Proactive Case Management of High-risk Patients With Type 2 Diabetes Mellitus by a Clinical Pharmacist: A Randomized Controlled Trial

, , , , ,
The American Journal of Managed Care, April 2005, Volume 11, Issue 4

Objective: To evaluate the effect of case management by a clinicalpharmacist on glycemic control and preventive measures inpatients with type 2 diabetes mellitus.

Study Design: Randomized controlled trial in a university-affiliatedprimary care internal medicine clinic.

Methods: We recruited 80 patients with poorly controlled type2 diabetes mellitus. A clinical pharmacist provided evaluation andmodification of pharmacotherapy, self-management diabetes education,and reinforcement of diabetes complications screeningprocesses through clinic visits and telephone follow-up. The mainclinical outcome was hemoglobin A1c (HbA1c) level; process measuresincluded HbA1c and low-density lipoprotein measurement,retinal examination, urine microalbumin testing (or use ofangiotensin-converting enzyme inhibitors), and monofilamentscreening for diabetic neuropathy.

P

P

P

Results: Patients in the intervention and control groups weresimilar in age, sex, mean HbA1c levels (10.1% and 10.2%, respectively;= .65), and current treatment regimens at baseline.Patients who received case management by the clinical pharmacistachieved greater reduction in HbA1c levels than those in the controlgroup (2.1% vs 0.9%, = .03). Three of the 5 process measureswere conducted more frequently in the intervention groupthan the control group, including low-density lipoprotein measurement(100.0% vs 85.7%, = .02), retinal examination (97.3% vs74.3%), and monofilament foot screening (92.3% vs 62.9%).

Conclusions: Proactive diabetes case management by a pharmacistsubstantially improved glycemic control and diabetesprocess-of-care measures. This approach, integrated with and basedin the primary care setting, was an effective and efficient approachto improving care, especially for those with poor glycemic controlat baseline.

(Am J Manag Care. 2005;11:253-260)

Type 2 diabetes mellitus is a highly prevalent conditionthat has substantial associated morbidity,mortality, and healthcare costs. Diabetes mellitusaffects approximately 18 million Americans,1 andthese numbers are expected to rise during the next severaldecades because of the increasing age and prevalenceof obesity in the US population.1,2 People withdiabetes mellitus die of cardiovascular disease at 2 to 3times the rate of age-and sex-matched control subjects,3-7 and the microvascular complications of diabetesmellitus make it the leading cause of preventableblindness, renal disease, and amputation in industrializednations.8-10 These complications have dramaticimplications for healthcare costs; indeed, diabetes mellitusled to about $132 billion in direct and indirectcosts in 2002 in the United States,11 and the cost tohealthcare plans is at least 2 to 3 times that of age-matchedpatients without diabetes mellitus.12,13

A substantial amount of diabetes-related morbiditycan be prevented through increased use of current managementguidelines and application and increased monitoringof available therapies, ranging from intensiveblood pressure14-16 and glycemic17-19 control to the useof preventive measures such as retinal screening.20-22Although there are effective therapies to prevent diabetescomplications, it is also clear that there are severalbarriers to providing optimal diabetes care in theprimary care setting.23-25 The complex and variednature of diabetes-related interventions makes it difficultto achieve multiple goals in limited visits, particularlyin the current environment in which providers areincreasingly allotted less time with patients. Furthermore,many diabetes treatment goals are heavilydependent on patient self-management and lifestylechanges. Therefore, providers are often left with the difficulttask of trying to manage a complex multisystemand patient-centered disease in a health system that ispoorly designed to manage chronic diseases that requirefrequent, intensive follow-up and detailed patient counselingand education.

