• costs;
  • diabetes;
  • economics;
  • medication adherence;
  • medication compliance


  1. Top of page
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References

Objectives:  Information on the health care costs associated with nonadherence to treatments for diabetes is both limited and inconsistent. We reviewed and critically appraised the literature to identify the main methodological issues that might explain differences among reports in the relationship of nonadherence and costs in patients with diabetes.

Methods:  Two investigators reviewed Medline, EMBASE, Cochrane library and CINAHL and studies with information on costs by level of adherence in patients with diabetes published between January 1, 1997 and September 30th 2007 were included.

Results:  A total of 209 studies were identified and ten fulfilled the inclusion criteria. All included studies analyzed claims data and 70% were based on non-Medicaid and non-Medicare databases. Low medication possession ratios were associated with higher costs. Important differences were found in the ICD-9/ICD-9 CM codes used to identify patients and their diagnoses, data sources, analytic window period, definitions of adherence measures, skewness in cost data and associated statistical issues, adjustment of costs for inflation, adjustment for confounders, clinical outcomes and costs.

Conclusions:  Important variation among cost estimates was evident, even within studies of the same population. Readers should be cautious when comparing estimated coefficients from various studies because methodological issues might explain differences in the results of costs of nonadherence in diabetes. This is particularly important when estimates are used as inputs to pharmacoeconomic models.


  1. Top of page
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References

Nonadherence has a significant impact on the cost-effectiveness of pharmaceuticals [1], and has been estimated to cost the US economy up to $100 billion per year [2]. In diabetes, nonadherence to oral hypoglycemic medications [3,4] may partly explain why only 43% of patients with diabetes mellitus have glycosylated hemoglobin (HbA1c) below the 7% level [5,6] recommended by the American Diabetes Association [7].

Studies of adherence in diabetes have focused on its economic burden [8–10], its complications [11,12] and the cost-effectiveness of antidiabetic drugs [13–18]. Many have reported wide variation in the percent of patients being “nonadherent,” ranging from 13% to 64% for oral agents and from 19% to 46% for users of insulin [19–21]. Additionally, important variations in the coefficient estimations for costs have been reported [21,22], which might be related to differences in the design, population, variables included in the analysis and statistical analyses. Therefore, we reviewed and critically appraised the literature to identify the main methodological issues that might explain differences among reports in the relationship of nonadherence and costs in patients with diabetes.


  1. Top of page
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References

Search Strategy

We conducted a systematic literature review using Medline, EMBASE, Cochrane Library, and the Cumulative Index to Nursing and Allied Health Literature (CINAHL) from January 1, 1997 to September, 30 2007.

The key terms used included: (compliance, adherence, persistence, nonadherence, concordance) AND (economics, costs, value, expenditures, resource utilization) AND (diabetes, hyperglycemia, diabetes-related complications, antidiabetic medications, insulin, oral hypoglycemic agents). We also hand-searched medical journals and reviewed the reference lists of other reviews.

Selection Criteria

Studies that reported costs by different levels of medication adherence or persistence were included. Adherence and persistence definitions were according to previous studies [23]. We also included studies that used HbA1c as a proxy of medication adherence because HbA1c is a well-established measure of glycemic control [22,24,25] and a proxy for adherence [26]. Non-English studies, articles with insufficient data, and those without costs or adherence information were excluded.

Extracted Information

Abstracts and full publications were reviewed by two researchers and disagreements were resolved by consensus. The extracted information included the study design, data source(s), methods of adherence measurement, statistical analysis, and results. Study designs were classified as trials, cohort, case-control, or cross-sectional studies. Data sources for patient demographics, adherence, resource utilization, and costs, as well as observation and follow-up periods, were recorded (Table 1). For statistical analysis, we included information on any statistical method used to assess the relationship or association between medication nonadherence and costs, sample size, adjustment for inflation and/or discounting, adjustment for confounders or for the days when patients were in institutionalized care settings such as hospital, and nursing home (Table 2).

Table 1.  Studies identified with costs reported by adherence level in diabetic patients
ReferenceDesignSource of dataInclusion criteriaObservation periodFollow-up period
  1. CVD, cardiovascular disease; HMO, Health Maintenance Organization; ICD-9, International Classification of Diseases, 9th Revision; ICD-9 CM, International Classification of Diseases Clinical Modification. HbA1c, glycosylated hemoglobin.

