Objective: To assess and compare healthcare utilization andcosts over a 2-year period in older patients (≥60 years) with type 2diabetes receiving combination therapy with rosiglitazone plus asulfonylurea (glipizide) or progressive up-titration of glipizidemonotherapy.
Study Design: Two-year, randomized, double-blind, parallel-groupclinical trial.
Patients and Methods: Older type 2 diabetic patients initiallyreceiving submaximal doses of a sulfonylurea were randomized toreceive rosiglitazone plus glipizide (n = 115) or up-titrated glipizidemonotherapy (n = 110). Information on patient self-reportedhealthcare utilization (hospitalizations, emergency department[ED] visits, physician office visits) was collected prospectively forthe duration of the trial. National average healthcare costs per unitwere applied to calculate direct medical costs.
Results: Demographic characteristics of the 2 groups were similar.At the study's end, glycemic values were better in the rosiglitazone-plus-glipizide group. Compared with the glipizide group,patients receiving rosiglitazone plus glipizide had significantlyfewer ED visits (= .0006) and hospitalizations (= .0263).Although the glipizide group had more unscheduled physicianoffice visits, the difference was not statistically significant.Estimated treatment costs per patient per month were significantlylower for the rosiglitazone-plus-glipizide group than for the glipizidegroup ($480 vs $645; < .05).
Conclusion: Addition of rosiglitazone to sulfonylurea therapy wasassociated with decreased use of medical resources, in particularhospitalizations and ED visits, compared with progressive sulfonylureaup-titration. Although causality could not be established, thistherapeutic approach could improve clinical outcomes in patientswith type 2 diabetes and reduce healthcare utilization and costs.
(Am J Manag Care. 2005;11:273-278)
Type 2 diabetes affects 18% to 20% of people morethan 65 years of age in the United States.1 Thecost of diabetes increases with age.2 In 2002, thetotal cost of diabetes in the United States was estimatedto be $132 billion, with direct medical costs of $92 billion.3 Direct medical costs of diabetes care, the chroniccomplications of diabetes, and the excess prevalenceof general medical conditions in people with diabetesaccounted for $23.2, $24.6, and $44.1 billion, respectively.Total per capita cost for patients with diabeteswas estimated to be $13 243 (compared with $2560 forthose without diabetes).2 The increasing prevalence ofdiabetes,4 the aging of the US population, and efforts atcost containment have raised concerns about thehealthcare resource utilization and costs associatedwith type 2 diabetes.
Comprehensive information on the cost of treatingdiabetes, including the costs of treatments, side effects,and outcomes, would help decision makers to makemore informed choices about how to manage the disease.A number of studies have analyzed costs associatedwith type 2 diabetes and its complications, usinghealth insurance claims, patient medical records, andpatient and provider surveys.5-8 The annual direct medicalcosts for HMO patients with diet-controlled type 2diabetes, a body mass index (BMI) of 30 kg/m2, and nomicrovascular, neuropathic, or cardiovascular complicationswere $1700 and $2100 for white men andwomen, respectively. A 10-kg/m2 increase in BMI, treatmentwith oral antidiabetic or antihypertensive agents,diabetic kidney disease, cerebrovascular disease, andperipheral vascular disease each were associated with a10% to 30% increase in cost. Insulin treatment, angina,and myocardial infarction each were associated with a60% to 90% increase in cost.5 Compared with personswho have no complications or comorbidities, personswho have 2 or more complications or comorbiditiesused moderately more primary care services (1.3-1.9times more) and markedly more specialty care services (5.8-6.3 times more), emergency department (ED)visits (3.3-5.5 times more), and hospital stays (3.3-11.9times more).9 The costs of managing diabetes complicationswere estimated to be $47 240 per patient overthe period of 30 years.10 Intensive glycemic control,although more expensive, was reported to significantlyreduce the risk of microvascular complications (eg,retinopathy, nephropathy, neuropathy) in patientswith type 2 diabetes. The greater costs of intensivetherapy were largely offset by the cost savings associatedwith fewer complications.11 Although type 2 diabetesand its complications have been recognized asimportant healthcare cost drivers, no clinical trialother than the United Kingdom Prospective DiabetesStudy has focused on the long-term healthcareresource use and costs associated with treatment andcomplications.12
The objective of this study was to analyze resourceutilization and cost of care in the Rosiglitazone Early vs.SULfonylurea Titration (RESULT) study, a 2-year clinicaltrial involving older patients (≥60 years) with type 2diabetes randomized to receive combination treatmentwith rosiglitazone plus a sulfonylurea (glipizide) or uptitrationof glipizide monotherapy.