Case management is one of the most commonly suggestedmethods to improve chronic disease management.26,27 Case management has been defined as a collaborativeprocess that assesses, plans, implements,coordinates, monitors, and evaluates the options andservices required to meet an individual's health needs,using communications and available resources to promotequality, cost-effective outcomes.28 Case managementcan be implemented in different forms, includingusing nurses (nurse practitioners or certified nurse specialists),physician assistants, social workers, and pharmacists,and the nature of the intervention can rangefrom autonomous medication management (in the caseof pharmacists or primary care personnel with prescribingprivileges) to telephone-based counseling oralgorithm-based disease management recommendationsto primary care providers. However, there hasbeen little rigorous evaluation of case management forpatients with diabetes mellitus, and results have beenvariable.29,30 Indeed, a recent randomized trial found nobenefit of case management in a high-risk population ofpatients with diabetes mellitus, raising concerns aboutthe optimal structure and implementation of case managementsystems.31,32

In particular, pharmacist-based case managementhas not been rigorously evaluated among patients withdiabetes mellitus, although results of short-term trialsand observational studies have been encouraging.33-36Pharmacist-based case management has, however, beenproven to be successful for management of congestiveheart failure and anticoagulation therapy.37-39 In thecase of diabetes, the complex nature of medicationmanagement would seem an ideal opportunity for usinga clinical pharmacist as a case manager. Pharmacistsoffer particular strengths in these situations, given thefrequent polypharmacy among patients with diabetesmellitus, risks of adverse drug events, and drug interactions.Therefore, we conducted a randomized controlledtrial of the effectiveness of a pharmacist-based casemanagement intervention in a general internal medicineclinic setting.

METHODS

Study Design and Setting

We conducted a randomized controlled trial of a diabetesquality improvement intervention using a clinicalpharmacist (HMC) to assist primary care providers inthe management of patients with type 2 diabetes mellitus.The study was conducted at a university-affiliatedambulatory care clinic that has 10 primary careinternists as its primary care staff, all of whom agreed tohave their patients contacted for possible participationin the study. The study design was intended to reflect,as much as possible, real-world effectiveness and tominimize the influence of the clinical pharmacist onusual practice. Therefore, the control group was kept asa natural control; that is, they received only regularcare, including regular follow-up visits with their primarycare physicians. The control group received nospecial contact during the intervention and did not haveexit interviews or process measurements at the end ofthe study. The study was approved by the University ofMichigan Medical School Institutional Review Board forHuman Subjects Research.

Patients and Randomization

P

We based our sample size calculations on theassumption that the effect size of the intervention wouldbe a change in hemoglobin A1c (HbA1c) level of 1.1%, thedifference found in a prior study30 of case management.An initial exploratory analysis of laboratory data at theintervention site showed that high-risk patients withdiabetes mellitus (ie, HbA1c levels, &#8805;8.0%) had a mean+ SD HbA1c level of 10.1% + 1.7%. Based on these numbers,we calculated that we would need 38 patients ineach study arm to have 80% power to detect a 1.1% differencein HbA level1c with a 2-tailed <.05.

International Classification of

Diseases, Ninth Revision, Clinical Modification

We identified 454 patients with diabetes mellitus atthe study site, using diagnosiscodes. Patients who were eligible for study enrollmentwere the 80 high-risk individuals whose mostrecent HbA1c levels were 8.0% or greater. We selectedthis group of patients because they are at particularlyhigh risk for adverse diabetes outcomes and are likely toreceive the most benefit from intensified glucose controland from preventive interventions such as eyescreening.19,20,40,41 Patients were excluded from thestudy if they had type 1 diabetes mellitus (based ondiagnosis before age 30 years), if they were older than70 years, or if they were diagnosed as having cancer,renal failure, severe cirrhosis, malignant hypertension,or a severe concurrent illness that would substantiallylimit life expectancy or require extensive systemictreatment. The trial flow diagram is shown in theFigure. Recruitment was coordinated through and withthe approval of the primary care physicians; this led toessentially no refusal to participate, so that the full sampleof eligible patients was enrolled, although severalpatients in each study arm dropped out as the studyprogressed. Eighty patients were enrolled and randomizedto the control group or the intervention group.To ensure balanced randomization across levels ofglycemic control, patients were stratified into 4 groupsbased on the baseline HbA1c levels (8.0%-8.9%, 9.0%-9.9%, 10.0%-10.9%, and &#8805;11.0%) and were randomizedto the interventiongroup or controlgroup within eachstratified group.42Randomizationwithin each stratumwas simple: becausethe study was small,randomization wasdone by hand,drawing numbersfrom a containerthat included "0"for the controlgroup or "1" for theinterventiongroup.43 Given thenature of the intervention,patients,providers, and thecase manager werenot blinded to theintervention.

