Current Medical Research & Opinion 0300-7995 doi:10.1185/03007995.2015.1047748
2015, 1–8
Article RT-0115.R1/1047748 All rights reserved: reproduction in whole or part not permitted
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 06/21/15 For personal use only.
Brief report An observational study of glycemic control in canagliflozin treated patients
L.M. Meckley G. Miyasato F. Kokkotos Trinity Partners LLC, Waltham, MA, USA
Abstract Objective: To evaluate changes in glycemic control following the initial canagliflozin pharmacy claim in a real-world population.
J. Bumbaugh Inovalon Inc., Bowie, MD, USA
R.A. Bailey Health Economics and Outcomes Research, Janssen Scientific Affairs LLC, Raritan, NJ, USA Address for correspondence: Robert A. Bailey MD, 1000 US Route 202 South, Raritan, NJ 08869, USA. [email protected] Keywords: Aged – Diabetes mellitus, type 2 – Quality indicators, health care – Sodium–glucose co-transporter 2 inhibitor Accepted: 27 April 2015; published online: 17 June 2015 Citation: Curr Med Res Opin 2015; 1–8
Research design and methods: A retrospective cohort analysis of adult patients with type 2 diabetes mellitus (T2DM) was conducted using 2013 medical, pharmacy and laboratory claims from the Inovalon MORE2 Registry. Patients with T2DM aged 18 years with 60 days of canagliflozin supply and HbA1c test results within 120 days before and 60 days after initial canagliflozin claim (defined as index date) were included. The differences between HbA1c levels pre- and post-index were assessed. Changes pre- and post-index in Healthcare Effectiveness Data and Information Set (HEDIS) glycemic control criteria of HbA1c57% and58% and poor control of HbA1c 49% were evaluated. Subgroup analyses of patients with HbA1c47% at baseline and patients aged 65 were also conducted. Results: Among the 268 patients meeting the study criteria, mean HbA1c pre-index was 8.3% and post-index was 7.6%; the mean reduction in HbA1c pre–post index was 0.7% (95% CI: 0.6%, 0.9%). The proportions of patients meeting the HEDIS glycemic control measures (HbA1c 57%, 58% and poor control of 49%) improved and was significantly different pre- and post-index (all p50.001). Of the patients with an HbA1c 47% prior to index (81% of the cohort; mean pre-index HbA1c ¼ 8.8%), HbA1c was reduced by 0.9% (95% CI: 0.8%, 1.1%). The aged 65 subgroup consisted of 15% of the cohort, with a pre-index HbA1c of 8.3%. The mean reduction in HbA1c test results pre- and post-canagliflozin index was 0.6% (95% CI: 0.4%, 0.9%). This analysis did not adjust for changes in antihyperglycemic agents during the study period. Conclusion: Patients with T2DM were observed to have improved glycemic control following initial canagliflozin pharmacy claim as measured by HbA1c change and attainment of specific glycemic control criteria.
Introduction Diabetes is one of the most common chronic conditions in the US, affecting 29.1 million people, approximately 9.3% of the population1. These patients incur approximately $245 million in direct and indirect costs annually, with the majority of these costs due to the treatment of complications associated with poorly controlled diabetes2,3. The mortality burden of diabetes is also high, as diabetes is the seventh leading cause of death in the US4. Glycemic control, as measured by glycosylated hemoglobin (HbA1c), is a key clinical measure of diabetes control. Lower HbA1c levels are associated with reduced risk of microvascular diabetes complications5. The current position of the American Diabetes Association (ADA) and the European Association for ! 2015 Informa UK Ltd www.cmrojournal.com
Observational study of glycemic control in canagliflozin pts Meckley et al.
1
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 06/21/15 For personal use only.
