• Center on Health Equity and Access
  • Clinical
  • Health Care Cost
  • Health Care Delivery
  • Insurance
  • Policy
  • Technology
  • Value-Based Care

Glycemic Response to Newly Initiated Diabetes Therapies

Publication
Article
The American Journal of Managed CareNovember 2007
Volume 13
Issue 11

OBJECTIVE: The glycemic response to antihyperglycemic therapies for type 2 diabetes has been thoroughly evaluated in randomized controlled trials, but inadequately studied in real-world settings.

STUDY DESIGN: We studied glycemic response among 15 126 type 2 diabetic patients who initiated any single new antihyperglycemic agent (metformin, sulfonylureas, thiazolidinediones, or insulin added to medical nutrition therapy or to existing diabetes therapies) during 1999-2000 within Kaiser Permanente of Northern California, an integrated healthcare delivery system.

METHODS: Pre–post (3-12 months after initiation) change in glycosylated hemoglobin (A1C) was analyzed using ANCOVA (analysis of covariance) models adjusted for baseline A1C, concurrent (ongoing) antihyperglycemic therapy, demographics, health behaviors, medication adherence, clinical factors, and processes of care.

RESULTS: Mean A1C was 9.01% (95% confidence interval [CI] 8.98%-9.04%) before therapy initiation and 7.87% (95% CI 7.85%-7.90%) 3 to 12 months after initiation (mean A1C reduction 1.14 percentage points; 95% CI 1.11-1.17). Overall, 30.2% (95% CI 29.2%-31.1%) of patients achieved glycemic target (A1C < 7%). Although baseline disease severity and concurrent therapies differed greatly across therapeutic classes, after adjustment for these baseline clinical characteristics, no significant differences were noted in glucose-lowering effect across therapeutic classes. Treatment effects did not differ by age, race, diabetes duration, obesity, or level of renal function.

CONCLUSIONS: Metformin, sulfonylures, thiazolidinediones, and insulin were equally effective in improving glucose control. Nonetheless, most patients failed to achieve the glycemic target. Findings suggest that, to keep up with progressive worsening of glycemic control, patients and providers must commit to earlier, more aggressive therapy intensification, triggered promptly after A1C exceeds the recommended glycemic target.

(Am J Manag Care. 2007;13:598-606)

About one third of patients with diabetes initiate new-therapy regimens annually. Thus, understanding the effect of newly added therapies in clinical practice has public health relevance.

Initiating metformin, sulfonylurea, thiazolidinediones, or insulin had a similar effect in adults with type 2 diabetes (~1 point lowering of A1C).

Therapy intensification occurred too late in the course of worsening glycemic control at which point the glucose-lowering effect of a single added therapy was insufficient to lower A1C under the glycemic target.

Findings indicate that greater patient and provider commitment to earlier therapy intensification is needed.

Maintaining tight blood glucose control is important in preventing microvascular complications of diabetes mellitus such as retinopathy and nephropathy.1,2 Although trends in glycemic control look favorable,3 national data suggest that a substantial proportion of patients remain poorly controlled.4-7 Moreover, as glycemic goals have been progressively lowered,8 achieving targets has become increasingly difficult.

While numerous randomized, placebo-controlled clinical trials have evaluated the efficacy of diabetes medications alone or in combinations,9 their effectiveness in usual clinical care settings has rarely been assessed. Furthermore, there is a concern that the effectiveness of diabetes medications may differ by clinical and demographic factors such as duration of diabetes, patient age, race, obesity, and chronic renal insufficiency.

We previously reported that, among patients with poorly controlled diabetes glycosylated hemoglobin [A1C] > 8%) who initiated a new antihyperglycemic therapy, only a very low proportion (18.4%; 95% confidence interval [CI] 17.3%-19.4%) achieved glycemic target (A1C < 7%) within 3 to 12 months.10 In the present study, we examined the effectiveness (A1C lowering) of the 4 antihyperglycemic therapies that are currently most commonly prescribed (metformin, sulfonylureas, insulin, thiazolidinediones), overall and by clinical subgroups, within a usual-care setting.