Outcome Measures

The primary outcomemeasure forthis study waschange in HbA1c level. The HbA1c level (referencerange, 3.8%-6.4%)was measured usinga high-performanceliquid chromatographymachine (Tosoh Corporation,Montgomeryville,Pa). Because thisproject was designedto measurereal-world effectiveness,there were nospecific exit interviews or follow-up laboratory testingmandated by the protocol; rather, the interactionbetween the clinical pharmacist and the primary carephysicians determined the frequency and timing of laboratorymeasures and interventions, based on their clinicaljudgment of patient needs. Therefore, we used thefirst HbA1c level measured after the 12-month interventionas the primary outcome measure. Patients wereallowed up to 24 months after enrollment to obtain thefinal HbA1c measurement.

Secondary outcome measures included rates of diabetesprocess measures, including low-density lipoproteinmeasurement, dilated retinal examination, urinemicroalbumin screening (or use of angiotensin-convertingenzyme inhibitors), and monofilament testing fordiabetic neuropathy. Performance of these tests withinthe 2-year time frame of the study was determined bymedical chart review by a single reviewer, using a standardizedabstraction form. The rate of HbA1c measurementwas also used as a process measure, althoughHbA1c level was assessed at 1 year because it was theprimary study outcome and measurement was dictatedby protocol in the intervention group after the 1-yearintervention.

Intervention

The intervention was conducted in a university-affiliatedprimary care clinic. The case manager was a clinicalpharmacist who was already established as apharmacotherapy consultant at the clinic before thestart of the intervention. It was hypothesized that herestablished relationship with the primary care physiciansand some of the patients would foster a level ofcooperation and trust that would allow the interventionto be a collaborative process. Of note, there is no on-sitediabetes education program in this clinical setting,although referral-based programs are available.

The clinical pharmacist evaluated patients' therapeuticregimens based on efficacy, safety, adverseeffects, drug interactions, drug costs, and monitoring.All therapeutic recommendations were discussed withthe primary care physicians before significant therapyalterations. The clinical pharmacist followed up on diseasemanagement and medication management protocolsapproved by the primary care physicians. Theprotocols were based on those used in a prior study30but were updated to account for new pharmacologicagents (ie, thiazolidinediones) and were intended toprovide guidance rather than to serve as inflexible protocols.This allowed the clinical pharmacist to havesome autonomy in decision making and to use herclinical knowledge and pharmacology expertise to carefor the patients. However, this approach to case managementdiffers from that in prior investigations inthat it also involved brief face-to-face consultationsbetween the pharmacist and the primary care physicians,creating a team-based approach to managementrather than having a centralized case managerunder the direction of 1 or 2 supervising specialists.Because the clinical pharmacist was on site as an activemember of the healthcare team, the primary care physicianshad access to discuss nonintervention cases; thismay have caused some cross-contamination in thegroups, although this would bias the study results towardthe null.

Intervention patients had an initial clinic visit withthe clinical pharmacist that lasted approximately 1hour. During this visit, in addition to an initial assessmentof medication management, the clinical pharmacistprovided patients with basic education regardingdiabetes self-management skills. This included anemphasis on the importance of self-care, medications,and screening processes.Subsequent visits werescheduled based on therapeuticalterations and educationalneeds. Generally,the clinical pharmacist contactedthe patients by telephoneon a monthly basis,unless more frequent assessmentor recommendationswere needed, and saw thepatients in conjunction withtheir routine primary carevisits. The clinical pharmacistalso periodicallyreviewed the status of allintervention patients andprovided condensed "diabetesstatus updates" toproviders using a standardizedform (available fromthe author).

Statistical Analysis

For the primary outcomemeasure, HbA1c level, wecompared the baseline levels, final levels, and change scores (calculated by subtractingthe final from the baseline HbA1c levels) usingWilcoxon rank sum tests. The change in HbA1c levelswas also evaluated using linear regression analysis.Because analysis of residuals demonstrated substantialheteroskedasticity, we used Huber-White estimators ofstandard errors.44,45 In these regression analyses, weused the final HbA1c level as the dependent variable;independent variables included an indicator variable forthe intervention and the baseline HbA1c level. We alsoconducted analyses controlling for differences in follow-upintervals (because of the nature of the controlgroup). Last, we tested an interaction term between thebaseline HbA1c level and the intervention to assesswhether the response to the intervention varied basedon the initial level of glycemic control.