The RESULT study was a 2-year, double-blind, randomized,parallel-group clinical trial. It has beendescribed in detail elsewhere (J. Rosenstock, MD, B. J.Goldstein, MD, PhD, A. I. Vinik, MD, PhD, et al.,unpublished data, 2004). In brief, after screening, eligiblepatients ≥60 years of age received 4 weeks oftreatment consisting of diet/exercise reinforcementand the prescribed half-maximum dose of glipizide (10mg twice daily). Patients whose diabetes remainedinsufficiently controlled were randomized in equalnumbers to 1 of 2 treatment groups: rosiglitazone plusglipizide and rosiglitazone-matched placebo plus glipizide.During the 2-year study period, physicians individualizedeach patient's treatment using a systematic,stepwise titration schedule and were encouraged totitrate medication to attain targets recommended bythe American Diabetes Association (ADA).13 Althoughstudy medications could be adjusted at the discretionof the study physicians, up-titration to the maximumlabeled dosages (glipizide 40 mg/day, rosiglitazone 8mg/day) was required if the fasting plasma glucoselevel was ≥180 mg/dL (10 mmol/L) and was recommendedif the fasting plasma glucose level was >140 mg/dL(7.8 mmol/L).
Eligible patients were men and women at least 60years of age who had a diagnosis of type 2 diabetes mellitusand who had been treated with sulfonylureamonotherapy for at least 3 months before screening, at1/4 to 1/2 of the maximum recommended dose of the sulfonylureafor a minimum of 2 of those 3 months. Inaddition, patients were required to have fasting plasmaglucose levels between 126 mg/dL and 250 mg/dL at thefirst study visit, and between 126 mg/dL and 179 mg/dLby the third study visit, during the 4-week run-in periodwith diet and exercise reinforcement and monotherapywith glipizide 10 mg twice daily.
Resource Utilization and Costs
Resource utilization data were collected prospectivelyfor all nonprotocol-related events requiring a healthcaresystem encounter. Resource utilization includedhospitalizations, ED visits, and unscheduled outpatientvisits to the study physician or other healthcareproviders. Dates of hospitalizations, ED visits, andphysician office visits were collected. These assessmentswere made at baseline (the point of randomization)and at all subsequent study visits. Patients wereasked to self-report any nonprotocol-related healthcareutilization since the last scheduled visit. In addition,patients were asked to self-report the number of beddays (days spent in bed at home for half a day or more)and restricted-activity days (days patients reduced theirusual activities, such as work, housework, and shopping)in the 7 days before the clinic visit. Patient-reportedutilization data were supplemented with data onprotocol-mandated outpatient visits, use of medications(both related and unrelated to the study), laboratorytests, and self-monitoring of blood glucose to estimatethe total cost of care in both study groups. Use of studymedications was recorded in the clinical trial, and useof nonstudy medications was self-reported. Nonstudymedication was defined as any concomitant medicationtaken by the patient during the period of study participationthat was not diabetes specific. Assumptionsabout types and frequency of the laboratory tests werebased on ADA recommendations.13 The latter alsoserved as a basis for estimates about the frequency ofself-monitoring of blood glucose.
Healthcare utilization was expressed as a rate per1000 patient-days. A Poisson regression model (PROCGENMOD) was used to estimate the event rate per 1000patient-days and to test for treatment differences foreach end point separately. Each model used the numberof events as the dependent variable; independent variablesincluded terms for treatment (rosiglitazone plusglipize versus glipizide) and baseline glycosylated hemoglobin (HbA1c), and accounted for the duration of therapy.SAS statistical software, version 8 (SAS InstituteInc, Cary, NC), was used in all analyses.
