- Email: [email protected]
Persistence and Adherence With Topical Glaucoma Therapy
Persistence and Adherence With Topical Glaucoma Therapy BETH L. NORDSTROM, PHD, DAVID S. FRIEDMAN, MD, ESSY MOZAFFARI, PHARMD, HARRY A. QUIGLEY, MD, AND ALEXANDER M. WALKER, MD, DRPH
● PURPOSE:
The present study describes the patterns and predictors of treatment persistence and adherence among patients who are diagnosed with glaucoma or as glaucoma suspects (based on claims codes). ● DESIGN: A retrospective cohort study using health insurance claims data. ● METHODS: Newly treated individuals with diagnosed glaucoma (n ⴝ 3623) and suspect glaucoma (n ⴝ 1677) were obtained from healthcare claims data in the Ingenix Research Database. For each of these two diagnostic groups, we calculated the duration of continuous treatment with the initially prescribed medication (persistence) and the prevalence of use of the initial medication at various time points (adherence). Four drug classes were included: -blockers, ␣-agonists, carbonic anhydrase inhibitors, and prostaglandin analogs. ● RESULTS: Nearly one half of the individuals who had filled a glaucoma prescription discontinued all topical ocular hypotensive therapy within six months, and just 37% of these individuals recently had refilled their initial medication at three years after the first dispensing. Prostaglandins were associated with better persistence than any other drug class, which was indicated by hazard ratios for discontinuation of prostaglandins compared with -blockers of 0.40 (95% confidence interval, 0.35– 0.44) for diagnosed patients and 0.44 (95% confidence interval, 0.37– 0.52) for patients with suspect glaucoma. Prostaglandins showed a similar advantage in adherence. Furthermore, patients with diagnosed glaucoma were more likely to adhere to therapy than patients with Accepted for publication Apr 22, 2005. From Ingenix Epidemiology, Auburndale, Massachusetts (B.L.N.; A.M.W.); Wilmer Institute, Johns Hopkins University, Baltimore, Maryland (D.S.F.; H.A.Q.); Pfizer, Inc, New York, NY (E.M.) Supported by Pfizer through a research contract with Ingenix (B.L.N.; A.M.W.) and through an unrestricted grant to Johns Hopkins University (D.S.F.;H.A.Q.). Pfizer provided support for this study through a research contract with Ingenix (B.N. and A.W.). Pfizer also provided an unrestricted grant to Johns Hopkins University (D.F. and H.Q.). One of the authors (E.M.) is employed by Pfizer; he was involved in the study design, interpretation, and writing only as a member of a steering committee; all final decisions were made by the authors not employed by Pfizer. Inquiries to Beth L. Nordstrom, PHD, Ingenix Epidemiology, Riverside Center, Suite 3–120 275 Grove St, Auburndale, MA 02466; fax: 617–244 –9669; e-mail: [email protected]
598.e1
©
2005 BY
suspect glaucoma (relative risk ⴝ 1.11; 95% confidence interval, 1.05–1.18). ● CONCLUSION: Persistence and adherence were substantially better with prostaglandins than with other drug classes, and patients with diagnosed open-angle glaucoma were more likely to adhere to treatment than suspected glaucoma. (Am J Ophthalmol 2005;140: 598 – 606. © 2005 by Elsevier Inc. All rights reserved.)
A
LTHOUGH THE PREFERRED PRACTICE PATTERN OF
the American Academy of Ophthalmology lists medical therapy, laser trabeculoplasty, and surgical treatment as reasonable options for the initial treatment of glaucoma, most patients initially receive topical ocular hypotensives.1 If topical treatment lowers intraocular pressure (IOP) adequately, the patient is intended to remain on therapy indefinitely to improve outcome.2 Although recent studies clearly have documented that the lowering of IOP decreases the risk of visual field loss3,4 and slows progression from ocular hypertension to glaucoma,5 many patients appear to discontinue their use of topical hypotensive agents.6 –13 Most studies of persistence with topical glaucoma medications that have been conducted to date have not restricted the study population to patients who were diagnosed with glaucoma or suspect glaucoma.6 –11 Of the two studies that have imposed diagnostic criteria, one study included only suspects,12 and the other study was fairly small (260 patients in total) and used prescribing records without ascertaining whether the prescriptions were filled.13 It is important to investigate persistence of treatment within groups of patients with similar diagnostic status; ocular hypotensives that are prescribed for the treatment of primary open angle glaucoma may be used differently from those given for ocular hypertension without optic nerve damage and certainly should be used differently for transient elevations in IOP. In addition, none of the studies that were found in the literature examined both the continuous use of glaucoma medications (persistence) and ongoing use, which allowed for gaps in therapy (adherence). We undertook the present study to investigate the patterns and predictors of treat-
ELSEVIER INC. ALL
RIGHTS RESERVED.
0002-9394/05/$30.00 doi:10.1016/j.ajo.2005.04.051
ment persistence in a large population of patients who were diagnosed in healthcare claims as suspect glaucoma and separately in patients who were diagnosed with open angle glaucoma.
METHODS ● DATA SOURCE:
Data for the study were obtained from the Ingenix Research Database, which consists of insurance claims that include demographic, medical, and prescription information from approximately 8,000,000 members of UnitedHealthcare. UnitedHealthcare is one of the largest healthcare companies in the United States, with ⬎300,000 physicians contracted to provide healthcare services to ⬎14 million members. The analyses for the present study used encrypted automated data that were de-identified with respect to protected health information. Under the current privacy rules of the Health Insurance Portability and Accountability Act, Institutional Review Board approval and informed consent of subjects were not required. ● COHORT DEFINITION:
The study cohort consisted of patients who were identified from the Ingenix Research Database between July 1, 1995, and December 31, 2001, as newly diagnosed with open angle (diagnosed glaucoma) or suspect glaucoma who received at least one dispensing of a topical ocular hypotensive medication. Patients were required to be at least 30 years of age and to have at least 365 days of continuous enrollment in UnitedHealthcare as of the initial diagnosis date. Patients were excluded if they had glaucoma-related claims within 365 days before the initial diagnosis date (ICD-9 code [365.xx or 377.14] in association with a physician visit or hospitalization), any dispensing of a topical ocular hypotensive, or any procedure code that indicated argon laser trabeculoplasty or glaucoma surgery in conjunction with a glaucoma diagnosis on the claim. We also excluded any patient with claims for visual field testing, fundus photography, or nerve fiber layer imaging during the 10 months of the baseline period that ranged from 2 to 12 months before the initial diagnosis date. Finally, to be included in the study cohort, newly diagnosed patients must have received at least one dispensing of a topical glaucoma medication on or after the date of initial diagnosis.
● BASELINE CHARACTERISTICS:
Patient characteristics that were evaluated at baseline included age, gender, and class of initial treatment, which included -blockers, ␣-agonists, miotics, carbonic anhydrase inhibitors (CAIs), prostaglandin analogs, and combination therapies (defined as a dispensing of the dorzolamide-timolol combination drops or of two or more different topical medications on the index date). We also divided the treatment location into four US regions: Midwest (Ohio, Illinois, Michigan,
VOL. 140, NO. 4
Arkansas, and Missouri), Northeast (Massachusetts and Rhode Island), Southeast (South Carolina, North Carolina, Florida, Virginia, Kentucky, and Georgia), and West (Nebraska, Arizona, and Utah). Patients with any ICD-9 diagnosis code for open angle glaucoma that indicated optic nerve damage (including ICD-9 codes 365.1x, 365.52, and 365.81) comprised the diagnosed glaucoma group. Patients with none of the diagnosed glaucoma codes from the date of initial diagnosis to the index date but who received a code for preglaucoma (365.00), borderline glaucoma findings (365.0 and 365.01), or ocular hypertension (365.04) were included in the suspect glaucoma group. Patients with both types of codes were classified as diagnosed glaucoma. ● DATA ANALYSIS: Patients were followed from the treatment initiation date through the study end date, which was defined as the earlier of March 31, 2002 (that is, 3 months after the end of the cohort identification period), or the end of enrollment in UnitedHealthcare. The days of therapy that were supplied in each dispensing were calculated on the basis of bottle size, with liberal assumptions regarding the duration of supply. For most medications, up to 5 ml were assumed to provide 60 days of therapy, up to 10 ml were assumed to provide 90 days of therapy, and above 10 ml were assumed to provided 120 days of therapy. Because prostaglandins are prescribed at fewer drops per day than the other classes, the corresponding bottle sizes for the prostaglandins were one-half those of the other drugs, with up to 2.5 ml assumed to provide 60 days of therapy, up to 5 ml assumed to provide 90 days of therapy, and over 5 ml assumed to provide 120 days of therapy. Pilocarpine, which requires the highest daily dosage, was assumed to supply 60 days of therapy for bottles of up to 10 ml, 90 days for bottles of up to 15 ml, and 120 days for bottles of ⬎15 ml. To examine persistence, which was defined as the duration of continuous medication use, we used survival analysis techniques. We obtained the time to discontinuation of therapy by finding the number of days from the index date to the end of the last dispensing of the initially prescribed topical medication before a gap in supply. Patients who reached their study end date (either by disenrolling from the UnitedHealthcare plan or by remaining in the plan until study closeout) without discontinuing therapy were censored as of the end date. A change in medication was considered a discontinuation of the originally prescribed drug. Kaplan-Meier plots showed the time to discontinuation of therapy by initial class of medication, separately for diagnosed glaucoma and suspect glaucoma patients. To investigate predictors of therapy persistence, we constructed Cox proportional hazards models of time from initial treatment to discontinuation of the initially prescribed treatment. This was calculated separately for diagnosed glaucoma and suspect glaucoma patients. An
TOPICAL GLAUCOMA THERAPY
598.e2
was highest in the Northeast (P ⬍ .001). The absolute number of patients with both diagnosed glaucoma and suspect glaucoma increased during the study period, although the relative proportions of suspect to glaucoma cases were stable. Timolol was the most common initial drug to be used, followed by latanoprost and brimonidine (Table 1). Because miotics and combinations of 2 or more drugs were rarely dispensed, these were not included in the analyses that are presented.
