Diabetes is a chronic and highly prevalent condition that

Similar documents
Supplementary Online Content

Diabetes affects 29 million Americans, imposing a substantial

Impact of Pharmacist Intervention on Diabetes Patients in an Ambulatory Setting

Effect on Glycemic, Blood Pressure, and Lipid Control according to Education Types

key words: chronic obstructive pulmonary disease, beta agonists, Medicare, health care costs, health care utilization

Community. Profile Yellowstone County. Public Health and Safety Division

Community. Profile Lewis & Clark County. Public Health and Safety Division

Community. Profile Missoula County. Public Health and Safety Division

Community. Profile Powell County. Public Health and Safety Division

A Four-System Comparison of Patients With Chronic Illness: The Military Health System, Veterans Health Administration, Medicaid, and Commercial Plans

Community. Profile Big Horn County. Public Health and Safety Division

Community. Profile Anaconda- Deer Lodge County. Public Health and Safety Division

Metformin and breast cancer stage at diagnosis: a population-based study

T.S. Kurki a, *,U.Häkkinen b, J. Lauharanta c,j.rämö d, M. Leijala c

Community. Profile Carter County. Public Health and Safety Division

PNEUMOVAX 23 is recommended by the CDC for all your appropriate adult patients at increased risk for pneumococcal disease 1,2 :

Assessment of Depression in Multiple Sclerosis. Validity of Including Somatic Items on the Beck Depression Inventory II

Trends in antihypertensive and lipidlowering therapy in subjects with type II diabetes: clinical effectiveness or clinical discretion?

XII. HIV/AIDS. Knowledge about HIV Transmission and Misconceptions about HIV

Clinical Study Report Synopsis Drug Substance Naloxegol Study Code D3820C00018 Edition Number 1 Date 01 February 2013 EudraCT Number

Mental health symptoms and patient-reported diabetes symptom burden: implications for medication regimen changes

Metabolic Syndrome and Health-related Quality of Life in Obese Individuals Seeking Weight Reduction

Chronic obstructive pulmonary disease (COPD) is the third

Relationship Between Hospital Performance on a Patient Satisfaction Survey and Surgical Quality

Body mass index, waist-to-hip ratio, and metabolic syndrome as predictors of middle-aged men's health

Supplementary Online Content

CheckMate 153: Randomized Results of Continuous vs 1-Year Fixed-Duration Nivolumab in Patients With Advanced Non-Small Cell Lung Cancer

BMI and Mortality: Results From a National Longitudinal Study of Canadian Adults

Estimating the impact of the 2009 influenza A(H1N1) pandemic on mortality in the elderly in Navarre, Spain

Metabolic syndrome (MetS) is defined by a group

Analysis of alternatives for insulinizing patients to achieve glycemic control and avoid accompanying risks of hypoglycemia

Management and Outcomes of Binge-Eating Disorder in Adults: Current State of the Evidence

Prostate cancer is among the most common malignancies

Urinary Tract Infection in Men

Inhaled Corticosteroid Is Associated With an Increased Risk of TB in Patients With COPD

Risks for All-Cause Mortality: Stratified by Age, Estimated Glomerular Filtration Rate and Albuminuria

The Effects of Small Sized Rice Bowl on Carbohydrate Intake and Dietary Patterns in Women with Type 2 Diabetes

Urinary Tract Infection in Women

Invasive Pneumococcal Disease Quarterly Report. July September 2017

The Association of Gender With Quality of Health in Peripheral Arterial Disease Following Peripheral Vascular Intervention

Seasonal influenza vaccination programme country profile: Ireland

Opioid Use and Survival at the End of Life: A Survey of a Hospice Population

Comparison of claims data on hospitalization rates and repeat procedures in patients receiving a bowel preparation prior to colonoscopy

URINARY incontinence is an important and common

Clinical statistics analysis on the characteristics of pneumoconiosis of Chinese miner population

Epilepsy & Behavior 20 (2011) Contents lists available at ScienceDirect. Epilepsy & Behavior. journal homepage:

Division of Endocrinology and Metabolism, Department of Internal Medicine, The Catholic University of Korea College of Medicine, Seoul, 2

The burden of cirrhosis and impact of universal coverage public health care system in Thailand: Nationwide study

Implications of iron deficiency/anemia on the classification of diabetes using HbA1c

The RUTHERFORD-2 trial in heterozygous FH: Results and implications

Original Article INTRODUCTION. Korean Diabetes J 2010;34: doi: /kdj pissn eissn

Potential for Interactions Between Dietary Supplements and Prescription Medications a

Appendix J Environmental Justice Populations

A Comparison of Serum Magnesium Level in Pregnant Women with and without Gestational Diabetes Mellitus (GDM)

Effectiveness of Belt Positioning Booster Seats: An Updated Assessment

A series of recent studies and meta-analyses confirm

Influence of the Duration of Diabetes on the Outcome of a Diabetes Self-Management Education Program

Health Coaching: A Preliminary Report on the Effects in Traumatic Brain Injury/Polytrauma Patients

Relationship between serum irisin, glycemic indices, and renal function in type 2 diabetic patients

Ethnic Disparities in Stroke Recognition in Individuals with Prior Stroke

Trends in Mortality From COPD Among Adults in the United States

Efficacy of Pembrolizumab in Patients With Advanced Melanoma With Stable Brain Metastases at Baseline: A Pooled Retrospective Analysis

Satisfaction with care is an important, patient-centered

Increased Relative Mortality in Women With Severe Oxygen-Dependent COPD

Abstract. Background. Aim. Patients and Methods. Patients. Study Design

Anemia in pediatric hemodialysis patients: Results from the 2001 ESRD Clinical Performance Measures Project

Original Article. T Akter 1, N Islam 2, MA Hoque 3, S Khanam 4, HA khan 5, BK Saha 6. Abstract:

Among US women 65 years and older, the annual

Introduction. These patients benefit less from conventional chemotherapy than patients identified as MMR proficient or microsatellite stable 3-5

