| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Faculties of Pharmacy (J.B., A.D., S.P.) and Medicine (L.-G.S.-M., J.C.F.), University of Montréal, Montréal, Québec, Canada H3C 3J7
Address all correspondence and requests for reprints to: Sylvie Perreault, Ph.D., P.O. Box 6128, Centre-Ville Station, Montréal, Québec, Canada H3C 3J7. E-mail: sylvie.perreault{at}umontreal.ca.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Objectives: We compared probability of discontinuation among women aged 70 yr or older with a diagnosis of osteoporosis or recent osteoporotic fracture having started ART (alendronate, risedronate, cyclical etidronate, raloxifene, nasal calcitonin) between 1998–2001 or 2002–2004.
Patients and Methods: We constructed two cohorts of women using Régie de l’Assurance Maladie du Québec databases. Discontinuation was defined as a lapse of 30 d or longer after completion of a refill. Switching from one ART to another was allowed. Probability of discontinuation was estimated using Kaplan-Meier analysis. Multivariate Cox models were used to identify potential determinants of ART discontinuation over 1 yr.
Results: After 1 yr, probability of discontinuation was slightly lower in the 2002–2004 cohort than the 1998–2001 cohort (52.2 vs. 57.5%; P < 0.001). This difference remained significant after adjusting for determinants [adjusted rate ratio (RR) 0.92, 95% confidence interval (CI) 0.87–0.98]. Significant determinants of ART discontinuation within 1 yr included bone mineral density testing (RR 0.77; CI 0.73–0.82) performed within 2 yr prior to initiation of therapy and having consulted more than two pharmacies (RR 1.15; CI 1.06–1.25) in the year before starting therapy. In the 2002–2004 cohort, when switching was allowed, women initiating a once-weekly regimen of alendronate or risedronate did not show a 1-yr risk of discontinuation different from women initiating daily regimens of the same drugs (RR 0.90; CI 0.82–1.00).
Conclusions: Even if new dosing regimens were introduced, discontinuation of ART among osteoporotic women remains high.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In the past few years, important changes have taken place in the pharmacotherapy of osteoporosis in Québec and around the world. First, once-daily alendronate was given provincial drug formulary status in January 1998. It was followed by raloxifene and nasal calcitonin, which were both given provincial drug formulary status in April 2000, and once-daily risedronate, as of January 2001. Then to improve tolerance and convenience, a once-weekly regimen of alendronate was listed in February 2002, followed by a once-weekly regimen of risedronate in June 2003. Furthermore, in July 2002, the Women’s Health Initiative (12) trial concluded that when used only for the prevention of postmenopausal osteoporosis, the risks of hormone replacement therapy may outweigh the benefits. Since 2002, a major decline in hormone replacement therapy and continued rise in bisphosphonate prescribing have been reported (13).
From a population-based study, we aimed to compare probability of discontinuation and nonadherence levels between women aged 70 yr or older having started ART for treatment of osteoporosis between 1998–2001 and 2002–2004. Our second objective was to identify potential determinants of discontinuation and nonadherence.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This cohort study used a 40% random sample of women aged 70 yr and over registered in the administrative databases of the Régie de l’Assurance Maladie du Québec (RAMQ), which administers public health care insurance programs in the province of Québec, Canada. RAMQ databases contain four types of files. First, the admissibility file lists date of the beginning and end of eligibility to the program. Second, the beneficiary file contains information about date of death, if such is the case. Third, the medical services file includes fee-for-service claims for inpatient and ambulatory medical services and contains demographic information (gender, age) as well as the nature of the medical act, date, site at which the act was performed (office, emergency room, hospital), and diagnostic code according to the International Classification of Diseases, ninth revision (ICD-9) classification (14). Codes for surgical procedures are assigned according to the Canadian classification of diagnostic, therapeutic, and surgical procedures (15). The medical services file provides information on services distributed to all residents of Québec. Fourth, the pharmaceutical file contains data on all drugs listed on the RAMQ drug formulary dispensed to community-living patients ensured by RAMQ for drugs. In 2004, about 94% of Québec citizens aged 65 yr and older were covered by the RAMQ drug plan (16). The pharmaceutical file includes the generic name, the drug identification number, strength, form, quantity, date, and duration of therapy, as indicated by the community pharmacist who filled the prescription. The RAMQ medical and pharmaceutical files have been validated for research and have been frequently used for pharmacoepidemiological studies (17, 18).
