Journal of Diabetes and Its Complications 29 (2015) 212–217
Contents lists available at ScienceDirect
Journal of Diabetes and Its Complications journal homepage: WWW.JDCJOURNAL.COM
Healthcare utilization and costs in diabetes relative to the clinical spectrum of painful diabetic peripheral neuropathy Alesia Sadosky a,⁎, Jack Mardekian a, Bruce Parsons a, Markay Hopps a, E. Jay Bienen b, John Markman c a b c
Pfizer, Inc., New York, NY, USA Outcomes Research Consultant, New York, NY, USA Translational Pain Research Program, Department of Neurosurgery, University of Rochester Medical Center, Rochester, NY, USA
a r t i c l e
i n f o
Article history: Received 4 September 2014 Received in revised form 30 October 2014 Accepted 31 October 2014 Available online 8 November 2014 Keywords: Diabetes Painful diabetic peripheral neuropathy Electronic medical records Pain severity Healthcare utilization Costs
a b s t r a c t Aims: Diabetic peripheral neuropathy (DPN) accompanied by painful symptoms is known as painful DPN (pDPN). This study characterized healthcare resource utilization and costs in patients with DPN, pDPN, and severe pDPN relative to diabetes only. Methods: Four adult cohorts were identified from the Humedica database: type 2 diabetes without DPN (n = 288,328); DPN (n = 35,050); pDPN (DPN subjects with a pain score ≥ 1 on a 0–10 numeric rating scale; n = 3449); and severe pDPN (pain scores 7–10; n = 1824). Resource utilization and costs for 12-months post-diagnosis were compared for diabetes relative to the other cohorts. Results: Demographic characteristics were different across cohorts. Relative to diabetes alone, DPN, pDPN, and severe pDPN were characterized by significantly higher proportions of patients with resource utilization for all resource categories (all P b 0.0001); the highest resource use generally observed for severe pDPN. Total annual direct medical costs were $6632 for diabetes only, with costs for DPN ($12,492), pDPN ($27,931), and severe pDPN ($30,755) significantly higher (all P b 0.0001); outpatient costs were consistently the primary driver of total costs. Conclusions: Patients with DPN, pDPN, and severe pDPN had significantly greater healthcare resource utilization and costs than patients with diabetes only, with the highest burden associated with severe pDPN. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Diabetic peripheral neuropathy (DPN) is a common neurologic sequela of diabetes, and when the resulting nerve damage is accompanied by painful symptoms it is known as painful DPN (pDPN). The presence of pDPN is associated with a substantial adverse impact on patient function, quality of life, and work productivity, and also results in an economic burden relative to the general population and to patients with diabetes without pDPN
Disclosures: Alesia Sadosky, Jack Mardekian, Bruce Parsons, and Markay Hopps are employees and shareholders of Pfizer, the sponsor of this study; E. Jay Bienen is an independent scientific consultant who was funded by Pfizer in connection with manuscript development; John Markman collaborated with Pfizer on the project but was not financially compensated for his involvement on the project, including manuscript development. Funding: This study was funded by Pfizer, Inc. Conflict of interest statement: There are no other financial relationships or relevant conflicts related to this manuscript. ⁎ Corresponding author at: Pfizer, Inc., 235 East 42nd Street, New York, NY 10017. Tel.: + 1 212 733 9491; fax: + 1 646 441 4757. E-mail address: alesia.sadosky@pfizer.com (A. Sadosky). http://dx.doi.org/10.1016/j.jdiacomp.2014.10.013 1056-8727/© 2015 Elsevier Inc. All rights reserved.
(Benbow, Wallymahmed, & Macfarlane, 1998; daCosta DiBonaventura, Cappelleri, & Joshi, 2011; Dworkin, Malone, Panarites, Armstrong, & Pham, 2010; Dworkin, Panarites, Armstrong, Malone, & Pham, 2011; Gore et al., 2005; Ritzwoller, Ellis, Korner, Hartsfield, & Sadosky, 2009; Stewart, Ricci, Chee, Hirsch, & Brandenburg, 2007). These effects have been reported to be greater as pain severity increases (Dibonaventura, Cappelleri, & Joshi, 2011; Gore et al., 2005; Sadosky et al., 2013). As integrated healthcare systems develop new care delivery models to manage the increasing burden of expensive chronic conditions, such as diabetes, in accord with Affordable Care Act, a deeper understanding of the clinical problems and types of care that disproportionately contribute to the high costs at the population level is required. However, there have been no studies comparing healthcare resource utilization and costs of diabetes relative to DPN, pDPN, and severe pDPN. Electronic medical records (EMR) capture real-world, patient-level data representing integral components of provider care that are not readily available in claims databases including patient-reported outcomes such as pain severity. The availability of these data enables identification of discrete populations and evaluation of resources and costs across inpatient and outpatient settings. This characterization is essential to managed care, especially accountable care organizations, providing the background and understanding required to implement
A. Sadosky et al. / Journal of Diabetes and Its Complications 29 (2015) 212–217
more targeted disease management strategies (Eggleston & Finkelstein, 2014). Therefore, the purpose of this study is to apply EMR-derived clinical information from the Humedica database to evaluate the direct medical costs of patients with diabetes relative to DPN, pDPN, and severe pDPN. This clinical database facilitates identification and management of patients with chronic conditions who are at risk for greater clinical complexity and higher costs of care. 2. Methods 2.1. Data source Data for this retrospective study were derived from the Humedica EMR database, which has broad geographic representation and includes information on demographics, diagnoses, inpatient and outpatient encounters, medications, procedures, lab results, vital signs, and select data derived from physicians' notes. Humedica does not mandate a particular EMR system, and in the more than 20 provider groups, many run multiple EMR installations for different sites of care. Records are linked using a unique patient identifier and are fully compliant with the Health Insurance Portability and Accountability Act (HIPAA). The current analysis utilized structured and unstructured data from the database. Structured data included demographic and clinical characteristics, and healthcare resource utilization (HCRU). Unstructured data were derived from the notes fields within the EMR, which were searched using Humedica's proprietary Natural Language Processing (NLP). The Humedica NLP system uses vocabulary from the Unified Medical Language System including multiple medical dictionaries such as the Logical Observation Identifiers Names and Codes (LOINC), the Systemized Nomenclature of Medicine–Clinical Terms (SNOMED-CT), and RxNorm, a listing of generic and branded drugs (among others). Each NLP concept included in the data is associated with a unique subject record and a date of observation, allowing longitudinal tracking. A specific part of the search of the unstructured data was for pain scores, which are collected by Humedica as a health measure, and are patient-reported using a 0 to 10 scale (0 = no pain, 10 = worst imaginable pain). The scores are recorded during discrete health care provider interactions, although providers may use different methods, including the Visual Analogue Scale (VAS) and the Verbal Numerical Rating Scale (VNRS). These scores may be recorded in the structured data elements of the EMR or harvested from notes using NLP. 2.2. Subjects Subjects ≥18 years of age with type 2 diabetes were identified from the database for the period January 1, 2008–September 30, 2013 based on ICD-9-CM diagnosis codes (250.00-250.93) in the most recent calendar year available (2008–2012) and having continuous data for 1 year pre- and post-diagnosis. Subjects were excluded if they had a diagnosis code for end-stage renal disease, cancer, and/or HIV any time during the study period, were resident in a nursing/inpatient facility in the year prior to the diabetes diagnosis, or had a pregnancy during the pre-index period. Only identified subjects who belonged to an integrated delivery network (IDN) were included. Among subjects who met all criteria, the DPN cohort was defined as subjects with diabetes who had an ICD-9 code for DPN (ICD-9 codes: 357.2 or 250.6). The pDPN cohort consisted of DPN subjects with a pain score ≥ 1 for current pain that was obtained within 15 days (before or after) the DPN diagnosis, and the severe pDPN cohort was pDPN subjects with pain scores 7–10, indicating severe pain based on established criteria (Zelman, Dukes, Brandenburg, Bostrom, & Gore, 2005). The index date, for evaluation of healthcare resource utilization and costs, was the date of ICD-9 diagnosis, i.e. date
213