P

All analyses were performed based on intention totreat and at a significance level of < .05. In the primaryanalyses, missing data were assumed to be missingat random. Notably, the only missing final resultswere among unavailable patients who dropped out ofthe trial because of loss to follow-up, transfer, or death.However, we also conducted sensitivity analyses for missingfinal HbA1c values by imputing these values based onthe baseline values and repeating the analyses. As anadditional sensitivity analysis, we repeated the analyses,assuming that, for those with missing final HbA1c data,there was no change in initial HbA1c values. All dataanalyses were conducted using Stata version 7.0 (StataCorporation, College Station, Tex).

RESULTS

The study sample consisted of 41 patients randomizedto the intervention group and 39 patients randomizedto the control group. The baseline characteristicsof the intervention and control groups are shown inTable 1. The mean baseline HbA1c level in the interventiongroup was 10.1% and in the control group was10.2%. There were no statistically significant differencesin the distribution of demographic characteristicsand medication use between the groups. Nearly allpatients had health insurance, mostly through privatethird-party payers.

P

The primary outcome measure of the study wasHbA1c level. Because of the nature of the control group,the follow-up interval for obtaining the final HbA1c measurement was slightly shorter in the interventiongroup than the control group (13.6 vs 14.9 months,= .046). Dropout rates were also slightly lower in theintervention group than the control group. In the interventiongroup, no patients died, 3 were lost to follow-up,and 2 transferred to a different healthcare provider. Inthe control group, 1 patient died, 6 were lost to followup,and 3 transferred to a different healthcare provider.

P

P

P

P

P

The effect of the clinical pharmacist-based managementon change in HbA1c levels is shown in Table 2. Themean decrease in HbA1c levels in the intervention groupwas -2.1% (from 10.1%-8.0%) and in the control groupwas -0.9% (from 10.2%-9.3%). Therefore, the mean differencein HbA1c change scores between the interventionand control groups was 1.2% (= .03), and the mean differencein final HbA1c values was 1.3% (= .01). Thesedifferences persisted after adjusting for differences induration of follow-up between the intervention and controlgroups using linear regression analysis (mean differencein HbA1c change scores between groups, 1.1%; =.04). Likewise, greater improvement among interventionpatients was still observed when we imputed missingfinal HbA1c values for patients who were lost to follow-up(mean difference in HbA1c change scores betweengroups, 1.0%; = .04) or when it was assumed that therewas no change in HbA1c level among those who were lostto follow-up (mean difference in HbA1c change scoresbetween groups, 1.2%; = .01).

P

We found a strong statistical interaction between theintervention and baseline HbA1c levels (< .001), suggestingthat patients with higher HbA1c levels at enrollmenthad a greater improvement in glycemic controlthan those with more moderate elevations. The meanchanges in HbA1c levels, stratified by the baseline HbA1c values, are shown in Table 3. These findings demonstratethat those with poor glycemic control at baselinereceived most of the benefit of the intervention; indeed,patients who had only moderate elevations in HbA1c levelshad minimal response. For example, patients withinitial HbA1c levels of 8.0% received virtually no benefitfrom the intervention, while those with initial HbA1c levelsof 10.0% had about a 1.5% decrease in HbA1c levelsattributed to the intervention (a decrease of 2.1% in theintervention group and 0.6% in the control group).

P

P

P

The effect of the clinical pharmacist's managementon rates of diabetes process-of-care measures is shownin Table 4. Low-density lipoprotein measurement(100.0% vs 85.7%, = .02), retinal examination within 2years (97.3% vs 74.3%, = .004), and documentedmonofilament examination for neuropathy (92.3% vs62.9%, = .002) occurred more frequently among thosein the intervention group compared with the controlgroup. Rates of HbA1c measurement and microalbuminuriascreening (or use of angiotensin-converting enzymeinhibitors) were not different between the 2 groups,although the trends favored the intervention patients.

DISCUSSION

Case management for chronic diseases such as diabetesmellitus is being recommended and adopted at arapid rate. Although there are clear needs for improveddisease management to improve outcomes of care, andpossibly to decrease costs related to diabetes-relatedcomplications, there is to date little rigorous evaluationof the effectiveness of disease management programsin real-world settings.