A standard unit-cost method was used to calculatedirect medical costs. Hospitalizations, ED visits, andphysician office visits reported by patients wereassigned monetary values based on published USnational average unit costs.14 US national average unitcosts were inflated using the consumer price index formedical care and all costs were expressed in 2002 USdollars. Cost of study medication was based on wholesaleacquisition cost, and the dose and frequency usedin each study arm. Cost of nonstudy medication wasbased on the reported national average annual cost ofprescription medications.14 Laboratory costs werederived from the published Medicaid fee schedule.15Cost of self-monitoring was based on the current marketprice of monitoring devices and test strips.16 Per patientper month (PPPM) costs were calculated for hospitalizations,ED visits, physician visits, medications, laboratorytests, and self-monitoring separately, and total costswere calculated for both study groups. All costs wereadjusted for the duration of therapy, which variedamong members in both groups.
To assess the economic impact of each therapy, costswere reported and analyzed as cost of hospitalizations,ED visits, physician visits, medication, laboratory tests,self-monitoring, cost per study group participant, andtotal cost. For comparison purposes, the costs wereexpressed as cost PPPM.
The distribution of cost data tends to be skewed dueto the presence of outliers. To estimate the impact ofoutliers, sensitivity analysis was conducted by identifyingoutliers and excluding them from the analyses.
The 2 treatment groups were comparable withrespect to age, sex, race, BMI, and smoking status. Themajority of patients were male (73%) and white (91%),with a mean BMI of 30.4 + 4.7 kg/m2. The mean age ofthe patients was 68.4 + 6.2 years, with approximately43% of patients older than 70 years of age. Table 1 summarizesthe demographic characteristics of the studypopulation.
The primary clinical outcome was time from randomizationto a final action point, defined as a fastingplasma glucose level of ≥180 mg/dL (upon confirmatorytesting) for a patient who was titrated to maximumdoses of glipizide and study medication. Only 2 of 115patients reached the final action point in the rosiglitazone-plus-glipizide group (2.0%) compared with 27 of110 patients in the glipizide monotherapy group(24.5%), a difference that was significant (<.0001).The average duration of therapy was 20.1 months for allselected subjects, with mean exposure of 21.5 and 18.8months in the rosiglitazone-plus-glipizide and glipizidegroups, respectively.
Compared with patients in the glipizide group,patients in the rosiglitazone-plus-glipizide group hadsignificantly fewer hospitalizations (= .0263), selfreportedbed days (= .0002), and ED visits (= .0006).In addition, patients in the rosiglitazone-plus-glipizidegroup had a significantly shorter mean length of stay perhospitalization than those in the glipizide group (4.5days vs 7.4 days; < .001). Although numericallygreater in the glipizide group, there were no significantdifferences between the 2 treatment groups in unscheduledvisits to study physician offices or restricted activitydays (Table 2).
Cost of Treatment
Based on the national average, the cost of 1 hospitalday is $2577.14 Total hospitalization costs were $311 684and $744 655 in the rosiglitazone-plus-glipizide andglipizide groups, respectively. In the group receivingrosiglitazone plus glipizide, the average hospitalizationcost per study group member was $2710, and theexposure-adjusted hospitalization cost PPPM was$126; comparable figures for the group receiving glipizidemonotherapy were $6770; and $361.
The national average Medicare reimbursement ratefor ED visits is $183.14 In the rosiglitazone-plus-glipizideand glipizide groups, ED visit costs totaled $8595 and$16 093, or $75 and $146 per patient, respectively. Therespective exposure-adjusted ED costs PPPM were $4and $8.
The national average Medicare outpatient visit reimbursementrate is $56.14 This figure was applied to allthe protocol-related and patient-reported physician visitsin the rosiglitazone-plus-glipizide and glipizidegroups. The total cost of physician visits was $92 878and $84 277 ($807.60 and $766.20 per patient) in therosiglitazone-plus-glipizide and glipizide groups, respectively.Exposure-adjusted PPPM costs for physician visitswere $38 and $41 in the 2 groups, respectively.