additional model that included both cohorts with diagnostic status as a predictor allowed a comparison of persistence between diagnostic groups. Risk factors that were entered into the model included class of initial medication (with -blockers as the reference group), age group, gender, region of the country, and index date divided into two periods (1995 to 1998 and 1999 to 2001). The index date dichotomy was included to take into account the possible effect of the entry of the first prostaglandin, latanoprost, by Food and Drug Administration approval in 1998. The coefficients of the proportional hazards model provided estimates of the relative daily probability of discontinuation (this is the “hazard ratio”) between people with and without characteristics that were represented by the variables in the model. To examine adherence (that is, prevalence of use of the initially prescribed medication at various time points), we obtained the proportion of patients who were still enrolled in the health plan with a current supply of the index medication at each 6-month interval after the index date. To be considered a current supply, the supply dates of a refill (the date of dispensing through that date plus the assumed days of supply) were required to include the given time point. These proportions describe point-in-time treatment without regard to previous gaps in therapy. Predictors of adherence were assessed with the use of the generalized estimating equations (GEE) procedure, which modeled the log of adherence as a linear function of the same covariates that were included in the Cox models. The GEE procedure was used to adjust the error estimates in the regression to account for repeated appearance of a single individual through the successive time points that were under evaluation. We used the log-binomial model with the GEE to allow the estimation of the adjusted relative risk of adherence between drug classes.14 We report both the persistence and adherence results that are grouped by medication class rather than by individual brands of medications, although persistence and adherence were defined only within the specific medication given on the index date.
● PERSISTENCE WITH TREATMENT: During the initial 3 years of follow-up, many patients received only one dispensing of the initial medication, which was indicated by the sharp drops in the curves at 60, 90, and 120 days (Figure 1). Almost one half of the patients who filled at least one prescription discontinued therapy within 6 months. Persistence with prostaglandins was substantially higher than with any of the other classes. Individuals who were treated initially with prostaglandins were approximately 40% as likely as those who were started on -blockers to discontinue use, based on Cox models that adjusted for age, gender, region, and year of index date (Table 2). ␣-Agonists and topical CAIs also had a lower risk of discontinuation than -blockers (Table 2). Among the other covariates in the model, only age group 40 to 49 years and female sex (among diagnosed glaucoma patients only) and Southeast region (in both cohorts) were significantly associated with higher risk of discontinuation. Overall, persistence was marginally greater among patients with diagnosed glaucoma than patients with suspect glaucoma (hazard ratio, 0.92; 95% CI, 0.86 to 1.00). ● ADHERENCE TO TREATMENT: The proportion of patients with a current refill, assuming a duration of supply of 60, 90, or 120 days depending on bottle size, was highest among individuals who used prostaglandins (Figure 2). Unlike in the persistence analyses, -blockers tended to be associated with higher proportions of adherent individuals than CAIs or ␣-agonists (Figure 2). In GEE models that adjusted for age, gender, region, and year (Table 3), patients with diagnosed glaucoma and suspect glaucoma who were filling prostaglandin analogs were 36% and 30% more likely, respectively, to adhere to the medication than patients who initially were dispensed -blockers. Patients with diagnosed glaucoma were 90% and with suspect glaucoma were 82% as likely to be adherent to ␣-agonists as patients receiving -blockers. Adherence to CAIs was similarly low, although the number of patients who were using CAIs was small, and the differences were not statistically significant (Table 3). Adherence decreased across time, with a greater decrease among patients with diagnosed glaucoma who were receiving ␣-agonists or CAIs and among patients with suspect glaucoma who were using ␣-agonists compared with the other classes (Figure 2). Patients with suspect glaucoma in the Southeast were
RESULTS ● BASELINE CHARACTERISTICS:
A total of 3623 patients with diagnosed glaucoma and 1677 patients with suspect glaucoma filled at least one prescription for topical ocular hypotensive agents. The age and gender distributions at baseline were similar, except for a slightly higher prevalence of 60- to 64-year-old patients in the diagnosed group (P ⫽ .01; Table 1). The mean ages of the patients with suspect glaucoma and diagnosed glaucoma were 55.1 and 55.8 years, respectively (P ⫽ .02). Compared with the other regions that were studied, the Midwest had a lower proportion of patients with diagnosed glaucoma to suspect glaucoma, although the diagnosed glaucoma proportion
598.e3
AMERICAN JOURNAL
OF
OPHTHALMOLOGY
OCTOBER 2005
TABLE 1. Baseline Characteristics and Initial Treatment of Patients With Diagnosed or Suspect Glaucoma Diagnosed Glaucoma (n ⫽ 3623) Variable
Age (y) at treatment initiation 30–39 40–49 50–59 60–64 65⫹ Gender Female Male Health plan region of the United States Midwest Northeast Southeast West Year of treatment initiation 1995 1996 1997 1998 1999 2000 2001 Initial glaucoma treatment Alpha-agonists Apraclonidine Brimonidine Dipivefrin Beta-blockers Betaxolol Carteolol Levobunolol Metipranolol Timolol Miotics Echothiophate Pilocarpine Carbonic anhydrase inhibitors Brinzolamide Dorzolamide Prostaglandins Bimatoprost Latanoprost Travoprost Unoprostone Combination product Dorzolamide and timolol Other combinations
Suspected Glaucoma (n ⫽ 1677)
N
%
N
%
235 824 1299 676 589
6.5 22.7 35.9 18.7 16.3
129 411 615 266 256
7.7 24.5 36.7 15.9* 15.3
1795 1828
49.5 50.5
855 822
51.0 49.0
1941 267 1195 220
53.6 7.4 33.0 6.1
989 63 513 112
59.0† 3.8† 30.6 6.7
203 445 451 584 615 647 678
5.6 12.3 12.4 16.1 17.0 17.9 18.7
103 210 198 298 290 282 296
6.1 12.5 11.8 17.8 17.3 16.8 17.7
514 7 489 18 1736 197 55 143 30 1311 32 0 32 94
14.2 0.2 13.5 0.5 47.9 5.4 1.5 3.9 0.8 36.2 0.9 0.0 0.9 2.6
244 8 230 6 883 115 24 56 12 676 28 1 27 59
14.5 0.5 13.7 0.4 52.7 6.9 1.4 3.3 0.7 40.3 1.7 0.1 1.6 3.5
21 73 966 85 810 65 6 64 64 217
0.6 2.0 26.7 2.3 22.4 1.8 0.2 1.8 1.8 6.0
15 44 386 37 330 16 3 27 27 50
0.9 2.6 23.0 2.2 19.7 1.0 0.2 1.6 1.6 3.0
*Differs from diagnosed group by chi-square analysis (P ⬍ .05). † Differs from diagnosed group by chi-square analysis (P ⬍ .001).
