HM Medical Clinic

Persistence Patterns with Oral Hypoglycemic
Agents in Type 2 Diabetes
Jennifer M. Stephens, PharmD, Betina T. Blak, MSc, Karen F. Gold, PhD, Marc F. Botteman, MSc, MA,
Chris L. Pashos, PhD, Caroline S. Walkinshaw, MBA, and R. Keith Campbell, RPh, MBA

Objectives: To determine persistence with therapy in affects more than 20% of the population older than age patients newly starting oral hypoglycemic treatment 65 rden of diabetes in the for management of type 2 diabetes, and to determine United States is more than $100 differences in persistence according to sex, age, and term glycemic control, achieved via a combination of life- style changes and the use of oral antidiabetic agents and • Design: Retrospective medication claims analysis.
insulin, has been found to significantly reduce morbidity • Setting and participants: 46,884 patients from a large re, effective management of U.S. pharmacy benefits manager (PBM) database who diabetes to achieve glycemic control improves patient quali- were older than 18 years of age and who newly start- ty of life and is cost-ef ed monotherapy with an oral agent for type 2 diabetes.
Although the clinical and economic value of long-term • Methods: All patients were followed for 10 months for glycemic control is clear and should be a priority, adherence refill persistence patterns with their initial drug. Agents to physician-prescribed oral therapy remains a challenge for included in the analysis were glimepiride, immediate- patients with type 2 diabetes. Depending on the patient pop- release glipizide, glipizide GITS (gastrointestinal ther- ulation, definitions, and methodology used, reported rates apeutic system), glyburide, metformin, pioglitazone, of adherence, compliance, or persistence with oral hypo- repaglinide, and rosiglitazone.
glycemic treatment range fr • Results: The overall average length of therapy was Limited information exists on how treatment persistence 167 days out of 300 days of follow-up. Women were differs across sex and age groups. For treatment with sulfonyl- significantly less persistent than men (161 days ver- ureas, Sclar and colleagues found that men and younger sus 173 days, respectively). Persistence significantly patients were less persistent with their treatment than were increased with patient age, from a low of 126 days (18–34 years) to 179 days (50–64 years), but then ings hold true for different patient populations. Furthermore, fell to 161 days (≥ 65 years). Significant differences although several different effective oral hypoglycemic treat- were also identified for the oral agents, with per- ments are currently used in the United States, very little is sistence ranging from 126 days with repaglinide to known regarding persistence rates with these agents. Per- 177 days with glipizide GITS. sistence may vary between agents that have different efficacy, • Conclusions: Overall persistence with the initial oral side effect, and dosing profiles.
hypoglycemic drug treatment appears to be low in Clinicians seeking to tailor the antidiabetic drug regimen type 2 diabetes and differs within sex, age, and indi- to the needs of their patients might benefit from increased vidual drug categories. Careful selection of the initial knowledge on how persistence with these agents varies by oral agent to minimize adverse events and reduce patient demographics and class of agents. The objectives of dosing frequency may improve persistence, enhance this study were to determine the persistence with oral hypo- glycemic control, and reduce the economic burden of glycemic agents in patients newly starting pharmacologic type 2 diabetes.
Diabetes is a costly and increasingly common chronic disease affecting more than 17 million persons in the From Abt Associates Clinical Trials Health Economics Research and Qualityof Life Evaluation Services (HERQuLES), Bethesda, MD (Drs. Stephens, United States; an estimated 5.9 million of these cases Gold, and Pashos, Ms. Blak, and Mr. Botteman); Pfizer Inc., New York, NY arType 2 diabetes accounts for approx- (Ms. Walkinshaw); and Washington State University, College of Pharmacy, imately 90% to 95% of all diagnosed diabetes cases and Pullman, WA (Mr. Campbell). Vol. 9, No. 9 September 2002 JCOM 491
treatment for type 2 diabetes, and whether differences in per- ed their treatment for more than 30 days, we determined sistence exist across patient sex and age groups and individ- whether they resumed therapy with another oral agent within ual drug therapies. 60 days after discontinuation of their initial drug. This behaviorwas defined as a switch in drug treatment, and we evaluat- ed the switching patterns within drug category. Evaluation/ tracking of add-on therapies (ie, combination and triple thera- A retrospective database analysis was conducted using pies) was beyond the scope of our primary objective (ie, persis- patient-specific drug refill information from a large national tence with the initial oral hypoglycemic agent). Thus, patients pharmacy benefits manager (PBM) in the United States, defined as persistent in our study may have been receiving which processes over 300 million prescription claims annual- more that one oral hypoglycemic agent (over time, a second ly. This prescription claims database contains the most recent agent may have been added to the initial drug therapy).