Lung cancer is the second-most common form of cancer

SYNOPSIS Final Abbreviated Clinical Study Report for Study CA ABBREVIATED REPORT

Factors influencing help seeking in mentally distressed young adults: a cross-sectional survey

In 2006, the prevalence of bipolar

Addendum to the Evidence Review Group Report on Aripiprazole for the treatment of schizophrenia in adolescents (aged years)

Single-Molecule Studies of Unlabelled Full-Length p53 Protein Binding to DNA

Emerging Options for Thromboprophylaxis After Orthopedic Surgery: A Review of Clinical Data

Fat intake in patients newly diagnosed with type 2 diabetes: a 4-year follow-up study in general practice

Employment Status and Depressive Symptoms in Koreans: Results From a Baseline Survey of the Korean Longitudinal Study of Aging

Health-Related Quality of Life and Symptoms of Depression in Extremely Obese Persons Seeking Bariatric Surgery

Nonpharmacologic Interventions for Treatment-Resistant Depression in Adults Executive Summary

Perforation Following Colorectal Endoscopy: What Happens Beyond the Endoscopy Suite?

Invasive Pneumococcal Disease Quarterly Report July September 2018

Between 45% and 85% of the approximately 4 million people

Comparison of three simple methods for the

Summary of Package Insert 1

Accuracy of a New Real-Time Continuous Glucose Monitoring Algorithm

Rising costs have led to a number of federal initiatives to

Effects of a 6-Month Exenatide Therapy on HbA1c and Weight in Korean Patients with Type 2 Diabetes: A Retrospective Cohort Study

A community-based comparison of trauma patient outcomes between D- and L-lactate fluids,

University of Cape Town

Positive Heparin-Platelet Factor 4 Antibody Complex and Cardiac Surgical Outcomes

Water fl uoridation and dental caries in 5- and 12-year-old children from Canterbury and Wellington

Different type 2 diabetes risk assessments predict dissimilar numbers at high risk :

Prevalence of Diabetes Mellitus and Prediabetes in Dalseong-gun, Daegu City, Korea

Summary. Effect evaluation of the Rehabilitation of Drug-Addicted Offenders Act (SOV)

Original Article. Diabetes Metab J 2011;35:26-33 doi: /dmj pissn eissn

Perioperative Hyperglycemia and Postoperative Infection after Lower Limb Arthroplasty

Rheumatoid arthritis (RA) is a chronic, progressive, and

Transcription:

Mediction Adherence nd Improved Outcomes Among Ptients With Type 2 Dibetes Srh E. Curtis, MPH; Kristin S. Boye, PhD; Mureen J. Lge, PhD; nd Luis-Emilio Grci-Perez, MD, PhD Dibetes is chronic nd highly prevlent condition tht my led to severe complictions, such s lower limb mputtions, blindness, kidney filure, stroke, hert disese, nd erly deth. 1 As of 2012, the estimted number of US dults (20 yers or older) dignosed with dibetes ws 28.9 million, or 12.3% of the popultion. 1 For these individuls, the totl cost of the disese in 2012 ws $245 billion, comprising $176 billion in direct medicl costs nd $69 billion in decresed productivity. 2 The mjority of ptients with dibetes (90% to 95%) hve type 2 dibetes (T2D), 1 the prevlence of which is growing so rpidly tht bout 40% of ll Americns re projected to be dignosed with the disese t some point in their dult lives. 3 Given the lrge nd incresing number of ptients with T2D nd the ssocited burden, severl recent studies hve exmined fctors tht my influence T2D tretment outcomes. One such fctor is dherence to prescribed glucose-lowering gents (GLAs). Improved dherence to GLAs hs been shown to be linked to reduction in hospitliztions nd/or emergency deprtment (ED) visits, 4-9 complictions, 6,10,11 nd costs. 5,9,12-15 However, limited reserch hs focused on ll clsses of GLAs, nd inclusion of the newer GLA clsses is even rrer. 5,7-9,11 The gol of the present study ws to further the literture exmining connections between ptient behvior nd T2D tretment outcomes. To this end, our retrospective, nturlistic investigtion used US clims dtbse to observe the medicl records ssocited with lrge popultion of Americns ged 18 to 64 yers with T2D. This study exmined the reltionships between GLA dherence nd ptient outcomes, including cute cre resource utiliztion, cute complictions, nd totl costs. ABSTRACT OBJECTIVES: Exmine the ssocition between dherence to glucose-lowering gents (GLAs) nd ptient outcomes in n dult type 2 dibetes (T2D) popultion. STUDY DESIGN: Retrospective nlysis. METHODS: Truven s Commercil Clims nd Encounters dtbse supplied dt from July 1, 2009, to June 30, 2014. Ptients 18 to 64 yers with T2D were included if they received GLA from July 1, 2010, through June 30, 2011. Multivrible nlyses exmined the reltionships mong 3-yer ptient outcomes nd dherence, defined s proportion of dys covered 80% or more. Outcomes included ll-cuse medicl costs, cute cre resource utiliztion, nd cute complictions. RESULTS: Although there ws no sttisticlly significnt difference in totl costs when compring dherent nd nondherent ptients ($38,633 vs $38,357; P =.0720), cute cre costs ($12,153 vs $8233; P <.0001) nd outptient costs ($16,964 vs $15,457; P <.0001) were significntly lower for dherent ptients. Adherence ws lso ssocited with lower probbility of hospitliztion (22.71% vs 17.65%; P <.0001) nd emergency deprtment (ED) visits (45.61% vs 38.47%; P <.0001), fewer hospitliztions (0.40 vs 0.27; P <.0001) nd ED visits (1.23 vs 0.83; P <.0001), nd shorter hospitl length of sty (2.16 vs 1.25 dys; P <.0001). Adherent ptients were lso less likely to be dignosed with n cute compliction in the 3-yer post period (12.54% vs 9.64%; P <.0001). CONCLUSIONS: Compred with nondherence, dherence to GLAs mong ptients with T2D ws ssocited with significnt reduction in cute cre costs nd resource utiliztion, outcomes tht my positively impct the welfre of ptients. METHODS Truven s Helth Anlytics MrketScn Commercil Clims nd Encounters dtbse, including dt from July 1, 2009, through June 30, 2014, ws used for this study. This dtbse consists of the helthcre records of millions of individuls who re covered Am J Mng Cre. 2017;23(7):e208-e214 e208 JULY 2017 www.jmc.com