Based on the wholesale acquisition cost, the durationof treatment, and the final dosage for study medication,study medication costs during the clinical trial totaled$243 550 and $22 689 in the rosiglitazone-plus-glipizideand glipizide groups, respectively. The respective costsof study medication per participant were $2118 and$206, and the exposure-adjusted costs of study medicationwere $99 and $11 PPPM.
Nonstudy medication was assessed by using thereported national average annual prescription drug costof $446 per medication.14 Based on study data, nonstudymedications cost a total of $449 876 and $402 698 inthe rosiglitazone-plus-glipizide and glipizide groups,respectively. The respective costs of nonstudy medicationper participant were $3912 and $3661, and theexposure-adjusted costs PPPM were $182 and $195.
The types of laboratory tests performed and their frequencywere based on ADA recommendations. Theyincluded a HbA1c measurementevery 3 months, and alipid panel and urine albuminassessment once peryear. A liver enzyme testwas performed every 2months for the first year oftherapy for the rosiglitazone-plus-glipizide group, aswas recommended for thiazolidinedionesat the time thisstudy was performed. (Subsequentto this study, liverenzyme testing was requiredonly prior to the initiation oftherapy with rosiglitazoneand periodically thereafterper the clinical judgment ofthe healthcare professional.)The cost of laboratory tests for both study groups wascalculated using appropriate procedure codes and theirrates as quoted in the Medicaid fee schedule.15 Lab testunit costs were $13.42, $18.50, $8.00, and $11.29 forHbA1c , the lipid panel, the urine albumin/creatinineratio, and the hepatic enzyme panel, respectively. Basedon these data, total lab costs were $23 827 and $13 792for the rosiglitazone-plus-glipizide and glipizide groups,respectively. The respective costs of lab tests per studygroup member totaled $207 and $125, with exposure-adjustedcosts of $10 and $7 PPPM.
Following ADA recommendations,13 all patients wereencouraged to self-monitor their blood glucose. It wasexpected that, on average, patients would perform self-monitoringonce a day. The cost of self-monitoring wasassumed to be $0.75 per day based on average retailprice of the meter and teststrips.16 The total cost of self-monitoringwas $55 582 and$46 455 in the rosiglitazone-plus-glipizide and glipizidegroups, respectively, or $483and $422 per study groupmember. The exposure-adjustedcost was $23 PPPM forboth groups.
An overall total treatmentcost was calculated for eachstudy group subject. Theoverall cost was $480 and$645 PPPM for the rosiglitazone-plus-glipizide and glipizidegroups, respectively. Asummary of total healthcarecosts by resource category,expressed as cost PPPM, isprovided in Table 3.
The distribution of costestimates was skewed in therosiglitazone-plus-glipizideand glipizide groups. Toevaluate the impact ofskewedness on cost estimates,a sensitivity analysiswas performed by excludingpatients with costs more than2 standard deviations aboveor below the mean. A total of10 outliers were identified: 6in the rosiglitazone-plus-glipizidegroup (PPPM cost range$1294 to $3297) and 4 in the glipizide group (PPPM costrange $3734 to $9850). PPPM costs generated after theexclusion of outliers were $418 and $588 for the rosiglitazone-plus-glipizide and glipizide groups, respectively,resulting in PPPM cost savings of $170 in the combination-therapy group. The difference in cost PPPMbetween the 2 study groups before excluding the outlierswas $165 [AU: 645 -480 = 165] ($480 vs $645 costPPPM in the rosiglitazone-plus-glipizide and glipizidegroups, respectively). This indicates that the presenceof outliers, while increasing PPPM values, did not have asignificant impact on relative cost estimates and distribution.
Prospectively collected economic data demonstratingthe effects of improved glycemic control are limited.This analysis of economic outcomes in the clinical-trialsetting establishes a direction of effect and contributesto the ongoing efforts to show economic benefits ofimproved glycemic control.