25 months of data covering the prescription claims of 50 mil-lion Americans in health plans throughout the country.
ANOVA testing was performed to compare differences in
average length of therapy between all of the agents. We also To be included in the study, a patient (1) had to be 18 years of conducted ANOVA for overall differences across age cate- age or older; (2) had to have been continuously enrolled in the gories and sex categories within treatment groups. When the prescription claims database, implying that the patient had omnibus F-test for the ANOVA was significant, an appropri- continuous drug insurance coverage; (3) had to have filled a ate post hoc analysis was conducted within each of the sex prescription for an oral hypoglycemic agent (defined as any and age categories. Each of the drugs was contrasted with agent that reduces blood glucose levels) during a 4-month the average treatment duration of all drugs en mass within index period spanning from 1 September 1999 to 31 December the sex or age category using a t test (one-tailed test for above 1999; and (4) must not have filled a prescription for an oral and below the mean) with Bonferroni correction for multiple hypoglycemic agent during a 6-month preperiod from comparisons. This is a slightly more conservative post hoc 1 March 1999 to 31 August 1999. Using these criteria, all pa- contrast (meaning that differences had to meet a higher tients included in the analysis were likely to be drug naive and threshold before they were considered to be statistically sig- newly diagnosed with type 2 diabetes. All patients were fol- nificant). These 8 contrasts were viewed as more informative lowed for 10 months after the initial prescription identified in than all 28 pairwise sets of contrasts within each of the sex the index period. Hence, the postperiod lasted through and age categories. The goal was to identify which treatment 31 October 2000. All patients included in the analysis had con- categories outperformed the whole group within each of the tinuous drug insurance coverage starting from the 6-month sex and age categories. Differences were considered statisti- preperiod through the entire analysis timeframe.
cally significant at the P = 0.05 level. Testing was performed Patients were categorized and analyzed according to using SAS (version 6.11, SAS Institute, Cary, NC).
sex and age (18–34 years, 35–49 years, 50–64 years, and ≥ 65 years) and the oral hypoglycemic prescriptions they filled during the index period. These hypoglycemic drugs A total of 46,884 patients met the criteria for inclusion in the were immediate-release glipizide (brand and generic), glip- analysis and were available for analysis. Their demographic izide GITS (gastrointestinal therapeutic system; Glucotrol characteristics and distribution across drug categories are XL), glimepiride (Amaryl), glyburide (brand and generic), re evenly represented in metformin (Glucophage), pioglitazone (Actos), repaglinide the sample. Seventy-eight percent of the patients were older (Prandin), and rosiglitazone (Avandia). We reviewed pro- than 50 years of age and 41% were older than 65 years. The prietary U.S. market research information (IMS data) to study demographics were generally consistent with other identify oral hypoglycemic agents with a market share major studies of demographics in type 2 diabetes (eg, the greater than 1%. Because miglitol and acarbose (alpha- National Health Interview Survey found that 45% of pa- glucosidase inhibitors) had less than 1% market share dur- tients with newly diagnosed type 2 diabetes are men and ing this period, they were not considered in the analysis.
Persistence was defined in terms of the number of days of More than one third of patients initially filled prescriptions uninterrupted treatment during the study's follow-up period for metformin, 21% for glyburide, 15% for glipizide GITS, 10% (ie, treatment with no more than a 30-day break in therapy).