Because alendronate, risedronate, cyclical etidronate, raloxifene, and nasal calcitonin are listed on the regular section of the RAMQ drug formulary, no specific criteria must be met for their prescription, and the medications can be prescribed by any physician. Persons covered by the RAMQ drug plan must pay an annual premium, depending on net family income. In 2004 the annual premium varied from $0 to $460 Canadian dollars (16). In addition, each month they purchase prescription drugs, ensured persons pay a deductible and a coinsurance corresponding to 25% of the cost of drugs until they reach a maximum monthly contribution which, in 2004, varied from $16.66 to $69.92 Canadian dollars, regardless of the number of drugs received (16). RAMQ’s usual policy for prescriptions for chronic medications such as osteoporosis therapies is a 30-d initial prescription followed by monthly refills over a 6-month to 1-yr period. There was no difference in copayment or in refill practice between the 1998–2001 and 2002–2004 cohorts.
Cohort definition
An initial cohort of 62,533 women was identified from the RAMQ pharmaceutical file. To be eligible, women needed to have been newly treated with alendronate, risedronate, cyclical etidronate, raloxifene, or nasal calcitonin between January 1, 1998, and June 30, 2001 (1998–2001 cohort) or between January 1, 2002, and December 31, 2003 (2002–2004 cohort). The date of the first prescription of ART was defined as the index date. New users were women with neither ART nor hormone replacement therapy dispensed in the year preceding the index date. Women were required to be at least 70 yr of age at the time of index date and to have been continuously ensured by the RAMQ drug plan for at least 5 yr before the index date.
Because we were interested in secondary prevention of osteoporosis, we included women with a diagnosis of osteoporosis (ICD-9 code 733.0) or an osteoporotic fracture (ICD-9 or medical procedure code for a spine, rib, humerus, radius, cubitus, carpal-hand, pelvis, hip, femoral shaft, ankle, tarsus, or metatarsus fracture) within 2 yr before the index date. The estimated sensitivity of the use of either diagnostic or procedure code in RAMQ medical file to detect any type of fracture is 85.0% (19). Women receiving 5 mg alendronate daily were not included, considering the little evidence supporting the efficacy of this dosage in secondary prevention of osteoporosis.
We excluded women having suffered from polytraumatism (more than three fractures in a 5 d lapse). We also excluded women who suffered from other disorders of bone metabolism such as hyperthyroidism (ICD-9 code 242 or 245.1 or drug markers), primary hyperparathyroidism (ICD-9 code 252 or medical procedure codes for exploration and excision of the parathyroid), Paget’s disease (ICD-9 code 731 or drug markers), severe renal impairment (ICD-9 codes 581–589, medical procedure codes for hemodialysis or peritoneal dialysis, or drug markers), malignant cancer (ICD-9 codes 140–159, 162, 170, 174, 188, 189, 193, 199, 203), celiac disease (ICD-9 code 579.0), Cushing’s disease (ICD-9 code 255.0 or a medical procedure code for adrenalectomy), or solid organ transplant (medical procedure code for a lung, cardiac, hepatic, or kidney transplant) in the 5 yr preceding the index date (20).
In addition, we excluded women taking medications that may cause secondary osteoporosis such as chronic use of oral glucocorticoids (
5 mg of equivalent prednisone per day during at least 3 months), continuous use of anticonvulsants (phenytoin or phenobarbital), standard or low-molecular-weight heparin, or immunosuppressive therapy (cyclosporine, or tacrolimus) during the year preceding the index date.
Finally, we excluded women with congestive heart failure (ICD-9 code 428 or drug markers) or dementia (ICD-9 codes 290, 294, 331 or drug markers) in the 5 yr preceding the index date because of the limited life expectancy associated with these diseases.
The two final cohorts consisted of 6032 women for the 1998–2001 cohort and 4130 women for the 2002–2004 cohort. For both cohorts, subjects were followed up for a maximum of 18 months or until the end of the study period, which was December 31, 2001, for the first cohort or June 30, 2004, for the second cohort or occurrence of an exclusion criterion, admission in a long-term public health care institution, loss of eligibility to the RAMQ drug plan, or death. The study protocol was approved by the University of Montréal’s Research and Ethics Committee and the Commission d’Accès à l’Information du Québec.