of diabetes diagnosis in the diabetes-only cohort, and date of DPN diagnosis in the other cohorts. 2.3. Healthcare resource utilization and costs All-cause resource utilization, estimated as the proportion of patients using each resource category, was identified from the claims data for the 12-month post-index period. Resource categories included emergency room (ER) visits, hospitalizations, outpatient resources stratified by office visits and other outpatient visits, laboratory tests, foot procedures and total procedures, and prescriptions. Additionally, prescriptions for pain-related medications were evaluated. All-cause costs were obtained for the subset of patients who are linked to the Optum claims database. The charged, or requested amount that is billed by the provider was used as the cost, although this number may have high variation since it often has little relationship to the actual amount paid. The Optum DataMart is an integrated database consisting of enrollment, inpatient and outpatient medical claims, pharmaceutical claims, and laboratory results. 2.4. Analyses Baseline demographic characteristics of the cohorts were summarized descriptively, and comparisons with the diabetes cohort were performed using likelihood ratio chi-square test for categorical variables and Student's t-test for continuous variables. Proportions of patients using each resource category were compared using likelihood ratio chi-square tests. Direct medical costs were compared using Wilcoxon two-sample rank sum tests. All analyses were conducted using SAS version 9.2 (SAS Institute Inc., Cary, NC). 3. Results 3.1. Cohort populations As shown in Table 1, the four cohorts derived from the 24,257,806 patients in the Humedica database for the specified time period consisted of 288,328 patients with diabetes only; 35,050 with DPN; 3,449 with pDPN; and 1824 with severe pDPN, which represents 52.9% of the patients with pDPN. Among the pDPN patients, there was no difference in mean (SD) pain scores collected in inpatient (n = 1724) and outpatient (n = 1725) settings, 6.4 (2.9) and 6.3 (2.6), respectively. Significant differences were observed for most demographic characteristics between the diabetes-only cohort and the other cohorts (Table 2). In particular, there were significantly lower proportions of females in the diabetes-only cohort (53.2%) relative to pDPN (55.6%; P = 0.005) and severe pDPN (59.7%; P b 0.0001), and whereas severe pDPN patients were significantly younger than diabetes-only (59.4 vs. 61.4 years; P b 0.0001), DPN patients were significantly older (64.8 years; P b 0.0001). The proportion of smokers was also lower in the diabetes-only cohort (28.9%) relative to the other cohorts (P b 0.0001), which appeared to increase from DPN (34.2%) to pDPN (37.8%) to severe pDPN (39.6%). The prevalence of comorbidities across all system classes was significantly higher with DPN, pDPN, and severe pDPN relative to diabetes-only (all P b 0.0001) (Table 2). The higher prevalence of neuropathic pain conditions in the DPN, pDPN, and severe pDPN cohorts was primarily driven by back and neck pain with neuropathic pain involvement, and causalgias (data not shown). Diabetes-only patients had the lowest CCI score, 1.3, with all other cohorts having significantly higher scores; 3.6, 3.9, and 3.9 for DPN, pDPN, and severe pDPN, respectively (all P b 0.0001 vs. diabetes-only). Diabetes-related comorbidities such as retinopathy and the CCI comorbidity of “Diabetes with chronic complications” were also significantly higher for DPN, pDPN,
214
A. Sadosky et al. / Journal of Diabetes and Its Complications 29 (2015) 212–217
Table 1 Sample attrition. Criteria
Number of subjects
Total patients in Humedica database during January 1, 2008–September 30, 2013 With any diabetes diagnosis from January 1, 2008–September 30, 2013 Age ≥18 in the year of the index diabetes diagnosis 12 months of continuous enrollment pre- and post-index Excluding patients with any diagnosis of cancer or diagnosis codes potentially associated with neuropathy of nondiabetic etiology during 1 year (360 days) pre-diabetes-index (Hartsfield et al., 2008; Ritzwoller et al., 2009) Excluding females with any pregnancy diagnosis 1 year pre-index Excluding patients with any end stage renal disease procedure 1 year pre-index Excluding patients who reside in a nursing/inpatient facility 1 year pre-index Excluding patients without continuous enrollment for 12 month pre- and post-DPN index periods for diabetes patients with DPN Excluding patients with any type I diabetes diagnosis on the diabetes index date Patients who belong to an Integrated Delivery Network With a DPN diagnosis on or after diabetes index date (DPN cohort) DPN and with pain score ≥1 within 15 days before or after DPN diagnosis DPN and with pain score ≥7 (severe pDPN cohort) Without any DPN diagnosis on or after diabetes index date (diabetes-only cohort)
and severe pDPN relative to diabetes-only (all P b 0.0001), but were generally similar across the DPN spectrum (Table 2). 3.2. Healthcare resource utilization Relative to diabetes-only, there was greater resource utilization during the 12-month post-index period among patients with DPN, pDPN, and severe pDPN (Table 3), as indicated by significantly higher
Percent of subjects
24,257,806 1,795,461 1,777,371 792,728 673,965
100.00 7.40
100.0 99.0 44.2 37.5
670,247 670,143 637,068 619,034
37.3 37.2 35.5 34.5
590,473 323,378 35,050 3449 1824 288,328
32.9 18.0
100.0 10.8 9.8 5.2 89.2
proportions of these patients using each of the resource categories (all P b 0.0001). While there appeared to be a tendency toward increased resource use from diabetes-only patients to DPN to pDPN, fewer differences were observed between pDPN and severe pDPN (Table 3). Relative to the pDPN cohort, higher proportions of severe pDPN patients had ER visits (57.0 vs. 50.7%) and hospitalizations (62.8 vs. 40.2%), with the proportion of patients using other resources generally similar between these two cohorts.
Table 2 Demographic and clinical characteristics of the cohorts. Variable
Diabetes only (n = 288,328)
DPN (n = 35,050)
Age, mean (SD) Gender, n (%) Female Male Race, n (%) African American Asian Caucasian Other/unknown Smoker, n (%) Region Midwest Northeast South West Unknown Average household income, mean (SD) Body mass index, kg/m2, mean (SD) Comorbidities, n (%) Cardiovascular disorders Gastrointestinal disorders Mental disorders Sleep disorders Musculoskeletal pain Neuropathic pain conditions Restless leg syndrome Fibromyalgia Myocardial infarction Retinopathy Charlson Comorbidity Index, mean (SD) Diabetes with chronic complications, n (%) Renal disease, n (%) Total cholesterol, mg/dL, mean (SD) Low density lipoprotein cholesterol, mg/dL, mean (SD) Diastolic blood pressure, mmHg, mean (SD) Fibrinogen activity, mg/dL, mean (SD) HgbA1C, %, mean (SD) a
All P values are for comparison with diabetes cohort.
pDPN (n = 3449)
Pa
Value
b0.0001 b0.0001
61.4 (13.4)
64.8 (12.2)
153,463 (53.2) 134,761 (46.8)
17,871 (51.0) 17,166 (49.0)
42,476 (14.7) 3209 (1.1) 219,016 (76.0) 23,627 (8.2) 83,220 (28.9)
5328 204 27,460 2058 11,980
(15.2) (0.6) (78.3) (5.9) (34.2)
140,213 (48.6) 12,515 (4.3) 123,907 (43.0) 6537 (2.3) 5156 (1.8) $39,740 (9918) 33.6 (7.4)
17,889 1926 13,736 809 690 $39,579 33.7
(51.0) (5.5) (39.2) (2.4) (2.0) (9690) (7.5)
0.0044 0.0642
102,183(35.4) 30,288 (10.5) 16,849 (5.8) 7840 (2.7) 60,348 (20.9) 9085 (3.2) 855 (0.3) 4135 (1.4) 1941 (0.7) 1452 (0.5) 1.3 (0.7) 7109 (2.5) 8078 (2.8) 179.8 (39.7) 101.4 (32.9) 76.4 (11.2) 355.3 (165.9) 7.2 (1.5)
24,194 8519 5168 2753 15,633 3827 534 1371 1403 1375 3.6 32,328 4995 170.0 91.7 73.4 404.3 7.6
(69.0) (4.3) (14.7) (7.9) (44.6) (10.9) (1.5) (3.9) (4.0) (3.9) (1.6) (92.2) (14.3) (40.1) (32.4) (11.4) (210.4) (1.7)
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 0.0002 b0.0001
Severe pDPN (n = 1824) Pa
Value 61.0 (12.9)
0.089 0.005
1918 (55.6) 1529 (44.4) b0.0001
b0.0001 b0.0001
Pa
Value 59.4 (13.2) 1088 (59.7) 735 (40.3)
b0.0001 630 (18.3) 7 (0.2) 2607 (75.6) 205 (5.9) 1302 (37.8) 1305 (37.8) 8 (0.2) 1788 (51.8) 260 (7.5) 88 (2.6) $36,746 (6913) 33.8 (7.9) 2354 1028 637 360 1789 465 62 178 200 62 3.9 3268 679 174.0 94.9 73.3 404.7 8.0
(68.3) (29.8) (18.5) (10.4) (51.9) (13.5) (1.8) (5.2) (5.8) (1.8) (1.8) (94.8) (19.7) (43.9) (35.6) (12.6) (204.8) (1.9)
b0.0001 b0.0001
b0.0001 b0.0001
b0.0001 386 4 1319 115 722
(21.2) (0.2) (72.3) (6.3) (39.6)
b0.0001 b0.0001
b0.0001 0.1926
606 2 1035 119 62 $36,118 33.8
(33.2) (0.1) (56.7) (6.5) (3.4) (6674) (7.9)
b0.0001 0.2426
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 0.0422 b0.0001
1215 586 355 192 1012 263 35 107 93 28 3.9 1743 1824 175.8 96.8 73.8 437.5 8.1
(66.6) (32.1) (19.5) (10.5) (55.5) (14.4) (1.9) (5.9) (5.1) (1.5) (1.8) (95.6) (18.9) (45.3) (36.4) (12.9) (223.8) (2.0)
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 0.0033 b0.0001 b0.0001 0.0112 b0.0001
A. Sadosky et al. / Journal of Diabetes and Its Complications 29 (2015) 212–217
215
Table 3 Healthcare resource utilization for the 12-month period after diagnosis. Resource
Diabetes only (n = 288,328)
DPN (n = 35,050)
Frequency of use
Number (%) of patients
Number (%) of patients
Pa
Number (%) of patients
Pa
Number (%) of patients
Pa
Emergency room visits Inpatient hospitalizations Office visits Outpatient visits (excluding office visits) Laboratory tests Foot exam Foot procedures Total procedures Prescriptions written
52,845 (18.2) 43,783 (15.2) 228,121 (79.1) 254,217 (88.2) 255,531 (88.6) 20,034 (6.9) 12,229 (4.2) 260,874 (90.5) 253,610 (88.0)
9537 (27.2) 10,174 (29.0) 30,390 (86.7) 32,517 (92.8) 32,330 (92.2) 5349 (15.3) 6575 (18.8) 32,911 (93.9) 33,020 (94.2)
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
1750 1386 3107 3346 3397 451 691 3391 3424
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
1040 1146 1638 1755 1803 225 388 1782 1812
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
Units per patient
Mean (SD)
Mean (SD)
Pa
Mean (SD)
Pa
Mean (SD)
Pa
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
1.4 1.3 20.2 8.3 280.8 0.2 0.5 108.5 102.9
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
1.7 (3.1) 1.4 (1.9) 21.6 (32.0) 7.5 (15.8) 313.9 (410.3) 0.2 (0.9) 0.5 (1.4) 117.3 (185.7) 114.3 (130.1)
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
Emergency room visits Inpatient hospitalizations Office visits Outpatient visits (excluding office visits) Laboratory tests Foot exam Foot procedures Total procedures Prescriptions written a
0.3 0.2 8.9 2.7 76.7 0.1 0.1 29.5 24.1
(1.0) (0.7) (12.7) (6.8) (131.1) (0.4) (0.5) (52.1) (47.1)
pDPN (n = 3449)
0.6 (1.6) 0.5 (1.2) 16.2 (20.6) 4.8 (11.2) 145.7 (234.4) 0.2 (0.7) 0.5 (1.3) 61.1 (109.5) 55.4 (100)
(50.7) (40.2) (90.1) (97.0) (98.5) (13.1) (20.0) (98.3) (99.3)
(2.6) (1.7) (29.2) (17.7) (372.7) (0.8) (1.3) (167.8) (117.9)
Severe pDPN (n = 1824)
(57.0) (62.8) (89.8) (96.2) (98.8) (12.3) (21.3) (97.7) (99.3)
All P values are for comparison with diabetes-only cohort.
