diabetes and
42
MANAGED CARE / JUNE 2002
DIABETES DISEASE MANAGEMENT IN A COMMUNITY-BASED SETTING
its associated conditions by aggressively monitoring and educating members about cardiovascular risk control,
glycemic control, and microvascular risk control The program has an intense focus on aggressive lipid management, appropriate aspirin use, beta-blocker use for diabetes patients with a history of coronary artery disease, and control of blood pressure, in addition to glucose control and preventive testing Rather than showing results from participants who were recruited in a clinic or hospital setting, this study shows community-based program effects for diabetes patients in an HMO and a PPO This diabetes-management program was developed to address some of the most serious challenges facing health care providers, health plans, and employers seeking to improve diabetes management Through an individualized approach to patient monitoring, education, and counseling, the CareEnhance Diabetes program seeks to improve patients selfmanagement practices and enhance communication with their providers, while improving clinical outcomes and quality of life, as well as reduce treatment costs associated with diabetes MHS developed this telephonebased, RN-delivered diabetes program to directly address the challenge of improving quality while decreasing cost that faces health care providers, health
plans, and employers A team of physicians, nurses, and nurse practitioners with extensive experience in diabetes developed the Diabetes CareEnhance program, which follows the most current clinical practice recommendations outlined in the national guidelines of the American Diabetes Association12 Specially trained registered nurses monitor, educate, and provide telephonic case-management services for members with diabetes On program enrollment, an RN conducts a comprehensive assessment of a program
participant Each participant is assigned a diabetes-severity level, which ranges from 1 stable to 3 high risk MHS provides a customized selfmanagement plan for each participant Interventions vary according to the members risk profile and individual need The interventions include written materials, informal and formal education, participant ability to contact the nurses with symptoms and concerns, and regular communication with participant physicians The customized nature of the program is an important feature, because program goals involve enabling behavioral changes that lead to reduced medical-service utilization Interventions are most concentrated in the first 3 months of the program
Thereafter, regular reinforcement of the core training is provided and compliance is assessed METHODS Study design The diabetes program began to identify diabetes patients in January 2000 from an HMO and PPO health plan Registered nurses called the members who were identified as having diabetes and asked if they were interested in enrolling in the program For those who agreed to participate, MHS provided a customized self-management intervention plan that varied according to individual needs and risk profiles The plan included written materials, informal and formal education, instructions on how to contact the nurses with symptoms and concerns, and encouragement to communicate regularly with participating physicians Once members were enrolled in the program, nurses conducted a telephonic survey at 6 and 12 months to assess each participants knowledge, behavior, and health status relative to their diabetic conditions The improvement in patient knowledge, behavior, and health status relative to diabetes was expected to lead to changes in their medical-
service utilization Six months after the program was implemented, a comparison of initial interview responses and 6-month interview
responses on behavioral and lifestyle changes was conducted A change in behavior and lifestyle is believed to affect medical-service utilization Twelve months after the program was implemented, a medical-claimsbased historical comparison study was conducted to reveal any changes in medical-service utilization for inpatient IP, emergency department ED, physician office MD, and outpatient facility FA services Study population The study group included members whose conditions were identified by the health plan as most severe In addition, all study participants were: health plan members for 12 months prior to their enrollment and for at least 6 months subsequent to their enrollment; not listed with deceased or does not have condition on the diabetes program report; had at least 3 months of diabetes-program participation This selection process resulted in a total of 127 study participants The requirements were chosen to provide a fair representation of medicalservice utilization at baseline and during the program The health plan referred members for participation if the member had diabetes and a hospitalization in the past year Historical control Changes in behavior and lifestyle were
compared at program intake and at 6 months after intake Behavioral changes were compared at program intake and 6 months after intake for sugar, microvascular, and macrovascular variables The program was designed to measure whether changes would result in reductions in medical-service utilization, which were compared during the baseline year and the year following program enrollment Since each program participant had a different enrollment date during the period from February 2000 through August 2000, a unique pre-
JUNE 2002 / MANAGED CARE
45
DIABETES DISEASE MANAGEMENT IN A COMMUNITY-BASED SETTING