The RESULT study is the first trial to prospectivelyassess resource utilization and estimate cost of care inolder patients with type 2 diabetes treated with a thiazolidinedione,the newest and most expensive class ofantidiabetic agents. Specifically, early addition of rosiglitazoneto a sulfonylurea was compared with maximaldose titration of an inexpensive sulfonylurea. Patientswere systematically and prospectively assessed for hospitalizations,ED visits, and physician office visits, aswell as bed days and restricted activity days. A standardunit-cost method, which had been previously applied toestimate costs associated with prevention and treatmentof type 2 diabetes,14 was used to compare budgetimplications of each therapy. The method provides acommon metric to sum the value of different categoriesof resources using approximate unit costs.
The results of the study indicate that despite greatermedication costs, rosiglitazone-plus-glipizide therapyprovided better clinical efficacy and reduced totalhealthcare resource utilization. The differences in favorof the rosiglitazone-plus-glipizide combination therapyappeared to be driven by decreased resource utilizationin terms of hospitalizations and ED visits.
The lower utilization patterns for healthcareresources in the rosiglitazone-plus-glipizide groupreflect significant economic benefits of the combinationtherapy. Rates of hospitalization and ED visits per 1000patient days in the rosiglitazone-plus-glipizide group(0.37 + 1.074 vs 0.59 + 1.324, respectively), comparedwith those in the glipizide group (0.76 + 1.816 vs 1.47 +3.905, respectively), clearly support the hypothesis thatthe combination therapy is associated with less intensiveuse of healthcare resources. In addition to thelower hospitalization rate in rosiglitazone-plus-glipizidegroup, the average length of stay per hospitalization inthat group was significantly shorter than that in the progressiveup-titrated glipizide group.
With clear indications of less intensive resource usein the rosiglitazone-plus-glipizide group, service costswere identified and assigned to the unique events withinthe trial to determine budgetary implications andcompare the economic outcomes of both therapies. Thefavorable budget impact of rosiglitazone-plus-glipizidetherapy due to lower resource utilization also was supportedby the lower cost of services used in rosiglita-zone-plus-glipizidegroup compared with the groupreceiving glipizide monotherapy. Hospitalization, ED,and physician office visit costs per person in rosiglitazone-plus-glipizide and glipizide groups ($2710 vs$6770, $75 vs $146, and $808 vs $766, respectively)suggest that the cost of healthcare services utilized waslower in the combination-therapy cohort, resulting intangible savings for healthcare decision makers andpayers, while ensuring quality and efficacious care forpatients with diabetes. Adjusted for duration of therapy,lower PPPM hospitalization costs, ED visit costs, andphysician office visit costs in the rosiglitazone-plus-glipizide,compared with the glipizide group, indicatedecrease in the level of consumption of healthcareresources in the combination-therapy group. From theperspective of a healthcare decision maker, these differencesdemonstrate potential cost savings resulting fromreduced healthcare utilization and support the earlyaddition of a thiazolidinedione to sulfonylurea therapy.
It is worth noting that, although the focus of thisanalysis was the total direct treatment costs, combinationtherapy with rosiglitazone and a sulfonylurea maybe associated with improvements in productivity, basedon the significant differences in self-reported bed days(0.85 vs 3.03 per 1000 patient days; = .0002). Applyinga published rate for time lost from doing usual activitydue to morbidity or mortality,14 the estimated per-patientcost of bed days was $65 and $202 in the rosiglitazone-plus-glipizide and glipizide groups, respectively.The total indirect-cost estimates based on patient-reportedbed days and restricted activity days were$1321 and $1722 per patient or $62 and $92 PPPM inthe rosiglitazone-plus-glipizide and glipizide groups,respectively. Therefore, the combination therapy maybe able to decrease not only the direct costs, but alsothe indirect costs associated with type 2 diabetes.Although the differences in healthcare resource utilizationcannot be directly attributed to the experimentaltreatments, the trial results demonstrate a strongassociation between rosiglitazone-plus-glipizide therapyand lower healthcare utilization. Lower use of healthcareservices and decreased cost of care have been reported tobe associated with improved glycemic control,17,18 whichwas achieved by the rosiglitazone-plus-glipizide therapy.The early intervention with rosiglitazone to optimizelong-term glycemic control was associated with substantialclinical and economic benefits and may presenta more efficient alternative to the traditional dose escalationof a sulfonylurea.
