PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Health Outcomes Res Med. Author manuscript; available in PMC 2011 July 1.
Published in final edited form as:
Health Outcomes Res Med. 2010 July; 1(1): e61–e66.
doi:  10.1016/j.ehrm.2010.03.002
PMCID: PMC2958673
NIHMSID: NIHMS206235

The effect of the Vermont Diabetes Information System on inpatient and emergency room use: results from a randomized trial

Shamima Khan, MBA, PhD, Assistant Professor,corresponding author Charles D. MacLean, MDCM, and Benjamin Littenberg, MD

Abstract

Objective

To describe the effect of the Vermont Diabetes Information System (VDIS) on hospital and emergency room use

Data Source

Statewide discharge database

Study Design

Randomized controlled trial of a decision support system for 7,412 adults with diabetes and their 64 primary care providers.

Data Collection/ Data Extraction

Charges and dates for hospital admissions and emergency room care in Vermont during an average of 32 months of observation. Data from New York hospitals were not available.

Principal Findings

Patients randomized to VDIS were admitted to the hospital less often than control subjects (0.17 admissions vs. 0.20; P=0.01) and generated lower hospital charges ($3,113 vs. $3,480; P=0.019). VDIS patients also had lower emergency room utilization (0.27 visits vs. 0.36; P<0.0001) and charges ($304 vs. $414; P<0.0001). The intervention was particularly effective in men and in older subjects.

Conclusions

In spite of data limitations that tended to reduce the apparent effect of the system; this randomized, controlled trial showed that VDIS reduces hospitalization and emergency room utilization and expenses.

Keywords: Diabetes Mellitus, Decision Support Systems, Clinical, Patient Care Management, Chronic Disease, Health Services Research, Primary Health Care, Cost of Care, Hospitalization, Emergency room use

The number of people in the United States with diagnosed diabetes is now 17.5 million, with total estimated health care costs in 2007 of $174 billion.1 Although a range of effective treatments are available, diabetes patients continue to receive sub-optimal care.2

The Vermont Diabetes Information System (VDIS) is a laboratory-based registry and decision support system that communicates directly with primary care providers and their adult patients with diabetes. It is designed for low cost and easy integration into primary care. It requires no data entry, additional staff, office space or capital investment by participating practices. Although VDIS can be easily integrated with office computers and/ or electronic medical records, these are not required.3

VDIS uses the Chronic Care Model as an organizing framework with daily data feeds from otherwise independent laboratories, automatic test interpretation using algorithms based on consensus guidelines, use of fax and mail to report to providers and patients not easily reached by electronic networks, and report formats that are accessible and useful to patients and providers. The primary function of the system is to collect pertinent clinical information (hemoglobin A1C, cholesterol, serum creatinine, and urine protein results) and generate five types of reports: flow sheets to providers with accurate and timely laboratory results, reminders of overdue laboratory tests to providers, overdue reminders to patients, alerts to patients with elevated test results, and summary population reports for providers regarding their diabetes roster.4-5

The intervention has been described in detail elsewhere, along with a prospective, cluster-randomized clinical trial in which 7,412 patients and 132 providers in 64 practices were randomized to receive VDIS or usual care.4-5 A random sample of patients completed questionnaires at the end of the study in which they recalled their use of medical services in the previous year including hospitalization and emergency room visits. Intervention subjects recalled significantly less utilization than control subjects with estimated savings of $2,426 per patient per year (95%CI -4647, -205; P=0.03).6 Although randomized to minimize bias, this analysis was limited by the possibility of patient recall error.

A second analysis of the cost impact of this decision support system was undertaken in a cohort of patients covered by a single insurer who used a commercially available version of the system called the Vermedx® Diabetes Information System. In this observational study, total insurance claims paid before and after institution of the system were compared for 153 intervention patients and 870 control subjects.3 Mean savings in the intervention group ranged from $504 per patient in year 1 of operation to $3,563 in year 4. The cumulative net savings reached $8,134 in 4 years. Although not subject to recall error, these data were limited by the non-randomized design and reliance on data from a single insurer.

These two studies have methodological limitations. In order to better understand the potential cost savings associated with the VDIS, we undertook the current study to describe the effect of VDIS on cost and utilization in the original randomized cohort using hospital claims paid across all insurers.

Methods

VDIS receives laboratory results (glycosolated hemoglobin A1C, cholesterol and kidney function) from clinical laboratories, maintains a registry, and produces reports for primary care providers and their patients. Reports are automatically generated whenever a laboratory test is completed. They include flow sheets with guideline-based recommendations for the providers and alert letters for the patient when results are above target. Patients and practices also receive reminders when test results are overdue. Population reports listings all the provider’s patients are sent to each provider quarterly, along with a report card indicating population-level performance. Reports are sent electronically or by fax to the practices, and mailed to patients. The system is not linked to any pay-for-performance incentive.