program period and a program period were established for each participant, eg, a member who enrolled on April 15, 2000 would have a preprogram period that was the year prior — April 15, 1999 to April 14, 2000, with the corresponding program period from April 15, 2000 to April 14, 2001, or to the date when that member disenrolled in the plan if that occurred prior to April 14, 2001 If the member disenrolled from the diabetes program but remained in the health plan, the time period after disenrollment was still included in the program evaluation period, up to 1 year This time period
was included because behaviors and lifestyle changes made during program participation were expected to carry over to later dates Medical-service utilization rates were reported as annualized rates If the member disenrolled from the health plan before 1 year of diabetes-program participation, no medical claims were available and utilization rates were then annualized Changes in behavior and lifestyle variables observed at 6 months were expected to lead to changes in medical-service utilization during the year following program enrollment Multivariate-regression analysis Multivariate-regression analysis was performed to account for possible confounders The list of possible confounders included age group, comorbidities, and the amount of time the member was a health plan member Four regression models were estimated — one for each type of utilization, including inpatient, emergency department, physician visits, and other visits, which are mostly facility visits The dependent variable was the number of visits that occurred in either the pre- or post-enrollment periods This dependent variable is a count variable 0,1,2, rather than continuous, which means that ordinary least squares OLS
statistical-
regression analysis is not appropriate 13, 14 OLS is an appropriate statisticalregression technique when the dependent variable is continuous Therefore, in this case we selected the Poisson regression technique, which is an appropriate statistical-regression technique when the dependent variable is a count variable Poisson models also account for possible confounders For each participant in the study , we calculated the actual number of visits in the 12-month period prior to enrollment and the actual number of visits in the 6- to 12months after intake during which the member was in the health plan Explanatory variables were a pre/post dummy variable, an age-group dummy variable, the number of congestive heart failure claims that the member had while with the health plan, the post-enrollment number of days that the member was with the health plan 612 months, and the post-enrollment number of days squared RESULTS Membership The total number of members who met the study criteria was 127 To be considered a study participant, the member must have been enrolled with the health plan for at least 1 year prior to enrollment, must have been enrolled with the health plan for at
least 6 months subsequent to enrollment, and must have participated in the diabetes program for at least 3 months The average length of participation in the diabetes program was 843 months for the 127 study participants Changes in sugars, microvascular, and macrovascular variables The evaluation period was from intake to the point at which each individuals 6-month reassessment was conducted Although the study population was 127, not all participants completed a 6-month evaluation Six-month evaluations were not performed for the 20 study participants who were not able to be
located for the interview The claims for these 20 study participants were included in the medical-service utilization analysis Table 1 shows measures of sugar, microvascular, and macrovascular variables at program intake and at the 6-month reassessment P-values were calculated using the McNemar test for all categorical variables, and the Wilcoxon signed-rank test for paired data was used for hemoglobin A1c values Statistically significant improvements in sugar variables were noted for symptoms of hyperglycemia during the 2-week periods prior to patient interview and for hemoglobin A1c testing during the year
prior to patient interview Members experienced a 532-percent decrease in symptoms of hyperglycemia p002 and the number of members with a hemoglobin A1c test during the prior year increased by 449 percent p001 Statistically significant improvements for microvascular variables were noted for those members who performed a daily foot exam and for those members who had a prescription for an ACE inhibitor The number of individuals who had daily foot exams increased by 206 percent p001, while the number of prescriptions for ACE inhibitors showed an increase of 342 percent p003 Statistically significant improvements for macrovascular variables were noted for members who took aspirin and for members with a diastolic blood pressure below 80 Members who took aspirin daily increased by 292 percent p001; members who reported that their most recent diastolic blood pressure was below 80 increased by 347 percent p003 The observed improvements in behavior and lifestyle variables between the intake and the 6-month reassessment are expected to lead to improvements in medical-service uti-
46
MANAGED CARE / JUNE 2002
DIABETES DISEASE MANAGEMENT IN A COMMUNITY-BASED SETTING
TABLE 1 Behavioral and
lifestyle changes Measure Sugars Use glucose meter daily Symptoms of hypoglycemia in past 2 weeks Symptoms of hyperglycemia in past 2 weeks Hemoglobin A1c test Hemoglobin A1c value Microvascular Foot exam Dilated eye exam Kidney function test Prescription for ACE inhibitor Macrovascular Prescription for beta blocker Takes aspirin Follows a low-salt diet Body mass index 30 Systolic blood pressure 130 Diastolic blood pressure 80 Intake 676 280 280 561 538 729 692 570 500 600 383 528 402 634 634 6 months 745 271 131 813 483 879 692 673 671 667 495 625 402 659 854 Difference 102 32 532 449 102 206 00 181 342 112 292 184 00 39 347 p-value 127 858 002 000 250 001 1000 093 003 317 001 194 1000 763 003 n 102 107 107 107 13 107 107 107 70 15 107 72 107 41 41
A p-value 05 indicates statistical significance at the 5-percent level
