Does Medication Persistence in Diabetes Affect the Risk Of Hospitalization In The Short-Term?
An evaluation of the impact of persistence to oral antidiabetic medication on reducing the risk of hospitalization in adult patients with type 2 diabetes.
Objective: To measure the association between persistence with oral antidiabetic (OAD) agents in patients with type-2 diabetes mellitus (T2DM) and risk of hospitalization.
Study Design: A retrospective cohort study using 2010-2014 commercial claims data.
Methods: The study cohort included adult T2DM patients, newly initiated on OAD therapy. Persistence to OAD medication was calculated by adding the number of uninterrupted days’ supply for 1 year, or until there was a ≥30 day gap in days’ supply. Multivariate extended Cox regression modeling evaluated the association between medication persistence and risk of hospitalization. Sensitivity analyses were conducted using propensity score adjusted and weighted extended Cox regression models.
Results: OAD therapy was initiated by 1026 adults; 57% were persistent with OAD therapy during the follow-up period, with a mean persistence of 232.3±138.0 days. There was no difference in the risk of all-cause hospitalization between the 2 groups within 108 days of starting OAD therapy. After 108 days, the non-persistent group experienced over a 2-fold greater risk of hospitalization (hazard ratio [HR] =2.25, 95% confidence interval [CI] =1.04—4.87, p-value=0.04). Results remained consistent in the sensitivity analyses.
Conclusions: Persistence with OAD medications was only 57%, and, 3.5 months after initiation of therapy, non-persistent patients had more than a 2-fold higher risk of hospitalization. These results suggest there may be value in diabetes-specific initiatives to improve medication compliance. Even in an environment in which long periods of continuous enrollment within an insurance plan are uncommon, short-term benefits may be appreciated through encouraging persistence to OAD medications and, in turn, avoiding costly inpatient admissions.
In the United States, 1 in 10 individuals has type 2 diabetes mellitus (T2DM), and the prevalence of this chronic disease is on the rise.1Due to high prevalence and associated complications, diabetes imposes a substantial drain on society. In 2012, the economic burden of diabetes in the United States was $245 billion with an estimated $176 billion in direct cost. On average, the medical expenditures of a patient with diabetes are 2.3 times higher than for a patient without diabetes.2T2DM is associated with poor clinical outcomes, quality of life, and health status.3Although rates of hospitalization of diabetic patients has been decreasing, in 2009, there were 267 diabetes-related hospital discharges per 1000 diabetic patients.4
There are various risk factors for hospitalization secondary to diabetic complications, which include high HbA1c, elevated urine albumin: creatinine ratio, high body mass index, elevated triglycerides, and low levels of high density lipoproteins.5 Tight blood glucose control is key to reducing morbidity and mortality associated with T2DM.6-9According to the American Diabetes Association, tight glucose control is defined as having blood glucose levels between 70 and 130 mg/dl before meals, and less than 180 two hours after starting a meal, with a glycated hemoglobin (A1C) level of less than 7%.10
In addition to lifestyle changes, pharmacologic therapy with oral antidiabetic (OAD) medications is important for tight blood glucose control and in reducing the incidence of diabetes-related complications, morbidity, and mortality.8,11-13 Better health outcomes in diabetic populations require medication compliance with OAD medications. Highlighting the importance of this problem, public and private payers in the US have recently incorporated measures of T2DM medication compliance as part of health plan performance measurement efforts.14-16
Health outcomes of the medication treatment are affected not only by compliance with day-to-day treatment (medication adherence), but also by the length of time treatment is continued during the prescribed duration (medication persistence). Medication adherence and persistence are 2 different constructs that are used to measure medication compliance. Medication adherence denotes the proportion of pills taken within a specific time interval. Medication persistence refers to the continuous use of the therapy for the recommended duration. It is defined as the duration of time from initiation to discontinuation of treatment.17Medication adherence to OAD therapy has been associated with reduced risk of hospitalization and associated cost.18
A previous study found that non-adherence to OAD medications was associated with 2.5 times higher odds of hospitalization.19Among patients with diabetes, inpatient care has been found to account for 35%-43% of total medical costs.2,20Improved compliance with antidiabetic medication has been shown to reduce emergency department (ED) visits by 26%.21In fact, better adherence to OAD therapy was estimated to save $4.7 billion in annual health care costs via reduced ED visits and hospitalizations.22Despite considerable efforts, adherence and persistence with OAD medications are sub-optimal. A recent meta-analysis found the proportion of adherent and persistent patients to be 67.9% (95% CI: 59.6%—76.3%) and 56.2% (95% CI: 46.1%–66.3%), respectively.23
In the current health care environment, patients are less likely to remain within a single health plan for long periods, making it difficult to convince health plans to initiate programs, such as those to improve medication persistence if results can only be seen in the long-term. Quantifying the impact of initial gaps in therapy on short-term outcomes for patients with T2DM is important for developing disease management strategies and interventions to address non-adherence at the health plan level. Thus, the objectives of this study were to measure medication persistence and its impact on hospitalization in a large cohort of commercially insured patients with T2DM.