Balkrishnan R, 2003 [28]Retrospective cohortMedicare HMO in North CarolinaICD-9 codes 250.xxPrescription refillsReimbursement by the HMOPatients aged ≥65 years, enrolled in a Medicare HMO in North Carolina who received ≥1 antidiabetic prescription dispensed every 6 months1996–2002Up to 5 years
Cobden D, 2007 [29]Retrospective cohortPharmetricsICD-9 CM code 250.xx excluding type 1 subcodesPrescription refillsPayments made by third-party payers to health care providers (reimbursement)≥18 years, type 2 diabetes who converted to BIAsp 70/30 pen device and previously treated with human or analog insulinJanuary 1, 2001 to April 30, 2005At least 2 years
Balkrishnan R, 2004 [30]Retrospective cohortNorth Carolina Medicaid programICD-9Prescription refillsReimbursementType 2 diabetes who were newly started on thiazolidinedione therapy or other oral antidiabetic drugJuly 2001 to June 20022 years
Hepke, 2004 [31]Retrospective cohortBlue Cross Blue Shield of MichiganICD-9 250, 352.2, 362, 366.41, 648Prescription refillsReimbursementNon-Medicare eligible Michigan residents enrolled continuously in 1999, at least 1 inpatient or Emergency room claim, ≥2 professional or outpatient facility claims with diabetes diagnosis and a filled prescription for antidiabetic drug.19991 year
Lee WC, 2006 [17]Retrospective cohort with pre and post analysisIntegrated medical and pharmacy claims database: PharmetricICD-9 code 250.xx excluding type 1 subcodesPrescription refillsPayments to the health insurance: reimbursement≥18 years of age, type 2 diabetes who initiated treatment with insulin analogue pen device between July 1, 2001 and Dec 31, 2002, and whose treatment was converted from conventional human or analogue insulin injection (vial/syringe) to a prefilled insulin analogue pen.January 2001–April 2005Up to 4 years
Shenolikar RA, 2006 [32]Retrospective cohortNorth Carolina Medicaid databaseICD-9 CM code 250.xxPrescription refillsTotal health care costs: medical and dental care, regular checkups, office visits, home health care, inpatient and outpatient care, long term care facility care and prescription drugs.At least one ICD9 code for diabetes, and one for antidiabetic medication and Medicaid eligibility for 36-month follow-up period. African Americans were analyzed vs. otherJuly 1, 2000 to June 30, 20031 year
Sokol MC, 2005 [33]Retrospective cohortAdministrative claims database maintained by a health plan organizationICD-9 codes 250.xx, 357.2, 362.0x, 366.41, 648.0Prescription refillsAll-cause costs and disease-related costs.Patients aged 65 and older with diagnosis of diabetesJune 1997 to May 19991 year
Wagner EH, 2001 [34]Retrospective cohortAutomated diabetes registry from the Group Health Cooperative of Puget Sound, Seattle WashingtonDiagnosis of diabetes and HbA1c from diabetes registryPrescriptions refills and HbA1cDecision support system that is automated, step-down cost accounting for health care provided to members.Diabetics older than 18 years, with at least one HBA1c, and continuously enrolled from 1992–1996January 1, 1992 to March 31, 19964 years
White, TJ 2004 [35]Retrospective cohortManaged care organization databaseICD-9 for type 2 diabetesPercentage of adherenceClaims dataPatients receiving an oral antidiabetic medication and have a diagnosis of CVD, continuously enrolled in the health plan, and ≥30 years of ageApril 1, 1998 to March 31, 20001 year
Shetty S, 2005 [36]Retrospective cohortUS Managed care organizationICD-9 CM codes 250.x0 or 250.x2Not reportedReimbursementHad ≥2 claims for type 2 diabetes in either the primary or secondary position, had at leas one prescription for an oral hypoglycemic agent and/or insulin, had at least one available HbA1c, were commercially insured with a drug benefit, and had at least 6 months of continuous enrollment.January–December 20021 year
Table 2.  Continuation of studies identified with costs reported by adherence level in diabetic patients
ReferenceNMeasures methodStatistical analysisResultsQuality score
AdherenceResource utilizationCosts/adjustment for inflationStatistical methodContinuous enrollment in the health insuranceAdjustment by confoundersAdjustment by hospitalization or other location
  1. ER, emergency room; HMO, health maintenance organization; MPR, medication possession ratio; NR, not reported. HbA1c, glycosylated hemoglobin.