We recruited 13 hospital-based clinical laboratories in our region, 64 practices with 128 primary care providers, and their 7,412 adult patients with diabetes. Practices were randomized to receive the VDIS intervention or usual care and were observed for at least 24 months.5

The Vermont Association of Hospitals and Health Systems discharge dataset includes data from all thirteen hospitals in Vermont. In a post hoc analysis, we linked individual records from the discharge data with records from the VDIS study using date of birth, sex, and address. Forty-four VDIS records linked to more than one patient in the discharge data and were dropped, leaving a final sample of 7,368 subjects. Patients were enrolled with a passive consent (“opt-out”) procedure.7

All data were analyzed on an intention-to-treat basis. All claims paid for inpatient hospitalizations or emergency department visits from each subject’s date of randomization through their date of censoring were included. Thirty-eight practices (59%) were located in Vermont and 26 (41%) in adjacent New York State. The claims data include only Vermont hospitals though some New York residents receive care in Vermont. At the time of the study, the cost of VDIS was $4 per patient per month.

The primary null hypothesis was that there is no difference between the intervention and control groups in hospital and emergency room charges. We also analyzed the number of hospital admissions, length of stay, and number of emergency department visits. We repeated the analyses for subgroups based on age and sex. Because utilization is heavily skewed, we used the non-parametric Wilcoxon Rank Sum Test for continuous variables and χ2 for categorical variables.

The research was approved by the University of Vermont Committee on Human Research in the Medical Sciences.

Results

Table 1 shows the demographic characteristics of the 3,856 intervention and 3,512 control subjects. The mean age was 63 years, with almost half over 65 years. Intervention subjects were about 1 year older than control subjects (P<0.001). Control subjects were more likely than intervention subjects to be Vermont residents (62.0 vs. 55.5%, P<0.0001). Intervention subjects were more likely than control subjects to have poor glucose control and evidence of diabetic kidney disease.

Table 1
Subject Characteristics at Baseline

Intervention subjects had 15% fewer hospital admissions during the study period than control subjects (0.17 vs. 0.20, P=0.01) and 11% lower hospital charges ($3,113 vs. $3,480, P=0.02; see Table 2). Emergency department visits were reduced by 25% (0.27 vs. 0.36, P<0.0001) and emergency charges by 27% ($304 vs. $414, P<0.0001). The net savings attributable to VDIS is the sum of hospital and ER charges avoided ($477.74 or $14.94 per patient per month over 32 months of observation) less the cost of the service ($4.00 per patient per month at the time of the study) or $10.94 per patient per month.

Table 2
Utilization

Significantly lower utilization with VDIS was observed among seniors (age ≥ 65). Although savings were also seen in younger subjects, the differences did not achieve statistical significance. Men in the intervention group had significantly lower utilization than male control subjects. For women, the intervention group had lower utilization in all categories, but the differences were statistically significant only for emergency room visits and charges.

Discussion

In this randomized controlled trial of a primary care based diabetes decision support system, patients in the intervention group had significant reductions in utilization of hospital and emergency care. The estimated return on investment for VDIS to a payer is at least 3.7 and is even greater among seniors and men.

Although this is a randomized study, randomization by cluster (practice) allowed the two groups to be somewhat different in age and gender at baseline. However, the subgroup analyses show that the overall effect is unlikely to be due to these differences.

The analysis is limited by the unavailability of data from hospitals and emergency rooms in New York State. This undoubtedly falsely lowers the estimates of utilization. However, in a randomized trial, this effect is similar in both groups. The higher proportion of control subjects among Vermont residents suggests that the relative difference between intervention and control groups would have been even larger if we had access to New York data.

All of the analyses of the effect of VDIS or Vermedx® on utilization have limitations. The initial analysis of the randomized population5 was free of bias, but subject to recall error. The insurance claims data were not subject to recall error, but were not randomized and applied to only one payer.3 The hospital data presented here are randomized and apply to all payers without recall error, but exclude an unknown number of admissions and emergency room visits. The current study excludes certain costs, such as physicians’ office visits, medications and supplies, and laboratory testing, which may be higher in the intervention group due to improved patient monitoring and care. Although randomized, the two groups did differ at baseline (Table 1). However, these differences favored the control group suggesting that any potential bias has the effect of reducing the apparent effect of the intervention.

Although each study is limited, together they comprise a consistent body of evidence that VDIS or Vermedx® reduces health care utilization of adults with diabetes. The results further support previous research that shows provider-centered interventions, such as diabetes registries and clinical decision support, improve outcomes and reduce costs 8-12 For instance, Bu et al. estimated that over a 10 year period, diabetic registries saved $14.5 billion ($1,016 per enrolled patient), and clinical decision support systems saved $10.7 billion ($752 per enrolled patient)9, suggesting a savings of similar magnitude to VDIS.