An administrative database was used to study pharmacy, inpatient, and other health care utilization from January 1, 2010, through December 31, 2014. The data consisted of individuals who received health coverage through self-insured employer groups and commercial insurers across the United States and included both health maintenance organizations and preferred provider organization models. Both eligibility and claims files were included, containing patient demographics, enrollment information, as well as pharmacy and medical claims. Prescription claims provided information regarding prescription drug dispensed (National Drug Code [NDC] number), date and quantity dispensed, and the number of days supplied.
Medical claims contained information about date of service, clinical diagnosis codes (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes), and procedure codes (Healthcare Common Procedure Coding System [HCPCS] codes). This study was conducted using fully de-identified data obtained through Health Advocate, Inc., received and managed in compliance with the Health Insurance Portability and Accountability Act of 1996. Given the use of de-identified data to provide aggregate results for this analysis, it was determined through past experience of the authors that approval by the Institutional Review Board would not be required.
Study Design and Study Cohort
The present study used a retrospective cohort design to examine the relationship between persistence with OAD therapy and the risk of all-cause hospitalization. The study cohort included adult patients (age ≥ 18 years) with T2DM. The presence of diabetes was ascertained using a nationally validated code set.24Our cohort included patients newly initiated on OAD therapy, defined as those who had no prescriptions for OAD medications in the previous 6 months and continuous health coverage during at least 6 months prior through 1 year after the date of the OAD medication start. The cohort only included new users of OAD medications in order to avoid selection bias associated with patients who were long-term users of OAD therapies, and to avoid prevalence bias that might be introduced by patients previously exposed to OAD medications.25Pregnant women, patients with gestational diabetes, type 1 diabetes, and steroid-induced diabetes were excluded from the analysis (Figure 1).
The primary outcome was all-cause hospitalization, measured using inpatient claims during the 1-year follow up period. Patients were censored if the follow-up period (1 year) ended without hospitalization, if they initiated insulin or other sub-cutaneous diabetes therapies, or if they had a gap of more than 30 days in the supply of OAD medications, whichever occurred earlier.
Primary Exposure Variable
The primary independent variable of interest was persistence to OAD medications, measured during the 1-year follow-up period. Persistence was determined using the service date and days’ supply information from the pharmacy claims for biguanides, sulfonylureas, meglitinides, thiazolidinediones (TZDs), dipeptidyl peptidase-4 (DPP-4) inhibitors, glucagon-like peptide-1 (GLP-1) receptor agonists, and other OAD combinations identified using relevant National Drug Codes (NDC).
Persistence was defined as the number of continuous days from the initiation of an OAD medication until discontinuation of therapy,17and was measured by adding the uninterrupted number of days supplied of any OAD medications during the follow up period (1 year following OAD start date).17A maximum gap of 30 days was allowed between consecutive periods of days supplied of the OAD medication. When the gap was more than 30 days, the treatment episode for the patient was terminated, despite the possibility of a patient resuming OAD therapy at a later date. Switching between OAD medications was permitted.26For multiple OAD medications with overlapping days’ supply intervals, being covered with at least 1 OAD medication was defined as being persistent with OAD therapy.
Patients who were persistent for the entire 1-year follow-up period or non-persistent during that year were clearly sorted into the persistent and non-persistent cohorts, respectively. There were, however, instances of patients who had a hospitalization or began utilizing injectable medication during the follow-up period. For these patients, we considered the time period prior to hospitalization or injectable medication. If the patient was persistent before this event, he or she was included in the persistent group and, in turn, if the patient was non-persistent prior to the event, then the patient was considered a part of the non-persistent cohort.