for glimepiride, and between 3% to 10% for each of the other Patients received prescriptions for a standard 30-day supply of agents. While the distribution of patients across the drug medication per refill, and persistence was calculated based on groups was almost evenly split between the sexes, women re- the number of days supplied. Among patients who interrupt- ceived metformin and rosiglitazone more frequently than 492 JCOM September 2002 Vol. 9, No. 9
Patient Demographics and Distribution of Initial Oral Hypoglycemic Drugs Sex, n (%)
Age, n (%)
18–34 yr
35–49 yr
50–64 yr
Glimepiride (Amaryl) Glipizide IR (brand/ Glipizide GITS (Glucotrol XL) 3525 (50.8) Glyburide (brand/ Metformin (Glucophage) 5301 (32.9) 16,115 (34.4) Pioglitazone (Actos) Rosiglitazone (Avandia) Repaglinide (Prandin) 19,349 (41.3) 46,884 (100.0) Note: Sex is not independent of treatment (χ2 = 98.47, df = 7, P < 0.0001). Age is not independent of treatment (χ2 = 1269.14, df = 21, P < 0.0001).
GITS = gastrointestinal therapeutic system; IR = immediate-release. men (54% versus 46% for metformin, and 53% versus 48% for 40% switched to a biguanide (metformin) and 23% switched rosiglitazone). Some differences in initial prescription filling to a thiazolidinedione (pioglitazone and rosiglitazone). Of were also observed between the 4 age groups. For instance, those patients that switched away from immediate-release the proportion of patients initially filling a prescription for sulfonylureas (glimepiride, glipizide, and glyburide), 19.4% metformin decreased steadily with increasing age (from 53% switched to the extended-release sulfonylurea (glipizide among the 18–34 year group to 27% among the ≥ 65 year GITS). When patients were switched from biguanide (met- group). In contrast, the proportion of patients initially filling a formin), they more typically were switched to one of the sul- prescription for glyburide or immediate-release glipizide in- fonylureas (55.5%) or one of the thiazolidinediones (41.7 %).
creased among older patients. Finally, of patients switched from the thiazolidinediones,33.8%, 31% and 5.1% were switched to a sulfonylurea, Overall Persistence and Persistence Across Demographic
biguanide, and meglitinide, respectively. Turnover from one thiazolidinedione to another was approximately 30%.
On average, the overall persistence across all oral hypo-glycemic agents was 166.7 days. Persistence was significant- Persistence Across Drug Groups
ly longer (11.6 days) among men than among women The average length of therapy across the 8 oral hypogly- (172.6 days versus 161.0 days, P < 0.001). There were consis- cemic treatment groups is shown in tent persistence differences among men and women for the repaglinide and immediate-release glipizide had the lowest persistence. Patients who initially filled immediate-release creased with age (P < 0.05), reaching a peak among patients glipizide or repaglinide were significantly less persistent aged 50 to 64 years (125.1 days among those aged 18–34 years, compared to patients who initially filled any other therapy 162.4 days among those aged 35–49 years, and 178.7 days (P < 0.05) (Figure 3). among those aged 50–64 years), but then decreased to Patients treated with glipizide GITS persisted longer with 161.4 days in the ≥ 65-year group. Persistence by age and drug their initial treatment than did patients treated with the other therapies. Statistically significant differences in length of ther- During the follow-up period, 13% of all patients switched apy were seen for glipizide GITS versus metformin, gly- from their initially assigned oral hypoglycemic drug treatment buride, immediate-release glipizide, and repaglinide (P < 0.05) to one of the other 7 drug therapies. The repaglinide group and the immediate-release glipizide group had the highest zone, and glimepiride was comparable to glipizide GITS (no turnover rates at 23% and 20%, respectively. Within the statistically significant differences). The drugs with the largest remaining drug categories, 12% to 15% of the patients difference in length of therapy were glipizide GITS and switched away from their initial assigned treatment. Of pa- repaglinide, with an average difference of 51.2 days (P < 0.05). tients that switched from any sulfonylurea (glimepiride, Persistence differences were also noted across sex and immediate-release glipizide, glipizide GITS, and glyburide), age categories within the various drug treatments. Women Vol. 9, No. 9 September 2002 JCOM 493
Persistence by sex. GITS = gas- trointestinal therapeutic system; IR =immediate-release.
Length of treatment, days Length of treatment, days Persistence by age. GITS = gastrointestinal therapeutic system; IR = immediate-release. were less persistent than men for all drug treatments by a 50-to-64-year age group, immediate-release glipizide range of 8 to 17 days, and these differences were statistically (150 days) and repaglinide (123 days) were significantly dif- significant within all treatment categories except for the gly- ferent from the age group average of 179 days (P < 0.05).
buride and immediate-release glipizide groups (P < 0.05).