Assessment of probability of discontinuation and nonadherence level
We developed an iterative model that used data on dispensing date, amount of medication supplied, prescription duration, and type of ART (21). Discontinuation was defined as having no ART dispensed within 30 d after the end date of a previous prescription for any type of ART. The discontinuation date was set at half the last prescription’s length of supply. Women who switched from one ART to another within 30 d were still considered persistent with nonexclusive use. Concomitant use of different ART was allowed. In sensitivity analyses, we examined the effect of using 45- and 60-d grace periods on probability of discontinuation. We also assessed 1-yr probability of discontinuation without allowing switching from the initial ART (exclusive use).
Nonadherence level was evaluated using a medication possession ratio (MPR) (21). MPR was defined as the percentage of days of ART supply during the first year of follow-up. MPR were calculated for women having had at least 1 yr of follow-up, regardless of whether they had discontinued therapy. A woman was considered nonadherent if she had a MPR less than 80%. This threshold was chosen because it had been frequently used in other studies (9, 10, 22, 23).
Determinants
We assessed several potential determinants of discontinuation and nonadherence with ART: age at the index date, BMD testing, or osteoporotic fracture within 2 yr before the index date as well as risk of fall, number of different therapeutic drug classes, use of antidepressants or anxiolytics, and health care services use in the year before index date. The risk of fall was defined as having an ICD-9 code or taking drugs for Parkinson’s disease; having an ICD-9 code for orthostatic hypotension, peripheral neuropathy, functional limitation such as paralysis, epilepsy, blindness, or stroke; or having a history of fall. The number of different therapeutic drug classes was defined according to the American Hospital Formulary Service classification. Health care services use included the number of pharmacies, physicians, outpatient medical visits, and any hospitalization. For the number of different therapeutic drug classes and health care services use, we compared probability of discontinuation and nonadherence among high users (those in the fourth quartile of the distribution) and others.
Statistical analysis
Women’s characteristics at baseline were compared using 
2 tests, Student t tests, and Wilcoxon rank sum tests. Probability of discontinuation was estimated using Kaplan-Meier failure time analysis. Multivariable stepwise Cox regression and logistic regression analyses were used to evaluate determinants of discontinuation and nonadherence at 1 yr, respectively. We used a criterion of P = 0.25 for entry and P = 0.15 for retention of a variable in the final models. Generalized R2 statistic was computed to evaluate how strongly the determinants selected in the final models were related with the risk of discontinuation and nonadherence. Ranging from 0 to 1, the R2 statistic is closer to 1 when covariates are strongly associated with the dependent variable (24). Statistical significance was defined at an alpha level of 0.05. All analyses were performed using SAS software, version 9.1.3 (SAS Institute Inc., Cary, NC).
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
As shown in Fig. 1
, a total of 6032 women for the 1998–2001 cohort and 4130 women for the 2002–2004 cohort were identified as being newly treated with ART for secondary prevention of osteoporosis. Characteristics of women at baseline in each cohort are presented in Table 1. For both cohorts, most of the women were included on basis of osteoporosis diagnosis alone. In the 1998–2001 cohort, 24.6% of women had sustained an osteoporotic fracture in the past 2 yr, compared with 19.7% in the 2002–2004 cohort (P < 0.001). The patterns of use of ART varied significantly between the two cohorts. The use of alendronate and risedronate rose from 60.7% in the 1998–2001 cohort to 84.2% in the 2002–2004 cohort (P < 0.001). Once-weekly regimens of alendronate and risedronate, not available in the 1998–2001 cohort, accounted for 58% of initial prescriptions of ART in the 2002–2004 cohort.
|
|
Women characteristics at baseline according to the initial ART are presented in Table 2. In the 1998–2001 cohort, compared with those having initiated other types of ART, women for whom the initial ART was nasal calcitonin were older and were 40% more likely to have sustained osteoporotic fractures but about 30% less likely to have had BMD testing in the past 2 yr. In the 2002–2004 cohort, compared with others, women having initiated ART with nasal calcitonin were also older, about 60% more likely to have sustained osteoporotic fractures, and about 50% less likely to have had BMD testing in the past 2 yr.