proportions of patients using each class from DPN to pDPN to severe pDPN. Opioids were the most frequently prescribed pain medication class in each cohort by a large margin, followed by anticonvulsants for DPN, pDPN, and severe pDPN, and NSAIDs for diabetes-only (Fig. 1).
Units of healthcare utilization per patient during the 12-month post-index period were also significantly higher with DPN, pDPN, and severe pDPN relative to diabetes for each resource category (all P b 0.0001) (Table 3). Except for outpatient visits, foot exams, and foot procedures, units used per patient incrementally increased across the DPN spectrum, with the highest number of units among patients with severe pDPN. The proportion of patients using any pain-related medication was substantially and significantly higher across the DPN cohorts relative to diabetes-only (all P b 0.0001) (Fig. 1). These proportions increased from diabetes-only (42.8%) to DPN (65.9%) to pDPN (91.4%) and severe pDPN (94.7%). Similar trends were observed across cohorts for all evaluated pain medication classes (Fig. 1), with increases in the
Diabetes (n=288,328)
3.3. Direct medical costs Parallel with observations for healthcare resource utilization, mean direct medical costs per patient for the post-index period were significantly higher for all evaluated resource categories in the DPN, pDPN, and severe pDPN cohorts relative to diabetes-only (P b 0.0001 for DPN and pDPN and P b 0.05 for severe pDPN) (Table 4). These higher category costs resulted in total direct costs
DPN (n=35,050)
pDPN (n=3449)
Severe pDPN (n=1824)
*P < 0.0001 versus diabetes
Percent of subjects 0
20
40
60
80
100
42.8%
Any pain-related medication
65.9%
*
91.4% 94.7%
*
31.7% 48.7%
Opioids
*
*
82.7% 88.4%
7.1% 31.9%
Anticonvulsants
18.2% 22.6%
NSAIDs
*
*
*
48.2% 54.3%
*
38.0% 41.8%
*
*
5.2%
Antidepressants
13.2%
*20.9%
* *
22.6%
Fig. 1. Pain medication prescriptions among the cohorts for the 12-month period after diagnosis.
*
*
216
A. Sadosky et al. / Journal of Diabetes and Its Complications 29 (2015) 212–217
Table 4 Direct medical costs for the 12-month period after diagnosis for the subset of patients with claims data in the Optum DataMart database. Resource
Emergency room Hospitalizations Office visits All outpatient Prescriptions Total costs a b
Cost per patient, $a Diabetes only (n = 41,928)
DPN (n = 5686)
Mean (SD)
Mean (SD)
Pb
Mean (SD)
Pb
Mean (SD)
Pb
303 (1368) 2068 (12,392) 940 (3265) 3721 (12,438) 843 (3183) 6632 (21,160)
552 3974 1361 7023 1495 12,492
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
1496 (6433) 10,632 (28,211) 1784 (3086) 15,496 (49,981) 1803 (4215) 27,931 (66,663)
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
2084 (8431) 11,324 (31,872) 1753 (3041) 17,337 (58,107) 2094 (4792) 30,755 (74,216)
b0.0001 b0.0001 b0.05 b0.01 b0.001 b0.001
(2395) (16,551) (2754) (26,424) (3869) (36,321)
pDPN (n = 466)
Severe pDPN (n = 260)
Dollar values represent charges rather than actual costs to patients. All P values are for comparison with diabetes cohort.
in the DPN cohort that were 1.8-fold higher than diabetes-only ($12,492 vs. $6,632; P b 0.0001), 4.2-fold higher with pDPN ($27,931; P b 0.0001), and 4.6-fold higher with severe pDPN ($30,755; P b 0.001) (Table 4). 4. Discussion This study is the first to evaluate and compare all-cause healthcare resource utilization and direct medical costs of diabetes relative to the spectrum of DPN, including DPN without pain, pDPN, and severe pDPN, the latter defined by a pain severity score ≥7. The results show that as a result of greater use of healthcare resources, total all-cause medical costs as well as costs for each resource category (based on charges rather than on patient costs) were significantly higher with DPN, pDPN, and severe pDPN relative to diabetes-only. The economic impact of resource utilization increased across cohorts from diabetes ($6632) to DPN ($12,492), pDPN ($27,931), and severe pDPN ($30,755), with the greatest burden among patients with severe pDPN. However, neither the statistical nor clinical relevance of this observation can be determined among the DPN cohorts (i.e. DPN, pDPN, and severe pDPN), since these cohorts were only compared with diabetes-only. Nevertheless, it could be expected that severe pDPN would be associated with higher resource use and costs across the DPN spectrum considering that previous studies suggested a relationship between pain severity and economic burden (Dibonaventura et al., 2011; Sadosky et al., 2013). The healthcare resource utilization and costs for pDPN and severe pDPN are different from those in other studies (daCosta DiBonaventura et al., 2011; Dibonaventura et al., 2011; Dworkin, Malone, Panarites, Armstrong, & Pham, 2010; Sadosky et al., 2013). In particular, the economic impact reported here was substantially higher than several studies that estimated costs based on patient-reported healthcare resource utilization (daCosta DiBonaventura et al., 2011; Dibonaventura et al., 2011; Sadosky et al., 2013). This greater economic impact may be a function of the use of charges rather than actual costs; charges are also likely to have greater variability and may not necessarily be related to the amount paid. However, it should be noted that comparisons in previous studies were generally based on pDPN vs. non-pDPN (daCosta DiBonaventura et al., 2011; Dworkin, Malone, Panarites, Armstrong, & Pham, 2010; Ritzwoller, Crounse, Shetterly, & Rublee, 2006), with the non-pDPN cohort thus reflecting a population that likely included patients with diabetes only as well as patients with DPN, and the pDPN cohort consisting of patients across the pain severity range. The major driver of the difference in resource use, on a percentage basis, for pDPN and severe pDPN relative to diabetes and DPN were ER visits and inpatient hospitalizations, with an especially large difference observed in hospitalizations between severe pDPN (62.8%) and pDPN (40.2%). In this regard, there did not appear to be a difference in pain scores between inpatient and outpatient settings. It may thus be proposed that this equivalence suggests that inpatient admissions were not necessarily attributable to pain flares or for the indication of pain control, highlighting that pDPN is associated with higher
utilization independent of the direct expenses that may be associated with pain control including hospitalization. Rates of office visits and outpatient visits across the groups did not separate clearly along the lines of painful symptoms relative to nonpainful symptoms. Despite the differences in hospitalizations, all outpatient costs were consistently higher than hospitalization costs across cohorts, and appeared to be the primary driver of total costs. Prescription costs provided the smallest contribution to overall costs among the major resource categories for diabetes, and office visits the smallest contribution for DPN, pDPN, and severe pDPN. These results are in contrast to another study by Ritzwoller et al. (Ritzwoller et al., 2009), who reported the primary cost driver was inpatient costs. While hospitalizations are associated with a higher cost per event than outpatient resources, the magnitude of outpatient use in the current study, including a high rate of office visits per patient, exceeded the rate of hospitalizations (Table 3). It was not surprising that the proportion of patients using pain medication prescriptions increased from diabetes to DPN, pDPN, and severe pDPN. However, the opioid use was surprising, not only because these drugs were the most frequently prescribed pain medication including in the diabetes-only cohort, but also for the magnitude of their use, which ranged from 31.7% in the diabetes-only cohort, to N80% in the pDPN and severe pDPN cohorts. While a study by Dworkin et al. (2011) also reported opioids were the most common pain medication in pDPN patients, studies by Hartsfield et al. (Hartsfield et al., 2008) and Sadosky et al. (Sadosky et al., 2013) reported anticonvulsants and antidepressants were the most common pain medications, respectively, followed by opioids. The data presented here indicate that there is discordance with current guidelines, which generally recommend opioids as second or third line treatment for neuropathic pain and pDPN in particular (Argoff et al., 2006; Attal et al., 2010; Dworkin, O'Connor, et al., 2010). Additionally, these data provide supporting evidence for a large presence of painful comorbid conditions and highlight the challenges of managing pain in pDPN patients. There were some notable differences in demographic characteristics among the cohorts. The age of DPN patients was significantly higher than diabetes-only. In contrast, those with severe pDPN were significantly younger than diabetes-only, and were also younger than the DPN and pDPN cohorts. These data are consistent with other studies that reported a younger age among severe pDPN relative to mild/moderate pDPN (Dibonaventura et al., 2011; Sadosky et al., 2013), and suggest that pDPN severity may be more complex than accounted for by disease progression. Further investigation into the relationship between pDPN severity and age is warranted. Not surprisingly, significantly higher proportions of the DPN, pDPN, and severe pDPN cohorts were smokers relative to diabetes. Smoking is well-recognized as a significant risk factor for DPN (Tesfaye et al., 1996, 2005), and it is worth exploring whether smoking cessation incentives should more clearly be aligned with diabetes management strategies, especially among healthcare providers using a pay-forperformance initiative to improve outcomes and reduce costs.