Within the ≥ 65 years and older group, glipizide GITS Across age groups, differences in persistence were also ob- (175 days) and repaglinide (137 days) were significantly dif- served (Figure 2). Of note is the variation in persistence ferent from the age group average of 161 days (P < 0.05). No across drug groups observed among the youngest age group significant differences within the 18-to-34-year age group (89 days) and the oldest age group (38 days). were identified, probably due to small sample sizes.
Within the 35-to-49-year age group, immediate-release glipizide (134 days), glipizide GITS (173 days), and repaglin- ide (111 days) were significantly different from the age group Lifelong treatment is usually necessary for persons diag- average persistence of 162.4 days (P < 0.05). Within the nosed with type 2 diabetes. Lack of persistence with oral 494 JCOM September 2002 Vol. 9, No. 9
Average Length of Therapy for Age Groups by Oral Hypoglycemic Drug Mean ± SD Length of Therapy, days
Age Group
Within Drug
18–34 yr*
35–49 yr
50–64 yr
P Value (F, df , df )
Glimepiride (Amaryl) 172.3 ± 137.2 < 0.05 (6.61, 3, 4665) Glipizide IR (brand/ 133.7† ± 127.5 149.9† ± 132.0 < 0.05 (5.01, 3, 3137) 172.9† ± 138.7 174.8† ± 140.3 < 0.05 (7.03, 3, 6932) Glyburide (brand/ < 0.05 (13.3, 3, 9711) < 0.05 (70.6, 3, 16111) Pioglitazone (Actos) < 0.05 (5.79, 3, 1181) < 0.05 (10.5, 3, 3828) 111.3† ± 114.0 122.5† ± 120.4 136.6† ± 130.8 < 0.05 (3.75, 3, 1287) (19.5, 7, 17106) (9.23, 7, 19341) Average length of < 0.05 (3.04, 3, 46880) therapy across drugs *Limited significance of pairwise differences among drugs in the 18–34 year age group likely due to small cell sizes.
†Test wise statistically significant from average across drugs at alpha = 0.005 so that family wise error of alpha < 0.05 (t critical value = 2.576). GITS = gatrointestinal therapeutic system; IR = immediate-release.
hypoglycemic drug treatment likely signals failure in a pa- tence, we selected this definition because it is commonly tient's management because of ineffectiveness, intolerance, and/or other factors. This study is the most current national The clinical implication of poor persistence is a lack of evaluation of oral drug persistence in type 2 diabetes across a glycemic control, poorer clinical outcomes, and ultimately wide range of drug classes (including the sulfonylureas, higher costs. Previous studies have shown that nonadherent biguanides, thiazolidinediones, and meglitinides). The results patients have higher mean blood glucose levels than adher- illustrate important issues and concerns with oral hypo- Within other diseases, poor refill compli- glycemic drug persistence in patients newly starting oral med- ance has been correlated with low drug serum levels and ication for management of type 2 diabetes. Overall, patients did reduced clinical ef not seem to obtain adequate refills of their hypoglycemic med- The current study results are both comparable to and ication, with medication available for only 167 out of 300 days.
expand on other analyses examining specific hypoglycemic Furthermore, the largest differences in persistence were seen agents. In a study using prescription claims information for across drug therapies rather than within demographic groups. Medicaid patients newly diagnosed with type 2 diabetes, The relatively poor performance across all drug classes is mean days of immediate-release glipizide and glyburide a clear signal to the clinical community that patients newly supply obtained reached 157 days out of a maximum of diagnosed with type 2 diabetes do not remain on their treat- ment or do not refill their prescription so that they have patients younger than 65 years treated with glyburide, ther- medication available on a regular basis. In fact, the actual apy obtained covered 210 days out of 360 number of days with available medication in this study may Most recently, another PBM claims analysis found 1-year have been even lower, as the study definition of persistence persistence rates of 60% for metformin, 56% for sulfonyl- allowed lapses as long as 30 days in obtaining refills before ureas, and 48% for r the patient was considered nonpersistent. While this 30-day persistence rates of 42% to 59%; however, it provides addi- window to obtain refills may underestimate actual persis- tional data on currently used thiazolidinediones as well as Vol. 9, No. 9 September 2002 JCOM 495
Average length of treatment, days Persistence by drug. Note: Pioglitazone sample size did not allow the difference between pioglitazone and glyburide toreach statistical significance. GITS = gastrointestinal therapeutic system; IR = immediate-release. *Average length of therapy signif-icantly different (P < 0.05) from repaglinide. †Average length of therapy significantly different (P < 0.05) from glipizide IR andrepaglinide. ‡Average length of therapy significantly different (P < 0.05) from glyburide, glipizide IR, and repaglinide. §Averagelength of therapy significantly different (P < 0.05) from metformin, glyburide, glipizide IR, and repaglinide.