lization in the year following a members diabetes program enrollment Changes in medical-service utilization The evaluation period for calculating utilization rates includes the full year of claims before enrollment, and 6 to 12 months of claims after enrollment, which depends on the length of member enrollment in the health plan Because all study participants were
members of the health plan for a full year before enrolling, no annual adjustment was needed for preprogram rates The number of members in the program evaluation period was 127 Table 2 shows annualized number of visits, annualized utilization rates, and the percentage change between the preprogram and program evaluation periods for inpatient admissions, emergency department visits, physician visits, and facility visits The p-values are from the Wilcoxon signed-rank test for paired data For inpatient admissions, members had a 237-percent reduction p027 in the post-enrollment period Such high utilization of inpatient services is due to the health
TABLE 2 Changes in medical-service utilization Annualized Utilization utilization rate/1,000 reduction Inpatient Emergency department MD Facility Pre-enrollment Post-enrollment Pre-enrollment Post-enrollment Pre-enrollment Post-enrollment Pre-enrollment Post-enrollment 1,110 847 457 431 9,850 9,043 8,685 7,321 237 p027 57 p109 82 p057 157 p012
A p-value 05 indicates statistical significance at the 5-percent level
plans selection of patients with the most severe cases of diabetes As such, the reduction in the post-enrollment period could
be influenced heavily by regression to the mean For emergency department visits, physician office visits, and facility visits, members experienced reductions in utilization of 57 percent p109, 82 percent p057, and 157 percent p012, respectively, in the post-enrollment period In all types of medical-service uti-
lization, the post-enrollment period is associated with a drop in utilization The medical-service utilization rates support the claim that observed behavior and lifestyle changes after 6 months of program participation led to observed reductions in medicalservice utilization in the year following enrollment Utilization changes Table 2 yielded a total savings of 214,486, and program costs totaled 49,429 Given the change that occurred in medical-service utilization
48
MANAGED CARE / JUNE 2002
DIABETES DISEASE MANAGEMENT IN A COMMUNITY-BASED SETTING
and disease-management program costs, a return on investment of 434:1 was calculated Multivariate-regression analysis Multivariate-regression analysis was estimated by a Poisson model, which accounts for the count nature of the dependent variable and allows for control of potential confounding variables The dependent variable
is the number of visits that occurred in either the pre- or post-enrollment periods Explanatory variables are a preenrollment/post-enrollment dummy variable, an age-group dummy variable, the number of CHF claims the member had while with the health plan, the post-enrollment number of days the member was with the health plan 612 months, and the postenrollment number of days squared Tables 3 and 4 show the Poisson regression analysis results for each type of visit Tables 3 and 4 show that the coefficient in the post-enrollment period is negative for all measures of utilization, with statistical significance occurring at p001, except for ED utilization What this means is that the number of visits in the postenrollment period, controlling for confounding variables, is less than the number of visits in the preenrollment period The Poisson multivariate-regression results support the claim that observed behavior and lifestyle changes after 6 months of program participation led to observed reductions in medical-service utilization in the year following enrollment DISCUSSION The current study has investigated behavior and lifestyle changes for 127 diabetes-program participants at the
initial interview and at 6 months following the initial interview for sugar, microvascular, and macrovascular variables Improvements in diabetes self-management, symptoms, and medication management were noted between the initial and 6-month
TABLE 3 Poisson multivariate-regression results — inpatient admissions and emergency department visits IP Explanatory variables Coefficient 1139 baseline 0391 baseline 0013 0011 00000 0025 p-value Coefficient 466 — 000 — 906 354 531 001 0289 baseline 0070 baseline 0227 00400 00001 0006 Intercept Pre-enrollment period Post-enrollment period Age 65 Age 65 Health plan enrollment days Health plan enrollment squared Number of comorbid CHF claims ED p-value 850 — 554 — 179 003 085 048
A p-value 05 indicates statistical significance at the 5-percent level TABLE 4 Poisson multivariate-regression results — MD and facility visits MD Explanatory variables Coefficient 3127 baseline 2960 baseline 1284 0076 00001 0081 Intercept Preenrollment period Post-enrollment period Age 65 Age 65 Health plan enrollment days Health plan enrollment squared Number of comorbid CHF claims 530 — 000 — 006 048 109 000 Facility p-value 064 — 000 — 128 906 760 000
7748 baseline 3110 baseline 0613 0004 00000 0070 p-value Coefficient
A p-value 05 indicates statistical significance at the 5-percent level
follow-up interviews Reductions in the number of hospitalizations also were noted in the year following the start of the diabetes program Presumably, the improvements in selfmanagement and medical management contributed to the reduced utilization of high-cost medical services Medical-service utilization for members was compared to baseline utilization along four dimensions: hospital inpatient, emergency department, physician visits, and outpatient facility visits Hospitalization and facility visits were lower and statistically significant Emergency department and office visits were lower but not statistically significant There are several inherent biases associated with pre-enrollment and post-enrollment historical control study design evaluations Patients included in this analysis voluntarily