A. Sadosky et al. / Journal of Diabetes and Its Complications 29 (2015) 212–217
While the proportion of pDPN patients with severe pain in the current study, 52.9%, was substantially higher than the 28.6% reported in a study by Sadosky et al. (2013) based on pain scores in patients presenting at community practices, it is similar to the 51.2% based on patient self-report of pain in a study using data derived from the National Health and Wellness Survey (Dibonaventura et al., 2011). Overall, these data not only provide information on the burden of diabetes and its sequela of DPN, but suggest opportunities for improving management strategies. Approaches such as optimized treatment matching, whereby management strategies are implemented based on a patient's specific characteristics, that target subpopulations of patients may help improve healthcare quality and outcomes. For example, younger female smokers with painful symptoms of diabetes could be stratified to programs emphasizing a multimodal approach to disease management that not only encompasses analgesic modalities but also lifestyle modification such as smoking reduction. Thus, early triage of patients may lead to improved clinical outcomes and reduced resource utilization. Furthermore, pDPN and severe pDPN are a particularly important clinical subset of patients to identify, since multiple diabetes comorbidities have insidious presentations. That is, unlike pain that is self-reported when it occurs, injury to end organs, which may be more likely in this population, does not generally result in self-report until serious complications are well underway such as retinal disease, renal impairment, or silent ischemia in patients with coronary artery disease. In this regard, it should be noted that no specific trends were observed across the DPN spectrum in the relevant comorbidities of retinopathy, renal impairment, and the CCI category of “Diabetes with chronic complications.” 4.1. Study limitations Interpretation and generalizability of these results should consider the study's strengths and limitations. An important strength is external validity based on “real world” EMR data from multiple sites. This validity was enhanced by requiring that patients be enrolled in an IDN, allowing capture of more complete information than other types of healthcare management systems. The dataset contained patient-level data that enabled tracking of individuals longitudinally and included types of data that are not generally available in claims databases, such as pain scores from NLP fields. In this regard, it should be noted that the source of the pain scores, i.e. whether due to pDPN or another painful condition, could not specifically be determined. However, the presence of the pain score to identify pDPN was required to be within 15 days of the DPN diagnosis to reduce the likelihood that pain was associated with a comorbid condition; most of the patients with a pain score had the evaluation on the DPN diagnosis date. This study could also be potentially criticized for focusing on type 2 diabetes. However, since type 1 occurs at an earlier age and is generally of greater severity than type 2, it was decided to evaluate a more homogenous population. Since this study was observational, causal inferences cannot be made, and all results should be considered inferential. As with all such database studies, potential errors in coding or record-keeping may exist at the point of the healthcare provider, although the use of large cohorts would likely dilute these errors. Additionally, patients were mainly from the South and Midwest geographic regions. The economic analysis was also based on charges rather than costs, which may overestimate the economic impact, and these data were obtained from a subset of the main sample, further limiting generalizability.
217
5. Conclusions This study provides an initial step in characterizing the spectrum from diabetes to severe pDPN. Patients with DPN, pDPN, and severe pDPN had significantly greater healthcare resource utilization and costs than diabetes-only patients, and the highest burden was associated with severe pDPN. Although additional studies are needed to identify characteristics that may be predictive of pDPN and its high healthcare burden, this study also provides a foundation for future research on developing and implementing management strategies to improve quality of care.
References Argoff, C. E., Backonja, M. M., Belgrade, M. J., Bennett, G. J., Clark, M. R., Cole, B. E., et al. (2006). Consensus guidelines: Treatment planning and options. Mayo Clinic Proceedings, 81, S12–S25. Attal, N., Cruccu, G., Baron, R., Haanpaa, M., Hansson, P., Jensen, T. S., et al. (2010). EFNS guidelines on the pharmacological treatment of neuropathic pain: 2010 revision. European Journal of Neurology, 17, 1113–1123. Benbow, S. J., Wallymahmed, M. E., & Macfarlane, I. A. (1998). Diabetic peripheral neuropathy and quality of life. Quarterly Journal of Medicine, 91, 733–737. daCosta DiBonaventura, M., Cappelleri, J. C., & Joshi, A. V. (2011). A longitudinal assessment of painful diabetic peripheral neuropathy on health status, productivity, and health care utilization and cost. Pain Medicine, 12, 118–126. Dibonaventura, M. D., Cappelleri, J. C., & Joshi, A. V. (2011). Association between pain severity and health care resource use, health status, productivity and related costs in painful diabetic peripheral neuropathy patients. Pain Medicine, 12, 799–807. Dworkin, R. H., Malone, D. C., Panarites, C. J., Armstrong, E. P., & Pham, S. V. (2010). Impact of postherpetic neuralgia and painful diabetic peripheral neuropathy on health care costs. The Journal of Pain, 11, 360–368. Dworkin, R. H., O'Connor, A. B., Audette, J., Baron, R., Gourlay, G. K., Haanpaa, M. L., et al. (2010). Recommendations for the pharmacological management of neuropathic pain: An overview and literature update. Mayo Clinic Proceedings, 85, S3–S14. Dworkin, R. H., Panarites, C. J., Armstrong, E. P., Malone, D. C., & Pham, S. V. (2011). Healthcare utilization in people with postherpetic neuralgia and painful diabetic peripheral neuropathy. Journal of the American Geriatrics Society, 59, 827–836. Eggleston, E. M., & Finkelstein, J. A. (2014). Finding the role of health care in population health. JAMA : The Journal of the American Medical Association, 311, 797–798. Gore, M., Brandenburg, N., Dukes, E., Hoffman, D., Tai, K. -S., & Stacey, B. (2005). Pain severity in diabetic peripheral neuropathy is associated with patient functioning, symptom levels of anxiety and depression, and sleep. Journal of Pain and Symptom Management, 30, 374–385. Hartsfield, C. L., Korner, E. J., Ellis, J. L., Raebel, M. A., Merenich, J., & Brandenburg, N. (2008). Painful diabetic peripheral neuropathy in a managed care setting: patient identification prevalence estimates and pharmacy utilization patterns. Population Health Management, 11, 317–328. Ritzwoller, D. P., Crounse, L., Shetterly, S., & Rublee, D. (2006). The association of comorbidities, utilization and costs for patients identified with low back pain. BMC Musculoskeletal Disorders, 7, 72. Ritzwoller, D. P., Ellis, J. L., Korner, E. J., Hartsfield, C. L., & Sadosky, A. (2009). Comorbidities, healthcare service utilization and costs for patients identified with painful DPN in a managed-care setting. Current Medical Research and Opinion, 25, 1319–1328. Sadosky, A., Schaefer, C., Mann, R., Bergstrom, F., Baik, R., Parsons, B., et al. (2013). Burden of illness associated with painful diabetic peripheral neuropathy among adults seeking treatment in the US: Results from a retrospective chart review and cross-sectional survey. Diabetes, Metabolic Syndrome and Obesity, 5, 1–14. Stewart, W. F., Ricci, J. A., Chee, E., Hirsch, A. G., & Brandenburg, N. (2007). Lost productive time and costs due to diabetes and diabetic neuropathic pain in the US workforce. Journal of Occupational and Environmental Medicine, 49, 672–679. Tesfaye, S., Chaturvedi, N., Eaton, S. E., Ward, J. D., Manes, C., Ionescu-Tirgoviste, C., et al. (2005). Vascular risk factors and diabetic neuropathy. The New England Journal of Medicine, 352, 341–350. Tesfaye, S., Stevens, L. K., Stephenson, J. M., Fuller, J. H., Plater, M., Ionescu-Tirgoviste, C., et al. (1996). Prevalence of diabetic peripheral neuropathy and its relation to glycaemic control and potential risks factors: The EURODIAB IDDM complications study. Diabetologia, 39, 1377–1384. Zelman, D., Dukes, E., Brandenburg, N., Bostrom, A., & Gore, M. (2005). Identification of cut-points for mild, moderate and severe pain due to diabetic peripheral neuropathy. Pain, 115, 29–36.