The baseline confounders and risk factors associated with the risk of hospitalization were identified from existing literature19,27-33and selected based on their availability in the data. These included demographic characteristics such as age, gender, type of health coverage (all patients were commercially insured), and year of entry into the cohort (2010, 2011, 2012, and 2013). Use of medications such as anti-hypertensives and anti-hyperlipidemics was identified using relevant NDCs in prescription claims during the 6-month baseline period.
Comorbidities were measured by the presence of a medical claim during the 6-month baseline period for hypertension, hyperlipidemia, neuropathy, retinopathy, and mental health conditions. Additionally, the Charlson comorbidity index (CCI) was included as a need factor. The CCI was calculated at the patient level by assigning a pre-defined weight to each of the patient’s comorbid conditions and summing the assigned weights for all conditions.34
Statistical differences in baseline covariates were examined using χ2 tests for the categorical variables and t-tests for the continuous variables. The unadjusted relationship between OAD persistence and all-cause hospitalization was evaluated using Kaplan-Meier (KM) survival plots. The pairwise log-rank test was used to compare survival distribution between the persistent and non-persistent OAD groups. The multivariate Cox proportional-hazard model was used to examine the adjusted relationship between persistence with OAD therapy and risk of hospitalization.
The proportional hazards (PH) assumption for the use of OAD therapy was checked by a plot of Schoenfeld-residuals across event time.35 A horizontal line at 0 indicates that the PH assumption is met, whereas a nonlinear trend across time indicates a violation of the PH assumption. The extended Cox model with a Heaviside function was employed to account for a violation of the PH assumption.
Results were presented as hazard ratios (HR), along with 95% confidence intervals (CI) for the adjusted analysis. All analyses were conducted at a priori 5% alpha level. SAS version 9.3 was used for all analyses (SAS Institute Inc, Cary, NC).
Multiple sensitivity analyses were conducted. In the first sensitivity analysis, a propensity score adjusted extended Cox Regression model was performed by using the propensity score as a covariate. The second sensitivity analysis, a propensity score weighted extended Cox Regression model, was conducted by using the inverse probability of treatment weighting (IPTW) technique. Individuals were weighted by the inverse probability of receiving the treatment that they actually received.36
Stabilized weights were generated to reduce bias due to a large degree of variability in the weights.36,37The stabilized weights were used in the extended Cox regression model to obtain the unbiased effect of medication persistence on hospitalization. Additionally, we conducted sensitivity analyses using longer gap periods (45- and 60-day gap) to examine the influence of the duration of the permissible gap on the robustness of the study results.
Figure 1 outlines the study sample selection. Of 3370 patients who filled at least 1 diabetes medication during the study period, 2568 patients initiated their medication between July 1, 2010, and December 31, 2013, allowing for adequate baseline and follow-up periods. Among these new users, 1920 patients had continuous health coverage for at least 6 months prior through 1 year after the index date.
Inclusion in the cohort was limited to patients who were ≥18 years old at the time of OAD medication start (n=1888). Among these patients, 1572 patients had T2DM, and had no evidence of gestational or steroid induced diabetes or pregnancy during the study period. Among the 1203 patients who received OAD agents, 171 patients had type-1 diabetes during the baseline period and were excluded. Finally, of the remaining 1032 patients, 6 did not have complete demographic information and were excluded from the study. Thus, the final cohort consisted of 1026 patients.
Table 1 presents characteristics of the study cohort. The majority of patients were male (66.8%); the mean age of the study population was 60.1±12.0 years, with a mean CCI score of 1.5±1.6. Most patients entered the cohort in 2012 (34.8%), and the most prevalent comorbid conditions were hyperlipidemia (55.8%) and hypertension (55.1%). Patients persistent to OAD therapy were similar to patients non-persistent to OAD therapy (Table 1) but patients persistent to OAD therapy were more likely to have entered the cohort in 2012 (34.8%, p=0.014).
As shown in table 1, mean persistence in the study cohort was 177.8±130.0 days. More than 57% of the T2DM patients were persistent with OAD therapy during the 1-year follow up period. As expected, the mean persistence in the persistent group (232.3±138.0 days) was significantly higher than that of the non-persistent group (104.8±68.9 days) at p<0.0001.