Balkrishnan R, 2003 [28]775MPR defined as the days of antidiabetic prescription supply dispensed divided by the number of days between prescription refills. The observation period began with the first date of dispensing within each year and ended as the dispensing date of the last prescriptionAdministrative claims data of the HMOTotal costs not specifiedSequential mixed-model and regression analysisYesCharlson index was used to adjust by severityNumber of days during hospitalization was subtracted from the denominatorMPR for 1 to 5 years of follow up were 0.70, 0.71, 0.75, 0.77, and 0.78; and mean health care costs were $8,306, $5,947, $5,821, $5,043, $5,118. 10% increase in antidiabetic MPR was associated with an 8.6% decrease in total annual health care costs (P < 0.001). After 5 years, high adherence = $4,000 while low adherence = $10,50014/30
Cobden D, 2007 [29]486MPR: sum of the days' supply of drug divided by the number of days between the first fill and the last refill plus the days' supply of the last refillPhysician visits, hospitalization, emergency department visits, pharmacy dataTotal health care costs, annual adjusted mean all-cause health care costs/adjustment for inflation to 2005 dollarsPerson-time and event-time analysis adjusted by length of follow-up. Poisson regression model and gamma regression modelYesHypoglycemia/adjustment for ComorbiditiesNot reportedMPR of 80% or greater was associated with significant reduction in all-cause health care costs (OR 0.55, 95% CI 0.31–0.80, P < 0.05). MPR of 68% was associated with total mean costs of $8,056 ± 8,559, while an MPR of 59% had total mean costs of $8,699 ± 9,268.14/30
Balkrishnan R, 2004 [30]3,483MPRClaims dataTotal annual health care costsMultivariate techniquesNRNRNR13% increase in MPR was associated with 16.1% lower total annual health care costs (P < 0.001).12/30
Hepke KL, 2004 [31]57,687Medication adherence rate calculated as percentage of days that the patient possessed any available diabetic drug during the yearInpatient hospitalization, outpatient care, emergency care, clinic visits, laboratory tests, professional services and pharmaceuticalsOverall cost of healthcare and cost related with diabetes careLeast squares regression model and multivariate logistic regressionYesIllness severity using diagnosis cost group.NR20% to 39% adherence level was needed before medical care costs were reduced. For diabetes related costs, the threshold was seen until 40% to 59% adherence level. Adherence-total average expenditures 0% = $6,500, 1–19% = $7,250, 20–39% = $7,750, 40%–59% = $7,500, 60%–79% = $7,700, 80%–99% = $7,300, 100% = $7,900.11/30
Lee WC, 2006 [17]1,156MPR: sum of the days' supply of medication divided by the number of days between the first fill and the last refill plus the days' supply of the last refill.Physician visits, hospitalizations, ER visitsTotal health care costs/Costs adjusted to 2005 dollars using the consumer price indexPerson-time and time-event analysis. Poisson regression models and incident rate ratios.Enrollment for at least 6 months before the index date and at least 2 years of continuous enrollment after the index dateHypoglycemic eventsNR62% MPR to insulin pen therapy = mean annual all-cause health care costs $14,76911/30
Shenolikar RA, 2006 [32]1,073MPR: Number of days of antidiabetic prescription supply dispensed (e.g., a 30-day supply) divided by the number of days between the first and last dispensation. Med-Total approach: ratio of total number of days the drug was supplied to the difference in the number of days between the first and last prescription dates.Medical and dental care, in patient and outpatient care, regular checkups, office visits, home health care, long-term facility care and prescription drugs.Annual total and diabetes-related health care costs/No discounted rate reportedMultivariate regression analysis adjusted by covariates. Costs were transformed using logarithm and they were transformed back using antilogarithms of the parameter estimateYesAdjusted by comorbiditiesNoMean rate of adherence to new medication of 59% = $9,546 ±  $14,861 mean total health care costs for year 2 and mean diabetes-related costs for year 2 of $4,576 ± $8,208; The estimated coefficients and standard errors for total annual health care cots as a function of covariates were: male sex 1,117.35 ± 1,001.69, high total number of prescriptions 8,223.48 ± 1,002.38; African American race 1,125.49 ± 914.39; rate of adherence—2,721.68 (932.50), constant 728.82 (1,180.29) and adjusted r2 = 0.06.13/30
Sokol MC, 2005 [33]137,277Percentage of days during the analysis period that patients had a supply of 1 or more maintenance medications for the conditionMedical and drug claims: hospitalization, ER service, outpatient services including physician office visits and outpatient visits. Nursing home and home care services were not includedTotal health care costs (Sum of medical—outpatient services, ER services, hospitalization- and drug costs), and disease-related costs. Net cost to the plan sponsor, patient copayments and deductibles were not included. Costs were adjusted for age, sex, comorbidity, disease subtype, employment group and medical plan type.Logistic regression model. No detail for costs transformation was provided.YesComorbidities were included in the analysis.NRAdherence level and total costs: 1–19% = $8,867; 20–39% =  $7,124; 40–59% = $6,522; 60–79% = $6,291; 80–100% =  $4,570. Differences were statistically significant for most adherence levels when compared with the highest level of adherence (P < 0.05).8/30
Wagner EH, 2001 [34]4,744Not included a measure of adherence and HbA1c was used as a proxy of medication adherenceAnnual utilization ratesTotal health care costs and mean costs per personCost data logarithmically transformed. Regression analysis.YesNoNRBaseline HbA1c, Level, % and mean annual costs$, (p values were calculated for the difference in log costs) <8 = $4,475 (P = 0.18); 8–10 = $5,898 (P = 0.32); >10 = $8,088 (P = 0.53)11/30
White TJ, 2004 [35]NRPercentage of adherenceHospitalization and ERAdjusted total healthcare costsRegression modelYesNRNRPatients with ≤75, >75 to ≤95 and 95% adherence, adjusted total healthcare costs were $US 5,706, $5,314 and $4,835 (P < 0.001).9/30
Shetty S, 2005 [36]3,121Not included but HbA1c was used as a proxy of medication adherence.Claims dataCosts of 6 months periodMultiple linear regression analysis. Logarithmic transformation of cost data was done prior analysis.YesAdjustment by age, gender, specialty of the physician, comorbidity and total baseline costs.NRPredicted total diabetes-related cost for target HbA1c level group during the first year of follow up was $1,540 per patient, 32% higher than the total diabetes related cost ($1,171) for the same target group (P < 0.001).11/30