VOL. 140, NO. 4
TOPICAL GLAUCOMA THERAPY
598.e4
FIGURE 1. Percentage of patients who remained continuously on the initially dispensed topical ocular hypotensive. Kaplan-Meier curves are shown separately by diagnostic status at treatment initiation (diagnosed vs suspect glaucoma) and class of initial glaucoma medication. Patients who disenrolled from the health plan were censored as of the disenrollment date.
that was ⬎60 to 120 days, depending on bottle size) over a period of 3 years. With a definition for adherence that allows individuals to intermittently refill prescriptions (which means medications are taken substantially less than prescribed), only 37% of patients had refilled their initially prescribed medication within 60 to 120 days at 3 years after their initial dispensing. Adherence estimates at 3 years ranged from 15% to 58%, depending on drug class. The fact that more persons met our criteria for adherence than for persistence demonstrates that patients either continually use fewer drops than would be calculated as appropriate for the volume that they receive or that they have periods of complete cessation of therapy but later restart the medication. Both the persistence and adherence analyses found that patients whose initial treatment regimen contained a prostaglandin analog were more likely to continue using the drug that had been prescribed originally compared with all other classes of medications. Because these data are derived from pharmacy claims, we can only hypothesize as to the reasons for the increased persistence with prostaglandins. One explanation is that prostaglandins lower IOP more effectively than the other classes of medicines.15–18 This greater efficacy would be associated with fewer changes of medication by the prescribing physician. Second, patients might persist or adhere better with
less likely to adhere to treatment than were patients in the Midwest. Adherence was greater with increasing age among patients with diagnosed glaucoma but not among patients with suspect glaucoma. Diagnosed patients were 11% more likely to be adherent than patients with suspect glaucoma (95% CI, 5% to 18%).
DISCUSSION PATIENTS IN THIS DATABASE WHO FILLED A FIRST PRE-
scription for a topical hypotensive agent were unlikely to use it consistently, and most of the patients had discontinued its use altogether over a median follow-up period of 1.2 years. These persistence results are poor and point to a major potential failing in our efforts to treat individuals who are at risk for or diagnosed with glaucoma. Furthermore, we believe that our assumptions about how long it would take the average person to use a certain volume of eye drop medication are liberal and, if anything, overstate how long the typical interval might be between refills for bilateral therapy. Despite this approach, we found that nearly one half of the patients had discontinued the initially prescribed therapy completely within 6 months. Furthermore, fewer than 10% of this cohort of patients with diagnosed or suspect glaucoma had refilled their topical medications continuously (that is, no gap in refills 598.e5
AMERICAN JOURNAL
OF
OPHTHALMOLOGY
OCTOBER 2005
TABLE 2. Cox Models of Time to Discontinuation* of the Initially Prescribed Medication Hazard Ratio (95% CI) Variable
Class of initial glaucoma treatment Alpha-agonist Carbonic anhydrase inhibitor Prostaglandin Beta-blocker Age at treatment initiation (y) 30–39 40–49 50–59 60–64 65⫹ Gender Female Male Health plan region of the United States Northeast Southeast West Midwest Year of treatment initiation 1995–1998 1999–2001
Diagnosed Glaucoma
Suspected Glaucoma
0.69 (0.60, 0.78) 0.76 (0.57, 1.01) 0.40 (0.35, 0.44) Reference group
0.80 (0.67, 0.96) 0.59 (0.42, 0.83) 0.44 (0.37, 0.52) Reference group
1.11 (0.92, 1.34) 1.13 (1.01, 1.27) Reference group 0.98 (0.87, 1.11) 0.93 (0.81, 1.06)
1.02 (0.80, 1.29) 1.01 (0.86, 1.18) Reference group 0.94 (0.77, 1.13) 0.94 (0.78, 1.14)
1.09 (1.00, 1.19) Reference group
0.99 (0.87, 1.12) Reference group
1.00 (0.84, 1.18) 1.16 (1.05, 1.28) 0.98 (0.81, 1.19) Reference group
0.93 (0.65, 1.32) 1.20 (1.04, 1.37) 0.89 (0.68, 1.17) Reference group
Reference group 0.99 (0.90, 1.08)
Reference group 1.05 (0.93, 1.20)
*Separate Cox models were computed for diagnosed patients and suspects. The outcome modeled was time to first gap in therapy of longer than the assumed duration of the dispensing.
prostaglandins if they have fewer side effects. Third, the dose regimen might favor prostaglandins. Once-a-day dosing is standard with most prostaglandins and is sometimes used with -blockers, which may result in greater persistence and adherence. Alternatively, patients who were prescribed prostaglandins may have had more severe disease than those patients who were prescribed prostaglandins in other classes and therefore were more likely to persist in treatment. Arguing against this is the fact that patients with suspect glaucoma persisted in their medication use in a similar fashion to patients with diagnosed glaucoma. Finally, latanoprost, which constituted ⬎80% of the prostaglandin use in the present data, has one of the highest copayment costs of any topical ocular hypotensive. Thus, better persistence and adherence with prostaglandins occurred despite the disincentive of greater cost. Our study was based entirely on health insurance claims data without validation through medical record review. Claims provide a complete listing of services that are rendered in a large insured population, but their accuracy in presenting the correct diagnoses and the timing of drug exposure information must be considered with caution. A pharmacy dispensing of a medication does not ensure that the patient used the drug, and samples given by the physician fail to appear in the claims data. A small VOL. 140, NO. 4
proportion of patients may have had more than one health insurance provider, and of those, some patients may have filled one or more prescriptions under the other insurance. If patients filled prescriptions but did not take the drops, our measures would overstate the likely persistence and adherence of treatment. If they went outside the plan to fill prescriptions, our adherence and persistence estimates would be lower than actual use. In our opinion, these factors are equally likely to affect our results for all drug classes and therefore might have had a small impact on our final estimates of persistence and adherence but should not have affected the differences in use between medications. Successful glaucoma surgery or laser treatment may have obviated the need for further topical therapy in some patients, but the occurrence of such treatments was rare in this population (approximately 5% had surgery or laser treatment at any time during the follow-up period) and cannot account for the low persistence estimates that we have observed. The distribution of free drug samples to patients typically occurs at the beginning of therapy for most patients. Patients with lower incomes, however, might use samples as a means of obtaining expensive drugs over a broader time scale. Patients who receive free sample medications repeatedly after receiving the initial pharmacy dispensing
TOPICAL GLAUCOMA THERAPY
598.e6
FIGURE 2. Percentage of patients with a current refill of the initially dispensed medication at each 6-month interval after treatment initiation. A current refill was defined as a refill with the date of dispensing through that date plus assumed days of supply crossing the given time point. These adherence plots are shown separately by drug class within diagnostic status at treatment initiation (diagnosed vs suspect glaucoma). The denominator for each time point includes only those patients who were still enrolled in the health plan.
will appear to have poor persistence and adherence. There is no way at present to determine how large an impact free samples have on our derived estimates, but certainly the use of free samples leads to an underestimate of drug use when relying on pharmacy databases. The diagnostic criteria depend on the accuracy of physician coding and may have variable regional application. However, in the present study, we attempted to create the most appropriate cohorts for an investigation of persistence and adherence with glaucoma treatment by imposing strict definitions for new-onset glaucoma and suspect glaucoma, unlike in most other studies of persistence.6 –11 We restricted the study group to those with no testing or treatment in the past year that suggested that they were prior suspects for glaucoma. Furthermore, therapy for glaucoma had to follow the diagnosis temporally. Without these requirements, patients who received an ocular hypotensive for an acute elevation of IOP after surgery or as the result of a condition other than primary glaucoma may have been included inappropriately. Studies that include patients in chronic care rather than only the newly diagnosed are likely to overestimate persistence, because individuals who have already stayed on treatment for some duration are more likely to stay on medications than the average patient who is just starting treatment. By 598.e7
AMERICAN JOURNAL
obtaining a cohort of newly diagnosed patients, we are more likely to capture the true persistence and adherence levels in those patients who are diagnosed with disease. It is possible that some patients with pre-existing disease did not receive care for glaucoma or suspect glaucoma during the year before the date of initial diagnosis in our database; these patients could have initiated topical therapy earlier than the initiation date that we identified. Treatment persistence and adherence could be greater in patients with higher IOP levels at baseline and greater reductions in IOP with therapy, but information on IOP measurements was not available in our claims database. It is important to note, however, that the one published study of adherence to glaucoma medications that included baseline IOP levels in the analysis found no effect of this factor on treatment compliance.19 The duration of supply of a bottle of eye drops is difficult to determine, unlike tablets, which can be counted easily and divided by the number prescribed per day to obtain the supply time. We have assumed a liberal duration for each dispensing on the basis of bottle sizes. It is likely therefore that even the very low estimates of persistence and adherence that we found were overestimations of the true levels. OF
OPHTHALMOLOGY
OCTOBER 2005
TABLE 3. Generalized Estimating Equations Model of Adherence to Initial Medication* Relative Risk (95% CI)
Variable
Class of initial glaucoma treatment Alpha-agonist Carbonic anhydrase inhibitor Prostaglandin Beta-blocker Age at treatment initiation (y) 30–39 40–49 50–59 60–64 65⫹ Gender Female Male Health plan region of the United States Northeast Southeast West Midwest Year of treatment initiation 1995–1998 1999–2001
Diagnosed Glaucoma
Suspected Glaucoma
0.90 (0.81, 0.99) 0.82 (0.66, 1.03)
0.82 (0.69, 0.98) 0.93 (0.70, 1.25)
1.36 (1.28, 1.45) Reference group
1.30 (1.16, 1.45) Reference group
0.79 (0.66, 0.93) 0.96 (0.88, 1.04) Reference group 1.06 (0.98, 1.15) 1.11 (1.02, 1.20)
1.10 (0.91, 1.33) 1.04 (0.92, 1.18) Reference group 1.07 (0.92, 1.25) 1.15 (1.00, 1.33)
0.95 (0.89, 1.00) Reference group
0.95 (0.86, 1.04) Reference group
1.06 (0.95, 1.18) 0.93 (0.87, 1.00) 1.06 (0.94, 1.20) Reference group
0.87 (0.62, 1.20) 0.88 (0.79, 0.99) 1.08 (0.91, 1.28) Reference group
Reference group 1.02 (0.96, 1.09)
Reference group 0.98 (0.89, 1.09)
*Separate models were computed for diagnosed patients and suspects. The outcome modeled was occurrence of a refill with days of supply covering each 6-month point after treatment initiation.