examines the individual sulfonylureas. Interestingly, our Age and Persistence
study found significant differences in persistence within the This study demonstrates that persistence is the lowest in sulfonylurea class. Nevertheless, these previous studies, along young patients and increases with age to a peak in patients with the current results, confirm that the overall hypogly- 50 to 64 years of age, before declining among patients aged cemic medication persistence is low among patients newly 65 years and older. Some studies of type 2 diabetes find that as diagnosed and treated for type 2 diabetes.
age increases, so does adherence to the oral antidiabetic drugtreatment ry, older age was not a Sex and Persistence
significant predictor of oral hypoglycemic drug treatment It has been suggested that women have better adherence to adherence in a study where adherence was self-r chronic drug treatment as they are more sensitive to illness, With respect to age and persistence in other chronic, initially more willing to seek health care, use more health care, and asymptomatic diseases, increased age has been found to be in general have a greater responsibility for family health associated with increased drug treatment adherence in newly y, women in our study were actually found to be somewhat less persistent (by approximately 12 days) Again, however, a number of other studies within these dis- than men almost consistently across all the oral hypo- eases report no significant association between age and adher- glycemic treatments evaluated. This finding, though unusu- al, is supported by findings of several prescription refill ed polypharmacy in elderly patients may help explain the adherence studies in differTo the con- significant fall in persistence that we observed in the ≥ 65-year trary, Sclar and colleagues reported newly diagnosed age group; however, the specific data set did not allow us to women receiving Medicaid to be more than twice as likely confirm this hypothesis.
Other diabetes studies, using different adherence measure- Drug Treatment and Persistence
ment methods, have found no significant difference by sex The largest differences in persistence were seen across drug results are found in oral treatment adherence therapies rather than within demographic groups. This finding is important given that clinicians cannot control the age or sex Clearly, the literature does not show a consistent association of their patients but do have many choices with regard to the between sex and persistence.
initial hypoglycemic therapy. Average persistence differences 496 JCOM September 2002 Vol. 9, No. 9
Differences in Average Days of Therapy of Index Drug versus Comparative Drug Glipizide IR
Glipizide GITS Pioglitazone
Index Drug
Pioglitazone (Actos) Rosiglitazone (Avandia) Glimepiride (Amaryl) Metformin (Glucophage) Glipizide IR (brand/generic) Repaglinide (Prandin) *Differences in average duration of therapy (index drug minus comparative drug) significant at the 0.05 level.
GITS = gastrointestinal therapeutic system; IR = immediate-release.