The mechanism by which VDIS or Vermedx® reduces utilization is uncertain. The cost savings are unlikely to be related to prevention of cardiovascular complications because glycemic control, cholesterol level, blood pressure, and self-care behavior remained unchanged between the control and intervention groups.6 However, the intervention facilitates communication between patient and PCP and may stimulate more scheduled contact with the PCP and thereby reduce the need for emergency care. It is also possible that communications from the provider (in the form of reminder and alert letters generated by VDIS) are reassuring to the patient, raising the threshold for urgent visits to the emergency room. Further studies are needed to better understand the mechanism of cost reduction.

Conclusion

In thishis randomized, controlled trial, VDIS was associated with reduced hospital and emergency room utilization and expenses.

Acknowledgments

The authors are grateful for the assistance and support of the Vermont Association of Hospitals and Health Systems.

Funded by: The National Institute of Diabetes and Digestive and Kidney Diseases (R01 DK61167 and K24 DK068380)

Footnotes

Conflict of interest: Drs. MacLean and Littenberg, along with the University of Vermont, hold equity in Vermedx, Inc. which distributes the Vermedx® Diabetes Information System, which is based on the work described in this manuscript. The authors comply with all conflict of interest policies of the National Institutes of Health and of the University of Vermont.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Clinical trials registration number: ClinicalTrials.gov-NCT00109369

Contributor Information

Shamima Khan, Pharmacy and Administrative Sciences, College of Pharmacy and Allied Health Professions, St. Albert’s Hall, Room 108, St. John’s University, Jamaica, NY, Phone: 718-990-2945, Fax: 802-847-0319, khans2/at/stjohns.edu.

Charles D. MacLean, Division of General Internal Medicine, University of Vermont, Burlington, Vermont.

Benjamin Littenberg, Division of General Internal Medicine, University of Vermont, Burlington, Vermont.

References

1. American Diabetes Association. Economic costs of diabetes in the U.S. In 2007. Diabetes Care. 2008;31(3):596–615. [PubMed]
2. Saaddine JB, Cadwell B, Gregg EW, Engelgau MM, Vinicor F, Imperatore G, Narayan KM. Improvements in diabetes processes of care and intermediate outcomes: United States, 1988-2002. Ann Intern Med. 2006;144(7):465–74. [PubMed]
3. Littenberg B, MacLean CD, Zygarowski K, Drapola BH, Duncan JA, Frank CR. The Vermedx Diabetes Information System reduces healthcare utilization. Am J Manag Care. 2009;15(3):166–70. [PubMed]
4. MacLean CD, Littenberg B, Gagnon M. Diabetes decision support: initial experience with the Vermont diabetes information system. Am J Public Health. 2006;96(4):593–5. [PubMed]
5. MacLean CD, Littenberg B, Gagnon M, Reardon M, Turner PD, Jordan C. The Vermont Diabetes Information System (VDIS): study design and subject recruitment for a cluster randomized trial of a decision support system in a regional sample of primary care practices. Clin Trials. 2004;1(6):532–44. [PMC free article] [PubMed]
6. MacLean CD, Gagnon M, Callas P, Littenberg B. The Vermont Diabetes Information System: A cluster randomized trial of a population based decision support system. J Gen Intern Med. 2009;24(12):1303–10. [PMC free article] [PubMed]
7. Littenberg B, MacLean CD. Passive consent for clinical research in the age of HIPAA. J Gen Intern Med. 2006;21(3):207–11. [PMC free article] [PubMed]
8. Renders CM, Valk GD, Griffin SJ, Wagner EH, Eijk Van JT, Assendelft WJ. Interventions to improve management of diabetes in primary care, outpatient, and community settings: a systematice review. Diabetes Care. 2001;24(10):1821–33. [PubMed]
9. Bu D, Pan E, Walker J, Adler-Milstein J, Kendrick D, Hook JM, Cusack CM, Bates DW, Middleton B. Benefits of information technology-enabled diabetes management. Diabetes Care. 2007;30(5):1137–42. [PubMed]
10. Dorr D, Bonner LM, Cohen AN, Shoai RS, Perrin R, Chaney E, Young AS. Informatics systems to promote improved care for chronic illness: a literature review. J Am Med Inform Assoc. 2007;14(2):156–63. [PMC free article] [PubMed]
11. Wyne K. Information technology for the treatment of diabetes: improving outcomes and controlling costs. J Manag Care Pharm. 2008;14(2 Suppl):S12–7. [PubMed]
12. Young AS, Chaney E, Shoai R, Bonner L, Cohen AN, Doebbeling B, Dorr D, Goldstein MK, Kerr E, Nicholas P, Perrin R. Information technology to support improved care for chronic illness. J Gen Intern Med. 2007;22(Suppl 3):425–30. [PMC free article] [PubMed]