Adherence nd Improved Outcomes by fully or prtilly cpitted fee-for-service helth plns. As such, the dtbse provides TAKEAWAY POINTS detiled costs, use, nd outcomes dt Compred with nondherent ptients, dherent ptients hd: for helthcre services performed in both Less likelihood of hospitliztion or n emergency deprtment (ED) visit. inptient nd outptient settings. Medicl Fewer hospitliztions nd fewer ED visits. clims re linked to outptient prescription Reduced cute cre costs nd outptient costs. drug clims nd person-level enrollment Incresed drug costs. Better ptient outcomes. informtion. The dt re fully deidentified No significnt difference in totl medicl costs. nd complint with the Helth Insurnce Portbility nd Accountbility Act. For inclusion in the study, ptients were required to hve received t lest 2 dignoses of T2D using were identified bsed on receipt of dignosis of hyperglycemi Interntionl Clssifiction of Diseses, Ninth Revision, Clinicl (ICD-9-CM code: 790.29), hypoglycemi (ICD-9-CM codes: 250.8, Modifiction (ICD-9-CM) codes 250.x0 or 250.x2 16,17 in the time 251.0, 251.1, or 251.2), or dibetic or hypoglycemic com (ICD-9-CM period of July 1, 2010, through June 30, 2011 (ie, the identifiction codes: 250.3, 251.0). window), nd to hve received t lest 1 prescription for GLA Multivrible nlyses were used to exmine the reltionship between outcomes of interest nd ptient dherence. All within the identifiction window; the dte of the first such prescription ws identified s the index dte. Additionlly, ptients nlyses controlled for the individul ptient s chrcteristics were required to be ged 18 to 64 yers s of the index dte nd to (ge, sex, region of residence, nd insurnce coverge), preperiod hve hd continuous insurnce coverge from 1 yer leding up to the index dte (the pre-period) through 3 yers following the index dte (the post period), s well s vlid ptient demogrphic FIGURE 1. Inclusion Exclusion Criteri nd Smple Size dt. Ptients were excluded from the nlyses if they received dignosis of type 1 dibetes (ICD-9-CM codes: 250.x1 or 250.x3) or Receipt of 2 dignoses of T2D pregnncy (ICD-9-CM codes: 630.xx-670.xx, V22.xx-V23.xx, V30. over identifiction window xx-v39.xx) t ny time from the strt of the pre-period through (n = 1,910,444) the end of the post period. Figure 1 illustrtes how ech of these criteri ffected smple size. Receipt of 1 prescriptions for GLA The nlyses focused on the reltionship between ptient over identifiction window (n = 1,158,151) dherence nd outcomes, where dherence ws proxied by the proportion of dys covered (PDC). The PDC is mesure of dherence used by both the Phrmcy Qulity Allince 18 nd CMS, nd No dignosis of T1D or pregnncy from strt of pre-period through end of post period it ws defined s the percentge of dys n individul received t (n = 907,067) lest 1 GLA in the first yer of the post period. For ll medictions except insulin, the PDC ws constructed using the dys supplied field provided in the dtbse. For insulin, this ws clculted Aged 18 to 64 yers s of index dte (n = 906,363) s the verge number of dys between fills for n insulin prescription. 19 Consistent with Helthcre Effectiveness Dt nd Informtion Set guidelines for the use of orl GLAs 20 nd with Continuous insurnce coverge from strt of previous reserch, ptients were ctegorized s dherent if they pre-period through end of post period chieved PDC threshold of t lest 80%. 6,8,21,22 (n = 236,340) Outcomes of interest included costs, cute cre resource utiliztion, nd cute complictions. Costs were subdivided into 4 Vlid ptient demogrphic dt ctegories cute cre (hospitliztion nd ED), outptient, drug, (n = 228,074) nd totl nd ll were converted to 2014 prices using the medicl component of the Consumer Price Index. In ll cses, costs were clculted using gross pyments to provider for service. Acute GLA indictes glucose-lowering gent; T1D, type 1 dibetes; T2D, type 2 dibetes. cre resource utiliztion ws defined s the probbility of hospitliztion or n ED visit, the number of hospitliztions or ED Index dte is defined s first receipt of GLA in identifiction window. Pre-period Identifiction window is defined s July 1, 2010, through June 20, 2011. is defined s 1 yer prior to index dte; post period is defined s index dte visits, nd the hospitl length of sty (LOS). Acute complictions through 3 yers post index dte. THE AMERICAN JOURNAL OF MANAGED CARE VOL. 23, NO. 7 e209