across all age groups were as high as 51 days. Within age repaglinide has a favorable side effect profile, it was the only groups, the differences were even larger. An example of this is agent that has to be taken with meals (typically resulting in seen in the 18-to-34-year age group, where differences in per- 3 administrations per day), which may have contributed to sistence were as high as 89 days (179 days for pioglitazone ver- it having the lowest persistence among the therapies evalu- sus 90 days for repaglinide). ated. These findings may highlight the importance of Although these data do not permit an analysis of reasons extended-release or once-daily formulations, which have for discontinuation, one could hypothesize that these persis- been suggested to improve medication-taking behavior and tence differences may be partially explained by the side also to partially compensate for delays and gaps in proper effect profiles of the drugs and to some extent the complexi- requency is an important ty of the treatment regimen. The agents with the highest per- factor to achieve increased oral hypoglycemic drug adher- sistence rates were those generally given once daily and that ence. In studies where adherence was measured by pill have relatively low rates of bothersome adverse events (eg, counts or refill behavior, there were associations or trends minimal/low gastrointestinal disturbances and hypoglycem- supporting the fact that reduction of the dosing frequency ia). Specifically, patients persisted longer on glipizide GITS, impry, a significant positive pioglitazone, rosiglitazone, and glimepiride as compared to association between a lower frequency dosing regimen and glyburide, immediate-release glipizide, and repaglinide. drug treatment adherence has been identified in other chron- Hypoglycemia may be an important factor associated with the poor persistence with glyburide and immediate-release glipizide therapies, often the most common side effect Limitations
experienced by patients on these agents and the most com- This study presents preliminary results that suggest trends in mon reason for discontinuation of these dr persistence differences across treatment categories. We trials have found that mild to moderate hypoglycemia oc- attempted to define the inclusion/exclusion criteria such that curred in 20% of glyburide and 19% of immediate-release the analysis population would represent a general type 2 dia- betes population, most likely those initiating oral therapy after hypoglycemia may be twice as high for patients on glyburide lifestyle modifications have failed. However, pharmacy as for patients being treated with immediate-release glipizide claims database research is limited for a number of reasons reported differences, this study found and must be interpreted with caution. Nevertheless, pharma- that patients treated with glyburide stayed approximately cy claims analyses often give valuable insights to the real- 14 days longer on treatment compared with patients treated world use of drug therapies and provide hypotheses for with immediate-release glipizide. future research. In the present study, we were able to remove The results regarding persistence of therapy cannot be obvious confounders, such as variations in the number explained by side effects alone, however. Even though of days medication was supplied and changes in insurance Vol. 9, No. 9 September 2002 JCOM 497
coverage. Changes in insurance coverage is an important nonadherent to their prescribed treatment (patient is in a issue for persistence studies, and we included only those newly diagnosed, contemplation stage and not actively tak- patients with continuous drug insurance coverage from the ing responsibility for their disease). These findings may not entire pre-period through the entire timeframe of the analysis; be surprising given that persistence is poor within other the number of days supplied was also uniform in the cohort.
chronic diseases such as hypertension, particularly for newly Because underlying reasons for differences in persistence, ro and colleagues reported such as severity of type 2 diabetes and concomitant diseases, that patients newly diagnosed with hypertension persisted could not be examined, selection bias is possibly present.
significantly less with their oral drug treatment at the end of Specifically, systematic differences in patient characteristics 1 year compared to patients with ongoing hypertension could conceivably exist across the drug treatment groups, (persistence rates of 78% versus 97%, r which might affect the persistence ratio and interfere with the Finally, examining prescription refill behavior provides interpretation of the comparison across treatment categories.
information on whether patients purchase their medications In addition, calculation of persistence based on number of but not whether the patients ar days supplied may be a limitation, as it does not reflect theless, we have no basis to believe that use varies different- changes in the use of medication on verbal instruction from ly from purchase (and therefore intended use). Adherence is the physician (eg, patients may be incorrectly classified in the secondary to availability, since it is impossible for patients to analysis as not persistent if they reduce the dose of their med- be persistent or adherent with a treatment regimen without ication to a half tablet once daily instead of a whole tablet).
having the drug available. A key limitation common to most pharmacy claims database analyses is the lack of additional clinical informa- Conclusion
tion regarding both patient and physician reasons for Overall persistence with the initial oral hypoglycemic drug changes in therapy. It would also be of interest to evaluate treatment is suboptimal in patients with type 2 diabetes. Dif- the association between oral hypoglycemic drug persis- ferences in hypoglycemic treatment persistence exist across tence with actual hemoglobin A levels to assess whether sex, age, and individual drug categories. These differences high drug persistence is correlated with controlled blood appear to be associated with initial drug choice rather than glucose. In addition, future research could also include with patient age or sex. Careful selection of the initial hypo- development of the clinical pathways for patients starting glycemic agent to minimize adverse events and reduce dosing oral agents to determine what agents are added as second frequency may improve persistence, enhance glycemic con- and third line therapies.
trol, and reduce the economic burden of type 2 diabetes.