Recent clinical trials document better preservation of visual field with lower IOP in patients with glaucoma.3,4 Poor persistence with ocular hypotensive medications is therefore likely to limit our ability to prevent vision loss from glaucoma. Prostaglandin analogs, which include latanoprost, bimatoprost, travoprost, and unoprostone, were associated with improved persistence compared with the other classes of drugs. Because latanoprost was the only prostaglandin that was available during most of the period of the present study, we cannot determine whether all of the prostaglandins show similar advantages with respect to persistence. Preliminary research suggests that persistence may be greater for latanoprost than for bimatoprost or travoprost,11,20 with the difference possibly related to tolerability,21 but additional studies that use appropriate diagnostic criteria are needed. In summary, we found that many patients fail to use topical medications as prescribed. Research is needed to VOL. 140, NO. 4
find ways to improve on this. Although physicians must monitor and encourage proper medication use among patients with glaucoma or ocular hypertension, we also need to evaluate novel approaches to help patients with this process if we are to improve on the currently poor persistence and adherence that are documented in this report.
REFERENCES 1. American Academy of Ophthalmology Preferred Practice Patterns Committee Glaucoma Panel. Preferred practice pattern: primary open angle glaucoma, limited revision. San Francisco: American Academy of Ophthalmology; 2003. 2. The AGIS Investigators. The Advanced Glaucoma Intervention Study (AGIS): 7, the relationship between control of intraocular pressure and visual field deterioration. Am J Ophthalmol 2000;130:429 – 440. 3. Heijl A, Leske MC, Bengtsson B, Hyman L, Bengtsson B, Hussein M; Early Manifest Glaucoma Trial Group. Reduction of intraocular pressure and glaucoma progression: results from the Early Manifest Glaucoma Trial. Arch Ophthalmol 2002;120:1268 –1279. 4. Anderson DR; Normal Tension Glaucoma Study. Collaborative normal tension glaucoma study. Curr Opin Ophthalmol 2003;14:86 –90. 5. Kass MA, Heuer DK, Higginbotham EJ, et al. The Ocular Hypertension Treatment Study: a randomized trial determines that topical ocular hypotensive medication delays or prevents the onset of primary open-angle glaucoma. Arch Ophthalmol 2002;120:701–713. 6. Gurwitz JH, Glynn RJ, Monane M, et al. Treatment for glaucoma: adherence by the elderly. Am J Public Health 1993;83:711–716. 7. Dasgupta S, Oates V, Bookhart BK, Vaziri B, Schwartz GF, Mozaffari E. Population-based persistency rates for topical glaucoma medications measured with pharmacy claims data. Am J Manag Care 2002;8(suppl 10):S255–S261. 8. Spooner JJ, Bullano MF, Ikeda LI, et al. Rates of discontinuation and change of glaucoma therapy in a managed care setting. Am J Manag Care 2002;8(suppl 10):S262–S270. 9. Shaya FT, Mullins CD, Wong W, Cho J. Discontinuation rates of topical glaucoma medications in a managed care population. Am J Manag Care 2002;8(suppl 10):S271–S277. 10. Reardon G, Schwartz GF, Mozaffari E. Patient persistency with pharmacotherapy in the management of glaucoma. Eur J Ophthalmol 2003;13(suppl 4):S44 –S52. 11. Reardon G, Schwartz GF, Mozaffari E. Patient persistency with topical ocular hypotensive therapy in a managed care population. Am J Ophthalmol 2004;137(suppl 1):S3–S12. 12. Schwartz GF, Reardon G, Mozaffari E. Persistency with latanoprost or timolol in primary open-angle glaucoma suspects. Am J Ophthalmol 2004;137(suppl 1):S13–S16. 13. Diestelhorst M, Schaefer CP, Beusterien KM, et al. Persistency and clinical outcomes associated with latanoprost and beta-blocker monotherapy: evidence from a European retrospective cohort study. Eur J Ophthalmol 2003;13(suppl 4):S21–S29.
TOPICAL GLAUCOMA THERAPY
598.e8
18. Niazi MK, Raja N. Comparison of latanoprost and dorzolamide in the treatment of patients with open angle glaucoma. J Ayub Med Coll Abbottabad 2004;16:50 – 53. 19. Gurwitz JH, Yeomans SM, Glynn RJ, Lewis BE, Levin R, Avorn J. Patient noncompliance in the managed care setting: the case of medical therapy for glaucoma. Med Care 1998;36:357–369. 20. Reardon G, Schwartz GF, Mozaffari E. Patient persistency with ocular prostaglandin therapy: a population-based, retrospective study. Clin Ther 2003;25:1172–1185. 21. Parrish RK, Palmberg P, Sheu WP, for the XLT study group. A comparison of latanoprost, bimatoprost, and travoprost in patients with elevated intraocular pressure: a 12-week, randomized, masked-evaluator multicenter study. Am J Ophthalmol 2003;135:688 –703.
14. McNutt L-A, Wu C, Xue X, Hafner JP. Estimating the relative risk in cohort studies and clinical trials of common outcomes. Am J Epidemiol 2003;157:940 –943. 15. Hedman K, Alm A, Gross RL. Pooled-data analysis of three randomized, double-masked, six-month studies comparing intraocular pressure-reducing effects of latanoprost and timolol in patients with ocular hypertension. J Glaucoma 2003; 12:463– 465. 16. Cohen JS, Gross RL, Cheetham JK, VanDenburgh AM, Bernstein P, Whitcup SM. Two-year double-masked comparison of bimatoprost with timolol in patients with glaucoma or ocular hypertension. Surv Ophthalmol 2004;49(suppl 1):S45–S52. 17. Thomas R, Parikh R, Muliyil J, George R, Paul P, Abraham LM. Comparison between latanoprost and brimonidine efficacy and safety in Indian eyes. Indian J Ophthalmol 2003; 51:123–128.
598.e9
AMERICAN JOURNAL
OF
OPHTHALMOLOGY
OCTOBER 2005
Biosketch David S. Friedman, MD, MPH, is an associate professor at the Wilmer Eye Institute of Johns Hopkins University School of Medicine, Baltimore, Maryland, and associate professor in the Department of International Health at Johns Hopkins Bloomberg School of Public Health. He has received numerous honors and awards, including a Clinician-Scientist Award from the NEI, the RPB Robert E. McCormick Award in 2000, and the American Geriatrics Society Douglas Jahnigen Scholars Award in 2002.
VOL. 140, NO. 4
TOPICAL GLAUCOMA THERAPY
598.e10
Biosketch Beth L. Nordstrom, PhD, MPH, is an epidemiologist with the Epidemiology Division of Ingenix i3 Magnifi in Auburndale, Massachusetts. Her current work consists primarily of drug safety and outcomes research using health care claims data. In addition to various investigations into glaucoma treatment, Dr Nordstrom’s research has focused on the epidemiology of nicotine addiction and treatments for infectious diseases.