Risk of Hospitalization
During the baseline period, 6.3% of the patients were hospitalized. During the follow-up period, 6.4% of the patients were hospitalized (n=66). Rate of hospitalization was similar between the persistent (5.0%) and non-persistent group (7.5%), p=0.1085. Mean time to hospitalization was 121.7±92.88 days. Mean time to hospitalization in the non-persistent group (89.6±64.7 days) was significantly lower than that of the persistent group (137.8±101.0 days) at p=0. 02. Patients hospitalized during follow-up were more likely to be older (64.5 vs 59.6 years of age, p=0.01), hospitalized at baseline (13.6% vs. 5.8%, p=0.02), and diagnosed with hypertension (68.2% vs. 54.2%, p=0.03) (Table 2).
The non-linear trend of the Schoenfeld-residual across event time indicated that the PH assumption was violated (Appendix 1). Additionally, Kaplan-Meier survival curves for the risk of hospitalization were intersecting each other, thereby indicating that the treatment effect was not constant over time (Figure 2). To adjust for time in the analysis, the extended Cox model with a Heaviside function was employed. Based on the intersection of the KM survival plots for the persistent and non-persistent group, a cut point of 108 days was chosen to perform the extended Cox model (Figure 2).
Table 3 shows results from the multivariate extended Cox model for the association between persistence to OAD therapy and the risk of all-cause hospitalization. The risk of all-cause hospitalization was higher for patients non-persistent to medication after 108 days of use (HR = 2.25; 95% CI= 1.04 — 4.87; p=0.038). There was no difference in the risk of hospitalization within 108 days of starting medications (HR = 0.94, 95% CI= 0.44 – 2.01, p=0.873).
Appendix 2 shows the distribution of propensity scores between the persistent and non-persistent groups. The figure shows a common region of overlap across the persistent and non-persistent groups indicating that the 2 groups are comparable conditioned on the propensity scores. Results from both the sensitivity analysis involving the propensity score adjusted extended Cox’s regression model (HR = 2.29, 95% CI = 1.07 — 4.89, p-value = 0.033) and the propensity score weighted extended Cox’s regression model (HR = 2.28, 95% CI = 1.02 – 5.11, p-value = 0.045) supported the main findings (Table 3).
Results from the additional sensitivity analyses using 45-day and 60-day gap periods showed that results were consistent with the main findings. There was no difference in the risk of all-cause hospitalization between the two groups within 108 days of starting OAD therapy. After 108 days, the non-persistent group experienced more than 2 times higher risk of hospitalization using a permissible gap of 45-day (HR = 2.63, 95% CI = 1.01 — 6.83, p-value = 0.047) and 60-day (HR = 2.59, 95% CI = 0.94 – 7.17, p-value = 0.066) (Table 3).
In this study of commercially insured patients with T2DM, patients newly started on OAD therapies were poorly persistent, with only 57% persistent during a 1-year follow-up period. There was no difference in the risk of all-cause hospitalization between the 2 groups within 108 days of starting OAD therapy. After 108 days, the non-persistent group had more than a 2 times higher risk of hospitalization compared with the persistent group.
Despite access to health insurance, only 57% of patients were persistent to OAD therapy after initiating treatment. This persistence rate is comparable to a recently conducted meta-analysis of patients with T2DM, which found the persistence to diabetes medication varies from 41% to 81%, with a mean persistence of 56%.23This high rate of treatment discontinuation is especially concerning in T2DM, where strict blood glucose control has been shown to reduce the risk of developing microvascular and macrovascular complications38and can be achieved with precise medication management.39
Improving the sub-optimal rate of medication persistence is invaluable in reducing hospitalization and, therefore, inpatient cost, which is the largest component of medical costs in diabetes patients.2 Results from the present study indicate that better persistence to diabetes medications is associated with lower risk of all-cause hospitalization, starting three months after initiation of OAD therapy after controlling for covariates. These results are consistent with previous studies that indicate higher levels of persistence are related to lower risk of hospitalization and health care cost in other chronic health conditions.40,41
The prevalence of diabetes is expected to increase from 11 million in 2000 to 29 million in 2050.42Given the dramatic increase in the number of patients with diabetes, high rates of non-persistence to OAD therapy among T2DM patients, and associated increased risk of hospitalization, payers may not only benefit from improved patient outcomes through better glycemic control, but also potentially achieve substantial cost-savings through reduced hospital utilization.