METHODS

Setting and Source Population

We identified all registry members who initiated a new diabetes medication between March 1, 1999, and July 31, 2002, and had a full year of membership with drug benefits after drug initiation. Given that individuals who initiate new diabetes therapies systematically differ from individuals who maintain ongoing therapies in terms of glycemic control, disease severity, patient characteristics, and behaviors,18 we used a “new user†design19 that assesses outcomes occurring after drug initiation in the subgroup of individuals initiating a single new therapy, while controlling for potential confounders existing before initiation. By design, we excluded patients with type 1 diabetes because they would not switch therapeutic classes once established on insulin; patients with end-stage renal disease (ESRD) because of its effects on clearance of insulin (and thus glycemic control) and sulfonylureas, and because ESRD is a contraindication for metformin20; and persons without continuous health plan membership (n=4752) or lacking drug benefits (n=2574) during the study period, to minimize misclassifying subjects who may have filled prescriptions in non—health-plan pharmacies. Of the 16 746 new users with pre—post A1Cs recorded, we excluded 1285 (6%) who simply switched medications (eg, switched from 1 monotherapy to another monotherapy) because we could not accurately ascertain each drug's washout period. We also excluded 145 (0.7%) patients who added more than 1 medication simultaneously. Finally, we excluded a small number (n=190) of new users of a-glucosidase inhibitors (eg, Acarbose), meglitinides (eg, Repaglinide), and incretins (glycolipoprotein [GLP]-1 analogues and dipeptidylpeptidase-4 inhibitors) because they were used too infrequently during the study period to provide sufficient statistical power.

After applying these exclusion criteria, our final analytic sample included 15 126 health plan members with diabetes who initiated a single, new diabetes therapy between March 1, 1999, and July 31, 2002, and had A1C measured both during the 12-month window before initiation and the 3 to 12 months after initiation and before discontinuing or modifying the index therapy.

Pharmacotherapeutic Exposure and Outcome

We used a “new-user†cohort study design,19 which includes only patients initiating a single new therapy (“index therapy”) and synchronizes cohorts of patients at the time of initiating the index therapy, thereby avoiding chronology bias.21 Demographic and clinical characteristics of patients initiating each index therapy were compared using standard descriptive statistics. We next estimated change in A1C in response to initiation of each diabetes therapy using ANCOVA (analysis of covariance) models. Among new users of diabetes therapies, glycemic control before the index date, disease severity, concurrent hypoglycemic therapies, and patient characteristics vary considerably by drug class,18 which could confound crude estimates of glycemic effect associated with initiation of different diabetes therapies. Therefore, we first examined the effect of initiating therapy in minimally adjusted models, which included the last A1C before initiation of the index therapy, concurrent therapies, age, and sex. Additional models were then specified by including relevant covariates available for the entire analysis sample.

Multivariate adjustment included the following factors assessed during the calendar year before index date: neighborhood- level socioeconomic status, number of annual visits to assigned primary care provider, primary care provider specialty (endocrinologist vs other primary care provider), an indicator variable for the Kaiser facility where most of the care was provided, rate of missed scheduled outpatient appointments, inpatient and outpatient comorbidity risk score,22 presence of chronic heart failure, diagnosed hypertension, smoking status (current, former, never), serum low-density lipoprotein cholesterol level (mg/dL), serum triglycerides level (mg/dL), and estimated glomerular filtration rate (in mL/min/1.73 m2) based on serum creatinine concentration.23 Neighborhood-level socioeconomic status was assessed by geocoding each member's address and linking to associated 2000 US census block group average annual per-capita income, percentage below poverty, and proportion in a working class occupation.11 Prevalent heart failure was based on primary discharge diagnoses or serial outpatient diagnoses found in hospital discharge and ambulatory visit databases. Diagnosed hypertension and smoking status were obtained from ambulatory visit databases. Cholesterol, triglyceride, and serum creatinine levels were captured from an automated laboratory database. Pharmacy records were also used to calculate an index of medication adherence for oral diabetes agents (defined as the percentage of days during follow-up without adequate supply or “continuous, multiple-interval measure of medication gaps”24), and self-monitoring of blood glucose frequency (based on glucose test strip utilization).12,17

Eighty-three percent (n = 77 726) of the 94 024 noninstitutionalized health plan members in the diabetes registry (as of 1995) responded to a self-administered questionnaire or computer-assisted telephone interview conducted in 1994-1997. The survey asked about daily number of insulin injections, use of exercise and diet as diabetes treatments, time since diabetes diagnosis, body mass index, smoking history, educational attainment, and self-identified race/ethnicity. Data on these variables were available for a subanalysis of 7605 members of our final analysis sample.