Contents lists available at ScienceDirect
Journal of Diabetes and Its Complications journal homepage: WWW.JDCJOURNAL.COM
Healthcare utilization and costs in diabetes relative to the clinical spectrum of painful diabetic peripheral neuropathy Alesia Sadosky a,⁎, Jack Mardekian a, Bruce Parsons a, Markay Hopps a, E. Jay Bienen b, John Markman c a b c
Pfizer, Inc., New York, NY, USA Outcomes Research Consultant, New York, NY, USA Translational Pain Research Program, Department of Neurosurgery, University of Rochester Medical Center, Rochester, NY, USA
a r t i c l e
i n f o
Article history: Received 4 September 2014 Received in revised form 30 October 2014 Accepted 31 October 2014 Available online 8 November 2014 Keywords: Diabetes Painful diabetic peripheral neuropathy Electronic medical records Pain severity Healthcare utilization Costs
a b s t r a c t Aims: Diabetic peripheral neuropathy (DPN) accompanied by painful symptoms is known as painful DPN (pDPN). This study characterized healthcare resource utilization and costs in patients with DPN, pDPN, and severe pDPN relative to diabetes only. Methods: Four adult cohorts were identified from the Humedica database: type 2 diabetes without DPN (n = 288,328); DPN (n = 35,050); pDPN (DPN subjects with a pain score ≥ 1 on a 0–10 numeric rating scale; n = 3449); and severe pDPN (pain scores 7–10; n = 1824). Resource utilization and costs for 12-months post-diagnosis were compared for diabetes relative to the other cohorts. Results: Demographic characteristics were different across cohorts. Relative to diabetes alone, DPN, pDPN, and severe pDPN were characterized by significantly higher proportions of patients with resource utilization for all resource categories (all P b 0.0001); the highest resource use generally observed for severe pDPN. Total annual direct medical costs were $6632 for diabetes only, with costs for DPN ($12,492), pDPN ($27,931), and severe pDPN ($30,755) significantly higher (all P b 0.0001); outpatient costs were consistently the primary driver of total costs. Conclusions: Patients with DPN, pDPN, and severe pDPN had significantly greater healthcare resource utilization and costs than patients with diabetes only, with the highest burden associated with severe pDPN. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Diabetic peripheral neuropathy (DPN) is a common neurologic sequela of diabetes, and when the resulting nerve damage is accompanied by painful symptoms it is known as painful DPN (pDPN). The presence of pDPN is associated with a substantial adverse impact on patient function, quality of life, and work productivity, and also results in an economic burden relative to the general population and to patients with diabetes without pDPN
Disclosures: Alesia Sadosky, Jack Mardekian, Bruce Parsons, and Markay Hopps are employees and shareholders of Pfizer, the sponsor of this study; E. Jay Bienen is an independent scientific consultant who was funded by Pfizer in connection with manuscript development; John Markman collaborated with Pfizer on the project but was not financially compensated for his involvement on the project, including manuscript development. Funding: This study was funded by Pfizer, Inc. Conflict of interest statement: There are no other financial relationships or relevant conflicts related to this manuscript. ⁎ Corresponding author at: Pfizer, Inc., 235 East 42nd Street, New York, NY 10017. Tel.: + 1 212 733 9491; fax: + 1 646 441 4757. E-mail address: alesia.sadosky@pfizer.com (A. Sadosky). http://dx.doi.org/10.1016/j.jdiacomp.2014.10.013 1056-8727/© 2015 Elsevier Inc. All rights reserved.
(Benbow, Wallymahmed, & Macfarlane, 1998; daCosta DiBonaventura, Cappelleri, & Joshi, 2011; Dworkin, Malone, Panarites, Armstrong, & Pham, 2010; Dworkin, Panarites, Armstrong, Malone, & Pham, 2011; Gore et al., 2005; Ritzwoller, Ellis, Korner, Hartsfield, & Sadosky, 2009; Stewart, Ricci, Chee, Hirsch, & Brandenburg, 2007). These effects have been reported to be greater as pain severity increases (Dibonaventura, Cappelleri, & Joshi, 2011; Gore et al., 2005; Sadosky et al., 2013). As integrated healthcare systems develop new care delivery models to manage the increasing burden of expensive chronic conditions, such as diabetes, in accord with Affordable Care Act, a deeper understanding of the clinical problems and types of care that disproportionately contribute to the high costs at the population level is required. However, there have been no studies comparing healthcare resource utilization and costs of diabetes relative to DPN, pDPN, and severe pDPN. Electronic medical records (EMR) capture real-world, patient-level data representing integral components of provider care that are not readily available in claims databases including patient-reported outcomes such as pain severity. The availability of these data enables identification of discrete populations and evaluation of resources and costs across inpatient and outpatient settings. This characterization is essential to managed care, especially accountable care organizations, providing the background and understanding required to implement
A. Sadosky et al. / Journal of Diabetes and Its Complications 29 (2015) 212–217
more targeted disease management strategies (Eggleston & Finkelstein, 2014). Therefore, the purpose of this study is to apply EMR-derived clinical information from the Humedica database to evaluate the direct medical costs of patients with diabetes relative to DPN, pDPN, and severe pDPN. This clinical database facilitates identification and management of patients with chronic conditions who are at risk for greater clinical complexity and higher costs of care. 2. Methods 2.1. Data source Data for this retrospective study were derived from the Humedica EMR database, which has broad geographic representation and includes information on demographics, diagnoses, inpatient and outpatient encounters, medications, procedures, lab results, vital signs, and select data derived from physicians' notes. Humedica does not mandate a particular EMR system, and in the more than 20 provider groups, many run multiple EMR installations for different sites of care. Records are linked using a unique patient identifier and are fully compliant with the Health Insurance Portability and Accountability Act (HIPAA). The current analysis utilized structured and unstructured data from the database. Structured data included demographic and clinical characteristics, and healthcare resource utilization (HCRU). Unstructured data were derived from the notes fields within the EMR, which were searched using Humedica's proprietary Natural Language Processing (NLP). The Humedica NLP system uses vocabulary from the Unified Medical Language System including multiple medical dictionaries such as the Logical Observation Identifiers Names and Codes (LOINC), the Systemized Nomenclature of Medicine–Clinical Terms (SNOMED-CT), and RxNorm, a listing of generic and branded drugs (among others). Each NLP concept included in the data is associated with a unique subject record and a date of observation, allowing longitudinal tracking. A specific part of the search of the unstructured data was for pain scores, which are collected by Humedica as a health measure, and are patient-reported using a 0 to 10 scale (0 = no pain, 10 = worst imaginable pain). The scores are recorded during discrete health care provider interactions, although providers may use different methods, including the Visual Analogue Scale (VAS) and the Verbal Numerical Rating Scale (VNRS). These scores may be recorded in the structured data elements of the EMR or harvested from notes using NLP. 2.2. Subjects Subjects ≥18 years of age with type 2 diabetes were identified from the database for the period January 1, 2008–September 30, 2013 based on ICD-9-CM diagnosis codes (250.00-250.93) in the most recent calendar year available (2008–2012) and having continuous data for 1 year pre- and post-diagnosis. Subjects were excluded if they had a diagnosis code for end-stage renal disease, cancer, and/or HIV any time during the study period, were resident in a nursing/inpatient facility in the year prior to the diabetes diagnosis, or had a pregnancy during the pre-index period. Only identified subjects who belonged to an integrated delivery network (IDN) were included. Among subjects who met all criteria, the DPN cohort was defined as subjects with diabetes who had an ICD-9 code for DPN (ICD-9 codes: 357.2 or 250.6). The pDPN cohort consisted of DPN subjects with a pain score ≥ 1 for current pain that was obtained within 15 days (before or after) the DPN diagnosis, and the severe pDPN cohort was pDPN subjects with pain scores 7–10, indicating severe pain based on established criteria (Zelman, Dukes, Brandenburg, Bostrom, & Gore, 2005). The index date, for evaluation of healthcare resource utilization and costs, was the date of ICD-9 diagnosis, i.e. date
213