generl helth sttus, comorbidities, mediction use, nd providers. Generl helth ws proxied by the Chrlson Comorbidity Index (CCI), which is scored on scle of 0 to 33 bsed on the presence of comorbidities such s crdiovsculr diseses, dementi, chronic pulmonry disese, hypertension, liver disese, nd mlignncies. 23,24 The severity of ny dibetes compliction ws proxied using the Dibetes Complictions Severity Index (DCSI), which is scored on scle of 0 to 13 bsed on the presence of conditions such s retinopthy, neuropthy, nephropthy, nd metbolic disorders. 25 In ddition, the models lso included comorbidities tht hve been shown to be common in ptients with dibetes nd were not cptured in either the CCI or DCSI score. Specificlly, nxiety hs been shown to be prevlent in ptients with dibetes, 26 comorbid depression hs been shown to be linked to nondherence, 27 nd hyperlipidemi hs been used in previous studies tht exmined the reltionship between dherence nd outcomes. 28 Therefore, this study included indictor vribles for pre-period dignoses of nxiety (ICD-9-CM codes: 300.xx except 300.3x nd 300.4x), depression (ICD-9-CM codes: 296.2, 296.3, 300.4, or 311.xx), nd hyperlipidemi (ICD-9-CM codes: 272.1-272.4). The nlyses lso controlled for provider visits/services during the pre-period, with indictor vribles for renl dilysis therpy nd for visits to crdiologist, endocrinologist, fmily medicine prctitioner, or internl medicine prctitioner. Also included s covrites in the model were the number nd type of GLAs prescribed nd overll mediction use in the first 3 months of the post period. These fctors were cptured by n indictor vrible for insulin use, the number of noninsulin clsses of GLAs prescribed, nd the number of non-gla medictions prescribed. The clsses of noninsulin GLAs included lph-glucosidse inhibitors, mylin nlogs, bigunides, dipeptidyl peptidse-4 inhibitors, dopmine gonists, glucgon-like peptide-1 receptor gonists, meglitinides, sodium-glucose co-trnsporter-2 inhibitors, sulfonylures, thizolidinediones, nd orl fixed combintions. Generl liner models with gmm distribution nd log link were used to exmine ll-cuse outptient, drug, nd totl costs. Two-prt models were used to exmine cute cre costs, in which the first prt cptured the probbility of n ll-cuse or dibetes-relted cute cre visit nd the second prt estimted costs mong users of the service. Resource utiliztion ws exmined by estimting negtive binomil regressions for the number of hospitliztions, the number of ED visits, nd hospitl LOS. Logistic regressions were used to exmine the probbility of hospitliztion, ED use, or cute compliction. Differences in the estimted outcomes were then exmined by dherence sttus. To clculte whether differences in outcomes between dherent nd nondherent ptients were sttisticlly significnt, t tests were used. All nlyses were conducted using SAS version 9.3 (SAS Institute, Inc; Cry, North Crolin). A P vlue of <.05 ws considered to be sttisticlly significnt. RESULTS Tble 1 presents descriptive sttistics for the 228,074 individuls included in the study. The mjority of ptients were mle (53.75%), nd the men ge ws 52 yers (stndrd devition [SD] = 7.1). Ptients most commonly resided in the South (42.08%) or North Centrl (24.42%) regions of the country, nd most were covered by preferred provider orgniztions (53.95%) or helth mintennce orgniztion (21.33%). More thn hlf the ptients visited fmily medicine prctitioner (54.24%), nd over one-third (40.37%) hd t lest 1 internl medicine visit in the pre-period; less frequently, they visited crdiologist (17.11%) or endocrinologist (8.23%). Approximtely 1 in 8 ptients (13.12%) received prescription for insulin in the first 3 months of the post period, nd ptients were prescribed n verge 1.42 noninsulin clsses of GLA, including orl fixed combintion drugs s distinct clss, nd 4.87 non-gla medictions over this sme time period. In ddition to providing chrcteristics for the entire popultion, Tble 1 lso presents descriptive sttistics compring dherent with nondherent ptients. Adherent ptients were significntly older (53.47 vs 51.18 yers; P <.0001), more likely to be mle (56.51% vs 50.80%; P <.0001), nd less likely to reside in the South (38.93% vs 45.45%; P <.0001) compred with nondherent ptients. Adherent ptients were lso in better generl helth compred with nondherent ptients, s evidenced by lower rtes of nxiety (2.79% vs 3.90%; P <.0001) nd depression (5.44% vs 7.15%; P <.0001), s well s significntly lower rtes of visits to crdiologist (16.76% vs 17.48%; P <.0001). Adherent ptients my hve less severe dibetes compred with nondherent ptients, s evidenced by the significntly lower DCSI score (0.50 vs 0.52; P <.0001) nd by less use of insulin in the first 3 months of the post period (9.84% vs 16.63%; P <.0001). However, dherent ptients received significntly more noninsulin clsses of medictions in the first 3 months of the post period (1.63 vs 1.20; P <.0001), s well s higher number of non-gla medictions (5.05 vs 4.68; P <.0001). Figure 2 focuses on the ssocition between dherence to GLAs nd 3-yer ll-cuse medicl costs. As shown in this figure, dherence ws ssocited with significnt reductions in both cute cre costs ($8223 vs $12,153; P <.0001) nd outptient costs ($15,457 vs $16,964; P <.0001) nd significnt increses in drug costs ($14,816 vs $9390; P <.0001). Consistent with significntly higher drug costs nd significntly lower cute cre costs nd outptient costs for dherent ptients compred with nondherent ptients, there ws no sttisticlly significnt difference in totl costs between these 2 groups ($38,357 vs $38,633; P =.0720). Tble 2 exmines the ssocitions between dherence nd resource utiliztion nd complictions, showing tht dherence ws ssocited with significnt improvements in ptient outcomes in ll cses. Both the probbility of hospitliztion (17.65% vs e210 JULY 2017 www.jmc.com