Current Medical Research & Opinion
2015
the Study of Diabetes, is that glycemic targets should be individualized to patients on the basis of the patient’s comorbidities, life expectancy and hypoglycemia history, with the goal for most patients of HbA1c 57%5. The Healthcare Effectiveness Data and Information Set (HEDIS) is a tool to measure health plan performance and includes quality measures for a variety of chronic illnesses. As part of its comprehensive diabetes care measures, HEDIS classifies HbA1c as a quality measure6. For the purpose of evaluating the quality of diabetes care, HEDIS has set HbA1c 58% as the glycemic control threshold. For the subgroup of patients who are aged 565 who do not have specific comorbidities, HEDIS has an additional HbA1c control threshold of HbA1c 57%. HEDIS measures also assess poor glycemic control, defined as HbA1c 49%. Canagliflozin is an antihyperglycemic agent that inhibits sodium–glucose co-transporter 2 (SGLT2) and was FDA approved on 29 March 2013. SGLT2 inhibition results in a lowering of the renal threshold for glucose and an increased urinary excretion of glucose, lowering blood glucose concentrations7. In clinical trials, canagliflozin has been shown to improve glycemic control as monotherapy, dual therapy and triple therapy in patients with type 2 diabetes mellitus (T2DM)7–13. In the trials, older patients were shown to have smaller reductions in HbA1c with canagliflozin14. As the efficacy of SGLT2 inhibitors is affected by renal function, the reduced efficacy of canagliflozin observed in older adults is consistent with the lower eGFR typically found in these patients15. In addition to glycemic control, canagliflozin lowers body weight and systolic blood pressure7–13. Furthermore, canagliflozin was generally well tolerated, without an increased risk of hypoglycemia. Canagliflozin is associated with adverse events related to the mechanism of action of SGLT2 inhibitors, such as genital mycotic infections and osmotic diuresis related adverse events7–13. The objectives of this study were to evaluate changes in glycemic control after the initial canagliflozin pharmacy claim in a real-world population as measured by HbA1c, and to assess whether the proportion of patients achieving glycemic control, as defined by HEDIS and the ADA, improved after the initial canagliflozin pharmacy claim.
Patients and methods Data source The data for this study were obtained from the MORE2 Registry* (Medical Outcomes Research for Effectiveness and Economics Registry) Research Edition, a national healthcare data warehouse created by Inovalon Inc. The *MORE2 Registry is a registered trademark of Inovalon Inc.
2
Observational study of glycemic control in canagliflozin pts Meckley et al.
MORE2 Registry contains data from more than 120 million de-identified individuals enrolled in over 100 commercial, Medicare and Medicaid health plans16. The database has been applied in support of quality measurement development and research aimed at improvements in population health. This analysis used diagnosis, procedure, prescription and laboratory data from the MORE2 Registry.
Study design and patient selection A retrospective cohort design was used to assess glycemic control in patients receiving canagliflozin. Adult patients with T2DM who had a canagliflozin claim during 2013 were included. Similar to the HEDIS definition of patients with diabetes, in this study, T2DM was defined as at least one face-to-face encounter in an acute inpatient or emergency department (ED) setting with a diagnosis of T2DM (ICD-9-CM diagnosis codes 250.x0 or 250.x2 reported in any position) or two face-to-face encounters in an outpatient setting or non-acute inpatient setting on separate dates with a diagnosis of T2DM, during 2012 or 2013. The index date was defined as the date of the first canagliflozin claim. Because some patients may have received canagliflozin samples prior to the first canagliflozin claim, the index claim is a proxy for canagliflozin initiation. In order to assess the change in glycemic control, patients were required to have obtained 60 days of canagliflozin supply and have an HbA1c test result within 120 days preand at least 60 days post-index claim (Figure 1). Thus, patients who initiated after 1 November 2013 were excluded. Patients aged518, with type 1 diabetes mellitus (T1DM), gestational diabetes, polycystic ovary syndrome, steroid induced diabetes, enrollment gaps of 445 days or missing data for key variables were also excluded.
Outcomes and statistical analyses Univariate descriptive statistics were calculated for demographic, clinical characteristics and medication usage at baseline. The diabetes complication severity index (DCSI) and Quan–Charlson comorbidity index (QCCI) were also calculated to assess comorbidities. The DCSI and QCCI predict mortality, and the DCSI also predicts the risk of hospitalization17,18. Both indexes are validated measures and can be assessed using claims data. Continuous outcomes were reported as means and standard deviations; frequencies were calculated for categorical data. HbA1c was evaluated pre- and post-index, with means, 95% confidence intervals and medians assessed. The preindex time period (baseline) was defined as 120 days prior to index. The post-canagliflozin time period was defined as 60 days from index to 31 December 2013 in www.cmrojournal.com ! 2015 Informa UK Ltd
Current Medical Research & Opinion
Post-Index HbA1c Test Period†
Baseline HbA1c Test Period*
Jan 2013
April 2013 Canagliflozin Launch
2015
Dec 2013 60 Days Canagliflozin Supply
Index Date: First Canagliflozin Claim
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 06/21/15 For personal use only.