Indeed, in the case of diabetes, rigorous evaluationof case management has been limited primarily to theuse of nurses as case managers. One randomizedstudy29 reported modest improvements in glycemiccontrol; this study was designed to be a low-intensityintervention, with the nurse case manager providingpatient education and facilitating compliance throughtelephone contacts. A second randomized study30found that more intensive case management led to asignificant improvement in glycemic control. In thisstudy, the nurse case manager had direct managementresponsibilities, including medication managementunder a protocol. However, the study waslimited by substantial differential dropout, with manymore patients lost to follow-up in the intervention armthan the control arm; therefore, the results may havebeen biased in favor of the intervention. A third investigation,and the most rigorously designed, found thatnurse practitioner-based case management of patientswith diabetes mellitus had no effect on any clinical outcomes,although patient satisfaction was improved.31,32

Other models of case management have not beenwell evaluated among patients with diabetes mellitus.Our study demonstrates that a pharmacist acting as acase manager can improve glycemic control and the useof recommended screening procedures among high-riskpatients with type 2 diabetes mellitus. The groupassigned to case management by a pharmacist achieveda 1.2% greater reduction in HbA1c levels than the controlgroup and had 15% to 30% improvements in the percentagesof patients completing recommended screening.Based on findings of the Diabetes Control andComplications Trial research group and the UKProspective Diabetes Study group, this improvement inHbA1c levels would be expectedto lead to 40% to 50% relativereductions in the risk of intermediateand advanced microvascularcomplications, althoughthis level of improvement wouldneed to be sustained for a longperiod before substantial improvementsin outcomes wouldbe seen.18,19,46,47 Interestingly,we found that the interventionwas mainly effective for thosewith poor baseline glycemiccontrol. This finding that it ismuch easier in real-world practiceto achieve moderate thantight glycemic control has been seen in other settingsas well.33,48 Still, our results imply that case managementinterventions could be successfully targeted atthe highest-risk patients, improving the efficiency ofthese interventions dramatically.

This intervention differs substantially in its structurefrom that of many other case management interventions.The intervention was based on management by aclinical pharmacist rather than a nurse. Although therehave been some nonrandomized or short-term studies34-36 of pharmacists as case managers of diabetic patients,the literature on this is limited. However, pharmacistscan play a critical role in the management of patientswith chronic diseases, such as congestive heart failureand anticoagulation, that require multiple medicationsor frequent medication adjustment,37-39 and we haveextended these findings to patients with type 2 diabetesmellitus. Another key difference between this and othermanagement interventions30 is that this was not a centralizedor referral-based program. Rather, the clinicalpharmacist was integrated directly into the clinic, whereface-to-face contact with patients and providers was possible.In informal discussions, providers indicated thatthey were satisfied with this type of intervention becausethere was a level of trust and interaction that they didnot perceive in previous centralized disease managementinvestigations. The increased level of trust led to agreat deal of autonomy for the clinical pharmacist, whowas thus able to function more independently and tomake medication adjustments (eg, dosages of insulin ororal hypoglycemic agents) in a more timely manner. Inaddition, the pharmacist provided direct feedback ondiabetes status to patients as a formal communication,and the ability to have informal contact to discuss casesprovides an opportunity for collaborative care that is lesslikely to occur in centralized programs.

There are several limitations to this study. This wasa small study involving only 80 patients at a single suburbanuniversity-affiliated clinic. The results of thestudy may not be generalizable to other settings orpatient populations. In fact, there is reason to believethat specific aspects of a case management interventionand the patient population are critical factors inwhether case management will be effective.31 Inaddition, we conducted the study as a qualityimprovement project aimed at measuring effectivenessin a real-world clinical setting rather than as astrict protocol-driven randomized trial of efficacy. Forexample, we did not insist on exit measurement ofHbA1c levels at a preset time; rather, we obtainedresults at the time of regularly scheduled follow-upvisits and only if the primary providers ordered theappropriate test. This presents some strengths in thatthe control group represents the underlying patientpopulation more accurately, but it introduced somesmall differences in duration of follow-up between treatmentarms. However, we controlled for differences induration of follow-up using regression models and stillfound that the case management intervention waseffective. We also do not have information onresource use at present.