Adherence nd Improved Outcomes TABLE 1. Descriptive Sttistics, Overll nd By Adherence Sttus Vrible (ptient chrcteristics) All (N = 228,074) N or Men % or SD Nondherent (PDC <80%) (n = 110,210) N or Men % or SD Adherent (PDC 80%) (n = 117,864) N or Men Age, yers (men nd SD) 52.36 7.09 51.18 7.61 53.47 6.37 Sex Mle 122,593 53.75 55,985 50.80 66,608 56.51 Femle 105,481 46.25 54,225 49.20 51,256 43.49 Region Northest 29,274 12.84 12,406 11.26 16,868 14.31 North Centrl 55,692 24.42 25,670 23.29 30,022 25.47 South 95,973 42.08 50,094 45.45 45,879 38.93 West 47,135 20.67 22,040 20.00 25,095 21.29 Insurnce pln type Comprehensive insurnce 11,465 5.03 4903 4.45 6562 5.57 Exclusive provider orgniztion 3498 1.53 1891 1.72 1607 1.36 HMO 48,644 21.33 24,269 22.02 24,375 20.68 Point-of-service 26,254 11.51 12,215 11.08 14,039 11.91 PPO 123,046 53.95 59,134 53.66 63,912 54.23 CDHP/HDHP 15,167 6.65 7798 7.08 7369 6.25 Generl helth nd comorbidities in pre-period Chrlson Comorbidity Index score (men nd SD) b 1.52 1.25 1.52 1.30 1.52 1.20 Dibetes Complictions Severity Index (men nd SD) 0.51 0.98 0.52 1.02 0.50 0.94 Anxiety 7585 3.33 4296 3.90 3289 2.79 Depression 14,289 6.27 7881 7.15 6408 5.44 Hyperlipidemi 101,469 44.49 47,411 43.02 54,058 45.86 Provider nd prescriber visited in pre-period Crdiologist 39,016 17.11 19,266 17.48 19,750 16.76 Endocrinologist 18,772 8.23 8920 8.09 9852 8.36 Fmily medicine 123,709 54.24 60,998 55.35 62,711 53.21 Internl medicine b 92,082 40.37 44,356 40.25 47,726 40.49 Renl dilysis therpy 184 0.08 103 0.09 81 0.07 Mediction use in first 3 months of post period Insulin 29,925 13.12 18,326 16.63 11,599 9.84 Number of noninsulin clsses (men nd SD) 1.42 0.74 1.20 0.64 1.63 0.76 Number of non-gla medictions (men nd SD) 4.87 3.35 4.68 3.44 5.05 3.25 Adherence in first yer of post period Proportion of dys covered (men nd SD) 70.10 28.71 46.33 24.05 92.34 5.61 CDHP indictes consumer-driven helth pln; GLA, glucose-lowering gent; HDHP, high-deductible helth pln; HMO, helth mintennce orgniztion; PDC, proportion of dys covered; PPO, preferred provider orgniztion; SD, stndrd devition. Differences in continuous vribles were exmined using t sttistics, nd differences in ctegoricl vribles were exmined using χ 2 sttistics. b Indictes tht the difference between dherence thresholds ws not sttisticlly significnt (P.05). % or SD THE AMERICAN JOURNAL OF MANAGED CARE VOL. 23, NO. 7 e211

Costs ($) FIGURE 2. All-Cuse Costs, by Adherence Threshold $45,000 $40,000 $35,000 $30,000 $25,000 $20,000 $15,000 $10,000 $5000 $0 $12,153 $8233 Acute cre b $16,964 $15,457 Outptient b Nondherent (PDC <80%) $9390 Drug b PDC indictes proportion of dys covered. Results of t tests tht exmined estimted outcomes from multivrible regression model tht controls for ptient chrcteristics, preperiod generl helth sttus, comorbidities, mediction use, providers, nd ptient dherence. b The t tests, which compred differences in costs between nondherent nd dherent ptients, were sttisticlly significnt (P <.05). TABLE 2. Resource Utiliztion nd Complictions by Adherence Sttus Nondherent (PDC <80%) (n = 110,210) $14,816 Adherent (PDC 80%) (0.27 vs 0.40; P <.0001), the number of ED visits (0.83 vs 1.23; P <.0001), nd hospitl LOS (1.25 vs 2.16; P <.0001) ll significntly diminished s dherence improved. The probbility of n cute compliction lso significntly decresed s dherence improved. Specificlly, dherence ws ssocited with significnt reduction in the probbility of n cute compliction being dignosed over the 3-yer post period (9.64% vs 12.54%; P <.0001). Two sensitivity nlyses were conducted to test the robustness of the results. First, ll nlyses were reestimted using the $38,633 $38,357 Adherent (PDC 80%) (n = 117,864) mediction possession rtio (MPR) rther thn the PDC s the mesure of dherence. Second, to control for the possibility of selection bis, n instrumentl vribles model 29 ws estimted using, s instruments, the copyments nd coinsurnce ssocited with GLAs prescribed in the first 3 months of the post period. The findings were generlly not sensitive to these lterntive specifictions, except tht there ws sttisticlly significnt difference between the ll-cuse totl costs of the dherent reltive to the nondherent ptients in both lterntive models. Specificlly, when MPR ws the mesure of dherence, totl costs were significntly lower for dherent reltive to nondherent ptients ($37,852 vs $39,282; P <.0001). Menwhile, in the instrumentl vribles model, totl costs were significntly higher for dherent compred with nondherent ptients ($38,755 vs $38,221; P =.0005). DISCUSSION Men SD Men SD The present study ws constructed to quntify Resource utiliztion the outcomes ssocited with dherence Probbility of hospitliztion 22.71% 12.59% 17.65% 9.52% to GLA therpy mong ptients with T2D. Probbility of ED visit 45.61% 12.09% 38.47% 10.49% Findings support lrge body of previous Number of hospitliztions 0.40 0.58 0.27 0.25 reserch tht hs reveled link between Number of ED visits 1.23 1.04 0.83 0.53 GLA dherence nd improved ptient outcomes. Hospitl length of sty, dys 2.16 4.18 1.25 1.43 4-9 Moreover, the current investigtion Complictions hs extended the literture by including ll Probbility of cute compliction 12.54% 7.32% 9.64% 5.52% clsses of GLAs nd tretment records of ED indictes emergency deprtment; PDC, proportion of dys covered; SD, stndrd devition. lrge (228,074) cross-ntionl popultion of Results of t tests, which exmined estimted outcomes from multivrible regression model tht dults with T2D over 3-yer post period. The controlled for ptient chrcteristics, pre-period generl helth sttus, comorbidities, mediction use, nd providers. study controlled for wide rnge of fctors b All differences between dherent nd nondherent ptients were sttisticlly significnt (P <.001). tht my ffect ptient outcomes nd exmined the robustness of results to lterntive 22.71%; P <.0001) nd of n ED visit (38.47% vs 45.61%; P <.0001) were significntly lower for dherent ptients compred with nondherent ptients. Similrly, the number of hospitliztions mesures of dherence nd methodologies. The following sections discuss the mjor findings of this study in the context of previous reserch. Totl Acute Cre Resource Utiliztion Consistent with erlier literture, 4,6,15,30,31 the dherent ptients in the present study hd significntly lower use of hospitl nd ED resources reltive to the nondherent ptients. This finding is importnt in both economic nd humnistic terms. From monetry stndpoint, the decresed use of cute cre ws ssocited with the substntilly reduced cute cre costs of the dherent cohort. In ddition to benefiting pyers, reduced cute cre costs e212 JULY 2017 www.jmc.com