Quality Criteria

A checklist for economic evaluation [27] was modified to assess the quality of studies. The original checklist contained 35 items, but 5 of them were related to health economic models (12, 14 15, 20, and 21), and were not considered applicable to the studies included in the review. We assigned a score of 1 if an article included the required item, and zero if it was not included. Therefore, the maximum score for an article that included all information related to study design, data collection, analysis and interpretation of results was 30.


  1. Top of page
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References

Search Results

Two hundred nine titles were identified and their abstracts were reviewed. Fifty abstracts included information on both adherence and costs in patients with diabetes, and their full articles were retrieved. Ten studies [17,28–36] fulfilled the inclusion criteria (Fig. 1). All studies analyzed US claims data using retrospective cohort studies designs [17,28–36] (Table 1). Three studies utilized Medicare or Medicaid databases [28,30,32], while all others used commercial or managed care organizations data sets.


Figure 1. Flow diagram showing the number of references identified, retrieved and included in the review.

Download figure to PowerPoint

Association between Medication Nonadherence and Cost

There were important variations in the items included in order to estimate costs. For example, one study included only claims for physician office visits, outpatient services, and hospital stays [29], while another was more comprehensive, and included: costs for hospitalization, outpatient care, emergency care, clinic visits, laboratory tests, professional services, and pharmaceuticals [31]. Two studies took into account the net cost to the plan but they did not include patients' copayments and deductibles [33,35], while a third study included copayments and deductibles [36]. The study by Wagner used its own internal accounting system that included overhead costs [34]. It was unclear in some studies as to which specific costs were included [17,28,32].

Low medication possession ratios (MPRs) were generally associated with higher costs. For example, one study reported an association of MPR of 60% with mean total costs of $8699 [29]. Balkrishnan et al. found that a 10% increase in MPR for an antidiabetic medication was associated with an 8.6% reduction in total annual health care costs [28]. Studies generally reported increments of mean annual costs according to baseline HbA1c values. For example, the mean annual costs for patients with baseline HbA1c < 8% were $4475, while for those with HbA1c > 10 were $8088 [34] (Table 2).