Although our intervention was successful, there areissues that need to be addressed on a broader scalebefore it can be disseminated into general practice.First, a pharmacist is likely to be more costly than anurse in providing this type of intervention, and clinicalpharmacists are often in short supply. However,some of the costs of pharmacist care may be offset byreductions in diabetes complications. Second, this wasnot a centralized case management system, and disseminatingit on a broad scale would generate substantialadditional resource requirements. The variance inresponse to the intervention based on initial HbA1c levels,however, indicates that the intervention may bestbe targeted to limited patient populations at high risk.Perhaps the best solution would be to integrate a pharmacistcase manager into clinics with high-risk patientsand to extend the activities of pharmacists who alreadyhave clinical duties such as monitoring and adjustinganticoagulation therapy.

This study shows that a pharmacist case managerintegrated into a suburban primary care clinic can substantiallyimprove glycemic control and adherence toscreening recommendations among high-risk patientswith type 2 diabetes mellitus. This offers a somewhat differentmodel of case management than other modelsthat have been tested and one that may offer solutions tosome of the concerns (such as fragmentation of care andlack of primary care physician involvement in decisionmaking) that primary care physicians have expressedabout current models of centralized care.49 The differencesbetween various case management strategies needto be more rigorously identified before wide-scale adoption;however, our study offers an effective and efficientapproach toward improving conventional diabetes care.

Acknowledgments

We would like to acknowledge Mary Beth Poole and Megan Brewer,PharmD, for their assistance with data collection.

From the College of Pharmacy (HMC) and Department of Internal Medicine (SM, DD,RAH, SLK, SV), University of Michigan; Ann Arbor Veterans Affairs Center for PracticeManagement and Outcomes Research (RAH, SLK, SV); and Michigan Diabetes Researchand Training Center (RAH, SLK, SV); Ann Arbor, Mich.

Dr Vijan was a Veterans Affairs Health Services Research and Development CareerDevelopment Awardee when this work was conducted. Funding for the clinical pharmacistwas provided by the University of Michigan College of Pharmacy.

Address correspondence to: Sandeep Vijan, MD, MS, Ann Arbor Veterans AffairsCenter for Practice Management and Outcomes Research, 2215 Fuller Road, Mailstop 11H,Ann Arbor, MI 48105. E-mail: svijan@umich.edu.

National Diabetes Fact Sheet:

General Information and National Estimates on Diabetes in the United States,

2003.

1. Centers for Disease Control and Prevention. Rev ed. Atlanta, Ga: Centers for Disease Control and Prevention, US Deptof Health and Human Services; 2004.

Diabetes Care.

2. Boyle JP, Honeycutt AA, Narayan KM, et al. Projection of diabetes burdenthrough 2050: impact of changing demography and disease prevalence in the US.2001;24:1936-1940.

Diabet Med.

3. 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 historicalcohort study in Dutch general practice. 1995;12:117-122.

Diabetes Care.

4. de Marco R, Locatelli F, Zoppini G, Verlato G, Bonora E, Muggeo M. Cause-specificmortality in type 2 diabetes: the Verona Diabetes Study. 1999;22:756-761.

Diabetes.

5. Garcia MJ, McNamara PM, Gordon T, Kannel WB. Morbidity and mortality indiabetics in the Framingham population: sixteen year follow-up study. 1974;23:105-111.

JAMA.

6. Kannel WB, McGee DL. Diabetes and cardiovascular disease: the FraminghamStudy. 1979;241:2035-2038.

Diabetologia.

7. Krolewski AS, Czyczyk A, Janeczko D, Kopczynski J. Mortality from cardiovasculardiseases among diabetics. 1977;13:345-350.

Diabetes in

America.

8. National Institute of Diabetes and Digestive and Kidney Diseases. 2nd ed. Bethesda, Md: National Institute of Diabetes and Digestive andKidney Diseases, National Institutes of Health; 1995.

J Gen Intern Med.

9. Vijan S, Stevens DL, Herman WH, Funnell MM, Standiford CJ. Screening, prevention,counseling, and treatment for the complications of type II diabetes mellitus:putting evidence into practice. 1997;12:567-580.

Br J Ophthalmol.