of diabetes diagnosis in the diabetes-only cohort, and date of DPN diagnosis in the other cohorts. 2.3. Healthcare resource utilization and costs All-cause resource utilization, estimated as the proportion of patients using each resource category, was identified from the claims data for the 12-month post-index period. Resource categories included emergency room (ER) visits, hospitalizations, outpatient resources stratified by office visits and other outpatient visits, laboratory tests, foot procedures and total procedures, and prescriptions. Additionally, prescriptions for pain-related medications were evaluated. All-cause costs were obtained for the subset of patients who are linked to the Optum claims database. The charged, or requested amount that is billed by the provider was used as the cost, although this number may have high variation since it often has little relationship to the actual amount paid. The Optum DataMart is an integrated database consisting of enrollment, inpatient and outpatient medical claims, pharmaceutical claims, and laboratory results. 2.4. Analyses Baseline demographic characteristics of the cohorts were summarized descriptively, and comparisons with the diabetes cohort were performed using likelihood ratio chi-square test for categorical variables and Student's t-test for continuous variables. Proportions of patients using each resource category were compared using likelihood ratio chi-square tests. Direct medical costs were compared using Wilcoxon two-sample rank sum tests. All analyses were conducted using SAS version 9.2 (SAS Institute Inc., Cary, NC). 3. Results 3.1. Cohort populations As shown in Table 1, the four cohorts derived from the 24,257,806 patients in the Humedica database for the specified time period consisted of 288,328 patients with diabetes only; 35,050 with DPN; 3,449 with pDPN; and 1824 with severe pDPN, which represents 52.9% of the patients with pDPN. Among the pDPN patients, there was no difference in mean (SD) pain scores collected in inpatient (n = 1724) and outpatient (n = 1725) settings, 6.4 (2.9) and 6.3 (2.6), respectively. Significant differences were observed for most demographic characteristics between the diabetes-only cohort and the other cohorts (Table 2). In particular, there were significantly lower proportions of females in the diabetes-only cohort (53.2%) relative to pDPN (55.6%; P = 0.005) and severe pDPN (59.7%; P b 0.0001), and whereas severe pDPN patients were significantly younger than diabetes-only (59.4 vs. 61.4 years; P b 0.0001), DPN patients were significantly older (64.8 years; P b 0.0001). The proportion of smokers was also lower in the diabetes-only cohort (28.9%) relative to the other cohorts (P b 0.0001), which appeared to increase from DPN (34.2%) to pDPN (37.8%) to severe pDPN (39.6%). The prevalence of comorbidities across all system classes was significantly higher with DPN, pDPN, and severe pDPN relative to diabetes-only (all P b 0.0001) (Table 2). The higher prevalence of neuropathic pain conditions in the DPN, pDPN, and severe pDPN cohorts was primarily driven by back and neck pain with neuropathic pain involvement, and causalgias (data not shown). Diabetes-only patients had the lowest CCI score, 1.3, with all other cohorts having significantly higher scores; 3.6, 3.9, and 3.9 for DPN, pDPN, and severe pDPN, respectively (all P b 0.0001 vs. diabetes-only). Diabetes-related comorbidities such as retinopathy and the CCI comorbidity of “Diabetes with chronic complications” were also significantly higher for DPN, pDPN,
214
A. Sadosky et al. / Journal of Diabetes and Its Complications 29 (2015) 212–217
Table 1 Sample attrition. Criteria
Number of subjects
Total patients in Humedica database during January 1, 2008–September 30, 2013 With any diabetes diagnosis from January 1, 2008–September 30, 2013 Age ≥18 in the year of the index diabetes diagnosis 12 months of continuous enrollment pre- and post-index Excluding patients with any diagnosis of cancer or diagnosis codes potentially associated with neuropathy of nondiabetic etiology during 1 year (360 days) pre-diabetes-index (Hartsfield et al., 2008; Ritzwoller et al., 2009) Excluding females with any pregnancy diagnosis 1 year pre-index Excluding patients with any end stage renal disease procedure 1 year pre-index Excluding patients who reside in a nursing/inpatient facility 1 year pre-index Excluding patients without continuous enrollment for 12 month pre- and post-DPN index periods for diabetes patients with DPN Excluding patients with any type I diabetes diagnosis on the diabetes index date Patients who belong to an Integrated Delivery Network With a DPN diagnosis on or after diabetes index date (DPN cohort) DPN and with pain score ≥1 within 15 days before or after DPN diagnosis DPN and with pain score ≥7 (severe pDPN cohort) Without any DPN diagnosis on or after diabetes index date (diabetes-only cohort)
and severe pDPN relative to diabetes-only (all P b 0.0001), but were generally similar across the DPN spectrum (Table 2). 3.2. Healthcare resource utilization Relative to diabetes-only, there was greater resource utilization during the 12-month post-index period among patients with DPN, pDPN, and severe pDPN (Table 3), as indicated by significantly higher
Percent of subjects
24,257,806 1,795,461 1,777,371 792,728 673,965
100.00 7.40
100.0 99.0 44.2 37.5
670,247 670,143 637,068 619,034
37.3 37.2 35.5 34.5
590,473 323,378 35,050 3449 1824 288,328
32.9 18.0
100.0 10.8 9.8 5.2 89.2
proportions of these patients using each of the resource categories (all P b 0.0001). While there appeared to be a tendency toward increased resource use from diabetes-only patients to DPN to pDPN, fewer differences were observed between pDPN and severe pDPN (Table 3). Relative to the pDPN cohort, higher proportions of severe pDPN patients had ER visits (57.0 vs. 50.7%) and hospitalizations (62.8 vs. 40.2%), with the proportion of patients using other resources generally similar between these two cohorts.
Table 2 Demographic and clinical characteristics of the cohorts. Variable
Diabetes only (n = 288,328)
DPN (n = 35,050)
Age, mean (SD) Gender, n (%) Female Male Race, n (%) African American Asian Caucasian Other/unknown Smoker, n (%) Region Midwest Northeast South West Unknown Average household income, mean (SD) Body mass index, kg/m2, mean (SD) Comorbidities, n (%) Cardiovascular disorders Gastrointestinal disorders Mental disorders Sleep disorders Musculoskeletal pain Neuropathic pain conditions Restless leg syndrome Fibromyalgia Myocardial infarction Retinopathy Charlson Comorbidity Index, mean (SD) Diabetes with chronic complications, n (%) Renal disease, n (%) Total cholesterol, mg/dL, mean (SD) Low density lipoprotein cholesterol, mg/dL, mean (SD) Diastolic blood pressure, mmHg, mean (SD) Fibrinogen activity, mg/dL, mean (SD) HgbA1C, %, mean (SD) a
All P values are for comparison with diabetes cohort.
pDPN (n = 3449)
Pa
Value
b0.0001 b0.0001
61.4 (13.4)
64.8 (12.2)
153,463 (53.2) 134,761 (46.8)
17,871 (51.0) 17,166 (49.0)
42,476 (14.7) 3209 (1.1) 219,016 (76.0) 23,627 (8.2) 83,220 (28.9)
5328 204 27,460 2058 11,980
(15.2) (0.6) (78.3) (5.9) (34.2)
140,213 (48.6) 12,515 (4.3) 123,907 (43.0) 6537 (2.3) 5156 (1.8) $39,740 (9918) 33.6 (7.4)
17,889 1926 13,736 809 690 $39,579 33.7
(51.0) (5.5) (39.2) (2.4) (2.0) (9690) (7.5)
0.0044 0.0642
102,183(35.4) 30,288 (10.5) 16,849 (5.8) 7840 (2.7) 60,348 (20.9) 9085 (3.2) 855 (0.3) 4135 (1.4) 1941 (0.7) 1452 (0.5) 1.3 (0.7) 7109 (2.5) 8078 (2.8) 179.8 (39.7) 101.4 (32.9) 76.4 (11.2) 355.3 (165.9) 7.2 (1.5)
24,194 8519 5168 2753 15,633 3827 534 1371 1403 1375 3.6 32,328 4995 170.0 91.7 73.4 404.3 7.6
(69.0) (4.3) (14.7) (7.9) (44.6) (10.9) (1.5) (3.9) (4.0) (3.9) (1.6) (92.2) (14.3) (40.1) (32.4) (11.4) (210.4) (1.7)
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 0.0002 b0.0001
Severe pDPN (n = 1824) Pa
Value 61.0 (12.9)
0.089 0.005
1918 (55.6) 1529 (44.4) b0.0001
b0.0001 b0.0001
Pa
Value 59.4 (13.2) 1088 (59.7) 735 (40.3)
b0.0001 630 (18.3) 7 (0.2) 2607 (75.6) 205 (5.9) 1302 (37.8) 1305 (37.8) 8 (0.2) 1788 (51.8) 260 (7.5) 88 (2.6) $36,746 (6913) 33.8 (7.9) 2354 1028 637 360 1789 465 62 178 200 62 3.9 3268 679 174.0 94.9 73.3 404.7 8.0
(68.3) (29.8) (18.5) (10.4) (51.9) (13.5) (1.8) (5.2) (5.8) (1.8) (1.8) (94.8) (19.7) (43.9) (35.6) (12.6) (204.8) (1.9)
b0.0001 b0.0001
b0.0001 b0.0001
b0.0001 386 4 1319 115 722
(21.2) (0.2) (72.3) (6.3) (39.6)
b0.0001 b0.0001
b0.0001 0.1926
606 2 1035 119 62 $36,118 33.8
(33.2) (0.1) (56.7) (6.5) (3.4) (6674) (7.9)
b0.0001 0.2426
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 0.0422 b0.0001
1215 586 355 192 1012 263 35 107 93 28 3.9 1743 1824 175.8 96.8 73.8 437.5 8.1
(66.6) (32.1) (19.5) (10.5) (55.5) (14.4) (1.9) (5.9) (5.1) (1.5) (1.8) (95.6) (18.9) (45.3) (36.4) (12.9) (223.8) (2.0)
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 0.0033 b0.0001 b0.0001 0.0112 b0.0001
A. Sadosky et al. / Journal of Diabetes and Its Complications 29 (2015) 212–217
215
Table 3 Healthcare resource utilization for the 12-month period after diagnosis. Resource
Diabetes only (n = 288,328)
DPN (n = 35,050)
Frequency of use
Number (%) of patients
Number (%) of patients
Pa
Number (%) of patients
Pa
Number (%) of patients
Pa
Emergency room visits Inpatient hospitalizations Office visits Outpatient visits (excluding office visits) Laboratory tests Foot exam Foot procedures Total procedures Prescriptions written
52,845 (18.2) 43,783 (15.2) 228,121 (79.1) 254,217 (88.2) 255,531 (88.6) 20,034 (6.9) 12,229 (4.2) 260,874 (90.5) 253,610 (88.0)
9537 (27.2) 10,174 (29.0) 30,390 (86.7) 32,517 (92.8) 32,330 (92.2) 5349 (15.3) 6575 (18.8) 32,911 (93.9) 33,020 (94.2)
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
1750 1386 3107 3346 3397 451 691 3391 3424
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
1040 1146 1638 1755 1803 225 388 1782 1812
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
Units per patient
Mean (SD)
Mean (SD)
Pa
Mean (SD)
Pa
Mean (SD)
Pa
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
1.4 1.3 20.2 8.3 280.8 0.2 0.5 108.5 102.9
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
1.7 (3.1) 1.4 (1.9) 21.6 (32.0) 7.5 (15.8) 313.9 (410.3) 0.2 (0.9) 0.5 (1.4) 117.3 (185.7) 114.3 (130.1)
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
Emergency room visits Inpatient hospitalizations Office visits Outpatient visits (excluding office visits) Laboratory tests Foot exam Foot procedures Total procedures Prescriptions written a
0.3 0.2 8.9 2.7 76.7 0.1 0.1 29.5 24.1
(1.0) (0.7) (12.7) (6.8) (131.1) (0.4) (0.5) (52.1) (47.1)
pDPN (n = 3449)
0.6 (1.6) 0.5 (1.2) 16.2 (20.6) 4.8 (11.2) 145.7 (234.4) 0.2 (0.7) 0.5 (1.3) 61.1 (109.5) 55.4 (100)
(50.7) (40.2) (90.1) (97.0) (98.5) (13.1) (20.0) (98.3) (99.3)
(2.6) (1.7) (29.2) (17.7) (372.7) (0.8) (1.3) (167.8) (117.9)
Severe pDPN (n = 1824)
(57.0) (62.8) (89.8) (96.2) (98.8) (12.3) (21.3) (97.7) (99.3)
All P values are for comparison with diabetes-only cohort.