Methodological Issues

The specific International Classification of Diseases (ICD-9) or ICD-9 Clinical Modification (ICD-9-CM) codes used to identify the study population were not mentioned in three studies [30,34,35], and among those that were reported, there were important variations in the codes included (Tables 1 and 2). For example, some studies included type 1 and type 2 diabetes [28,31,33], while others excluded type 1 diabetes [17,29,36]. The population varied by study, as well by period of observation. The maximum follow-up was 5 years, and half of the studies followed patients for only 1 year.

Table 2 presents measures of adherence, costs, statistical analysis, results of each study, and quality score. All studies used claims data to collect drug utilization information. Five studies used MPR as a measure of medication adherence [17,28–30,32], two studies did not report a specific medication adherence measure [34,36], and three used various measures of medication adherence such as medication adherence rate [31], percentage days supplies [33], and percentage of adherence [35]. All studies used the total follow-up period to calculate adherence and costs, and used charges as proxy for costs. In terms of type of costs, some studies reported total health-care costs [17,28–30,34–36], while others focused on overall costs of health care [31], or costs related with diabetes care [32,33]. Two studies used Poisson regression models for costs [17,29], and the remainder used multivariate regression analysis for costs. Few studies log-transformed costs [32,34,36], and only one study [22] tried to deal with the skewed distribution of both health-care costs and MPRs. Seven studies were able to adjust for some potential confounders [17,28,29,31–33,36], while only one adjusted costs for inflation and duration of hospitalizations [28]. Most studies were assigned a low quality score (<50% of required information), ranging from 8/30 to 14/30.


  1. Top of page
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References

We identified various methodological issues that hinder comparisons from being made across studies, and which might result in significant differences in the reported associations between nonadherence to medicines and costs in patients with diabetes.

Based on the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) recommendations on improving the quality of adherence studies [37,38], we found that the type of study design was not clearly established, and studies were unable to distinguish prevalent from incident cases. Incident cases are more expensive than prevalent cases in terms of hospitalization rates, length of stay, case mix, and service intensity, and have higher discontinuation rates [39–41]. Studies included different population groups, which has an impact on costs: some focused on codes for type 2 diabetes only, type 1 and type 2, gestational diabetes, and/or diabetes-related complications. For example, gestational diabetes is more expensive than type 2 diabetes because of the frequency and duration of hospitalizations [42]. None of the studies described if primary, secondary, or both codes were used. Previous studies have shown an increase in costs by up to twofold when both primary and secondary ICD-9 codes were used [31].

Contrary to accepted recommendations, none of the studies validated ICD-9/ICD-9-CM codes [43]. Wilchesky et al. showed 64% sensitivity of claims data to detect patients with diabetes [44], which means that an appreciable number of cases may be missed. Similarly, none of the studies validated prescription claims data that are vulnerable to errors from sampling, misidentification of newly treated patients, and misclassification of added versus switched medications [45,46]. ICD-9/ICD-9-CM codes to measure utilization and costs also requires validation, because some studies have found that 9% of discharges incorrectly omit codes for diabetes, and 13% of discharges are registered without any foot-related diagnosis code [47].

Most studies used medication possession ratios, but there were important variation in the definition. For some, MPR was the sum of days of antidiabetic prescription supply dispensed divided by the number of days between prescription refills, from the first date of dispensing within each year until the dispensing date of the last prescription [28,32]. Others added the days' supply of the last refill to the denominator [17,29], or they used the percentage of days that the patient possessed any available diabetic drug during the year [31]. None of the studies considered the effects of censoring, which is important, because six filled prescriptions evaluated over 12 months equals an MPR of 50%, but if they are evaluated over 6 months, the six filled prescriptions equals an MPR of 100%.

The non-MPR measures included were: Med-total approach defined as the ratio of total number of days the drug was supplied to the difference in the number of days between the first and last prescription dates [32]; the percentage of days during the analysis period that patients had a supply of one or more maintenance medications for the condition [33], and the percentage of adherence [35]. The problem is that these measures are not comparable. Hess [48] analyzed various adherence measures and found that only 4—Continuous Measure of Medication Acquisition; Continuous Multiple Interval Measure of Oversupply; MPR; and Medication Refill Adherence—out of 11 measures were identical for measuring adherence to prescription refills throughout the study period.