Using cross-product terms (index therapy × clinical strata), we then assessed whether statistical differences (effect modification) existed in the glycemic response differences between initiated therapies, across the clinical substrata of interest: prevalent chronic heart failure (yes or no); age class (<45, 45-65, or =65 years); body mass index (nonobese [=30 kg/m2] vs obesity class 1-3 [=30 kg/m2]); estimated glomerular filtration rate (>60 vs =60 mL/min/1.73 m2); time since diabetes diagnosis (<10 vs =10 years); and race/ethnicity (African American, Asian, Latino, non-Latino White). We calculated least-squares means to estimate adjusted A1C difference for each index therapy, and again for each clinical substratum. All analyses were performed using SAS, version 9.1.3 (Cary, NC) and a 2-sided P < .05 was considered significant.

RESULTS

Metformin and insulin initiators were younger, while sulfonylurea initiators were older (Table 1). No relevant differences were noted in the therapy distributions by sex or socioeconomic status categories. Asian participants were least likely, while multiethnic and White participants were the most likely, to initiate insulin among the 4 therapeutic categories. Initiation of either thiazolidinediones or insulin was associated with substantially greater severity of diabetes or associated complications (including longer duration of diabetes, obesity, chronic kidney disease, and peripheral neuropathy), higher comorbidity scores compared with those initiating either metformin or a sulfonylurea, and a much greater likelihood of already being on at least 2 oral hypoglycemic agents. In contrast, persons initiating sulfonylureas were much more likely to be transitioning from medical nutrition therapy alone, and metformin users were most likely to have previously been on a single oral agent, sulfonylureas. Excessive alcohol consumption was low overall, but somewhat more common among participants initiating metformin or sulfonylureas. revalent diagnosed heart failure was much less common among patients who initiated metformin and most common among patients who initiated insulin. Patients initiating sulfonylureas were the most likely to have poorer medication adherence for oral agents and the least likely to be selfmonitoring their blood glucose. Smokers were more likely to initiate insulin.

DISCUSSION

We examined glycemic response (absolute change in A1C) 3 to 12 months after adding a single diabetes therapy to an existing regimen and evaluated whether initiation brought a patient under the glycemic target (A1C < 7%). Glycemic response (mean change: -1.14 points; 95% CI -1.11 to -1.17) did not differ significantly across the 4 therapeutic classes, and was similar in magnitude to that reported for randomized trials of the individual therapies.25,26 Moreover, glycemic response did not differ by age, race, duration of diabetes, obesity, or level of kidney function. Our finding of uniform glycemic response is consistent with a systematic review (based on 216 controlled trials and cohort studies, and 2 previous systematic reviews),27 which reported similar findings of comparative effectiveness (about a 1-point improvement in A1C) among oral medications for type 2 diabetes. Similar findings were also reported by an Agency for Health Care Research and Quality review of randomized trials, which reported no substantive differences between therapy efficacy across oral diabetes therapies.28 We acknowledge that our findings are based on 1-year follow-up and may not be durable over a longer time frame. A randomized trial with a median follow-up of 4 years, the study known as A Diabetes Outcome Progression Trial (ADOPT), showed differences in the cumulative incidence of monotherapy failure at 5 years (15% for rosiglitazone, 21% for metformin, and 34% for glyburide).29 Although few studies have evaluated differences in effectiveness among special clinical subgroups, our findings are also consistent with a recent study reporting no differences in metformin effectiveness between obese and nonobese patients.30

Despite the observed favorable and uniform glycemic response, only about 30% of new users achieved glycemic target (<7%) 3 to 12 months after initiation. However, since the time of the study, the prevalence of well-controlled patients has increased in the source population for this study. In a previous study of the Kaiser Permanente Northern California Diabetes Registry,10 we reported that only 18% of patients with poor glycemic control (A1C > 8%) achieved target 3 to 12 months after initiating new therapy. In the current study, we included all patients initiating new therapy, regardless of their baseline glycemic control, and found that even patients with a seemly reasonable level of control (baseline A1C between 7% and 8%), less than 50% achieved the glycemic target within 1 year. Therapy was typically initiated only after A1C had advanced too far beyond the threshold at which the expected therapeutic benefit (ie, about 1% lowering) would bring A1C under target. These findings are consistent with an earlier study31 demonstrating that intensification was generally not initiated until after A1C exceeded 9%.