Figure 1. Study design. *Baseline period consists of the 4 months prior to canagliflozin index claim. yPost-index period begins 60 days post canagliflozin index claim and ends 31 December 2013.
order to allow the post-canagliflozin measurement to primarily reflect time on canagliflozin (Figure 1). For patients with multiple pre- or post-index HbA1c laboratory results, the most recent results were used for analysis. The difference between HbA1c levels pre- and post-index was tested using a paired t-test. Additionally, the proportion of patients meeting the HEDIS6 and ADA19 recommendations for HbA1c thresholds pre- and post-index were compared using McNemar’s test for significance, specifically: HbA1c control 58.0% HbA1c control 57.0% HbA1c poor control 49.0% Subgroup analyses of all outcome measures were conducted to assess glycemic control in patients receiving canagliflozin who had HbA1c 47% prior to index and those aged 65. The subgroup analysis of patients with pre-index HbA1c47% was conducted to evaluate consistency of results with the clinical trial population, as this threshold was typically used as a lower limit for inclusion. Older patients were shown to have smaller reductions in HbA1c with canagliflozin in clinical trials14, which is consistent with older adults having lower eGFR15, and thus reduced efficacy of agents that inhibit SGLT2. Therefore, the purpose of the subgroup analysis of patients aged 65 was to descriptively assess changes in HbA1c after canagliflozin index in this population in a real-world setting. HbA1c measures glucose levels over approximately 90 days, so we conducted a descriptive analysis of the temporal distribution of HbA1c tests pre- and postcanagliflozin index to assist with the interpretation of HbA1c in the context of patients’ treatment. Statistical analysis was performed using SAS 9.2 (SAS Institute, Cary, NC, USA).
Results Demographics and cohort The MORE2 Registry for the 2013 calendar year contained 1,918,778 patients with a diagnosis of T2DM. ! 2015 Informa UK Ltd www.cmrojournal.com
After clinical and administrative exclusions, 2619 patients with canagliflozin prescription claims were identified. After restrictions for canagliflozin usage and HbA1c results, 268 patients met the inclusion criteria for the study (Figure 2). Baseline demographics and clinical characteristics are presented in Table 1. Most patients (81%) had HbA1c 47% at baseline. Approximately 15% of patients were aged 65 years. Most patients in the cohort were treated with multiple antihyperglycemic agents at baseline, with metformin (48%) and sulfonylureas (46%) as the most common agents (Table 2). The average days of canagliflozin supplied was 166.2. The dose distribution of the initial canagliflozin claim was 70% of patients receiving the 100 mg dose, with the remaining 30% of patients receiving the 300 mg dose.
HbA1c test timing The mean time between pre-index HbA1c testing and index date was 27.5 days (Table 3), with 68% of the patients having a pre-index HbA1c test within a month of index canagliflozin claim. Mean time to HbA1c testing after the canagliflozin index claim was 119.9 days, with 71% of patients having an HbA1c test more than 90 days after canagliflozin index.
Glycemic control Mean HbA1c decreased from 8.3% at baseline to 7.6% post-index. The mean difference pre- and post-index was 0.7% (p50.001) (Table 4). Median HbA1c results were similar to the mean results. The proportion of patients who met the glycemic control definition of HbA1c 58% increased from 41% (110 patients) at baseline to 69% (185 patients) in the post-index period (p50.001). Similarly, the proportion of patients meeting HbA1c control of 57% increased from 15% (39 patients) at baseline to 31% (84 patients) after index (p50.001). The proportion of patients defined as having poor glycemic control according to HEDIS (HbA1c 49%) decreased from 28% Observational study of glycemic control in canagliflozin pts Meckley et al.
3
Current Medical Research & Opinion
2015
At least one T2DM diagnosis in acute inpatient or ED setting OR at least 2 T2DM diagnoses in outpatient or non-acute setting N = 1,918,778 Patients without a diagnosis of T1DM, gestational diabetes, polycystic ovaries, steroid induced diabetes or gaps in 2013 coverage N = 657,568 ≥1 canagliflozin claims N = 2619
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 06/21/15 For personal use only.
HbA1c measured pre-and post-canagliflozin index claim N = 420
≥60 day canagliflozin supply N = 355 HbA1c measured ≥60 days post-canagliflozin index claim N = 268
Figure 2. Sample selection flowchart.
Table 1. Baseline demographic and clinical characteristics. Characteristic N Age, Years (Mean, SD) 18–64 65–74 75þ Gender, n (%) Female Male Geographic Region, n (%) Northeast Midwest South West Insurance Type, n (%) Commercial Medicaid Medicare Unknown Diabetes Related Comorbiditiesa, n (%) Cardiovascular disease Chronic heart failure Chronic kidney disease Hyperlipidemia Hypertension Hypoglycemia Nephropathy Neuropathy Obesity Proteinuria Quan-Charlson Comorbidity Index (mean, SD) Diabetes Complications Severity Index (mean, SD) a
4
Table 2. Antihyperglycemic agents at index.