Another consideration with any pharmacy claims analy- The results of this study may be useful for educating clini- sis is defining an adequate preperiod to ensure that new cians about the lack of persistence with the initial oral medica- medications truly represent a new treatment for the patient.
tion in type 2 diabetes and serve as an impetus to enhance pa- It would be unlikely that a patient switching insurance plans tient self-management education programs. Self-management would carry over a 6-month supply of diabetes drug; thus, education programs should consider incorporating addition- we believe that the 6-month preperiod was sufficient to al follow-up regarding the importance of adhering to a pre- identify patients that may have already been receiving oral scribed treatment. Future research should examine the impact diabetes drugs under other insurance carriers.
of oral hypoglycemic drug persistence on achieving good In addition, the economic, psychosocial, and behavioral glycemic control across age, sex, and drug classes. factors that play a role in drug adherence were not ad-dressed due to absence of such data. However, it is interest- Corresponding author: Jennifer Stephens, PharmD, Abt Associates ing to note that patients in the 65 year and older age group Clinical Trials, 4800 Montgomery Lane, Suite 600, Bethesda, MD were receiving more of the generic sulfonylureas than the other age groups. This may indicate that there was someselection bias related to drug affordability issues (lower Funding: This research was sponsored in part by Pfizer Inc. copayments for generic agents). In addition, this analysis Author contributions: conception and design, CSW; analysis and inter- could not assess patients with public insurance coverage pretation of data, JMS, BTB, KFG, MFB, CLP, RKC; drafting of the arti- only (eg, Medicaid or Medicare only).
cle, JMS, BTB, MFB; critical revision of the article for important intel- From the psychosocial and behavioral aspect, the overall lectual content, JMS, BTB, KFG, MFB, CLP, CSW, RKC; final approval low treatment persistence found in this study may partly be of the article, JMS, CLP, CSW; provision of study materials or patients, explained by the fact that the cohort likely consists of newly CSW; statistical expertise, KFG; administrative, technical, or logistic diagnosed patients who may be more likely to switch thera- support, JMS, CSW; collection and assembly of data, JMS, BTB, KFG. pies (until a tolerable agent is found) and/or just simply be Financial disclosures: None. 498 JCOM September 2002 Vol. 9, No. 9
tions. J Clin Epidemiol 1997;50:105–16.
American Diabetes Association. Basic diabetes information: Steiner JF, Koepsell TD, Fihn SD, Inui TS. A general method of Newly diagnosed. Accessed compliance assessment using centralized pharmacy records.
23 Aug 2002.
Description and validation. Med Care 1988;26:814–23. Ratner RE. Type 2 diabetes mellitus: the grand overview.
Fitzgerald JT, Anderson RM, Davis WK. Gender differences Diabet Med 1998;15 Suppl 4:S4–7.
in diabetes attitudes and adherence. Diabetes Educ 1995;21: Gaster B, Hirsch IB. The effects of improved glycemic control on complications in type 2 diabetes. Arch Intern Med 1998; Sung JC, Nichol MB, Venturini F, et al. Factors affecting pa- tient compliance with antihyperlipidemic medications in an The effect of intensive treatment of diabetes on the develop- HMO population. Am J Manag Care 1998;4:1421–30. ment and progression of long-term complications in insulin- Shaw E, Anderson JG, Maloney M, et al. Factors associated dependent diabetes mellitus. The Diabetes Control and Com- with noncompliance of patients taking antihypertensive plications Trial Research Group. N Engl J Med 1993;329: medications. Hosp Pharm 1995;30:201–3,206–7.