Specifically, for a commercially-insured population, self-insured employers and insurers have particular incentive to manage patient care and minimize hospitalizations. Thus, effective interventions that improve persistence or address barriers to medication persistence need to be developed. Medication persistence is a multifaceted issue involving patient, provider, and caregiver characteristics.43,44
For example, increased patient knowledge through educational interventions, phone, text, or mail reminders, as well as simplification of the medication regimen chosen by the physician may be effective to increase persistence.43,45-47 Greater understanding of these factors can aid in developing tailored interventions aimed at improving OAD medication persistence in patients with T2DM. In addition, persistence to OAD therapy is found to differ by OAD classes.29,48,49
Thus, future studies might examine whether or not risk of hospitalization differs by different OAD classes, and how factors that are associated with persistence, such as out-of-pocket cost, adverse effects, and medication availability, relate to OAD class.
Strengths and limitations
Better improvement in health outcomes entails both conformity with the treatment (adherence) and continuous use of the therapy for the recommended duration (persistence). Medication adherence to OAD therapy has been associated with reduced risk of hospitalization and associated cost.18 To the authors’ knowledge, this is the first study to examine association between medication persistence and risk of hospitalization in the short-term among commercially insured patients with T2DM.
Use of commercial claims data provides a large sample of privately insured patients who can be followed longitudinally to examine the association between hospitalization and persistence to OAD therapy in a real world setting.50 Medication claims represent a rich source of pharmacotherapy-related information, including medication persistence.51-53To minimize baseline differences between the treatment groups and possible confounding, propensity score weighted analysis was conducted using inverse probability of treatment weighting and findings were consistent with the main analysis.
This study has some limitations. The present retrospective cohort study was conducted in commercially-insured patients and results may not be generalizable to patients with other types of health coverage. As in any nonrandomized study design, assumptions were made regarding unobserved factors in the data, limiting conclusions regarding causality. The timing of medication interruptions and hospitalizations was carefully reviewed to reduce the likelihood of reverse causality. Although we used previously validated algorithms to identify patients with T2DM using administrative claims,54 administrative data submitted primarily for billing purposes may misidentify some patients.
We used well-validated methods of measuring persistence with OAD medications using pharmacy claims data;26 however, we were unable to assess whether or not patients were actually taking medications, if multiple medications were being taken at the same time, or if pill-cutting techniques were used to “stretch” the days’ supply of a prescription. Discontinuation of oral therapies to start insulin may represent appropriate intensification of a diabetes regimen for some patients; however, since the dose of insulin is dependent on blood glucose measurements, it is more difficult to calculate persistence from the administrative claims data and can introduce error. Thus, these patients were defined as having discontinued OAD therapy.
Consequently, patients taking insulin might have different clinical characteristics from those taking OAD medications only. Although censoring those taking insulin assisted in attributing the treatment effect to OAD therapy and reducing selection bias, it may have concealed the inherent dynamic nature of diabetes treatment. Claims data does not provide information regarding HbA1c, which is an important risk factor for hospitalization due to diabetes complications.
Lastly, this study focused on all-cause hospitalization, as administrative claims data can have less than ideal information on the causes of hospitalization. It is likely that patients who are non-persistent to diabetic medications are more likely to be non-persistent to other medications. Therefore, measuring the all-cause hospitalization rate can provide useful information regarding the overall health status of the patient.
In summary, patients who were non-persistent with OAD therapy experienced more than a 2 times higher risk of hospitalization after the initial 108 days; risk of hospitalization for patients who were persistent compared to those who were not persistent with OAD therapy was similar within 108 days of medication start date. Investment in programs to improve medication persistence may not only improve glycemic control and outcomes of patients with T2DM, but also produce cost-savings through reduced hospitalization.
Even in an environment in which it is uncommon for patients to remain with a single health plan on a long-term basis, the current study shows that there are short-term real-world benefits in encouraging patients to be persistent to OAD medication.
Vishal Bali, Senior Health Outcomes Researcher, West Corporation, Westlake Village, CA; email: VBali2@west.com
Irina Yermilov, Vice President, Clinical Development, West Corporation, Westlake Village, CA; IYermilov@west.com
Alain Koyama, Senior Health Outcomes Researcher, West Corporation, Westlake Village, CA; AKoyama@west.com
Antonio P. Legorreta, Adjunct Professor, University of California, Los Angeles, School of Public Health, Los Angeles, CA; email@example.com
Please send all correspondence to Antonio P. Legorreta at:
Department of Health Policy and Management
UCLA Fielding School of Public Health
650 Charles Young Dr. S.
31-269 CHS Box 951772
Los Angeles, CA, 90095
Tel: (805) 367-7260
Fax: (805) 379-1549
Funding: No funding was received for this study.