598.e11
AMERICAN JOURNAL
OF
OPHTHALMOLOGY
OCTOBER 2005
● PURPOSE:
The present study describes the patterns and predictors of treatment persistence and adherence among patients who are diagnosed with glaucoma or as glaucoma suspects (based on claims codes). ● DESIGN: A retrospective cohort study using health insurance claims data. ● METHODS: Newly treated individuals with diagnosed glaucoma (n ⴝ 3623) and suspect glaucoma (n ⴝ 1677) were obtained from healthcare claims data in the Ingenix Research Database. For each of these two diagnostic groups, we calculated the duration of continuous treatment with the initially prescribed medication (persistence) and the prevalence of use of the initial medication at various time points (adherence). Four drug classes were included: -blockers, ␣-agonists, carbonic anhydrase inhibitors, and prostaglandin analogs. ● RESULTS: Nearly one half of the individuals who had filled a glaucoma prescription discontinued all topical ocular hypotensive therapy within six months, and just 37% of these individuals recently had refilled their initial medication at three years after the first dispensing. Prostaglandins were associated with better persistence than any other drug class, which was indicated by hazard ratios for discontinuation of prostaglandins compared with -blockers of 0.40 (95% confidence interval, 0.35– 0.44) for diagnosed patients and 0.44 (95% confidence interval, 0.37– 0.52) for patients with suspect glaucoma. Prostaglandins showed a similar advantage in adherence. Furthermore, patients with diagnosed glaucoma were more likely to adhere to therapy than patients with Accepted for publication Apr 22, 2005. From Ingenix Epidemiology, Auburndale, Massachusetts (B.L.N.; A.M.W.); Wilmer Institute, Johns Hopkins University, Baltimore, Maryland (D.S.F.; H.A.Q.); Pfizer, Inc, New York, NY (E.M.) Supported by Pfizer through a research contract with Ingenix (B.L.N.; A.M.W.) and through an unrestricted grant to Johns Hopkins University (D.S.F.;H.A.Q.). Pfizer provided support for this study through a research contract with Ingenix (B.N. and A.W.). Pfizer also provided an unrestricted grant to Johns Hopkins University (D.F. and H.Q.). One of the authors (E.M.) is employed by Pfizer; he was involved in the study design, interpretation, and writing only as a member of a steering committee; all final decisions were made by the authors not employed by Pfizer. Inquiries to Beth L. Nordstrom, PHD, Ingenix Epidemiology, Riverside Center, Suite 3–120 275 Grove St, Auburndale, MA 02466; fax: 617–244 –9669; e-mail: [email protected]
598.e1
©
2005 BY
suspect glaucoma (relative risk ⴝ 1.11; 95% confidence interval, 1.05–1.18). ● CONCLUSION: Persistence and adherence were substantially better with prostaglandins than with other drug classes, and patients with diagnosed open-angle glaucoma were more likely to adhere to treatment than suspected glaucoma. (Am J Ophthalmol 2005;140: 598 – 606. © 2005 by Elsevier Inc. All rights reserved.)
A
LTHOUGH THE PREFERRED PRACTICE PATTERN OF
the American Academy of Ophthalmology lists medical therapy, laser trabeculoplasty, and surgical treatment as reasonable options for the initial treatment of glaucoma, most patients initially receive topical ocular hypotensives.1 If topical treatment lowers intraocular pressure (IOP) adequately, the patient is intended to remain on therapy indefinitely to improve outcome.2 Although recent studies clearly have documented that the lowering of IOP decreases the risk of visual field loss3,4 and slows progression from ocular hypertension to glaucoma,5 many patients appear to discontinue their use of topical hypotensive agents.6 –13 Most studies of persistence with topical glaucoma medications that have been conducted to date have not restricted the study population to patients who were diagnosed with glaucoma or suspect glaucoma.6 –11 Of the two studies that have imposed diagnostic criteria, one study included only suspects,12 and the other study was fairly small (260 patients in total) and used prescribing records without ascertaining whether the prescriptions were filled.13 It is important to investigate persistence of treatment within groups of patients with similar diagnostic status; ocular hypotensives that are prescribed for the treatment of primary open angle glaucoma may be used differently from those given for ocular hypertension without optic nerve damage and certainly should be used differently for transient elevations in IOP. In addition, none of the studies that were found in the literature examined both the continuous use of glaucoma medications (persistence) and ongoing use, which allowed for gaps in therapy (adherence). We undertook the present study to investigate the patterns and predictors of treat-
ELSEVIER INC. ALL
RIGHTS RESERVED.
0002-9394/05/$30.00 doi:10.1016/j.ajo.2005.04.051
ment persistence in a large population of patients who were diagnosed in healthcare claims as suspect glaucoma and separately in patients who were diagnosed with open angle glaucoma.
METHODS ● DATA SOURCE:
Data for the study were obtained from the Ingenix Research Database, which consists of insurance claims that include demographic, medical, and prescription information from approximately 8,000,000 members of UnitedHealthcare. UnitedHealthcare is one of the largest healthcare companies in the United States, with ⬎300,000 physicians contracted to provide healthcare services to ⬎14 million members. The analyses for the present study used encrypted automated data that were de-identified with respect to protected health information. Under the current privacy rules of the Health Insurance Portability and Accountability Act, Institutional Review Board approval and informed consent of subjects were not required. ● COHORT DEFINITION:
The study cohort consisted of patients who were identified from the Ingenix Research Database between July 1, 1995, and December 31, 2001, as newly diagnosed with open angle (diagnosed glaucoma) or suspect glaucoma who received at least one dispensing of a topical ocular hypotensive medication. Patients were required to be at least 30 years of age and to have at least 365 days of continuous enrollment in UnitedHealthcare as of the initial diagnosis date. Patients were excluded if they had glaucoma-related claims within 365 days before the initial diagnosis date (ICD-9 code [365.xx or 377.14] in association with a physician visit or hospitalization), any dispensing of a topical ocular hypotensive, or any procedure code that indicated argon laser trabeculoplasty or glaucoma surgery in conjunction with a glaucoma diagnosis on the claim. We also excluded any patient with claims for visual field testing, fundus photography, or nerve fiber layer imaging during the 10 months of the baseline period that ranged from 2 to 12 months before the initial diagnosis date. Finally, to be included in the study cohort, newly diagnosed patients must have received at least one dispensing of a topical glaucoma medication on or after the date of initial diagnosis.
● BASELINE CHARACTERISTICS:
Patient characteristics that were evaluated at baseline included age, gender, and class of initial treatment, which included -blockers, ␣-agonists, miotics, carbonic anhydrase inhibitors (CAIs), prostaglandin analogs, and combination therapies (defined as a dispensing of the dorzolamide-timolol combination drops or of two or more different topical medications on the index date). We also divided the treatment location into four US regions: Midwest (Ohio, Illinois, Michigan,
VOL. 140, NO. 4
Arkansas, and Missouri), Northeast (Massachusetts and Rhode Island), Southeast (South Carolina, North Carolina, Florida, Virginia, Kentucky, and Georgia), and West (Nebraska, Arizona, and Utah). Patients with any ICD-9 diagnosis code for open angle glaucoma that indicated optic nerve damage (including ICD-9 codes 365.1x, 365.52, and 365.81) comprised the diagnosed glaucoma group. Patients with none of the diagnosed glaucoma codes from the date of initial diagnosis to the index date but who received a code for preglaucoma (365.00), borderline glaucoma findings (365.0 and 365.01), or ocular hypertension (365.04) were included in the suspect glaucoma group. Patients with both types of codes were classified as diagnosed glaucoma. ● DATA ANALYSIS: Patients were followed from the treatment initiation date through the study end date, which was defined as the earlier of March 31, 2002 (that is, 3 months after the end of the cohort identification period), or the end of enrollment in UnitedHealthcare. The days of therapy that were supplied in each dispensing were calculated on the basis of bottle size, with liberal assumptions regarding the duration of supply. For most medications, up to 5 ml were assumed to provide 60 days of therapy, up to 10 ml were assumed to provide 90 days of therapy, and above 10 ml were assumed to provided 120 days of therapy. Because prostaglandins are prescribed at fewer drops per day than the other classes, the corresponding bottle sizes for the prostaglandins were one-half those of the other drugs, with up to 2.5 ml assumed to provide 60 days of therapy, up to 5 ml assumed to provide 90 days of therapy, and over 5 ml assumed to provide 120 days of therapy. Pilocarpine, which requires the highest daily dosage, was assumed to supply 60 days of therapy for bottles of up to 10 ml, 90 days for bottles of up to 15 ml, and 120 days for bottles of ⬎15 ml. To examine persistence, which was defined as the duration of continuous medication use, we used survival analysis techniques. We obtained the time to discontinuation of therapy by finding the number of days from the index date to the end of the last dispensing of the initially prescribed topical medication before a gap in supply. Patients who reached their study end date (either by disenrolling from the UnitedHealthcare plan or by remaining in the plan until study closeout) without discontinuing therapy were censored as of the end date. A change in medication was considered a discontinuation of the originally prescribed drug. Kaplan-Meier plots showed the time to discontinuation of therapy by initial class of medication, separately for diagnosed glaucoma and suspect glaucoma patients. To investigate predictors of therapy persistence, we constructed Cox proportional hazards models of time from initial treatment to discontinuation of the initially prescribed treatment. This was calculated separately for diagnosed glaucoma and suspect glaucoma patients. An
TOPICAL GLAUCOMA THERAPY
598.e2
was highest in the Northeast (P ⬍ .001). The absolute number of patients with both diagnosed glaucoma and suspect glaucoma increased during the study period, although the relative proportions of suspect to glaucoma cases were stable. Timolol was the most common initial drug to be used, followed by latanoprost and brimonidine (Table 1). Because miotics and combinations of 2 or more drugs were rarely dispensed, these were not included in the analyses that are presented.