proportions of patients using each class from DPN to pDPN to severe pDPN. Opioids were the most frequently prescribed pain medication class in each cohort by a large margin, followed by anticonvulsants for DPN, pDPN, and severe pDPN, and NSAIDs for diabetes-only (Fig. 1).
Units of healthcare utilization per patient during the 12-month post-index period were also significantly higher with DPN, pDPN, and severe pDPN relative to diabetes for each resource category (all P b 0.0001) (Table 3). Except for outpatient visits, foot exams, and foot procedures, units used per patient incrementally increased across the DPN spectrum, with the highest number of units among patients with severe pDPN. The proportion of patients using any pain-related medication was substantially and significantly higher across the DPN cohorts relative to diabetes-only (all P b 0.0001) (Fig. 1). These proportions increased from diabetes-only (42.8%) to DPN (65.9%) to pDPN (91.4%) and severe pDPN (94.7%). Similar trends were observed across cohorts for all evaluated pain medication classes (Fig. 1), with increases in the
Diabetes (n=288,328)
3.3. Direct medical costs Parallel with observations for healthcare resource utilization, mean direct medical costs per patient for the post-index period were significantly higher for all evaluated resource categories in the DPN, pDPN, and severe pDPN cohorts relative to diabetes-only (P b 0.0001 for DPN and pDPN and P b 0.05 for severe pDPN) (Table 4). These higher category costs resulted in total direct costs
DPN (n=35,050)
pDPN (n=3449)
Severe pDPN (n=1824)
*P < 0.0001 versus diabetes
Percent of subjects 0
20
40
60
80
100
42.8%
Any pain-related medication
65.9%
*
91.4% 94.7%
*
31.7% 48.7%
Opioids
*
*
82.7% 88.4%
7.1% 31.9%
Anticonvulsants
18.2% 22.6%
NSAIDs
*
*
*
48.2% 54.3%
*
38.0% 41.8%
*
*
5.2%
Antidepressants
13.2%
*20.9%
* *
22.6%
Fig. 1. Pain medication prescriptions among the cohorts for the 12-month period after diagnosis.
*
*
216
A. Sadosky et al. / Journal of Diabetes and Its Complications 29 (2015) 212–217
Table 4 Direct medical costs for the 12-month period after diagnosis for the subset of patients with claims data in the Optum DataMart database. Resource
Emergency room Hospitalizations Office visits All outpatient Prescriptions Total costs a b
Cost per patient, $a Diabetes only (n = 41,928)
DPN (n = 5686)
Mean (SD)
Mean (SD)
Pb
Mean (SD)
Pb
Mean (SD)
Pb
303 (1368) 2068 (12,392) 940 (3265) 3721 (12,438) 843 (3183) 6632 (21,160)
552 3974 1361 7023 1495 12,492
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
1496 (6433) 10,632 (28,211) 1784 (3086) 15,496 (49,981) 1803 (4215) 27,931 (66,663)
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
2084 (8431) 11,324 (31,872) 1753 (3041) 17,337 (58,107) 2094 (4792) 30,755 (74,216)
b0.0001 b0.0001 b0.05 b0.01 b0.001 b0.001
(2395) (16,551) (2754) (26,424) (3869) (36,321)
pDPN (n = 466)
Severe pDPN (n = 260)
Dollar values represent charges rather than actual costs to patients. All P values are for comparison with diabetes cohort.
in the DPN cohort that were 1.8-fold higher than diabetes-only ($12,492 vs. $6,632; P b 0.0001), 4.2-fold higher with pDPN ($27,931; P b 0.0001), and 4.6-fold higher with severe pDPN ($30,755; P b 0.001) (Table 4). 4. Discussion This study is the first to evaluate and compare all-cause healthcare resource utilization and direct medical costs of diabetes relative to the spectrum of DPN, including DPN without pain, pDPN, and severe pDPN, the latter defined by a pain severity score ≥7. The results show that as a result of greater use of healthcare resources, total all-cause medical costs as well as costs for each resource category (based on charges rather than on patient costs) were significantly higher with DPN, pDPN, and severe pDPN relative to diabetes-only. The economic impact of resource utilization increased across cohorts from diabetes ($6632) to DPN ($12,492), pDPN ($27,931), and severe pDPN ($30,755), with the greatest burden among patients with severe pDPN. However, neither the statistical nor clinical relevance of this observation can be determined among the DPN cohorts (i.e. DPN, pDPN, and severe pDPN), since these cohorts were only compared with diabetes-only. Nevertheless, it could be expected that severe pDPN would be associated with higher resource use and costs across the DPN spectrum considering that previous studies suggested a relationship between pain severity and economic burden (Dibonaventura et al., 2011; Sadosky et al., 2013). The healthcare resource utilization and costs for pDPN and severe pDPN are different from those in other studies (daCosta DiBonaventura et al., 2011; Dibonaventura et al., 2011; Dworkin, Malone, Panarites, Armstrong, & Pham, 2010; Sadosky et al., 2013). In particular, the economic impact reported here was substantially higher than several studies that estimated costs based on patient-reported healthcare resource utilization (daCosta DiBonaventura et al., 2011; Dibonaventura et al., 2011; Sadosky et al., 2013). This greater economic impact may be a function of the use of charges rather than actual costs; charges are also likely to have greater variability and may not necessarily be related to the amount paid. However, it should be noted that comparisons in previous studies were generally based on pDPN vs. non-pDPN (daCosta DiBonaventura et al., 2011; Dworkin, Malone, Panarites, Armstrong, & Pham, 2010; Ritzwoller, Crounse, Shetterly, & Rublee, 2006), with the non-pDPN cohort thus reflecting a population that likely included patients with diabetes only as well as patients with DPN, and the pDPN cohort consisting of patients across the pain severity range. The major driver of the difference in resource use, on a percentage basis, for pDPN and severe pDPN relative to diabetes and DPN were ER visits and inpatient hospitalizations, with an especially large difference observed in hospitalizations between severe pDPN (62.8%) and pDPN (40.2%). In this regard, there did not appear to be a difference in pain scores between inpatient and outpatient settings. It may thus be proposed that this equivalence suggests that inpatient admissions were not necessarily attributable to pain flares or for the indication of pain control, highlighting that pDPN is associated with higher
utilization independent of the direct expenses that may be associated with pain control including hospitalization. Rates of office visits and outpatient visits across the groups did not separate clearly along the lines of painful symptoms relative to nonpainful symptoms. Despite the differences in hospitalizations, all outpatient costs were consistently higher than hospitalization costs across cohorts, and appeared to be the primary driver of total costs. Prescription costs provided the smallest contribution to overall costs among the major resource categories for diabetes, and office visits the smallest contribution for DPN, pDPN, and severe pDPN. These results are in contrast to another study by Ritzwoller et al. (Ritzwoller et al., 2009), who reported the primary cost driver was inpatient costs. While hospitalizations are associated with a higher cost per event than outpatient resources, the magnitude of outpatient use in the current study, including a high rate of office visits per patient, exceeded the rate of hospitalizations (Table 3). It was not surprising that the proportion of patients using pain medication prescriptions increased from diabetes to DPN, pDPN, and severe pDPN. However, the opioid use was surprising, not only because these drugs were the most frequently prescribed pain medication including in the diabetes-only cohort, but also for the magnitude of their use, which ranged from 31.7% in the diabetes-only cohort, to N80% in the pDPN and severe pDPN cohorts. While a study by Dworkin et al. (2011) also reported opioids were the most common pain medication in pDPN patients, studies by Hartsfield et al. (Hartsfield et al., 2008) and Sadosky et al. (Sadosky et al., 2013) reported anticonvulsants and antidepressants were the most common pain medications, respectively, followed by opioids. The data presented here indicate that there is discordance with current guidelines, which generally recommend opioids as second or third line treatment for neuropathic pain and pDPN in particular (Argoff et al., 2006; Attal et al., 2010; Dworkin, O'Connor, et al., 2010). Additionally, these data provide supporting evidence for a large presence of painful comorbid conditions and highlight the challenges of managing pain in pDPN patients. There were some notable differences in demographic characteristics among the cohorts. The age of DPN patients was significantly higher than diabetes-only. In contrast, those with severe pDPN were significantly younger than diabetes-only, and were also younger than the DPN and pDPN cohorts. These data are consistent with other studies that reported a younger age among severe pDPN relative to mild/moderate pDPN (Dibonaventura et al., 2011; Sadosky et al., 2013), and suggest that pDPN severity may be more complex than accounted for by disease progression. Further investigation into the relationship between pDPN severity and age is warranted. Not surprisingly, significantly higher proportions of the DPN, pDPN, and severe pDPN cohorts were smokers relative to diabetes. Smoking is well-recognized as a significant risk factor for DPN (Tesfaye et al., 1996, 2005), and it is worth exploring whether smoking cessation incentives should more clearly be aligned with diabetes management strategies, especially among healthcare providers using a pay-forperformance initiative to improve outcomes and reduce costs.