|
Cumulative probability of discontinuation of ART increased rapidly during the first year of treatment for the two cohorts (Fig. 2). Kaplan-Meier analysis showed that the unadjusted probability of discontinuation was lower in the 2002–2004 cohort than the 1998–2001 cohort (P < 0.001). After 6 months, 45.8% of women in the 1998–2001 cohort had discontinued ART, compared with 39.4% in the 2002–2004 cohort. After 1 yr, probabilities were 57.5 and 52.2%, respectively. Cumulative probability of discontinuation tended to stabilize over time and reached 64.3% (1998–2001 cohort) and 60.5% (2002–2004 cohort) after 18 months. Sensitivity analyses using different grace periods showed similar patterns of discontinuation between 12 and 18 months. After 18 months, using gaps of 45 or 60 d, probabilities of discontinuation were 57.3 and 53.1%, respectively, for the 1998–2001 cohort, compared with 54.5 and 51.6%, respectively, for the 2002–2004 cohort.
|
. Not allowing switching increased the overall probability of discontinuation by 3.3% in the 1998–2001 cohort and 8.8% in the 2002–2004 cohort. Most switches in the 2002–2004 cohort were due to women changing from a daily regimen of alendronate or risedronate to a weekly regimen of the same drug. Indeed, among women who switched from alendronate once-daily to another ART, 91.4% of them switched to once-weekly alendronate. Among women who switched from risedronate once-daily, 75.4% of them switched to once-weekly risedronate.
|
, after controlling for determinants, women initiating ART in the 2002–2004 period still showed a slightly lower rate ratio (RR) of stopping therapy (RR 0.92; 95% confidence interval 0.87–0.98) than women in the 1998–2001 cohort.
|
In the 1998–2001 cohort, compared with those whom initial ART was once-daily alendronate or risedronate, women initiating ART with nasal calcitonin had a 2.23 higher risk to stop therapy, whereas no statistically significant difference was found for women initiating ART with raloxifene or cyclical etidronate (Table 4
, model 1). In the 2002–2004 cohort, compared with those whom initial ART was once-daily alendronate or risedronate, women initiating ART with cyclical etidronate had a 1.34 higher risk to stop therapy; those initiating with nasal calcitonin had a 1.97 higher risk, whereas no statistically significant difference was found for women initiating ART with once-weekly alendronate or risedronate or raloxifene (Table 4, model 2).
Other determinants included in the final Cox models yielded similar RR for both periods. Considering both cohorts together (Table 4, model 3), BMD testing performed within 2 yr before index date was associated with a 23% lower risk of discontinuation. Having used more than two dispensing pharmacies in the year before index date was associated with a 15% higher risk of discontinuation. Age on the index date, risk of fall, and number of outpatient medical visits were not related to the risk of stopping ART. Even if some of them were statistically significant, in concert, determinants included in the final models were weakly associated with discontinuation (generalized R2 statistic varied from 0.012 to 0.036%).
Nonadherence
Unadjusted 2 test showed that the percentage of women with a MPR less than 80% was slightly lower in the 2002–2004 cohort, compared with the 1998–2001 cohort (39.2 vs. 42.3%, P = 0.010), respectively. However, after controlling for potential determinants, this difference was no longer significant (odds ratio 0.93; 95% CI 0.84–1.03).
Determinants of nonadherence
Determinants included in the final stepwise logistic regression models showed similar associations with those observed in the Cox models. Considering both cohorts together, BMD testing performed within 2 yr before index date was associated with a 33% lower risk of nonadherence. Age on the index date, a risk of fall, and having used more than two dispensing pharmacies in the year before index date were not related to the risk of nonadherence. Even if some of them were statistically significant, determinants included in the final logistic regression model were weakly associated with nonadhesion (R2 for the model including both cohorts together was 0.01%).
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Compared with once-daily regimens, initiating ART with once-weekly regimens of alendronate and risedronate was not associated with a significantly lower probability of discontinuation. Women initiating raloxifene had a similar risk to stop their therapy than women initiating once-daily alendronate or risedronate. We have shown that women initiating ART with nasal calcitonin tended to be frailer and had the highest probability of discontinuation. It is possible that, in some cases, nasal calcitonin was used for analgesia instead of the treatment of osteoporosis (25).
Other studies having evaluated discontinuation and nonadherence with ART since the introduction of once-weekly regimens of bisphosphonates using administrative databases showed varying results according to their definition of end of continuation, the grace period or adherence threshold used, and whether switching to another drug was allowed (22, 23, 26, 27). Lo et al. (26) found that using 30- or 120-d gaps in the same cohort of women yielded to a 16% variation in the probability of discontinuation, compared with a 60-d gap.