With regard to confounders, 6 out of 10 studies made some effort to adjust their estimates by disease severity, but most did not adjust by comorbidities, thereby potentially underestimating the real costs. None of the databases used by analysts contain information of behavioral variables such as smoking and alcohol that are closely related to adherence [49–53]. There was also lack of information on adverse drug events, such as hypoglycemia, which has been shown to be a costly component of diabetes-related treatment [54]. None of the studies were able to measure the direct consequences of either nonadherence (e.g., hyperosmolar coma) or associated utilization-based outcomes. Costs were, therefore, not disaggregated according to the main drivers that are a consequence of loss of therapeutic effect through nonadherence.

All studies used charges as proxy for costs. However, charges have been criticized because they do not reflect real costs [55], and they do not take into account the various levels of copayment, deductibles, and coinsurance for prescriptions and other medical services, including physician office care, medical emergency care, and inpatient hospitalization.

Only one study tried to deal with skewed distribution of health-care costs and MPR [22]. This is important, because inappropriate analysis of costs will produce biased estimates for the mean. For costs, nonparametric bootstrap techniques or GLM regression analyses are recommended [56,57].


  1. Top of page
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References

The research assessing the association between medication adherence/nonadherence and health-care costs is limited and of poor quality. There are important methodological differences among studies of costs of adherence/nonadherence in patients with diabetes, making robust comparisons difficult; and those differences might explain the inconsistency in the reported associations between medication adherence and costs. Readers should be cautious when interpreting or comparing the results of such studies. More research is needed to validate measures of medication adherence using claims data and to determine the impact of nonadherence on health-care costs.


  1. Top of page
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References

This article is written by members of the International Society for Pharmacoeconomics & Outcomes Research (ISPOR) Economics of Medication Compliance Working Group; part of the Medication Compliance and Persistence Special Interest Group.

Source of financial support: None.

Supporting information for this article can be found at:


  1. Top of page
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References
  • 1
    Hughes DA, Bagust A, Haycox A, Walley T. The impact of non-compliance on the cost-effectiveness of pharmaceuticals: a review of the literature. Health Econ 2001;10:60115.
  • 2
    Lewis A. Noncompliance: a $100 billion problem. Remington Report 1997;5:145.
  • 3
    World Health Organization. Adherence to long-term therapies: evidence for action. 2003. Available from:[Accessed December 12, 2007].
  • 4
    Osterberg L, Blaschke T. Adherence to medication. N Engl J Med 2005;353:48797.
  • 5
    Saaddine JB, Engelgau MM, Beckles GL, et al. A diabetes report card for the United States: quality of care in the 1990s. Ann Intern Med 2002;136:56574.
  • 6
    Kerr EA, Gerzoff RB, Krein SL, et al. Diabetes care quality in the Veterans Affairs Health Care System and commercial managed care: the TRIAD study. Ann Intern Med 2004;141:27281.
  • 7
    American Diabetes Association. Standards of medical care in diabetes—2007. Diabetes Care 2007;30:S441.
  • 8
    Abegunde DO, Mathers CD, Adam T, et al. The burden and costs of chronic diseases in low-income and middle-income countries. Lancet 2007;370:192938.
  • 9
    Clarke P, Gray A, Legood R, et al. The impact of diabetes-related complications on healthcare costs: results from the United Kingdom Prospective Diabetes Study (UKPDS Study No. 65). Diabet Med 2003;20:44250.
  • 10
    Pohar SL, Majumdar SR, Johnson JA. Health care costs and mortality for Canadian urban and rural patients with diabetes: population-based trends from 1993–2001. Clin Ther 2007;29:131624.
  • 11
    Stratton IM, Adler AI, Neil HA, et al. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ 2000;321:40512.
  • 12
    UK Prospective Diabetes Study (UKPDS) Group. 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;352:85465.
  • 13
    Salas M, Ward A, Caro J. Health and economic effects of adding nateglinide to metformin to achieve dual control of glycosylated hemoglobin and postprandial glucose levels in a model of type 2 diabetes mellitus. Clin Ther 2002;24:1690705.
  • 14
    Balu S. Incremental treatment expenditure of diabetes in the United States. Manag Care Interface 2007;20:207.
  • 15
    Valentine WJ, Erny-Albrecht KM, Ray JA, et al. Therapy conversion to insulin among patients with type 2 diabetes treated with oral agents: a modeling study of cost-effectiveness in the United States. Adv Ther 2007;24:27390.
  • 16
    Pawaskar MD, Camacho FT, Anderson RT, et al. Health care costs and medication adherence associated with initiation of insulin pen therapy in Medicaid-enrolled patients with type 2 diabetes: a retrospective database analysis. Clin Ther 2007;29:1294305.
  • 17
    Lee WC, Balu S, Cobden D, et al. Medication adherence and the associated health-economic impact among patients with type 2 diabetes mellitus converting to insulin pen therapy: an analysis of third-party managed care claims data. Clin Ther 2006;28:171225.
  • 18
    Kalsekar I, Iyer S, Mody R, et al. Utilization and costs for compliant patients initiating therapy with pioglitazone or rosiglitazone versus insulin in a Medicaid fee-for-service population. J Manag Care Pharm 2006;12:1219.
  • 19
    Lee WC, Balu S, Cobden D, et al. Prevalence and economic consequences of medication adherence in diabetes: a systematic literature review. Manag Care Interface 2006;19:3141.
  • 20
    Cramer JA. A systematic review of adherence with medications for diabetes. Diabetes Care 2004;27:121824.
  • 21
    Ho PM, Rumsfeld JS, Masoudi FA, et al. Effect of medication nonadherence on hospitalization and mortality among patients with diabetes mellitus. Arch Intern Med 2006;166:183641.
  • 22
    Gerstein HC. Glycosylated hemoglobin: finally ready for prime time as a cardiovascular risk factor. Ann Int Med 2004;141: 4756.
  • 23
    Cramer JA, Roy A, Burrell A, et al. Medication compliance and persistence: terminology and definitions. Value Health 2008;11: 447.
  • 24
    American Diabetes Association. Standards of medical care in diabetes—2009. Diabetes Care 2009;32(Suppl. 1):S1361. Available from:[Accessed January 5, 2009.
  • 25
    Selvin E, Marinopoulos S, Berkenblit G, et al. Meta-analysis: glycosylated hemoglobin and cardiovascular disease in diabetes mellitus. Ann Intern Med 2004;141:42131.
  • 26
    Rozenfeld Y, Hunt JS, Plauschinat C, Wong KS. Oral antidiabetic medication adherence and glycemic control in managed care. Am J Manag Care 2008;14:715.
  • 27
    Drummond MF, Jefferson TO. Guidelines for authors and peer reviewers of economic submissions to the BMJ. BMJ 1996;313:27583.
  • 28
    Balkrishnan R, Rajagopalan R, Camacho FT, et al. Predictors of medication adherence and associated health care costs in an older population with type 2 diabetes mellitus: a longitudinal cohort study. Clin Ther 2003;25:295871.
  • 29
    Cobden D, Lee WC, Balu S, et al. Health outcomes and economic impact of therapy conversion to a biphasic insulin analog pen among privately insured patients with type 2 diabetes mellitus. Pharmacotherapy 2007;27:94862.
  • 30
    Balkrishnan R, Rajagopalan R, Shenolikar RA, et al. Healthcare costs and prescription adherence with introduction of thiazolidinedione therapy in Medicaid type 2 diabetic patients: a retrospective data analysis. Curr Med Res Opin 2004;20:163340.