additional model that included both cohorts with diagnostic status as a predictor allowed a comparison of persistence between diagnostic groups. Risk factors that were entered into the model included class of initial medication (with -blockers as the reference group), age group, gender, region of the country, and index date divided into two periods (1995 to 1998 and 1999 to 2001). The index date dichotomy was included to take into account the possible effect of the entry of the first prostaglandin, latanoprost, by Food and Drug Administration approval in 1998. The coefficients of the proportional hazards model provided estimates of the relative daily probability of discontinuation (this is the “hazard ratio”) between people with and without characteristics that were represented by the variables in the model. To examine adherence (that is, prevalence of use of the initially prescribed medication at various time points), we obtained the proportion of patients who were still enrolled in the health plan with a current supply of the index medication at each 6-month interval after the index date. To be considered a current supply, the supply dates of a refill (the date of dispensing through that date plus the assumed days of supply) were required to include the given time point. These proportions describe point-in-time treatment without regard to previous gaps in therapy. Predictors of adherence were assessed with the use of the generalized estimating equations (GEE) procedure, which modeled the log of adherence as a linear function of the same covariates that were included in the Cox models. The GEE procedure was used to adjust the error estimates in the regression to account for repeated appearance of a single individual through the successive time points that were under evaluation. We used the log-binomial model with the GEE to allow the estimation of the adjusted relative risk of adherence between drug classes.14 We report both the persistence and adherence results that are grouped by medication class rather than by individual brands of medications, although persistence and adherence were defined only within the specific medication given on the index date.
● PERSISTENCE WITH TREATMENT: During the initial 3 years of follow-up, many patients received only one dispensing of the initial medication, which was indicated by the sharp drops in the curves at 60, 90, and 120 days (Figure 1). Almost one half of the patients who filled at least one prescription discontinued therapy within 6 months. Persistence with prostaglandins was substantially higher than with any of the other classes. Individuals who were treated initially with prostaglandins were approximately 40% as likely as those who were started on -blockers to discontinue use, based on Cox models that adjusted for age, gender, region, and year of index date (Table 2). ␣-Agonists and topical CAIs also had a lower risk of discontinuation than -blockers (Table 2). Among the other covariates in the model, only age group 40 to 49 years and female sex (among diagnosed glaucoma patients only) and Southeast region (in both cohorts) were significantly associated with higher risk of discontinuation. Overall, persistence was marginally greater among patients with diagnosed glaucoma than patients with suspect glaucoma (hazard ratio, 0.92; 95% CI, 0.86 to 1.00). ● ADHERENCE TO TREATMENT: The proportion of patients with a current refill, assuming a duration of supply of 60, 90, or 120 days depending on bottle size, was highest among individuals who used prostaglandins (Figure 2). Unlike in the persistence analyses, -blockers tended to be associated with higher proportions of adherent individuals than CAIs or ␣-agonists (Figure 2). In GEE models that adjusted for age, gender, region, and year (Table 3), patients with diagnosed glaucoma and suspect glaucoma who were filling prostaglandin analogs were 36% and 30% more likely, respectively, to adhere to the medication than patients who initially were dispensed -blockers. Patients with diagnosed glaucoma were 90% and with suspect glaucoma were 82% as likely to be adherent to ␣-agonists as patients receiving -blockers. Adherence to CAIs was similarly low, although the number of patients who were using CAIs was small, and the differences were not statistically significant (Table 3). Adherence decreased across time, with a greater decrease among patients with diagnosed glaucoma who were receiving ␣-agonists or CAIs and among patients with suspect glaucoma who were using ␣-agonists compared with the other classes (Figure 2). Patients with suspect glaucoma in the Southeast were
RESULTS ● BASELINE CHARACTERISTICS:
A total of 3623 patients with diagnosed glaucoma and 1677 patients with suspect glaucoma filled at least one prescription for topical ocular hypotensive agents. The age and gender distributions at baseline were similar, except for a slightly higher prevalence of 60- to 64-year-old patients in the diagnosed group (P ⫽ .01; Table 1). The mean ages of the patients with suspect glaucoma and diagnosed glaucoma were 55.1 and 55.8 years, respectively (P ⫽ .02). Compared with the other regions that were studied, the Midwest had a lower proportion of patients with diagnosed glaucoma to suspect glaucoma, although the diagnosed glaucoma proportion
598.e3
AMERICAN JOURNAL
OF
OPHTHALMOLOGY
OCTOBER 2005
TABLE 1. Baseline Characteristics and Initial Treatment of Patients With Diagnosed or Suspect Glaucoma Diagnosed Glaucoma (n ⫽ 3623) Variable
Age (y) at treatment initiation 30–39 40–49 50–59 60–64 65⫹ Gender Female Male Health plan region of the United States Midwest Northeast Southeast West Year of treatment initiation 1995 1996 1997 1998 1999 2000 2001 Initial glaucoma treatment Alpha-agonists Apraclonidine Brimonidine Dipivefrin Beta-blockers Betaxolol Carteolol Levobunolol Metipranolol Timolol Miotics Echothiophate Pilocarpine Carbonic anhydrase inhibitors Brinzolamide Dorzolamide Prostaglandins Bimatoprost Latanoprost Travoprost Unoprostone Combination product Dorzolamide and timolol Other combinations
Suspected Glaucoma (n ⫽ 1677)
N
%
N
%
235 824 1299 676 589
6.5 22.7 35.9 18.7 16.3
129 411 615 266 256
7.7 24.5 36.7 15.9* 15.3
1795 1828
49.5 50.5
855 822
51.0 49.0
1941 267 1195 220
53.6 7.4 33.0 6.1
989 63 513 112
59.0† 3.8† 30.6 6.7
203 445 451 584 615 647 678
5.6 12.3 12.4 16.1 17.0 17.9 18.7
103 210 198 298 290 282 296
6.1 12.5 11.8 17.8 17.3 16.8 17.7
514 7 489 18 1736 197 55 143 30 1311 32 0 32 94
14.2 0.2 13.5 0.5 47.9 5.4 1.5 3.9 0.8 36.2 0.9 0.0 0.9 2.6
244 8 230 6 883 115 24 56 12 676 28 1 27 59
14.5 0.5 13.7 0.4 52.7 6.9 1.4 3.3 0.7 40.3 1.7 0.1 1.6 3.5
21 73 966 85 810 65 6 64 64 217
0.6 2.0 26.7 2.3 22.4 1.8 0.2 1.8 1.8 6.0
15 44 386 37 330 16 3 27 27 50
0.9 2.6 23.0 2.2 19.7 1.0 0.2 1.6 1.6 3.0
*Differs from diagnosed group by chi-square analysis (P ⬍ .05). † Differs from diagnosed group by chi-square analysis (P ⬍ .001).