The study limitations derive from the nature of self-reportedhealthcare utilization data. These data are subjectto recall bias and were not independentlyreconciled with medical-record data because no subjects'medical records were available to verify the reasonsfor hospital admissions, ED visits, or physicianoffice visits in either of the study arms.
This 2-year prospective study in older patients withtype 2 diabetes demonstrated that treatment with acombination of rosiglitazone and a sulfonylurea mayprovide a health economic benefit through use of fewerhealthcare resources compared with progressive dosetitration of sulfonylurea monotherapy, even afteraccounting for the increase in treatment costs. In clinicalpractice, the earlier addition of rosiglitazone to a sulfonylureain a patient with type 2 diabetes maysubstantially reduce the costs associated with managingthe disease, in addition to providing improved glycemiccontrol. Economic evaluations of therapeutic approachesand health policy should consider costs and sideeffects of treatment, and changes in health outcomesresulting from therapy.
From the Michigan Diabetes Research and Training Center, Ann Arbor, Mich (WHH);and GlaxoSmithKline, King of Prussia, Pa (RGD, RH, MCO, BK, MAH, AB, MIF).
This study was funded by GlaxoSmithKline.
Address correspondence to: William H. Herman, MD, MPH, Michigan DiabetesResearch and Training Center, 3920 Taubman Center, Ann Arbor, MI 48109-0354. E-mail:firstname.lastname@example.org.
1. Wallace JI. Management of diabetes in the elderly. 1999;17(1):19-25.
2. American Diabetes Association. Economic costs of diabetes in the US in 2002.Report. 2003;26:917-932.
3. Centers for Disease Control and Prevention. National diabetes fact sheet: generalinformation and national estimates on diabetes in the United States, 2003.Rev ed. Atlanta, GA: US Department of Health and Human Services, Centers forDisease Control and Prevention; 2004.
4. King H, Aubert RE, Herman WH. Global burden of diabetes, 1995-2025: prevalence,numerical estimates, and projections. 1998;21:1414-1431.
5. Brandle M, Zhou H, Smith B, et al. The direct medical cost of type 2 diabetes.2003;26:2300-2304.
6. Engelgau MM, Geiss LS, Manninen DL, et al. Use of services by diabetespatients in managed care organizations. 1998;21:2062-2068.
7. Harris MI. Health care and health status and outcomes for patients with type 2diabetes. 2000;23:754-758.
8. O'Brien JA, Patrick AR, Caro J. Estimates of direct medical costs for microvascularand macrovascular complications resulting from type 2 diabetes mellitus in theUnited States in 2000. 2003;25:1017-1038.
Am J Manag Care.
9. CDC Diabetes in Managed Care Work Group. Diabetes mellitus in managedcare: complications and resource utilization. 2001;7:501-508.
10. Caro JJ, Ward AJ, O'Brien JA. Lifetime costs of complications resulting fromtype 2 diabetes in the US. 2003;25:476-481.
Am J Health Syst Pharm.
11. White JR. Economic considerations in treating patients with type 2 diabetesmellitus. 2002;59(suppl 9):S14-S17.
12. American Diabetes Association. Implications of the United KingdomProspective Diabetes Study. 1998;21:2180-2184.
13. American Diabetes Association. Standards of medical care for patients withdiabetes mellitus. 2002;25:S33-S49.
14. Herman WH, Brandle M, Zhang P, et al. Costs associated with the primaryprevention of type 2 diabetes mellitus in the diabetes prevention program. 2003;26:36-47.
15. Texas Medicaid fee schedule information. Report D (PRCR403C). NHIC, Austin, Tex. September 2002. http://www.tmhp.com/default.aspx. AccessedSeptember 25, 2003.
16. Rite Aid's Online Weekly Circular. Available at: http://www.riteaid.com/weekly_ad. Accessed September 25, 2003.
17. Wagner EH, Sandhu N, Newton KM, McCulloch DK, Ramsey SD, GrothausLC. Effect of improved glycemic control on health care costs and utilization. 2001;285:182-189.
18. Gilmer TP, O'Connor PJ, Manning WG, Rush WA. The cost to health plans ofpoor glycemic control. 1997;20:1847-1853.