N or Mean
% or SD
268 56.8 229 34 5
8.7 85 13 2
116 152
43 57
45 22 173 28
17 8 65 10
232 11 22 104
63 3 6 28
43 1 14 210 206 1 1 31 48 11 0.65
16 0 5 78 77 0 0 12 18 4 1.07
0.68
1.08
Codes available in Supplementary Table 1.
Observational study of glycemic control in canagliflozin pts Meckley et al.
Number of antihyperglycemic agentsa 0 1 2 3þ Medication class Alpha-glucosidase inhibitors Amylin analogs Antihyperglycemic fixed dose combinationsb Metformin Bile acid sequestrants Dipeptidyl peptidase 4 (DPP-4) inhibitors Glucagon-like peptide-1 (GLP-1) receptor agonists Insulin Meglitinides Sulfonylureas Thiazolidinediones
N
%
18 50 73 127
7 19 27 47
4 1 79 129 6 68 62 70 6 122 30
1 0 29 48 2 25 23 26 2 46 11
a
Antihyperglycemic fixed dose combinations were counted as two medications. b The component medications were not counted in their respective class – they were only counted as an antihyperglycemic fixed dose combination.
pre-index to 12% in the post-index period (p50.001) (Figure 3).
Subgroup analyses In the 81% of patients who had HbA1c 47% at canagliflozin index, HbA1c was reduced from 8.8% at baseline to 7.9% post-canagliflozin index (Table 4). The mean www.cmrojournal.com ! 2015 Informa UK Ltd
Current Medical Research & Opinion
to 7.6% post-index. The mean HbA1c difference was 0.6% (p50.001) (Table 4). Post-canagliflozin index, the proportion of patients achieving glycemic control (at the HbA1c 58% threshold), improved from 41% to 67% of patients (p ¼ 0.012). Using the tighter control threshold of HbA1c 57%, glycemic control improved from 10% to 31% of patients (p ¼ 0.016). Fewer patients were poorly controlled post-canagliflozin index as well, with 23% of patients having HbA1c 49% pre-index compared to 8% of patients post-index (p ¼ 0.031).
Discussion Table 3. Timing of HbA1c testing.
Days Pre-Index Days Post-Index
Mean (95% CI)
Median (Range)
27.5 (24.3, 30.6) 119.9 (114.6, 125.2)
18.0 (1.0, 112.0) 106.0 (60.0, 253.0)
Overall, patients had better glycemic control postcanagliflozin index compared to baseline, with an average absolute reduction in HbA1c of 0.7%. Post-canagliflozin index, patients were significantly more likely to meet the glycemic control goals set by the ADA and HEDIS, of HbA1c 57% and HbA1c 58%, respectively, compared
Table 4. Changes in glycemic control pre- and post-index. Population
N
Time period
Mean (95% CI)
Mean Pre–Post (95% CI)
p value
Overall Cohort Analysis Total
268
Pre-Index Post-Index
8.3% (8.2%, 8.5%) 7.6% (7.5%, 7.8%)
0.7% (0.6%, 0.9%)
50.001
Subgroup Analyses HbA1c 47% Pre-Index
216
Pre-Index Post-Index Pre-Index Post-Index
8.8% (8.6%, 9.0%) 7.9% (7.7%, 8.0%) 8.3% (7.9%, 8.6%) 7.6% (7.3%, 7.9%)
0.9% (0.8%, 1.1%)
50.001
0.6% (0.4%, 0.9%)
50.001
Age 65 Years
39
N = 268
100% 12%
90% 28%
80% % of Patients at HbA1c Level
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 06/21/15 For personal use only.
difference in HbA1c test results at baseline to post-index was 0.9%, p50.001. Post-canagliflozin index, the proportion of patients achieving glycemic control (at the HbA1c 58% threshold), improved from 33% of patients to 69% of patients (p50.001). Post-index, 24% of patients attained the more stringent glycemic control threshold of HbA1c 57%. Fewer patients were poorly controlled post-index as well, with 35% of patients having HbA1c 49% pre-index compared to 14% of patients post-index (p50.001). The subgroup analysis of patients aged 65 years consisted of 39 (15%) patients (Table 4), with a mean age of 69.7 years. Mean HbA1c decreased from 8.3% at baseline
2015
19%
Pre vs Post HbA1c >9% p < 0.001
70% HbA1c Level
60%
HbA1c > 9%
31% 38%
50% 40% 30%
8% ≤ HbA1c ≤ 9% 7% ≤ HbA1c < 8% HbA1c < 7%
26%
20% 31%
10%
Pre vs Post HbA1c
2015, 1–8
Article RT-0115.R1/1047748 All rights reserved: reproduction in whole or part not permitted
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 06/21/15 For personal use only.