Findings suggest that, to keep up with the expected progressive, age-related worsening of glycemic control,32,33 patients and providers must commit to more aggressive therapy intensification. Clinical “targets†should be considered synonymous with “action levels,†with therapy intensified as soon as target values are exceeded. In the present study, initiation was delayed longer for second-line therapies; thiazolidinediones and insulin were started in patients with mean A1C of 9.3% and 9.6%, respectively, whereas first-line therapies, sulfonylureas and metformin, were initiated at 8.3% and 9.2%, respectively. While therapy intensification algorithms26,34-36 vary somewhat and are evolving, typically the history of previous therapy, concurrent therapy, current level of glycemic control, therapy cost, and patient factors (age, competency, and preferences) are considered when determining the next therapy to be prescribed. Typically metformin or a sulfonylurea is prescribed as a first step, followed by thiazolidinediones, insulin, or other agents (eg, nonsulfonylurea secretagogues, a-glucosidase inhibitors, or incretins), for more difficult-to-control glycemia. We show that with each step of intensification, the added therapy has a similar glycemic-lowering effect, but steps are initiated at progressively higher A1C levels. Thus intensification is not keeping pace with worsening glycemic control.

Our study had several limitations. Nonrandomized exposures preclude us from ruling out the effect of residual confounding (by indication37 or contraindication) or treatment selection bias. However, controlling for preinitiation A1C, ongoing therapy, and duration of diabetes did not alter our findings of similar medication effect sizes. Also, although no differences in response by therapy were detected, differential response to pharmacologic agents at the individual patient level cannot be excluded. It is important to note that selective prescribing renders our findings uninformative for standard cost-effectiveness assessments, given that lower cost, first-line therapies (eg, generic sulfonylureas) would simplistically appear substantially more cost-effective than more expensive, second-line therapies (eg, thiazolidinediones). The similar effectiveness across select demographic and clinical subgroups may be due to a selection process whereby physicians are prescribing optimal agents for a given patient's characteristics. While a randomized trial provides an estimate of efficacy in selected volunteers, these observational study findings should reflect outcomes in typical clinical practice; thus this study provides an important complement to trial findings.38,39 Also, it is important to note that both patient and provider may play a role in delayed intensification. The relative role of each person remains undetermined because we were unable to distinguish whether the provider failed to prescribe a needed therapy or the patient failed to obtain the prescribed prescription from the pharmacy.

We excluded patients who discontinued 1 medication and substituted another (ie, switchers), because analysis of response to therapy was intractable. Although we did not attempt to directly ascertain the glycemic response to newly initiated medications in switchers, we suspect it would not be substantively different.

Finally, we were unable to ascertain pharmacy utilization at non-Kaiser pharmacies. However, we limited our analysis to members with a drug benefit, which is only recognized at Kaiser pharmacies; thus there is a substantial economic incentive for members with a drug benefit to use only Kaiser pharmacies. In a 2005-2006 survey of 20 188 diabetic Kaiser members (DISTANCE Study), those patients with drug benefits reported using non-Kaiser pharmacies on average less than a single time (0.4 times) in the past 12 months.40

Our study had certain unique strengths. Our source population of more than 3 million individuals represents approximately one third of the population of northern California.Our population had broad age, sex, and racial/ethnic diversity and uniform access to care. The Kaiser Permanente Northern California population is representative of the local surrounding and statewide population, except for slight underrepresentation of the extremes of income.11,13,41 Our rich clinical databases also facilitated statistical adjustment for relevant confounding variables usually unavailable in administrative or claims databases. Levels of glycemic control in our population are similar to those in other published studies from a similar time period,42 suggesting that our findings are likely generalizable to the large number of insured individuals with diabetes throughout the United States. Additionally, whereas this study focused on the glycemic target of A1C less than 7%, a target used by Kaiser Permanente at the time the study was carried out (and recommended by the American Diabetes Association), findings are likely applicable to other targets (eg, less aggressive targets may be desirable for frail or geriatric patients43).

Author Affiliation: From the Division of Research, Kaiser Permanente, Oakland, Calif (AJK, HHM, JL, MMP, ATA, ASG, JVS); the Department of Epidemiology, School of Public Health & Community Medicine, University of Washington, Seattle, Wash (AJK); and the Departments of Epidemiology, Biostatistics, and Medicine, University of California, San Francisco (ASG).

Funding: This research was funded by an unrestricted grant from Novartis Pharmaceuticals. The lead author was also supported by grants from the National Institute of Diabetes and Digestive and Kidney Diseases (#R01 DK65664) and the National Institute of Child Health and Human Development (#R01 HD046113-02). By contract, the authors retained the right to publish without approval from Novartis or National Institutes of Health.

Author Disclosure: Dr Karter reports receiving a grant from Novartis for this study. The authors (HHM, JL, MMP, ATA, ASG, JVS) report no relationship of financial interest with any entity that would pose a conflict of interest with the subject matter discussed in this manuscript.