Table 1: Characteristics of Patients Who Initiated Diabetes Treatment with OAD Therapy
Overall Cohort (N=1,026)
Persistent to OAD therapy (N=587)
Non-Persistent to OAD therapy (N=439)
Persistence in days (mean ± SD)
Age in years (mean ± SD)
60.1 ± 12.0
60.5 ± 12.0
59.6 ± 12.1
Year of Cohort Entry, %
Hospitalization at the baseline period, %
Charlson Comorbidity Index
1.5 ± 1.6
1.6 ± 1.7
1.5 ± 1.5
Mental health conditions
Use of Comedications, %
Note: a indicates statistical significance at p <0.0001, b indicates statistical significance at p <0.05
Table 2: Characteristics of Patients by Hospitalization
Not- Hospitalized (N=960)
Age in years (mean ± SD)
64.5 ± 14.2
59.6 ± 11.8
Year of Cohort Entry, %
Hospitalization at the baseline period, %
Charlson Comorbidity Index
2.9 ± 2.0
2.1 ± 1.8
Mental health conditions
Use of Comedications, %
Note: a indicates statistical significance at p <0.01, b indicates statistical significance at p <0.05
Table 3: Risk of All Cause hospitalization in Patients Who Initiated Diabetes Treatment with Oral Antidiabetic Therapy (N=1,026)
Adjusted Hazard Ratio
95% Confidence Interval
Multivariable Extended Cox Regression Analysisa
Less than 108 days
0.47 — 1.88
More than 108 days
1.26 — 7.05
First sensitivity analysis: Propensity Score Adjusted Analysis
Less than 108 days
More than 108 days
Second sensitivity Analysis: Propensity Score Weighted (IPTW) Analysis
Less than 108 days
0.45 — 1.97
More than 108 days
1.19 — 7.10
Third sensitivity Analysis: Permissible gap >= 45 daysa
Less than 108 days
0.54 — 2.32
More than 108 days
1.01 — 6.83
Fourth sensitivity Analysis: Permissible gap >= 60 daysa
Less than 108 days
0.58 — 2.54
More than 108 days
0.94 — 7.17
Note: a indicates model adjusted for age, gender, year of entry into the cohort, baseline hospitalization, comedications such as antihypertensives and antihyperlipidemics. Comorbidities such as hypertension, hyperlipidemia, neuropathy, retinopathy, mental health conditions, Charlson comorbidity index; b indicates statistical significance at p <0.05.
1. National Diabetes Statistics Report: Estimates of Diabetes and Its Burden in the United States. 2014. http://www.cdc.gov/diabetes/pubs/statsreport14/national-diabetes-report-web.pdf. Accessed 10 september, 2015.
2. American Diabetes A. Economic costs of diabetes in the U.S. in 2012. Diabetes Care. 2013;36(4):1033-1046.
3. Sundaram M, Kavookjian J, Patrick JH, Miller LA, Madhavan SS, Scott VG. Quality of life, health status and clinical outcomes in Type 2 diabetes patients. Qual Life Res. 2007;16(2):165-177.
4. Crude and Age-Adjusted Hospital Discharge Rates for Diabetes as Any-Listed Diagnosis per 1,000 Diabetic Population, United States, 1988—2009. 2009. https://www.cdc.gov/diabetes/statistics/dmany/fig3.htm.
5. Tomlin AM, Dovey SM, Tilyard MW. Risk factors for hospitalization due to diabetes complications. Diabetes Res Clin Pract. 2008;80(2):244-252.
6. Effect of intensive blood-glucose control with metformin on complications in overweight patients with type 2 diabetes (UKPDS 34). UK Prospective Diabetes Study (UKPDS) Group. Lancet. 1998;352(9131):854-865.
7. Genuth S, Eastman R, Kahn R, et al. Implications of the United kingdom prospective diabetes study. Diabetes Care. 2003;26 Suppl 1:S28-32.
8. 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(9131):837-853.
9. Knowler WC, Barrett-Connor E, Fowler SE, et al. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med. 2002;346(6):393-403.
10. American Diabetes A. What Does Tight Control Mean? 2016.
11. Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HA. 10-year follow-up of intensive glucose control in type 2 diabetes. N Engl J Med. 2008;359(15):1577-1589.
12. Group AC, Patel A, MacMahon S, et al. Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl J Med. 2008;358(24):2560-2572.