Adherence nd Improved Outcomes my lessen the economic burden of dibetes for society s whole. For instnce, hospitliztion, just 1 component of cute cre, ccounted for 43% of ll direct dibetes spending in the United Sttes in 2012. 2 Regrding humnistic benefits, cute cre costs my indicte ptient disenggement nd poor helthcre mngement in contrst to drug nd some outptient spending, which hve been shown to be ssocited with improved disese control. 32 In ddition, cute cre is ssocited with unexpected loss of time, productivity, nd workdys for ptients nd cregivers, s well s other indirect costs. 2 Hospitliztion lso crries intrinsic helth risks, such s the chnce of contrcting helthcre-ssocited infection. 33 Thus, mediction spending tht leds to decresed use of cute cre my be considered worthwhile, prticulrly when such expenditures re offset by other cost reductions, s in our study. Complictions The results of previous reserch tht exmined chronic microvsculr nd mcrovsculr complictions hve indicted tht the odds of developing such conditions diminished s dherence improved. 6,10,11 Given tht chronic complictions my best be exmined over significntly longer time horizon thn the 3-yer post period of this study, the present investigtion focused on the reltionships mong dherence nd cute complictions, including hyperglycemi, hypoglycemi, nd dibetic or hypoglycemic com. Like the chronic complictions exmined in erlier reserch, the cute complictions were tken s indictors of the qulity of dibetes mngement. For instnce, dignosis of hyperglycemi indictes tht the body hs too little insulin or is not using insulin properly, 34 hypoglycemi hs been reported to be ssocited with GLA therpy, 35,36 nd severe hypoglycemi results in com, reduced consciousness, or prolonged or recurrence or hypoglycemi for up to 60 hours. 37 Furthermore, clinicl tril dt hve shown, nd the results of severl previous observtionl studies hve indicted, tht the risk of hypoglycemic events incresed s sulfonylure or insulin therpy ws intensified. 36 Although intensified tretment is different from dherence to prescribed mediction, fers of hypoglycemi nd its consequences hve been shown to be brrier to ptient dherence to GLA therpy. 35 Contrry to such fers, the dherent ptients in the current investigtion, reltive to the nondherent ptients, hd substntilly lower rtes of hyperglycemi, hypoglycemi, nd dibetic or hypoglycemic com. These results re consistent with the diminished need for cute cre observed mong the dherent ptients. Tken together, the reductions in both cute complictions nd cute cre indicte tht the dherent ptients in this study hd better-mnged T2D reltive to the nondherent individuls. By extension, it might lso be rgued tht their dherence contributed to better qulity of life. Costs As noted in the introductory section, number of previous investigtions hve demonstrted link between GLA dherence nd lower totl costs. 5,9,12-15 However, n extensive review of the literture, encompssing 37 studies, found tht this link is inconsistent given tht there re incresed drug costs ssocited with better dherence. 30 In concert with the findings of the literture review, dherence in the present study ws ssocited with greter drug costs. However, these higher costs were entirely offset by decreses in cute cre nd outptient services, nd there ws no significnt difference in totl medicl costs mong this ptient popultion when compring dherent nd nondherent ptients. However, it should be noted tht this finding ws not robust to lterntive model specifictions. This study found tht there ws no sttisticlly significnt difference in totl costs when compring dherent (PDC 80%) with nondherent (PDC <80%) ptients, most likely due to the incresed mediction costs ssocited with dherence. However, there were significnt reductions in outptient costs nd cute cre costs ssocited with dherent versus nondherent ptients. In ddition, there re lso potentilly lrge offsets in cute cre nd outptient costs ssocited with even smll chnges in dherence. For exmple, 1% increse in dherence ws ssocited, on verge, with cute cre cost svings of $25,160 per 1000 persons, or $5,738,276 mong ll 228,074 individuls over the 3-yer post period. Such increses in cute cre costs suggest more hospitliztions mong nondherent ptients compred with dherent ptients, which hs been shown to be ssocited with poorer long-term outcomes nd decresed helth-relted qulity of life. 38 Limittions The findings of this study must be interpreted within the context of the limittions. First, the nlyses were bsed on observtionl helth insurnce clims dt tht described popultion of commercilly insured ptients with T2D who my or my not be representtive of the mjority of Americns with T2D. Secondly, the use of dignostic codes ws not s rigorous s forml ssessments nd my underrepresent certin conditions, such s hypoglycemi. Third, the use of clims dt precluded the nlyses from directly controlling for undocumented fctors, such s glycted hemoglobin levels, rce, durtion of dibetes, or socioeconomic clss, wheres ny of these fctors my be ssocited with ptient outcomes. Fourth, lthough the clims dt fcilitted the observtion of prescriptions filled, they were unble to provide insight into whether, or in wht mnner, the mediction ws tken. Fifth, the ssocition between dherence to specific GLA clsses or medictions nd ptient outcomes, s well s the ssocition between ptient glycemic control nd outcomes, re beyond the scope of this reserch. Finlly, the study focused on sttisticl significnce nd ws unble to determine whether differences in outcomes represented miniml cliniclly importnt differences. THE AMERICAN JOURNAL OF MANAGED CARE VOL. 23, NO. 7 e213