Authorship Information: Concept and design (AJK, HHM, ATA); acquisition of data (AJK, JL, MMP, JVS); analysis and interpretation of data (AJK, JL, MMP, ATA, ASG, JVS); drafting of the manuscript (AJK, HHM); critical revision of the manuscript for important intellectual content (AJK, HHM, ATA, ASG, JVS); statistical analysis (AJK, JL, MMP); provision of study materials or patients (AJK); obtaining funding (AJK); administrative, technical, or logistic support (HHM, JVS); supervision (AJK, JVS).

Address correspondence to: Andrew J. Karter, PhD, Epidemiology & Health Services Research, Division of Research, Kaiser Permanente, 2000 Broadway, Oakland, CA 94612. E-mail: andy.j.karter@kp.org.1. Diabetes Control and Complications Trial (DCCT): results of feasibility study. The DCCT Research Group. Diabetes Care. 1987;10:1-19.

3. Saaddine JB, Cadwell B, Gregg EW, et al. Improvements in diabetes processes of care and intermediate outcomes: United States, 1988-2002. Ann Intern Med. 2006;144:465-474.

5. Harris MI, Eastman RC, Cowie CC, Flegal KM, Eberhardt MS. Racial and ethnic differences in glycemic control of adults with type 2 diabetes. Diabetes Care. 1999;22:403-408.

7. Kirk JK, D’Agostino RB Jr, Bell RA, et al. Disparities in HbA1c levels between African-American and non-Hispanic white adults with diabetes: a meta-analysis. Diabetes Care. 2006;29:2130-2136.

9. Johnson JL, Wolf SL, Kabadi UM. Efficacy of insulin and sulfonylurea combination therapy in type II diabetes. A meta-analysis of the randomized placebo-controlled trials. Arch Intern Med. 1996;156:259-264.

11. Krieger N. Overcoming the absence of socioeconomic data in medical records: validation and application of a census-based methodology. Am J Public Health. 1992;82:703-710.

13. Hiatt RA, Friedman GD. Characteristics of patients referred for treatment of end-stage renal disease in a defined population. Am J Public Health. 1982;72:829-833.

15. Karter AJ, Parker MM, Moffet HH, et al. Missed appointments and poor glycemic control: an opportunity to identify high-risk diabetic patients. Med Care. 2004;42:110-115.

17. Karter AJ, Ferrara A, Darbinian J, Ackerson LM, Selby JV. Selfmonitoring of blood glucose: language and financial barriers in a managed care population with diabetes. Diabetes Care. 2000;23:477-483.

19. Ray WA. Evaluating medication effects outside of clinical trials: new-user designs. Am J Epidemiol. 2003;158:915-920.

21. Feinstein AR. Clinical biostatistics. XI. Sources of ‘chronology bias’ in cohort statistics. Clin Pharmacol Ther. 1971;12:864-879.

23. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med. 1999;130:461-470.

25. Inzucchi SE. Oral antihyperglycemic therapy for type 2 diabetes: scientific review. JAMA. 2002;287:360-372.

27. Bolen S, Feldman L,Vassy J, et al. Systematic review: comparative effectiveness and safety of oral medications for type 2 diabetes mellitus. Ann Intern Med. 2007;147:386-399.

29. Kahn SE, Haffner SM, Heise MA, et al; for the ADOPT Study Group. Glycemic durability of rosiglitazone, metformin, or glyburide monotherapy. N Engl J Med. 2006;355:2427-2443.

31. Brown JB, Nichols GA. Slow response to loss of glycemic control in type 2 diabetes mellitus. Am J Manag Care. 2003;9:213-217.

33. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group. Lancet. 1998;352:837-853.

35. Riddle MC. Glycemic management of type 2 diabetes: an emerging strategy with oral agents, insulins, and combinations. Endocrinol Metab Clin North Am. 2005;34:77-98.

37. Poses RM, Smith WR, McClish DK, Anthony M. Controlling for confounding by indication for treatment. Are administrative data equivalent to clinical data? Med Care. 1995;33(suppl):AS36-AS46.

39. Colwell JA. Controlling type 2 diabetes: are the benefits worth the costs? JAMA. 1997;278:1700.

41. Krieger N. Analyzing socioeconomic and racial/ethnic patterns in health and health care. Am J Public Health. 1993;83:1086-1087.

43. Aspray TJ, Unwin N. Clinical guidelines for older adults with diabetes mellitus. JAMA. 2006;296:1839-1840.

Related Videos
Related Content
© 2024 MJH Life Sciences
AJMC®
All rights reserved.