Among studies having used a grace period of 30 d, Penning-van Beest et al. (27) found a 1-yr overall probability of discontinuation among new users of bisphosphonates (alendronate, etidronate, and risedronate) of 57.1% for nonexclusive use and 63.5% for exclusive use. These results are comparable with ours for both nonexclusive and exclusive use. Cramer et al. (23) reported 1-yr discontinuation probability with exclusive use of 55.8% for once-weekly alendronate and 68.3% for once-daily alendronate and risedronate. This difference with our results for once-daily regimens can be explained by the fact that they excluded women who switched from the analyses. Both Penning-van Beest et al. (27) and Cramer et al. (23) found in multivariate analyses that women on weekly alendronate were less likely to discontinue therapy, compared with those on daily regimen of alendronate or risedronate. However, these studies did not include women on once-weekly risedronate, which showed a higher 1-yr probability of discontinuation then once-weekly alendronate in our cohort. Moreover, we have shown that evaluating discontinuation with only exclusive use overestimates probability of discontinuation among women initiating once-daily bisphosphonates because a considerable percentage of women on these drugs will switch to a once-weekly regimen of the same drugs. Lo et al. (26) also reported that a considerable subset of women restarted bisphosphonate or other osteoporosis therapy within 6 months of presumed discontinuation.
Our nonadherence results differ from those of two other studies (22, 23), although we used the same MPR threshold of 80% over a 1-yr period. Cramer et al. (23) found that 44.7% of women on weekly bisphosphonates had a MPR less than 80%, compared with 59.3% of women on daily regimens. Recker et al. (22) reported that 74.8% of women starting on weekly bisphosphonates and 86.8% of those starting on daily bisphosphonates had a MPR less than 80%. However, these studies assessed only nonadherence with exclusive use instead of allowing switching as we did.
Similar to other studies (26, 28), we identified determinants that, even if statistically significant, appeared to be weakly associated with the risk of discontinuation or nonadherence. However, we believe that these determinants can provide insights into implementing strategies to reduce discontinuation and improve adherence with ART. Our results confirm that BMD testing performed before initiation of therapy is associated with a lower risk to stop therapy and to be nonadherent (26, 28, 29). We found that women consulting the highest number of different pharmacies had a higher risk of discontinuing therapy. This highlights the importance of promoting continuity of care and health care professional-patient communication. Studies have reported that a better monitoring of patients by health care professionals could enhance adherence and persistence to ART (30) and that patients who had received a good education about their disease were less likely to discontinue treatment (31).
Our study had some limitations. First, there was a lack of some clinical data, especially concerning BMD findings. Also, RAMQ databases did not allow us to assess the reasons for ART discontinuation. Thus, we could not know whether ART was discontinued following advice from physicians. However, results from a large survey carried out in 9851 postmenopausal women indicated that among women having stopped ART, less than 10% reported that it was after recommendation of a physician (32).
Also worth mentioning, we used different markers to identify conditions that might have been miscoded or simply not recorded in the RAMQ databases. Finally, the assessment of discontinuation and nonadherence was based on supplies of medication, which is an indirect measure of medication-taking behavior. However, although they only approximate medication-taking behavior, administrative databases offer many advantages over other means of data collection such as patient self-reported measures, mainly because they avoid the problem of reporting bias (33). Moreover, a systematic review showed that administrative databases are particularly suited for the evaluation of drugs intended for long-term therapy (34).
In summary, we have shown that after 18 months, in both cohorts, about two thirds of women who had initiated osteoporosis therapies had stopped. Clinicians need to be aware that even after the important changes that have taken place in the pharmacotherapy of osteoporosis, including the introduction of once-weekly regimens of bisphosphonates, discontinuation and nonadherence with ART remain high. Further research is needed to better understand determinants of discontinuation and nonadherence with ART.
|
Footnotes |
|---|
The authors have nothing to disclose.
First Published Online January 2, 2007
Abbreviations: ART, Antiresorptive therapies; BMD, bone mineral density; ICD-9, International Classification of Diseases, ninth revision; MPR, medication possession ratio; RAMQ, Régie de l’Assurance Maladie du Québec; RR, rate ratio.
Received October 24, 2006.
Accepted December 21, 2006.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