CONCLUSIONS Generlly robust to wide rnge of sensitivity nlyses, the results of this study indicte tht GLA dherence is ssocited with significnt improvements in cute cre outcomes, s mesured by the probbility of hospitliztion, the probbility of n ED visit, the number of hospitliztions, the number of ED visits, nd hospitl LOS. The odds of n cute compliction lso declined s dherence improved. Consistent with these findings, improved dherence ws lso ssocited with decresed cute cre nd outptient costs, with no significnt chnge in totl costs. Furthermore, even smll chnges in ptient dherence were found to hve potentilly lrge cost implictions, with 1% increse in dherence ssocited with cute cre cost reductions of $25,160 for 1000 individuls, or $5,738,276 for the entire popultion, over the 3-yer post period. The findings of this study suggest tht dherence to GLAs mong dults with T2D my led to significnt benefits for ptients without incresing pyer costs. n Acknowledgments The uthors would like to thnk Ptrici Pltt for her ssistnce in the writing of the mnuscript. Author Affilitions: Globl Ptient Outcomes nd Rel World Evidence (SC, KSB), nd Globl Medicl Affirs (LEG-P), Eli Lilly nd Compny, Indinpolis, IN; HelthMetrics Outcomes Reserch, LLC (MJL), Bonit Springs, FL. Source of Funding: Ms Curtis, Dr Boye, nd Dr Grci-Perez completed this reserch s employees of Eli Lilly nd Compny. Dr Lge ws compensted by Eli Lilly nd Compny for her work on this reserch. Author Disclosures: Ms Curtis, Dr Boye, nd Dr Grci-Perez re employed by Eli Lilly nd Compny, nd Dr Boye nd Dr Grci-Perez re minor stockholders. Dr Lge ws pid by Eli Lilly nd Compny for work on this project. The uthors report no other reltionship or finncil interest with ny entity tht would pose conflict of interest with the subject mtter of this rticle. Authorship Informtion: Concept nd design (SC, KSB, LEG-P); cquisition of dt (KSB); nlysis nd interprettion of dt (KSB, MJL, LEG-P); drfting of the mnuscript (MJL); criticl revision of the mnuscript for importnt intellectul content (SC, KSB, LEG-P); sttisticl nlysis (MJL); provision of ptients or study mterils (KSB); obtining funding (KSB); dministrtive, technicl, or logistic support (SC, KSB); nd supervision (LEG-P). Address Correspondence to: Mureen J. Lge, PhD, HelthMetrics Outcomes Reserch, 27576 River Rech Dr, Bonit Springs, FL 34134. E-mil: lgemj@hlthmetrics.com. REFERENCES 1. Ntionl dibetes sttistics report: estimtes of dibetes nd its burden in the United Sttes, 2014. CDC website. http://www.cdc.gov/dibetes/pubs/sttsreport14/ntionl-dibetes-report-web.pdf. Published 2014. Accessed Februry 20, 2016. 2. Americn Dibetes Assocition. Economic costs of dibetes in the U.S. in 2012. Dibetes Cre. 2013;36(4):1033-1046. doi: 10.2337/dc12-2625. 3. Gregg EW, Zhuo X, Cheng YJ, Albright AL, Nryn KMV, Thompson TJ. Trends in lifetime risk nd yers of life lost due to dibetes in the USA, 1985-2011: modelling study. Lncet Dibetes Endocrinol. 2014;2(11):867-874. doi: 10.1016/S2213-8587(14)70161-5. 4. Cheng SH, Chen CC, Tseng CH. Does mediction dherence led to lower helthcre expenses for ptients with dibetes? Am J Mng Cre. 2013;19(8):662-670. 5. Encinos WE, Bernrd D, Dor A. Does prescription drug dherence reduce hospitliztions nd costs? the cse of dibetes. Adv Helth Econ Helth Serv Res. 2010;22:151-173. 6. Gibson TB, Song X, Alemyehu B, et l. Cost shring, dherence, nd helth outcomes in ptients with dibetes. Am J Mng Cre. 2010;16(8):589-600. 7. Jh AK, Aubert RE, Yo J, Tegrden JR, Epstein RS. Greter dherence to dibetes drugs is linked to less hospitl use nd could sve nerly $5 billion nnully. Helth Aff (Millwood). 2012;31(8):1836-1846. doi: 10.1377/hlthff.2011.1198. 8. Jurez DT, Tn C, Dvis J, Mu M. Fctors ffecting sustined mediction dherence nd its impct on helth cre utiliztion in ptients with dibetes. J Phrm Helth Serv Res. 2013;4(2):89-94. 9. Sokol MC, McGuign KA, Verbrugge RR, Epstein RS. Impct of mediction dherence on hospitliztion risk nd helthcre cost. Med Cre. 2005;43(6):521-530. 10. Prk IS, Sohn HS. Effect of medicine dherence on the occurrence of cerebrovsculr disorders in dibetes mellitus ptients. Epidemiol Helth. 2011;33:e2011001. doi: 10.4178/epih/e2011001. 11. Yu AP, Yu YF, Nichol MB. Estimting the effect of mediction dherence on helth outcomes mong ptients with type 2 dibetes n ppliction of mrginl structurl models. Vlue Helth. 2010;13(8):1038-1045. doi: 10.1111/j.1524-4733.2010.00787.x. 12. Hnsen RA, Frley JF, Droege M, Mciejewski ML. A retrospective cohort study of economic outcomes nd dherence to monotherpy with metformin, pioglitzone, or sulfonylure mong ptients with type 2 dibetes mellitus in the United Sttes from 2003 to 2005. Clin Ther. 2010;32(7):1308-1319. doi: 10.1016/j. clinther.2010.07.011. 