VOL. 140, NO. 4
TOPICAL GLAUCOMA THERAPY
598.e4
FIGURE 1. Percentage of patients who remained continuously on the initially dispensed topical ocular hypotensive. Kaplan-Meier curves are shown separately by diagnostic status at treatment initiation (diagnosed vs suspect glaucoma) and class of initial glaucoma medication. Patients who disenrolled from the health plan were censored as of the disenrollment date.
that was ⬎60 to 120 days, depending on bottle size) over a period of 3 years. With a definition for adherence that allows individuals to intermittently refill prescriptions (which means medications are taken substantially less than prescribed), only 37% of patients had refilled their initially prescribed medication within 60 to 120 days at 3 years after their initial dispensing. Adherence estimates at 3 years ranged from 15% to 58%, depending on drug class. The fact that more persons met our criteria for adherence than for persistence demonstrates that patients either continually use fewer drops than would be calculated as appropriate for the volume that they receive or that they have periods of complete cessation of therapy but later restart the medication. Both the persistence and adherence analyses found that patients whose initial treatment regimen contained a prostaglandin analog were more likely to continue using the drug that had been prescribed originally compared with all other classes of medications. Because these data are derived from pharmacy claims, we can only hypothesize as to the reasons for the increased persistence with prostaglandins. One explanation is that prostaglandins lower IOP more effectively than the other classes of medicines.15–18 This greater efficacy would be associated with fewer changes of medication by the prescribing physician. Second, patients might persist or adhere better with
less likely to adhere to treatment than were patients in the Midwest. Adherence was greater with increasing age among patients with diagnosed glaucoma but not among patients with suspect glaucoma. Diagnosed patients were 11% more likely to be adherent than patients with suspect glaucoma (95% CI, 5% to 18%).
DISCUSSION PATIENTS IN THIS DATABASE WHO FILLED A FIRST PRE-
scription for a topical hypotensive agent were unlikely to use it consistently, and most of the patients had discontinued its use altogether over a median follow-up period of 1.2 years. These persistence results are poor and point to a major potential failing in our efforts to treat individuals who are at risk for or diagnosed with glaucoma. Furthermore, we believe that our assumptions about how long it would take the average person to use a certain volume of eye drop medication are liberal and, if anything, overstate how long the typical interval might be between refills for bilateral therapy. Despite this approach, we found that nearly one half of the patients had discontinued the initially prescribed therapy completely within 6 months. Furthermore, fewer than 10% of this cohort of patients with diagnosed or suspect glaucoma had refilled their topical medications continuously (that is, no gap in refills 598.e5
AMERICAN JOURNAL
OF
OPHTHALMOLOGY
OCTOBER 2005
TABLE 2. Cox Models of Time to Discontinuation* of the Initially Prescribed Medication Hazard Ratio (95% CI) Variable
Class of initial glaucoma treatment Alpha-agonist Carbonic anhydrase inhibitor Prostaglandin Beta-blocker Age at treatment initiation (y) 30–39 40–49 50–59 60–64 65⫹ Gender Female Male Health plan region of the United States Northeast Southeast West Midwest Year of treatment initiation 1995–1998 1999–2001
Diagnosed Glaucoma
Suspected Glaucoma
0.69 (0.60, 0.78) 0.76 (0.57, 1.01) 0.40 (0.35, 0.44) Reference group
0.80 (0.67, 0.96) 0.59 (0.42, 0.83) 0.44 (0.37, 0.52) Reference group
1.11 (0.92, 1.34) 1.13 (1.01, 1.27) Reference group 0.98 (0.87, 1.11) 0.93 (0.81, 1.06)
1.02 (0.80, 1.29) 1.01 (0.86, 1.18) Reference group 0.94 (0.77, 1.13) 0.94 (0.78, 1.14)
1.09 (1.00, 1.19) Reference group
0.99 (0.87, 1.12) Reference group
1.00 (0.84, 1.18) 1.16 (1.05, 1.28) 0.98 (0.81, 1.19) Reference group
0.93 (0.65, 1.32) 1.20 (1.04, 1.37) 0.89 (0.68, 1.17) Reference group
Reference group 0.99 (0.90, 1.08)
Reference group 1.05 (0.93, 1.20)
*Separate Cox models were computed for diagnosed patients and suspects. The outcome modeled was time to first gap in therapy of longer than the assumed duration of the dispensing.
prostaglandins if they have fewer side effects. Third, the dose regimen might favor prostaglandins. Once-a-day dosing is standard with most prostaglandins and is sometimes used with -blockers, which may result in greater persistence and adherence. Alternatively, patients who were prescribed prostaglandins may have had more severe disease than those patients who were prescribed prostaglandins in other classes and therefore were more likely to persist in treatment. Arguing against this is the fact that patients with suspect glaucoma persisted in their medication use in a similar fashion to patients with diagnosed glaucoma. Finally, latanoprost, which constituted ⬎80% of the prostaglandin use in the present data, has one of the highest copayment costs of any topical ocular hypotensive. Thus, better persistence and adherence with prostaglandins occurred despite the disincentive of greater cost. Our study was based entirely on health insurance claims data without validation through medical record review. Claims provide a complete listing of services that are rendered in a large insured population, but their accuracy in presenting the correct diagnoses and the timing of drug exposure information must be considered with caution. A pharmacy dispensing of a medication does not ensure that the patient used the drug, and samples given by the physician fail to appear in the claims data. A small VOL. 140, NO. 4
proportion of patients may have had more than one health insurance provider, and of those, some patients may have filled one or more prescriptions under the other insurance. If patients filled prescriptions but did not take the drops, our measures would overstate the likely persistence and adherence of treatment. If they went outside the plan to fill prescriptions, our adherence and persistence estimates would be lower than actual use. In our opinion, these factors are equally likely to affect our results for all drug classes and therefore might have had a small impact on our final estimates of persistence and adherence but should not have affected the differences in use between medications. Successful glaucoma surgery or laser treatment may have obviated the need for further topical therapy in some patients, but the occurrence of such treatments was rare in this population (approximately 5% had surgery or laser treatment at any time during the follow-up period) and cannot account for the low persistence estimates that we have observed. The distribution of free drug samples to patients typically occurs at the beginning of therapy for most patients. Patients with lower incomes, however, might use samples as a means of obtaining expensive drugs over a broader time scale. Patients who receive free sample medications repeatedly after receiving the initial pharmacy dispensing
TOPICAL GLAUCOMA THERAPY
598.e6
FIGURE 2. Percentage of patients with a current refill of the initially dispensed medication at each 6-month interval after treatment initiation. A current refill was defined as a refill with the date of dispensing through that date plus assumed days of supply crossing the given time point. These adherence plots are shown separately by drug class within diagnostic status at treatment initiation (diagnosed vs suspect glaucoma). The denominator for each time point includes only those patients who were still enrolled in the health plan.
will appear to have poor persistence and adherence. There is no way at present to determine how large an impact free samples have on our derived estimates, but certainly the use of free samples leads to an underestimate of drug use when relying on pharmacy databases. The diagnostic criteria depend on the accuracy of physician coding and may have variable regional application. However, in the present study, we attempted to create the most appropriate cohorts for an investigation of persistence and adherence with glaucoma treatment by imposing strict definitions for new-onset glaucoma and suspect glaucoma, unlike in most other studies of persistence.6 –11 We restricted the study group to those with no testing or treatment in the past year that suggested that they were prior suspects for glaucoma. Furthermore, therapy for glaucoma had to follow the diagnosis temporally. Without these requirements, patients who received an ocular hypotensive for an acute elevation of IOP after surgery or as the result of a condition other than primary glaucoma may have been included inappropriately. Studies that include patients in chronic care rather than only the newly diagnosed are likely to overestimate persistence, because individuals who have already stayed on treatment for some duration are more likely to stay on medications than the average patient who is just starting treatment. By 598.e7
AMERICAN JOURNAL
obtaining a cohort of newly diagnosed patients, we are more likely to capture the true persistence and adherence levels in those patients who are diagnosed with disease. It is possible that some patients with pre-existing disease did not receive care for glaucoma or suspect glaucoma during the year before the date of initial diagnosis in our database; these patients could have initiated topical therapy earlier than the initiation date that we identified. Treatment persistence and adherence could be greater in patients with higher IOP levels at baseline and greater reductions in IOP with therapy, but information on IOP measurements was not available in our claims database. It is important to note, however, that the one published study of adherence to glaucoma medications that included baseline IOP levels in the analysis found no effect of this factor on treatment compliance.19 The duration of supply of a bottle of eye drops is difficult to determine, unlike tablets, which can be counted easily and divided by the number prescribed per day to obtain the supply time. We have assumed a liberal duration for each dispensing on the basis of bottle sizes. It is likely therefore that even the very low estimates of persistence and adherence that we found were overestimations of the true levels. OF
OPHTHALMOLOGY
OCTOBER 2005
TABLE 3. Generalized Estimating Equations Model of Adherence to Initial Medication* Relative Risk (95% CI)
Variable
Class of initial glaucoma treatment Alpha-agonist Carbonic anhydrase inhibitor Prostaglandin Beta-blocker Age at treatment initiation (y) 30–39 40–49 50–59 60–64 65⫹ Gender Female Male Health plan region of the United States Northeast Southeast West Midwest Year of treatment initiation 1995–1998 1999–2001
Diagnosed Glaucoma
Suspected Glaucoma
0.90 (0.81, 0.99) 0.82 (0.66, 1.03)
0.82 (0.69, 0.98) 0.93 (0.70, 1.25)
1.36 (1.28, 1.45) Reference group
1.30 (1.16, 1.45) Reference group
0.79 (0.66, 0.93) 0.96 (0.88, 1.04) Reference group 1.06 (0.98, 1.15) 1.11 (1.02, 1.20)
1.10 (0.91, 1.33) 1.04 (0.92, 1.18) Reference group 1.07 (0.92, 1.25) 1.15 (1.00, 1.33)
0.95 (0.89, 1.00) Reference group
0.95 (0.86, 1.04) Reference group
1.06 (0.95, 1.18) 0.93 (0.87, 1.00) 1.06 (0.94, 1.20) Reference group
0.87 (0.62, 1.20) 0.88 (0.79, 0.99) 1.08 (0.91, 1.28) Reference group
Reference group 1.02 (0.96, 1.09)
Reference group 0.98 (0.89, 1.09)
*Separate models were computed for diagnosed patients and suspects. The outcome modeled was occurrence of a refill with days of supply covering each 6-month point after treatment initiation.