13. Gaede P, Lund-Andersen H, Parving HH, Pedersen O. Effect of a multifactorial intervention on mortality in type 2 diabetes. N Engl J Med. 2008;358(6):580-591.
14. 2016 Part C & D Star Ratings Measures. https://www.cms.gov/Medicare/Prescription-Drug-Coverage/PrescriptionDrugCovGenIn/Downloads/2016_Star_Ratings_Measure_List.pdf. Accessed 11 September, 2015.
15. NCQA. Addressing the Quality Gaps in Diabetes Prevention and Care. http://www.ncqa.org/PublicationsProducts/OtherProducts/QualityProfiles/FocusonDiabetes/AddressingtheQualityGaps.aspx#sthash.Z4SKE2tq.dpuf. Accessed 11 September, 2011.
16. Seabury SA, Lakdawalla DN, Dougherty JS, Sullivan J, Goldman DP. Medication adherence and measures of health plan quality. Am J Manag Care. 2015;21(6):e379-389.
17. Cramer JA, Roy A, Burrell A, et al. Medication compliance and persistence: terminology and definitions. Value Health. 2008;11(1):44-47.
18. Breitscheidel L, Stamenitis S, Dippel FW, Schoffski O. Economic impact of compliance to treatment with antidiabetes medication in type 2 diabetes mellitus: a review paper. J Med Econ. 2010;13(1):8-15.
19. Lau DT, Nau DP. Oral antihyperglycemic medication nonadherence and subsequent hospitalization among individuals with type 2 diabetes. Diabetes Care. 2004;27(9):2149-2153.
20. Zhuo X, Zhang P, Barker L, Albright A, Thompson TJ, Gregg E. The lifetime cost of diabetes and its implications for diabetes prevention. Diabetes Care. 2014;37(9):2557-2564.
21. Mahoney JJ. Reducing patient drug acquisition costs can lower diabetes health claims. Am J Manag Care. 2005;11(5 Suppl):S170-176.
22. Jha AK, Aubert RE, Yao J, Teagarden JR, Epstein RS. Greater adherence to diabetes drugs is linked to less hospital use and could save nearly $5 billion annually. Health Aff (Millwood). 2012;31(8):1836-1846.
23. Iglay K, Cartier SE, Rosen VM, et al. Meta-analysis of studies examining medication adherence, persistence, and discontinuation of oral antihyperglycemic agents in type 2 diabetes. Curr Med Res Opin. 2015;31(7):1283-1296.
24. NCQA. Comprehensive Diabetes Care. 2015; http://www.ncqa.org/report-cards/health-plans/state-of-health-care-quality/2015-table-of-contents/diabetes-care. Accessed December 15, 2015.
25. Ray WA. Evaluating medication effects outside of clinical trials: new-user designs. Am J Epidemiol. 2003;158(9):915-920.
26. Leslie SR, Gwadry-Sridhar F, Thiebaud P, Patel BV. Calculating medication compliance, adherence and persistence in administrative pharmacy claims databases. Pharmaceutical programming. 2008;1(1):13-19.
27. Bali V, Kamble PS, Aparasu RR. Predictors of concomitant use of antipsychotics and stimulants and its impact on stimulant persistence in pediatric attention deficit hyperactivity disorder. J Manag Care Spec Pharm. 2015;21(6):486-498.
28. Bali V, Kamble PS, Aparasu RR. Cardiovascular Safety of Concomitant Use of Atypical Antipsychotics and Long-Acting Stimulants in Children and Adolescents With ADHD. Journal of Attention Disorders. 2015.
29. Gregoire JP, Sirois C, Blanc G, Poirier P, Moisan J. Persistence patterns with oral antidiabetes drug treatment in newly treated patients--a population-based study. Value Health. 2010;13(6):820-828.
30. Ho PM, Magid DJ, Masoudi FA, McClure DL, Rumsfeld JS. Adherence to cardioprotective medications and mortality among patients with diabetes and ischemic heart disease. BMC Cardiovasc Disord. 2006;6:48.
31. Jermendy G, Wittmann I, Nagy L, et al. Persistence of initial oral antidiabetic treatment in patients with type 2 diabetes mellitus. Med Sci Monit. 2012;18(2):CR72-77.
32. Khalid JM, Raluy-Callado M, Curtis BH, Boye KS, Maguire A, Reaney M. Rates and risk of hospitalisation among patients with type 2 diabetes: retrospective cohort study using the UK General Practice Research Database linked to English Hospital Episode Statistics. Int J Clin Pract. 2014;68(1):40-48.