10. Kocur I, Resnikoff S. Visual impairment and blindness in Europe and their prevention.2002;86:716-722.

Diabetes Care.

11. Hogan P, Dall T, Nikolov P; American Diabetes Association. Economic costsof diabetes in the US in 2002. 2003;26:917-932.

Diabetes Care.

12. Selby JV, Ray GT, Zhang D, Colby CJ. Excess costs of medical care for patientswith diabetes in a managed care population. 1997;20:1396-1402.

Pharmacoeconomics.

13. Ramsey SD, Newton K, Blough D, McCulloch DK, Sandhu N, Wagner EH.Patient-level estimates of the cost of complications in diabetes in a managed carepopulation. 1999;16:285-295.

Lancet.

14. Hansson L, Zanchetti A, Carruthers SG, et al; HOT Study Group. Effects ofintensive blood-pressure lowering and low-dose aspirin in patients with hypertension:principal results of the Hypertension Optimal Treatment (HOT) randomisedtrial. 1998;351:1755-1762.

BMJ.

BMJ.

15. UK Prospective Diabetes Study Group. Tight blood pressure control and riskof macrovascular and microvascular complications in type 2 diabetes: UKPDS 38[published correction appears in 1999;318:29]. 1998;317:703-713.

Ann

Intern Med.

16. Vijan S, Hayward RA. Treatment of hypertension in type 2 diabetes mellitus:blood pressure goals, choice of agents, and setting priorities in diabetes care. 2003;138:593-602.

N Engl J Med.

17. Diabetes Control and Complications Trial Research Group. The effect of intensivetreatment of diabetes on the development and progression of long-term complicationsin insulin-dependent diabetes mellitus. 1993;329:977-986.

BMJ.

Lancet.

18. UK Prospective Diabetes Study (UKPDS) Group. Intensive blood-glucose controlwith sulphonylureas or insulin compared with conventional treatment and riskof complications in patients with type 2 diabetes (UKPDS 33) [published correctionappears in 1999;318:29]. 1998;352:837-853.

Ann Intern Med.

19. Vijan S, Hofer TP, Hayward RA. Estimated benefits of glycemic control inmicrovascular complications in type 2 diabetes. 1997;127:788-795.

JAMA.

20. Vijan S, Hofer TP, Hayward RA. A cost-utility analysis of screening intervalsfor diabetic retinopathy. 2000;283:889-896.

Ann Intern Med.

21. Singer DE, Nathan DM, Fogel HA, Schachat AP. Screening for diabeticretinopathy. 1992;116:660-671.

Med Care.

22. Dasbach EJ, Fryback DG, Newcomb PA, et al. Cost-effectiveness of strategiesfor detecting diabetic retinopathy. 1991;29:20-39.

Ann Intern Med.

23. Von Korff M, Gruman J, Schaefer J, Curry SJ, Wagner EH. Collaborative managementof chronic illness. 1997;127:1097-1102.

Health Aff

(Millwood).

24. Wagner EH, Austin BT, Davis C, Hindmarsh M, Schaefer J, Bonomi A.Improving chronic illness care: translating evidence into action. 2001;20(6):64-78.

Health Aff (Millwood).

25. Vladeck BC. You can't get there from here: obstacles to improving care of thechronically ill. 2001;20(6):175-179.

Arch Intern

Med.

26. DeBusk RF, West JA, Miller NH, Taylor CB. Chronic disease management:treating the patient with disease(s) vs treating disease(s) in the patient. 1999;159:2739-2742.

Gen Hosp

Psychiatry.

27. Katon W, Von Korff M, Lin E, Simon G. Rethinking practitioner roles in chronicillness: the specialist, primary care physician, and the practice nurse. 2001;23:138-144.

CCM Certification Guide:

Certified Case Manager

28. Commission for Case Manager Certification. [abstract]. Rolling Meadows, Ill: Commission for CaseManager Certification; 1996.

J Gen Intern Med.

29. Weinberger M, Kirkman MS, Samsa GP, et al. A nurse-coordinated interventionfor primary care patients with non-insulin-dependent diabetes mellitus: impacton glycemic control and health-related quality of life. 1995;10:59-66.

Ann Intern Med.