Brief report An observational study of glycemic control in canagliflozin treated patients
L.M. Meckley G. Miyasato F. Kokkotos Trinity Partners LLC, Waltham, MA, USA
Abstract Objective: To evaluate changes in glycemic control following the initial canagliflozin pharmacy claim in a real-world population.
J. Bumbaugh Inovalon Inc., Bowie, MD, USA
R.A. Bailey Health Economics and Outcomes Research, Janssen Scientific Affairs LLC, Raritan, NJ, USA Address for correspondence: Robert A. Bailey MD, 1000 US Route 202 South, Raritan, NJ 08869, USA. [email protected] Keywords: Aged – Diabetes mellitus, type 2 – Quality indicators, health care – Sodium–glucose co-transporter 2 inhibitor Accepted: 27 April 2015; published online: 17 June 2015 Citation: Curr Med Res Opin 2015; 1–8
Research design and methods: A retrospective cohort analysis of adult patients with type 2 diabetes mellitus (T2DM) was conducted using 2013 medical, pharmacy and laboratory claims from the Inovalon MORE2 Registry. Patients with T2DM aged 18 years with 60 days of canagliflozin supply and HbA1c test results within 120 days before and 60 days after initial canagliflozin claim (defined as index date) were included. The differences between HbA1c levels pre- and post-index were assessed. Changes pre- and post-index in Healthcare Effectiveness Data and Information Set (HEDIS) glycemic control criteria of HbA1c57% and58% and poor control of HbA1c 49% were evaluated. Subgroup analyses of patients with HbA1c47% at baseline and patients aged 65 were also conducted. Results: Among the 268 patients meeting the study criteria, mean HbA1c pre-index was 8.3% and post-index was 7.6%; the mean reduction in HbA1c pre–post index was 0.7% (95% CI: 0.6%, 0.9%). The proportions of patients meeting the HEDIS glycemic control measures (HbA1c 57%, 58% and poor control of 49%) improved and was significantly different pre- and post-index (all p50.001). Of the patients with an HbA1c 47% prior to index (81% of the cohort; mean pre-index HbA1c ¼ 8.8%), HbA1c was reduced by 0.9% (95% CI: 0.8%, 1.1%). The aged 65 subgroup consisted of 15% of the cohort, with a pre-index HbA1c of 8.3%. The mean reduction in HbA1c test results pre- and post-canagliflozin index was 0.6% (95% CI: 0.4%, 0.9%). This analysis did not adjust for changes in antihyperglycemic agents during the study period. Conclusion: Patients with T2DM were observed to have improved glycemic control following initial canagliflozin pharmacy claim as measured by HbA1c change and attainment of specific glycemic control criteria.
Introduction Diabetes is one of the most common chronic conditions in the US, affecting 29.1 million people, approximately 9.3% of the population1. These patients incur approximately $245 million in direct and indirect costs annually, with the majority of these costs due to the treatment of complications associated with poorly controlled diabetes2,3. The mortality burden of diabetes is also high, as diabetes is the seventh leading cause of death in the US4. Glycemic control, as measured by glycosylated hemoglobin (HbA1c), is a key clinical measure of diabetes control. Lower HbA1c levels are associated with reduced risk of microvascular diabetes complications5. The current position of the American Diabetes Association (ADA) and the European Association for ! 2015 Informa UK Ltd www.cmrojournal.com
Observational study of glycemic control in canagliflozin pts Meckley et al.
1
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 06/21/15 For personal use only.