participated in the program and, furthermore, elected to participate for at least 3 months and up to 1 year Such members may be predisposed to improving their self-management skills This may suggest that the difference in outcomes is not attributable
to the diabetes program but to the motivation levels among those who elect to enroll in a diabetesmanagement program, which is then enabled by the diabetes program Clearly, however, if the diabetes program had not been available, members would not have had this opportunity to focus their motivation to manage their condition better Another bias inherent to this design is regression to the mean The study participants were high users of medical services, as denoted by their high baseline inpatient-admission rates High medical-service usage
JUNE 2002 / MANAGED CARE
49
DIABETES DISEASE MANAGEMENT IN A COMMUNITY-BASED SETTING
members may return to mean utilization over time, given the natural history of their condition The health plan selected the most severe diabetes patients for recommendation to the diabetes program As such, without a similar comparison group, this study may suffer from regression to the mean where patients cycle in and out of high-severity status Without a comparison group, the degree of regression to the mean cannot be known Nevertheless, unlike other chronic diseases such as asthma, diabetes is a progressive disease Therefore, regression to the mean is not as
likely to play a highly attributive role relative to the results of the study Though the sample size of 127 is not large, the statistical model demonstrates many significant findings The 3-month minimum participation exclusion was designed to include patients who did complete the core curriculum but who may not have participated in the full-year intervention The highly positive return on investment points to the effect of intervention in a population of diabetes patients who are high users of health care services It will be of further interest to extend these results to moderate and mild populations, as well, to determine the associated return on investment Yet the ability to show a return on investment in 1 year for any group of diabetes patients is of great interest, particularly given the dearth of published results demonstrating this finding in a population of diabetes patients The fact that the model emphasized macrovascularrisk reduction may account for the rapid return on the investment Moreover, it is of interest to evaluate the duration of behavior change and reductions in utilization for periods greater than 1 year All these limitations must be considered in light of the
findings of a decrease in utilization and high return on investment The diabetes program has provided comprehensive information
and counseling about diabetes selfmanagement practices through a number of integrated program components The current study strongly suggests that implementation of such a program is associated with improvements in behavior and lifestyle variables, leading to substantial reductions in medical-service utilization when administered in a community-based setting REFERENCES
1 Centers for Disease Control and Prevention CDC National Diabetes Fact Sheet 1998 Available at http://wwwcdcgov/diabetes/pubs/ pdf/bw_engpdf Accessed February 21, 2002 de Grauw WJ, van de Lisdonk EH, van den Hoogen HJ, van Weel C Cardiovascular morbidity and mortality in type 2 diabetic patients: A 22 year historical cohort study in Dutch general practice Diabetic Medicine 1995;122:117122 Centers for Disease Control and Prevention CDC National Vital Statistics System 1999 Available at http://wwwcdcgov/diabetes/ statistics/survl99/chap3/table1htm Accessed February 21, 2002 Weiner JP, Parente ST, Garnick DW, Fowles J, Lawthers AG, Palmer RH Variation in office-based quality: A claims-based
profile of care provided to Medicare patients with diabetes JAMA 1995;273:15031508 Janes GR Ambulatory care for diabetes In: Harris MI, Cowie CC, Stern MP, Boyko EJ, Reiber GE, Bennet PH, eds Diabetes in America 2nd ed Washington: US Department of Health and Human Services, National Institutes of Health; 1995: DHHS publication no NIH 951468 Aubert RE, Geiss LS, Ballard DJ, Cocanougher B, Herman WH Diabetes-related hospitalization and hospital utilization In: Harris MI, Cowie CC, Stern MP, Boyko EJ, Reiber GE, Bennet PH, eds Diabetes in America 2nd ed Washington: US Department of Health and Human Services, National Institutes of Health; 1995: DHHS publication no NIH 951468 America Diabetes Association Direct and indirect costs of diabetes in the United States in 1992 Alexandria, Va: American Diabetes Association, 1993
8
9
10
11
2
12
13 14
3
American Diabetes Association ADA Economic consequences of diabetes mellitus in the United States in 1997 Diabetes Care 1998:269309 Sidorov J, Shull R, Tomcavage J, Girolami S, Lawton N, Harris R Does diabetes disease management save money and improve outcomes? Diabetes Care 2002;25:684689 Sidorov J, Gabbay R, Harris R, et al Disease
management for diabetes mellitus: Impact on hemoglobin A1c Am J Managed Care 2000;6:1217 1226 Olivarius NF, Beck-Nielsen H, Andreasen AH, Horder M, Pedersen PA Randomized controlled trial of structured personal care of type 2 diabetes mellitus Br Med J 2001; 323:970975 American Diabetes Association ADA Clinical Practice Recommendations 2001 Available at http://journal diabetesorg/CareSup1Jan01htm Accessed February 21, 2002 Greene W Econometric analysis 4th ed New Jersey: Prentice Hall: 2000:931 Rothman KJ, Greenland S Modern Epidemiology 2nd ed Philadelphia: Lippincott Williams Wilkins: 1998:404405
4
5
6
7
50
MANAGED CARE / JUNE 2002