33. Sokol MC, McGuigan KA, Verbrugge RR, Epstein RS. Impact of medication adherence on hospitalization risk and healthcare cost. Med Care. 2005;43(6):521-530.
34. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373-383.
35. Ahmed FE, Vos PW, Holbert D. Modeling survival in colon cancer: a methodological review. Mol Cancer. 2007;6:15.
36. Harder VS, Stuart EA, Anthony JC. Propensity score techniques and the assessment of measured covariate balance to test causal associations in psychological research. Psychol Methods. 2010;15(3):234-249.
37. Robins JM, Hernan MA, Brumback B. Marginal structural models and causal inference in epidemiology. Epidemiology. 2000;11(5):550-560.
38. Clement M, Bhattacharyya O, Conway JR. Is tight glycemic control in type 2 diabetes really worthwhile?: YES. Can Fam Physician. 2009;55(6):580-582.
39. Krentz AJ, Bailey CJ. Oral antidiabetic agents: current role in type 2 diabetes mellitus. Drugs. 2005;65(3):385-411.
40. Breekveldt-Postma NS, Siiskonen SJ, Penning-van Beest FJ, et al. Persistent use of antihypertensive drugs leads to increased risk of hospitalizations for acute myocardial infarction or stroke. Value in Health.9(6):A339.
41. Jiang Y, Ni W. Estimating the Impact of Adherence to and Persistence with Atypical Antipsychotic Therapy on Health Care Costs and Risk of Hospitalization. Pharmacotherapy. 2015;35(9):813-822.
42. Boyle JP, Honeycutt AA, Narayan KMV, et al. Projection of Diabetes Burden Through 2050. Impact of changing demography and disease prevalence in the U.S. 2001;24(11):1936-1940.
43. Krueger KP, Felkey BG, Berger BA. Improving adherence and persistence: a review and assessment of interventions and description of steps toward a national adherence initiative. J Am Pharm Assoc (2003). 2003;43(6):668-678; quiz 678-669.
44. Touchette DR, Shapiro NL. Medication compliance, adherence, and persistence: current status of behavioral and educational interventions to improve outcomes. J Manag Care Pharm. 2008;14(6):S2-S10.
45. Vervloet M, van Dijk L, Santen-Reestman J, et al. SMS reminders improve adherence to oral medication in type 2 diabetes patients who are real time electronically monitored. Int J Med Inform. 2012;81(9):594-604.
46. Legorreta AP, Hasan MM, Peters AL, Pelletier KR, Leung KM. An intervention for enhancing compliance with screening recommendations for diabetic retinopathy. A bicoastal experience. Diabetes Care. 1997;20(4):520-523.
47. Halbert RJ, Leung KM, Nichol JM, Legorreta AP. Effect of multiple patient reminders in improving diabetic retinopathy screening. A randomized trial. Diabetes Care. 1999;22(5):752-755.
48. Farr AM, Sheehan JJ, Curkendall SM, Smith DM, Johnston SS, Kalsekar I. Retrospective analysis of long-term adherence to and persistence with DPP-4 inhibitors in US adults with type 2 diabetes mellitus. Adv Ther. 2014;31(12):1287-1305.
49. Boccuzzi SJ, Wogen J, Fox J, Sung JC, Shah AB, Kim J. Utilization of oral hypoglycemic agents in a drug-insured U.S. population. Diabetes Care. 2001;24(8):1411-1415.
50. Wilson J, Bock A. The benefit of using both claims data and electronic medical record data in health care analysis. Optum Insight. 2012.
51. Tamblyn R, Lavoie G, Petrella L, Monette J. The use of prescription claims databases in pharmacoepidemiological research: the accuracy and comprehensiveness of the prescription claims database in Quebec. J Clin Epidemiol. 1995;48(8):999-1009.
52. Dezii CM. Persistence with drug therapy: a practical approach using administrative claims data. Manag Care. 2001;10(2):42-45.
53. Kirking D, Ammann M, Harrington C. Comparison of medical records and prescription claims files in documenting prescription medication therapy. Journal of Pharmacoepidemiology. 1996;5:3-16.
54. Comprehensive Diabetes Care (CDC) National Committee for Quality Assurance (NCQA). 2009. http://www.ncqa.org/portals/0/PolicyUpdates/HEDIS%20Technical%20Updates/09_CDC_Spec.pdf. Accessed September 10, 2015.