Current Medical Research & Opinion
2015
the Study of Diabetes, is that glycemic targets should be individualized to patients on the basis of the patient’s comorbidities, life expectancy and hypoglycemia history, with the goal for most patients of HbA1c 57%5. The Healthcare Effectiveness Data and Information Set (HEDIS) is a tool to measure health plan performance and includes quality measures for a variety of chronic illnesses. As part of its comprehensive diabetes care measures, HEDIS classifies HbA1c as a quality measure6. For the purpose of evaluating the quality of diabetes care, HEDIS has set HbA1c 58% as the glycemic control threshold. For the subgroup of patients who are aged 565 who do not have specific comorbidities, HEDIS has an additional HbA1c control threshold of HbA1c 57%. HEDIS measures also assess poor glycemic control, defined as HbA1c 49%. Canagliflozin is an antihyperglycemic agent that inhibits sodium–glucose co-transporter 2 (SGLT2) and was FDA approved on 29 March 2013. SGLT2 inhibition results in a lowering of the renal threshold for glucose and an increased urinary excretion of glucose, lowering blood glucose concentrations7. In clinical trials, canagliflozin has been shown to improve glycemic control as monotherapy, dual therapy and triple therapy in patients with type 2 diabetes mellitus (T2DM)7–13. In the trials, older patients were shown to have smaller reductions in HbA1c with canagliflozin14. As the efficacy of SGLT2 inhibitors is affected by renal function, the reduced efficacy of canagliflozin observed in older adults is consistent with the lower eGFR typically found in these patients15. In addition to glycemic control, canagliflozin lowers body weight and systolic blood pressure7–13. Furthermore, canagliflozin was generally well tolerated, without an increased risk of hypoglycemia. Canagliflozin is associated with adverse events related to the mechanism of action of SGLT2 inhibitors, such as genital mycotic infections and osmotic diuresis related adverse events7–13. The objectives of this study were to evaluate changes in glycemic control after the initial canagliflozin pharmacy claim in a real-world population as measured by HbA1c, and to assess whether the proportion of patients achieving glycemic control, as defined by HEDIS and the ADA, improved after the initial canagliflozin pharmacy claim.
Patients and methods Data source The data for this study were obtained from the MORE2 Registry* (Medical Outcomes Research for Effectiveness and Economics Registry) Research Edition, a national healthcare data warehouse created by Inovalon Inc. The *MORE2 Registry is a registered trademark of Inovalon Inc.
2
Observational study of glycemic control in canagliflozin pts Meckley et al.
MORE2 Registry contains data from more than 120 million de-identified individuals enrolled in over 100 commercial, Medicare and Medicaid health plans16. The database has been applied in support of quality measurement development and research aimed at improvements in population health. This analysis used diagnosis, procedure, prescription and laboratory data from the MORE2 Registry.
Study design and patient selection A retrospective cohort design was used to assess glycemic control in patients receiving canagliflozin. Adult patients with T2DM who had a canagliflozin claim during 2013 were included. Similar to the HEDIS definition of patients with diabetes, in this study, T2DM was defined as at least one face-to-face encounter in an acute inpatient or emergency department (ED) setting with a diagnosis of T2DM (ICD-9-CM diagnosis codes 250.x0 or 250.x2 reported in any position) or two face-to-face encounters in an outpatient setting or non-acute inpatient setting on separate dates with a diagnosis of T2DM, during 2012 or 2013. The index date was defined as the date of the first canagliflozin claim. Because some patients may have received canagliflozin samples prior to the first canagliflozin claim, the index claim is a proxy for canagliflozin initiation. In order to assess the change in glycemic control, patients were required to have obtained 60 days of canagliflozin supply and have an HbA1c test result within 120 days preand at least 60 days post-index claim (Figure 1). Thus, patients who initiated after 1 November 2013 were excluded. Patients aged518, with type 1 diabetes mellitus (T1DM), gestational diabetes, polycystic ovary syndrome, steroid induced diabetes, enrollment gaps of 445 days or missing data for key variables were also excluded.
Outcomes and statistical analyses Univariate descriptive statistics were calculated for demographic, clinical characteristics and medication usage at baseline. The diabetes complication severity index (DCSI) and Quan–Charlson comorbidity index (QCCI) were also calculated to assess comorbidities. The DCSI and QCCI predict mortality, and the DCSI also predicts the risk of hospitalization17,18. Both indexes are validated measures and can be assessed using claims data. Continuous outcomes were reported as means and standard deviations; frequencies were calculated for categorical data. HbA1c was evaluated pre- and post-index, with means, 95% confidence intervals and medians assessed. The preindex time period (baseline) was defined as 120 days prior to index. The post-canagliflozin time period was defined as 60 days from index to 31 December 2013 in www.cmrojournal.com ! 2015 Informa UK Ltd
Current Medical Research & Opinion
Post-Index HbA1c Test Period†
Baseline HbA1c Test Period*
Jan 2013
April 2013 Canagliflozin Launch
2015
Dec 2013 60 Days Canagliflozin Supply
Index Date: First Canagliflozin Claim
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 06/21/15 For personal use only.