Recent clinical trials document better preservation of visual field with lower IOP in patients with glaucoma.3,4 Poor persistence with ocular hypotensive medications is therefore likely to limit our ability to prevent vision loss from glaucoma. Prostaglandin analogs, which include latanoprost, bimatoprost, travoprost, and unoprostone, were associated with improved persistence compared with the other classes of drugs. Because latanoprost was the only prostaglandin that was available during most of the period of the present study, we cannot determine whether all of the prostaglandins show similar advantages with respect to persistence. Preliminary research suggests that persistence may be greater for latanoprost than for bimatoprost or travoprost,11,20 with the difference possibly related to tolerability,21 but additional studies that use appropriate diagnostic criteria are needed. In summary, we found that many patients fail to use topical medications as prescribed. Research is needed to VOL. 140, NO. 4
find ways to improve on this. Although physicians must monitor and encourage proper medication use among patients with glaucoma or ocular hypertension, we also need to evaluate novel approaches to help patients with this process if we are to improve on the currently poor persistence and adherence that are documented in this report.
REFERENCES 1. American Academy of Ophthalmology Preferred Practice Patterns Committee Glaucoma Panel. Preferred practice pattern: primary open angle glaucoma, limited revision. San Francisco: American Academy of Ophthalmology; 2003. 2. The AGIS Investigators. The Advanced Glaucoma Intervention Study (AGIS): 7, the relationship between control of intraocular pressure and visual field deterioration. Am J Ophthalmol 2000;130:429 – 440. 3. Heijl A, Leske MC, Bengtsson B, Hyman L, Bengtsson B, Hussein M; Early Manifest Glaucoma Trial Group. Reduction of intraocular pressure and glaucoma progression: results from the Early Manifest Glaucoma Trial. Arch Ophthalmol 2002;120:1268 –1279. 4. Anderson DR; Normal Tension Glaucoma Study. Collaborative normal tension glaucoma study. Curr Opin Ophthalmol 2003;14:86 –90. 5. Kass MA, Heuer DK, Higginbotham EJ, et al. The Ocular Hypertension Treatment Study: a randomized trial determines that topical ocular hypotensive medication delays or prevents the onset of primary open-angle glaucoma. Arch Ophthalmol 2002;120:701–713. 6. Gurwitz JH, Glynn RJ, Monane M, et al. Treatment for glaucoma: adherence by the elderly. Am J Public Health 1993;83:711–716. 7. Dasgupta S, Oates V, Bookhart BK, Vaziri B, Schwartz GF, Mozaffari E. Population-based persistency rates for topical glaucoma medications measured with pharmacy claims data. Am J Manag Care 2002;8(suppl 10):S255–S261. 8. Spooner JJ, Bullano MF, Ikeda LI, et al. Rates of discontinuation and change of glaucoma therapy in a managed care setting. Am J Manag Care 2002;8(suppl 10):S262–S270. 9. Shaya FT, Mullins CD, Wong W, Cho J. Discontinuation rates of topical glaucoma medications in a managed care population. Am J Manag Care 2002;8(suppl 10):S271–S277. 10. Reardon G, Schwartz GF, Mozaffari E. Patient persistency with pharmacotherapy in the management of glaucoma. Eur J Ophthalmol 2003;13(suppl 4):S44 –S52. 11. Reardon G, Schwartz GF, Mozaffari E. Patient persistency with topical ocular hypotensive therapy in a managed care population. Am J Ophthalmol 2004;137(suppl 1):S3–S12. 12. Schwartz GF, Reardon G, Mozaffari E. Persistency with latanoprost or timolol in primary open-angle glaucoma suspects. Am J Ophthalmol 2004;137(suppl 1):S13–S16. 13. Diestelhorst M, Schaefer CP, Beusterien KM, et al. Persistency and clinical outcomes associated with latanoprost and beta-blocker monotherapy: evidence from a European retrospective cohort study. Eur J Ophthalmol 2003;13(suppl 4):S21–S29.
TOPICAL GLAUCOMA THERAPY
598.e8
18. Niazi MK, Raja N. Comparison of latanoprost and dorzolamide in the treatment of patients with open angle glaucoma. J Ayub Med Coll Abbottabad 2004;16:50 – 53. 19. Gurwitz JH, Yeomans SM, Glynn RJ, Lewis BE, Levin R, Avorn J. Patient noncompliance in the managed care setting: the case of medical therapy for glaucoma. Med Care 1998;36:357–369. 20. Reardon G, Schwartz GF, Mozaffari E. Patient persistency with ocular prostaglandin therapy: a population-based, retrospective study. Clin Ther 2003;25:1172–1185. 21. Parrish RK, Palmberg P, Sheu WP, for the XLT study group. A comparison of latanoprost, bimatoprost, and travoprost in patients with elevated intraocular pressure: a 12-week, randomized, masked-evaluator multicenter study. Am J Ophthalmol 2003;135:688 –703.
14. McNutt L-A, Wu C, Xue X, Hafner JP. Estimating the relative risk in cohort studies and clinical trials of common outcomes. Am J Epidemiol 2003;157:940 –943. 15. Hedman K, Alm A, Gross RL. Pooled-data analysis of three randomized, double-masked, six-month studies comparing intraocular pressure-reducing effects of latanoprost and timolol in patients with ocular hypertension. J Glaucoma 2003; 12:463– 465. 16. Cohen JS, Gross RL, Cheetham JK, VanDenburgh AM, Bernstein P, Whitcup SM. Two-year double-masked comparison of bimatoprost with timolol in patients with glaucoma or ocular hypertension. Surv Ophthalmol 2004;49(suppl 1):S45–S52. 17. Thomas R, Parikh R, Muliyil J, George R, Paul P, Abraham LM. Comparison between latanoprost and brimonidine efficacy and safety in Indian eyes. Indian J Ophthalmol 2003; 51:123–128.
598.e9
AMERICAN JOURNAL
OF
OPHTHALMOLOGY
OCTOBER 2005
Biosketch David S. Friedman, MD, MPH, is an associate professor at the Wilmer Eye Institute of Johns Hopkins University School of Medicine, Baltimore, Maryland, and associate professor in the Department of International Health at Johns Hopkins Bloomberg School of Public Health. He has received numerous honors and awards, including a Clinician-Scientist Award from the NEI, the RPB Robert E. McCormick Award in 2000, and the American Geriatrics Society Douglas Jahnigen Scholars Award in 2002.
VOL. 140, NO. 4
TOPICAL GLAUCOMA THERAPY
598.e10
Biosketch Beth L. Nordstrom, PhD, MPH, is an epidemiologist with the Epidemiology Division of Ingenix i3 Magnifi in Auburndale, Massachusetts. Her current work consists primarily of drug safety and outcomes research using health care claims data. In addition to various investigations into glaucoma treatment, Dr Nordstrom’s research has focused on the epidemiology of nicotine addiction and treatments for infectious diseases.
598.e11
AMERICAN JOURNAL
OF
OPHTHALMOLOGY
OCTOBER 2005