  • 31
    Hepke KL, Martus MT, Share DA. Costs and utilization associated with pharmaceutical adherence in a diabetic population. Am J Manag Care 2004;10:14451.
  • 32
    Shenolikar RA, Balkrishnan R, Camacho FT, et al. Comparison of medication adherence and associated health care costs after introduction of pioglitazone treatment in African Americans versus all other races in patients with type 2 diabetes mellitus: a retrospective data analysis. Clin Ther 2006;28:1199207.
  • 33
    Sokol MC, McGuigan KA, Verbrugge RR, Epstein RS. Impact of medication adherence on hospitalization risk and healthcare cost. Med Care 2005;43:52130.
  • 34
    Wagner EH, Sandhu N, Newton KM, et al. Effect of improved glycemic control on health care costs and utilization. JAMA 2001;285:1829.
  • 35
    White TJ, Vanderplas A, Chang E, et al. The costs of non-adherence to oral antihyperglycemic medication in individuals with diabetes mellitus and concomitant diabetes mellitus and cardiovascular disease in a managed care environment. Dis Manage Health Outcomes 2004;12:1818.
  • 36
    Shetty S, Secnik K, Oglesby AK. Relationship of glycemic control to total diabetes-related costs for managed care health plan members with type 2 diabetes. J Manag Care Pharm 2005;11:55964.
  • 37
    Peterson AM, Nau DP, Cramer JA, et al. A checklist for medication compliance and persistence studies using retrospective databases. Value Health 2007;10:312.
  • 38
    Hughes D, Cowell W, Koncz T, Cramer J, International Society for Pharmacoeconomics & Outcomes Research Economics of Medication Compliance Working Group. Methods for integrating medication compliance and persistence in pharmacoeconomic evaluations. Value Health 2007;10:498509.
  • 39
    Johnson JA, Pohar SL, Majumdar SR. Health care use and costs in the decade after identification of type 1 and type 2 diabetes: a population-based study. Diabetes Care 2006;29:24038.
  • 40
    Pohar SL, Majumdar SR, Jacobs P, Johnson JA. Healthcare Utilization and Direct Healthcare Costs of Diabetes in Urban and Rural Saskatchewan, 1991–2001. Edmonton: Alliance for Canadian Health Outcomes Research in Diabetes, 2006 (Institute of Health Economics Working Paper WP 06-1).
  • 41
    Liu L, Hader J, Brossart B, et al. Impact of rural hospital closures in Saskatchewan, Canada. Soc Sci Med 2001;52:1793804.
  • 42
    Phillips PJ. Gestational diabetes. Aust Fam Physician 2006;35:7013.
  • 43
    Newton KM, Wagner EH, Ramsey SD, et al. The use of automated data to identify complications and comorbidities of diabetes: a validation study. J Clin Epidemiol 1999;52:199207.
  • 44
    Wilchesky M, Tamblyn RM, Huang A. Validation of diagnostic codes within medical services claims. J Clin Epidemiol 2004;57:13141.
  • 45
    Grymonpre R, Cheang M, Fraser M, et al. Validity of a prescription claims database to estimate medication adherence in older persons. Med Care 2006;44:4717.
  • 46
    Friedman DS, Quigley HA, Gelb L, et al. Using pharmacy claims data to study adherence to glaucoma medications: methodology and findings of the Glaucoma Adherence and Persistency Study (GAPS). Invest Ophthalmol Vis Sci 2007;48:50527.
  • 47
    Ragnarson Tennvall G, Apelqvist J, Eneroth M. The inpatient care of patients with diabetes mellitus and foot ulcers. A validation study of the correspondence between medical records and the Swedish Inpatient Registry with the consequences for cost estimations. J Intern Med 2000;248:397405.
  • 48
    Hess LM, Raebel MA, Conner DA, Malone DC. Measurement of adherence in pharmacy administrative databases: a proposal for standard definitions and preferred measures. Ann Pharmacother 2006;40:128088.
  • 49
    Zeller A, Schroeder K, Peters T. Cigarette smoking and adherence to antihypertensive medication in patients from primary care. Eur J Gen Pract 2007;13:1612.
  • 50
    Vik SA, Hogan DB, Patten SB, et al. Medication nonadherence and subsequent risk of hospitalisation and mortality among older adults. Drugs Aging 2006;23:34556.
  • 51
    Cook RL, Sereika SM, Hunt SC, et al. Problem drinking and medication adherence among persons with HIV infection. J Gen Intern Med 2001;16:836.
  • 52
    Palepu A, Horton NJ, Tibbetts N, et al. Uptake and adherence to highly active antiretroviral therapy among HIV-infected people with alcohol and other substance use problems: the impact of substance abuse treatment. Addiction 2004;99:3618.
  • 53
    Tucker JS, Burnam MA, Sherbourne CD, et al. Substance use and mental health correlates of nonadherence to antiretroviral medication in a sample of patients with Human Immunodeficiency Virus Infection. Am J Med 2003;114:57380.
  • 54
    Jonsson L, Bolinder B, Lundkvist J. Cost of hypoglycemia in patients with type 2 diabetes in Sweden. Value Health 2006;9:1938.
  • 55
    Finkler SA. The distinction between cost and charges. Ann Intern Med 1982;96:1029.
  • 56
    Rutten-van Mölken MP, Van Doorslaer EK, Van Vliet RC. Statistical analysis of cost outcomes in a randomized controlled clinical trial. Health Econ 1994;3:33345.
  • 57
    Dodd S, Bassi A, Bodger K, Williamson P. A comparison of multivariable regression models to analyze cost data. J Eval Clin Pract 2006;12:7686.