Figure 1. Study design. *Baseline period consists of the 4 months prior to canagliflozin index claim. yPost-index period begins 60 days post canagliflozin index claim and ends 31 December 2013.
order to allow the post-canagliflozin measurement to primarily reflect time on canagliflozin (Figure 1). For patients with multiple pre- or post-index HbA1c laboratory results, the most recent results were used for analysis. The difference between HbA1c levels pre- and post-index was tested using a paired t-test. Additionally, the proportion of patients meeting the HEDIS6 and ADA19 recommendations for HbA1c thresholds pre- and post-index were compared using McNemar’s test for significance, specifically: HbA1c control 58.0% HbA1c control 57.0% HbA1c poor control 49.0% Subgroup analyses of all outcome measures were conducted to assess glycemic control in patients receiving canagliflozin who had HbA1c 47% prior to index and those aged 65. The subgroup analysis of patients with pre-index HbA1c47% was conducted to evaluate consistency of results with the clinical trial population, as this threshold was typically used as a lower limit for inclusion. Older patients were shown to have smaller reductions in HbA1c with canagliflozin in clinical trials14, which is consistent with older adults having lower eGFR15, and thus reduced efficacy of agents that inhibit SGLT2. Therefore, the purpose of the subgroup analysis of patients aged 65 was to descriptively assess changes in HbA1c after canagliflozin index in this population in a real-world setting. HbA1c measures glucose levels over approximately 90 days, so we conducted a descriptive analysis of the temporal distribution of HbA1c tests pre- and postcanagliflozin index to assist with the interpretation of HbA1c in the context of patients’ treatment. Statistical analysis was performed using SAS 9.2 (SAS Institute, Cary, NC, USA).
Results Demographics and cohort The MORE2 Registry for the 2013 calendar year contained 1,918,778 patients with a diagnosis of T2DM. ! 2015 Informa UK Ltd www.cmrojournal.com
After clinical and administrative exclusions, 2619 patients with canagliflozin prescription claims were identified. After restrictions for canagliflozin usage and HbA1c results, 268 patients met the inclusion criteria for the study (Figure 2). Baseline demographics and clinical characteristics are presented in Table 1. Most patients (81%) had HbA1c 47% at baseline. Approximately 15% of patients were aged 65 years. Most patients in the cohort were treated with multiple antihyperglycemic agents at baseline, with metformin (48%) and sulfonylureas (46%) as the most common agents (Table 2). The average days of canagliflozin supplied was 166.2. The dose distribution of the initial canagliflozin claim was 70% of patients receiving the 100 mg dose, with the remaining 30% of patients receiving the 300 mg dose.
HbA1c test timing The mean time between pre-index HbA1c testing and index date was 27.5 days (Table 3), with 68% of the patients having a pre-index HbA1c test within a month of index canagliflozin claim. Mean time to HbA1c testing after the canagliflozin index claim was 119.9 days, with 71% of patients having an HbA1c test more than 90 days after canagliflozin index.
Glycemic control Mean HbA1c decreased from 8.3% at baseline to 7.6% post-index. The mean difference pre- and post-index was 0.7% (p50.001) (Table 4). Median HbA1c results were similar to the mean results. The proportion of patients who met the glycemic control definition of HbA1c 58% increased from 41% (110 patients) at baseline to 69% (185 patients) in the post-index period (p50.001). Similarly, the proportion of patients meeting HbA1c control of 57% increased from 15% (39 patients) at baseline to 31% (84 patients) after index (p50.001). The proportion of patients defined as having poor glycemic control according to HEDIS (HbA1c 49%) decreased from 28% Observational study of glycemic control in canagliflozin pts Meckley et al.
3
Current Medical Research & Opinion
2015
At least one T2DM diagnosis in acute inpatient or ED setting OR at least 2 T2DM diagnoses in outpatient or non-acute setting N = 1,918,778 Patients without a diagnosis of T1DM, gestational diabetes, polycystic ovaries, steroid induced diabetes or gaps in 2013 coverage N = 657,568 ≥1 canagliflozin claims N = 2619
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 06/21/15 For personal use only.
HbA1c measured pre-and post-canagliflozin index claim N = 420
≥60 day canagliflozin supply N = 355 HbA1c measured ≥60 days post-canagliflozin index claim N = 268
Figure 2. Sample selection flowchart.
Table 1. Baseline demographic and clinical characteristics. Characteristic N Age, Years (Mean, SD) 18–64 65–74 75þ Gender, n (%) Female Male Geographic Region, n (%) Northeast Midwest South West Insurance Type, n (%) Commercial Medicaid Medicare Unknown Diabetes Related Comorbiditiesa, n (%) Cardiovascular disease Chronic heart failure Chronic kidney disease Hyperlipidemia Hypertension Hypoglycemia Nephropathy Neuropathy Obesity Proteinuria Quan-Charlson Comorbidity Index (mean, SD) Diabetes Complications Severity Index (mean, SD) a
4
Table 2. Antihyperglycemic agents at index.