A. Sadosky et al. / Journal of Diabetes and Its Complications 29 (2015) 212–217
While the proportion of pDPN patients with severe pain in the current study, 52.9%, was substantially higher than the 28.6% reported in a study by Sadosky et al. (2013) based on pain scores in patients presenting at community practices, it is similar to the 51.2% based on patient self-report of pain in a study using data derived from the National Health and Wellness Survey (Dibonaventura et al., 2011). Overall, these data not only provide information on the burden of diabetes and its sequela of DPN, but suggest opportunities for improving management strategies. Approaches such as optimized treatment matching, whereby management strategies are implemented based on a patient's specific characteristics, that target subpopulations of patients may help improve healthcare quality and outcomes. For example, younger female smokers with painful symptoms of diabetes could be stratified to programs emphasizing a multimodal approach to disease management that not only encompasses analgesic modalities but also lifestyle modification such as smoking reduction. Thus, early triage of patients may lead to improved clinical outcomes and reduced resource utilization. Furthermore, pDPN and severe pDPN are a particularly important clinical subset of patients to identify, since multiple diabetes comorbidities have insidious presentations. That is, unlike pain that is self-reported when it occurs, injury to end organs, which may be more likely in this population, does not generally result in self-report until serious complications are well underway such as retinal disease, renal impairment, or silent ischemia in patients with coronary artery disease. In this regard, it should be noted that no specific trends were observed across the DPN spectrum in the relevant comorbidities of retinopathy, renal impairment, and the CCI category of “Diabetes with chronic complications.” 4.1. Study limitations Interpretation and generalizability of these results should consider the study's strengths and limitations. An important strength is external validity based on “real world” EMR data from multiple sites. This validity was enhanced by requiring that patients be enrolled in an IDN, allowing capture of more complete information than other types of healthcare management systems. The dataset contained patient-level data that enabled tracking of individuals longitudinally and included types of data that are not generally available in claims databases, such as pain scores from NLP fields. In this regard, it should be noted that the source of the pain scores, i.e. whether due to pDPN or another painful condition, could not specifically be determined. However, the presence of the pain score to identify pDPN was required to be within 15 days of the DPN diagnosis to reduce the likelihood that pain was associated with a comorbid condition; most of the patients with a pain score had the evaluation on the DPN diagnosis date. This study could also be potentially criticized for focusing on type 2 diabetes. However, since type 1 occurs at an earlier age and is generally of greater severity than type 2, it was decided to evaluate a more homogenous population. Since this study was observational, causal inferences cannot be made, and all results should be considered inferential. As with all such database studies, potential errors in coding or record-keeping may exist at the point of the healthcare provider, although the use of large cohorts would likely dilute these errors. Additionally, patients were mainly from the South and Midwest geographic regions. The economic analysis was also based on charges rather than costs, which may overestimate the economic impact, and these data were obtained from a subset of the main sample, further limiting generalizability.
217
5. Conclusions This study provides an initial step in characterizing the spectrum from diabetes to severe pDPN. Patients with DPN, pDPN, and severe pDPN had significantly greater healthcare resource utilization and costs than diabetes-only patients, and the highest burden was associated with severe pDPN. Although additional studies are needed to identify characteristics that may be predictive of pDPN and its high healthcare burden, this study also provides a foundation for future research on developing and implementing management strategies to improve quality of care.
References Argoff, C. E., Backonja, M. M., Belgrade, M. J., Bennett, G. J., Clark, M. R., Cole, B. E., et al. (2006). Consensus guidelines: Treatment planning and options. Mayo Clinic Proceedings, 81, S12–S25. Attal, N., Cruccu, G., Baron, R., Haanpaa, M., Hansson, P., Jensen, T. S., et al. (2010). EFNS guidelines on the pharmacological treatment of neuropathic pain: 2010 revision. European Journal of Neurology, 17, 1113–1123. Benbow, S. J., Wallymahmed, M. E., & Macfarlane, I. A. (1998). Diabetic peripheral neuropathy and quality of life. Quarterly Journal of Medicine, 91, 733–737. daCosta DiBonaventura, M., Cappelleri, J. C., & Joshi, A. V. (2011). A longitudinal assessment of painful diabetic peripheral neuropathy on health status, productivity, and health care utilization and cost. Pain Medicine, 12, 118–126. Dibonaventura, M. D., Cappelleri, J. C., & Joshi, A. V. (2011). Association between pain severity and health care resource use, health status, productivity and related costs in painful diabetic peripheral neuropathy patients. Pain Medicine, 12, 799–807. Dworkin, R. H., Malone, D. C., Panarites, C. J., Armstrong, E. P., & Pham, S. V. (2010). Impact of postherpetic neuralgia and painful diabetic peripheral neuropathy on health care costs. The Journal of Pain, 11, 360–368. Dworkin, R. H., O'Connor, A. B., Audette, J., Baron, R., Gourlay, G. K., Haanpaa, M. L., et al. (2010). Recommendations for the pharmacological management of neuropathic pain: An overview and literature update. Mayo Clinic Proceedings, 85, S3–S14. Dworkin, R. H., Panarites, C. J., Armstrong, E. P., Malone, D. C., & Pham, S. V. (2011). Healthcare utilization in people with postherpetic neuralgia and painful diabetic peripheral neuropathy. Journal of the American Geriatrics Society, 59, 827–836. Eggleston, E. M., & Finkelstein, J. A. (2014). Finding the role of health care in population health. JAMA : The Journal of the American Medical Association, 311, 797–798. Gore, M., Brandenburg, N., Dukes, E., Hoffman, D., Tai, K. -S., & Stacey, B. (2005). Pain severity in diabetic peripheral neuropathy is associated with patient functioning, symptom levels of anxiety and depression, and sleep. Journal of Pain and Symptom Management, 30, 374–385. Hartsfield, C. L., Korner, E. J., Ellis, J. L., Raebel, M. A., Merenich, J., & Brandenburg, N. (2008). Painful diabetic peripheral neuropathy in a managed care setting: patient identification prevalence estimates and pharmacy utilization patterns. Population Health Management, 11, 317–328. Ritzwoller, D. P., Crounse, L., Shetterly, S., & Rublee, D. (2006). The association of comorbidities, utilization and costs for patients identified with low back pain. BMC Musculoskeletal Disorders, 7, 72. Ritzwoller, D. P., Ellis, J. L., Korner, E. J., Hartsfield, C. L., & Sadosky, A. (2009). Comorbidities, healthcare service utilization and costs for patients identified with painful DPN in a managed-care setting. Current Medical Research and Opinion, 25, 1319–1328. Sadosky, A., Schaefer, C., Mann, R., Bergstrom, F., Baik, R., Parsons, B., et al. (2013). Burden of illness associated with painful diabetic peripheral neuropathy among adults seeking treatment in the US: Results from a retrospective chart review and cross-sectional survey. Diabetes, Metabolic Syndrome and Obesity, 5, 1–14. Stewart, W. F., Ricci, J. A., Chee, E., Hirsch, A. G., & Brandenburg, N. (2007). Lost productive time and costs due to diabetes and diabetic neuropathic pain in the US workforce. Journal of Occupational and Environmental Medicine, 49, 672–679. Tesfaye, S., Chaturvedi, N., Eaton, S. E., Ward, J. D., Manes, C., Ionescu-Tirgoviste, C., et al. (2005). Vascular risk factors and diabetic neuropathy. The New England Journal of Medicine, 352, 341–350. Tesfaye, S., Stevens, L. K., Stephenson, J. M., Fuller, J. H., Plater, M., Ionescu-Tirgoviste, C., et al. (1996). Prevalence of diabetic peripheral neuropathy and its relation to glycaemic control and potential risks factors: The EURODIAB IDDM complications study. Diabetologia, 39, 1377–1384. Zelman, D., Dukes, E., Brandenburg, N., Bostrom, A., & Gore, M. (2005). Identification of cut-points for mild, moderate and severe pain due to diabetic peripheral neuropathy. Pain, 115, 29–36.