30. Aubert RE, Herman WH, Waters J, et al. Nurse case management to improveglycemic control in diabetic patients in a health maintenance organization: a randomized,controlled trial. 1998;129:605-612.

Am J Med.

31. Krein SL, Klamerus ML, Vijan S, et al. Case management for patients withpoorly controlled diabetes: a randomized trial. 2004;116:732-739.

Am J Med.

32. Shojania KG, Grimshaw JM. Still no magic bullets: pursuing more rigorousresearch in quality improvement. 2004;116:778-780.

J Am Pharm Assoc (Wash).

33. Cranor CW, Bunting BA, Christensen DB. The Asheville Project: long-termclinical and economic outcomes of a community pharmacy diabetes care program.2003;43:173-184.

Ann Pharmacother.

34. Coast-Senior EA, Kroner BA, Kelley CL, Trilli LE. Management of patients withtype 2 diabetes by pharmacists in primary care clinics. 1998;32:636-641.

Am J Med Qual.

35. Davidson MB, Karlan VJ, Hair TL. Effect of a pharmacist-managed diabetescare program in a free medical clinic. 2000;15:137-142.

Ann Pharmacother.

36. Jaber LA, Halapy H, Fernet M, Tummalapalli S, Diwakaran H. Evaluation of apharmaceutical care model on diabetes management. 1996;30:238-243.

Arch Intern Med.

37. Chiquette E, Amato MG, Bussey HI. Comparison of an anticoagulation clinicwith usual medical care: anticoagulation control, patient outcomes, and health carecosts. 1998;158:1641-1647.

Arch Intern Med.

38. Gattis WA, Hasselblad V, Whellan DJ, O'Connor CM. Reduction in heart failureevents by the addition of a clinical pharmacist to the heart failure managementteam: results of the Pharmacist in Heart Failure Assessment Recommendation andMonitoring (PHARM) Study. 1999;159:1939-1945.

Pharmacotherapy.

39. Wilt VM, Gums JG, Ahmed OI, Moore LM. Outcome analysis of a pharmacist-managedanticoagulation service. 1995;15:732-739.

BMJ.

40. Adler AI, Stratton IM, Neil HA, et al. Association of systolic blood pressurewith macrovascular and microvascular complications of type 2 diabetes (UKPDS36): prospective observational study. 2000;321:412-419.

BMJ.

41. Stratton IM, Adler AI, Neil HA, et al. Association of glycaemia with macrovascularand microvascular complications of type 2 diabetes (UKPDS 35): prospectiveobservational study. 2000;321:405-412.

Designing Clinical Research.

42. Hulley SB, Feigal D, Martin M, Cummings SR. Randomization procedures forstudies of small or moderate sample size. In: Hulley SB, Cummings SR, eds.Baltimore, Md: Williams & Wilkins; 1988:210-211.

Designing Clinical Research.

43. Hulley SB, Feigal DE, Martin M, Cummings SR. Designing a new study, IV:experiments. In: Hulley SB, Cummings SR, eds. Baltimore, Md: Williams & Wilkins; 1988:110-127.

Anonymous Proceedings of the Fifth Berkeley Symposium in

Mathematical Statistics and Probability.

44. Huber PJ. The behavior of maximum likelihood estimates under non-standardconditions. In: Berkeley: University of California Press;1967:221-233.

J Econometrics.

45. MacKinnon JG, White H. Some heteroskedasticity consistent covariancematrix estimators with improved finite sample properties. 1985;29:305-325.

Diabetes Care.

46. Eastman RC, Javitt JC, Herman WH, et al. Model of complications of NIDDM,I: model construction and assumptions. 1997;20:725-734.

Diabetes Care.

47. Eastman RC, Javitt JC, Herman WH, et al. Model of complications of NIDDM,II: analysis of the health benefits and cost-effectiveness of treating NIDDM with thegoal of normoglycemia. 1997;20:735-744.

JAMA.

48. Hayward RA, Manning WG, Kaplan SH, Wagner EH, Greenfield S. Startinginsulin therapy in patients with type 2 diabetes: effectiveness, complications, andresource utilization. 1997;278:1663-1669.

J Gen Intern Med.

49. Fernandez A, Grumbach K, Vranizan K, Osmond DH, Bindman AB. Primarycare physicians' experience with disease management programs. 2001;16:163-167.