N or Mean
% or SD
268 56.8 229 34 5
8.7 85 13 2
116 152
43 57
45 22 173 28
17 8 65 10
232 11 22 104
63 3 6 28
43 1 14 210 206 1 1 31 48 11 0.65
16 0 5 78 77 0 0 12 18 4 1.07
0.68
1.08
Codes available in Supplementary Table 1.
Observational study of glycemic control in canagliflozin pts Meckley et al.
Number of antihyperglycemic agentsa 0 1 2 3þ Medication class Alpha-glucosidase inhibitors Amylin analogs Antihyperglycemic fixed dose combinationsb Metformin Bile acid sequestrants Dipeptidyl peptidase 4 (DPP-4) inhibitors Glucagon-like peptide-1 (GLP-1) receptor agonists Insulin Meglitinides Sulfonylureas Thiazolidinediones
N
%
18 50 73 127
7 19 27 47
4 1 79 129 6 68 62 70 6 122 30
1 0 29 48 2 25 23 26 2 46 11
a
Antihyperglycemic fixed dose combinations were counted as two medications. b The component medications were not counted in their respective class – they were only counted as an antihyperglycemic fixed dose combination.
pre-index to 12% in the post-index period (p50.001) (Figure 3).
Subgroup analyses In the 81% of patients who had HbA1c 47% at canagliflozin index, HbA1c was reduced from 8.8% at baseline to 7.9% post-canagliflozin index (Table 4). The mean www.cmrojournal.com ! 2015 Informa UK Ltd
Current Medical Research & Opinion
to 7.6% post-index. The mean HbA1c difference was 0.6% (p50.001) (Table 4). Post-canagliflozin index, the proportion of patients achieving glycemic control (at the HbA1c 58% threshold), improved from 41% to 67% of patients (p ¼ 0.012). Using the tighter control threshold of HbA1c 57%, glycemic control improved from 10% to 31% of patients (p ¼ 0.016). Fewer patients were poorly controlled post-canagliflozin index as well, with 23% of patients having HbA1c 49% pre-index compared to 8% of patients post-index (p ¼ 0.031).
Discussion Table 3. Timing of HbA1c testing.
Days Pre-Index Days Post-Index
Mean (95% CI)
Median (Range)
27.5 (24.3, 30.6) 119.9 (114.6, 125.2)
18.0 (1.0, 112.0) 106.0 (60.0, 253.0)
Overall, patients had better glycemic control postcanagliflozin index compared to baseline, with an average absolute reduction in HbA1c of 0.7%. Post-canagliflozin index, patients were significantly more likely to meet the glycemic control goals set by the ADA and HEDIS, of HbA1c 57% and HbA1c 58%, respectively, compared
Table 4. Changes in glycemic control pre- and post-index. Population
N
Time period
Mean (95% CI)
Mean Pre–Post (95% CI)
p value
Overall Cohort Analysis Total
268
Pre-Index Post-Index
8.3% (8.2%, 8.5%) 7.6% (7.5%, 7.8%)
0.7% (0.6%, 0.9%)
50.001
Subgroup Analyses HbA1c 47% Pre-Index
216
Pre-Index Post-Index Pre-Index Post-Index
8.8% (8.6%, 9.0%) 7.9% (7.7%, 8.0%) 8.3% (7.9%, 8.6%) 7.6% (7.3%, 7.9%)
0.9% (0.8%, 1.1%)
50.001
0.6% (0.4%, 0.9%)
50.001
Age 65 Years
39
N = 268
100% 12%
90% 28%
80% % of Patients at HbA1c Level
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 06/21/15 For personal use only.
difference in HbA1c test results at baseline to post-index was 0.9%, p50.001. Post-canagliflozin index, the proportion of patients achieving glycemic control (at the HbA1c 58% threshold), improved from 33% of patients to 69% of patients (p50.001). Post-index, 24% of patients attained the more stringent glycemic control threshold of HbA1c 57%. Fewer patients were poorly controlled post-index as well, with 35% of patients having HbA1c 49% pre-index compared to 14% of patients post-index (p50.001). The subgroup analysis of patients aged 65 years consisted of 39 (15%) patients (Table 4), with a mean age of 69.7 years. Mean HbA1c decreased from 8.3% at baseline
2015
19%
Pre vs Post HbA1c >9% p < 0.001
70% HbA1c Level
60%
HbA1c > 9%
31% 38%
50% 40% 30%
8% ≤ HbA1c ≤ 9% 7% ≤ HbA1c < 8% HbA1c < 7%
26%
20% 31%
10%
Pre vs Post HbA1c