13. Hong JS, Kng HC. Reltionship between orl ntihyperglycemic mediction dherence nd hospitliztion, mortlity, nd helthcre costs in dult mbultory cre ptients with type 2 dibetes in South Kore. Med Cre. 2011;49(4):378-384. doi: 10.1097/MLR.0b013e31820292d1. 14. Roebuck MC, Libermn JN, Gemmill-Toym M, Brennn TA. Mediction dherence leds to lower helth cre use nd costs despite incresed drug spending. Helth Aff (Millwood). 2011;30(1):91-99. doi: 10.1377/ hlthff.2009.1087. 15. Wong ES, Bryson CL, Hebert PL, Liu CF. Estimting the impct of orl dibetes mediction dherence on medicl costs in VA. Ann Phrmcother. 2014;48(8):978-985. 16. Kheirbek RE, Alemi F, Zrgoush M. Comprtive effectiveness of hypoglycemic medictions mong veterns. J Mng Cre Phrm. 2013;19(9):740-744. 17. Mlmenäs M, Bouchrd JR, Lnger J. Retrospective rel-world dherence in ptients with type 2 dibetes inititing once-dily lirglutide 1.8 mg or twice-dily exentide 10 μg. Clin Ther. 2013;35(6):795-807. doi: 10.1016/j.clinther.2013.03.021. 18. PQA mediction qulity mesures in the helth insurnce mrket. Phrmcy Qulity Allince website. http://pqllince.org/mesures/qrs.sp. Published 2016. Accessed Februry 12, 2016. 19. Bser O, Bouchrd J, DeLuzio T, Henk H, Agren M. Assessment of dherence nd helthcre costs of insulin device (FlexPen) versus conventionl vil/syringe. Adv Ther. 2010;27(2):94-104. doi: 10.1007/s12325-010-0009-6 20. Helth insurnce mrketplce: qulity rting system mesure technicl specifictions. CMS website. https://www.cms.gov/medicre/qulity-inititives-ptient-assessment-instruments/qulityinititivesgeninfo/ Downlods/2016-QRS-Mesure-Technicl-Specifictions.pdf. Published September 2015. Accessed June 9, 2017. 21. Choudhry NK, Shrnk WH, Levin RL, et l. Mesuring concurrent dherence to multiple relted medictions. Am J Mng Cre. 2009;15(7):457-464. 22. Ho PM, Rumsfeld JS, Msoudi FA, et l. Effect of mediction nondherence on hospitliztion nd mortlity mong ptients with dibetes mellitus. Arch Intern Med. 2006;166(17):1836-1841. 23. Deyo RA, Cherkin DC, Ciol MA. Adpting clinicl comorbidity index for use with ICD-9-CM dministrtive dtbses. J Clin Epidemiol. 1992;45(6):613-619. 24. Qun H, Sundrrjn V, Hlfon P, et l. Coding lgorithms for defining comorbidities in ICD-9-CM nd ICD-10 dministrtive dt. Med Cre. 2005;43(11):1130-1139. 25. Chng HY, Weiner JP, Richrds TM, Bleich SN, Segl JB. Vlidting the dpted Dibetes Complictions Severity Index in clims dt. Am J Mng Cre. 2012;18(11):721-726. 26. Grigsby AB, Anderson RJ, Freedlnd KE, Clouse RE, Lustmn PJ. Prevlence of nxiety in dults with dibetes: systemtic review. J Psychosom Res. 2002;53(6):1053-1060. 27. Gonzlez JS, Peyrot M, McCrl LA, et l. Depression nd dibetes tretment nondherence: metnlysis. Dibetes Cre. 2008;31(12):2398-2403. doi: 10.2337/dc08-1341. 28. Ayygri R, Wei W, Cheng D, Pn C, Signorovitch J, Wu EQ. Effect of dherence nd insulin delivery system on clinicl nd economic outcomes mong ptients with type 2 dibetes inititing insulin tretment. Vlue Helth. 2015;18(2):198-205. doi: 10.1016/j.jvl.2014.12.016. 29. McClelln M, McNeil BJ, Newhouse JP. Does more intensive tretment of cute myocrdil infrction in the elderly reduce mortlity? nlysis using instrumentl vribles. JAMA. 1994;272(11):859-866. 30. Asche C, LFleur J, Conner C. A review of dibetes tretment dherence nd the ssocition with clinicl nd economic outcomes. Clin Ther. 2011;33(1):74-109. doi: 10.1016/j.clinther.2011.01.019. 31. Hn E, Suh DC, Lee SM, Jng S. The impct of mediction dherence on helth outcomes for chronic metbolic diseses: retrospective cohort study. Res Socil Adm Phrm. 2014;10(6):e87-e98. doi: 10.1016/j. sphrm.2014.02.001. 32. Bodenheimer T, Wgner EH, Grumbch K. Improving primry cre for ptients with chronic illness: the chronic cre model, prt 2. JAMA. 2002;288(15):1909-1914. 33. Sydnor ER, Perl TM. Hospitl epidemiology nd infection control in cute-cre settings. Clin Microbiol Rev. 2011;24(1):141-173. doi: 10.1128/CMR.00027-10. 34. Hyperglycemi (high blood glucose). Americn Dibetes Assocition website. http://www.dibetes.org/ living-with-dibetes/tretment-nd-cre/blood-glucose-control/hyperglycemi.html. Updted September 16, 2014. Accessed July 13, 2016. 35. Ahrén B. Avoiding hypoglycemi: key to success for glucose-lowering therpy in type 2 dibetes. Vsc Helth Risk Mng. 2013:9:155-163. doi: 10.2147/VHRM.S33934. 36. Zmmitt NN, Frier BM. Hypoglycemi in type 2 dibetes: pthophysiology, frequency, nd effects of different tretment modlities. Dibetes Cre. 2005;28(12):2948-2961. 37. Asplund K, Wiholm BE, Lundmn B. Severe hypoglycemi during tretment with glipizide. Dibetic Med. 1991;8(8):726-731. 38. Jvnbkht M, Abolhsni F, Mshyekhi A, Brdrn HR, Jhngirinoudeh, Y. Helth relted qulity of life in ptients with type 2 dibetes mellitus in Irn: ntionl survey. PLoS One. 2012;7(8):e44526. doi: 10.1371/journl.pone.0044526. Full text nd PDF t www.jmc.com e214 JULY 2017 www.jmc.com