Weingarten MA, Cannon BS. Age as a major factor affecting Testa MA, Simonson DC. Health economic benefits and adherence to medication for hypertension in a general prac- quality of life during improved glycemic control in patients tice population. Fam Pract 1988;5:294–6. with type 2 diabetes mellitus: a randomized, controlled, Bailey JE, Lee MD, Somes GW, Graham RL. Risk factors for double-blind trial. JAMA 1998;280:1490–6. antihypertensive medication refill failure by patients under Wake N, Hisashige A, Katayama T, et al. Cost-effectiveness Medicaid managed care. Clin Ther 1996;18:1252–62. of intensive insulin therapy for type 2 diabetes: a 10-year Caro JJ, Salas M, Speckman JL, et al. Persistence with treat- follow-up of the Kumamoto study. Diabetes Res Clin Pract ment for hypertension in actual practice. CMAJ 1999;160: 2000;48:201–10. Sclar DA, Robison LM, Skaer TL, et al. Sulfonylurea phar- Avorn J, Monette J, Lacour A, et al. Persistence of use of macotherapy regimen adherence in a Medicaid population: lipid-lowering medications: a cross-national study. JAMA influence of age, gender, and race. Diabetes Educ 1999;25: 2001;279:1458–62. Vaur L, Vaisse B, Genes N, et al. Use of electronic pill boxes to Skaer TL, Sclar DA, Markowski DJ, Won JK. Effect of value- assess risk of poor treatment compliance: results of a large- added utilities on prescription refill compliance and Medi- scale trial. Am J Hypertens 1999;12:374–80. caid health care expenditures—a study of patients with Garay-Sevilla ME, Malacara HJM, Gonzalez-Parada F, non-insulin-dependent diabetes mellitus. J Clin Pharm Ther Jordan-Gines L. The belief in conventional medicine and 1993;18:295–9. adherence to treatment in non-insulin-dependent diabetes Mason BJ, Matsuyama JR, Jue SG. Assessment of sulfony- mellitus patients. J Diabetes Complications 1998;12:239–45. lurea adherence and metabolic control. Diabetes Educ 1995; Caro JJ, Speckman JL, Salas M, et al. Effect of initial drug choice on persistence with antihypertensive therapy: the Catalan VS, Couture JA, LeLorier J. Predictors of persistence importance of actual practice data. CMAJ 1999;160:41–6.
of use of the novel antidiabetic agent acarbose. Arch Intern Farmer KC, Jacobs EW, Phillips CR. Long-term patient com- pliance with prescribed regimens of calcium channel block- Diehl AK, Bauer RL, Sugarek NJ. Correlates of medication ers. Clin Ther 1994;16:316–26. compliance in non-insulin-dependent diabetes mellitus.
Traub YM, McDonald RH Jr, Shapiro AP. Patient characteris- South Med J 1987;80:332–5. tics and their influence on therapy and its outcome in hyper- Brown JB, Nichols GA, Glauber HS, Bakst A. Ten-year follow- tension of the elderly. Cardiology 1988;75:45–55. up of antidiabetic drug use, nonadherence, and mortality in a Cada DJ, Covington TR, Generali JA, et al. Antidiabetic defined population with type 2 diabetes mellitus. Clin Ther agents. In: Drug facts and comparisons. St. Louis (MO): Facts 1999;21:1045–57. and Comparisons; 2001:287–312. Paes AH, Bakker A, Soe-Agnie CJ. Impact on dosage fre- DeFronzo RA. Pharmacologic therapy for type 2 diabetes quency on patient compliance. Diabetes Care 1997;20:1512–7. mellitus. Ann Intern Med 1999;131:281–303. Boccuzzi SJ, Wogen J, Fox J, et al. Utilization of oral hypo- van Staa T, Abenhaim L, Monette J. Rates of hypoglycemia glycemic agents in a drug-insured U.S. population. Diabetes in users of sulfonylureas. J Clin Epidemiol 1997;50:735–41. Shorr RI, Ray WA, Daugherty JR, Griffin MR. Individual sul- Kenny SJ, Aubert RE, Geiss LS. Prevalence and incidence of fonylureas and serious hypoglycemia in older people. J Am non-insulin-dependent diabetes. In: Diabetes in America.
Geriatr Soc 1996;44:751–5. 2nd ed. Bethesda (MD): National Institutes of Health; 1995: Rudd P. Partial compliance: implications for clinical practice.
47–67; NIH Publication No. 95–1468.
J Cardiovasc Pharmacol 1993;22 (Suppl A):S1–5. Steiner JF, Prochazka AV. The assessment of refill compliance Dezii CM. Persistence with drug therapy: a practical approach using pharmacy records: methods, validity, and applica- using administrative claims data. Manag Care 2001;10:42–5.
Copyright 2002 by Turner White Communications Inc., Wayne, PA. All rights reserved.
Vol. 9, No. 9 September 2002 JCOM 499


Access Details: Free Access Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Publication details, including instructions for authors and subscription information: Phototherapy with Narrowband vs Broadband UVB Mark Berneburg a; Martin Röcken a; Frauke Benedix a