Memorandum

Differences in Antidepressant Treatment Continuation Associated with Escitalopram and alternative SSRIs 1 Mathematica Policy Research, Inc., Princeton NJ 2 HealthCore Inc., Wilmington DE 3 Department of Health Services Administration and Population Health, Georgetown University School of Nursing and Health Sciences, Washington DC 4 Forest Research Institute, Jersey City, NJ 5 Mathematica Policy Research, Inc. and Departments of Medicine and Psychiatry, Georgetown University School of Medicine, Washington DC Corresponding author: Dominick Esposito, Ph.D., Senior Researcher, Mathematica Policy Research, Inc., P.O. Box 2393, Princeton, NJ 08543-2393, Phone: 609-799-3535, Fax: 609-799-0005, email: desposito@mathematica-mpr.com.

Acknowledgments. This research was supported by a contract with Forest Research Institute,
Jersey City, NJ. Portions of the work presented here have been accepted for presentation the annual meetings of AcademyHealth, Orlando FL, June 3-5, 2007, and the New Clinical Drug Evaluation Unit, Boca Raton, FL, June 11-14, 2007 ABSTRACT
Introduction: Escitalopram has been associated with more rapid antidepressant action compared
with alternative medications, a finding that could result in better adherence to treatment. The objective of this study was to compare (adjusting for selection factors) the percentage of patients initiating treatment on escitalopram and those initiating treatment on one of three generic antidepressants (fluoxetine, citalopram, and paroxetine) that continue on initial treatment for at least 2 and 6 months after starting antidepressant medication. Methods: We used administrative claims data from the HealthCore Integrated Research
Database. Patients were included in the study sample if they had evidence of depression, filled a prescription for a study antidepressant (escitalopram, citalopram, fluoxetine, or paroxetine), and had at least six months of eligibility in their health plan before and after the date of the first prescription fill of a study antidepressant (the index date). Antidepressant therapy was categorized into four mutually-exclusive use patterns: continuation, discontinuation, switch, and augmentation based on prescriptions filled after the index date. To account for potential selection bias, logistic regression was used to estimate a propensity score for initial antidepressant choice (escitalopram versus other antidepressants). The propensity score was then used as a weight in series of second stage logistic regressions that examined the association between antidepressant choice and subsequent treatment patterns. Results: In both the two- and six-month use pattern analyses, patients who initiated
antidepressant therapy with escitalopram were more likely (30% [95% CI: 24% to 35%] and 42% [95% CI: 34% to 50%], respectively) to continue initial antidepressant treatment. Among patients who discontinue, switch, or augment their initial therapy, patients initiating with escitalopram were more likely to discontinue than switch or augment (60% [95% CI: 47% to 75%] at two months and 79% [95% CI: 62% to 97%] at six months). Conclusions: Consistent with the hypothesis that escitalopram is associated with more rapid
onset of action, we find that patients who initiate treatment with escitalopram are more likely to continue treatment and, given changes in treatment, are less likely to switch or augment their antidepressant treatment when compared to those who initiate treatment on alternative, generic INTRODUCTION
Although selective serotonin reuptake inhibitors (SSRIs) are sometimes considered to be similar, there are several characteristics that distinguish one SSRI from another, such as reuptake inhibition potency,time to response time, degree of selectivity and specificity allowing for side effects (e.g., daytime sedation, insomnia, agitation, loss of libido), potential for drug-drug interactions, elimination half-life, and antidepressant activity of metabolites (1). Until recently, these pharmacological differences (not all the above cited differences were pharmacological, that’s why I deleted the clinical ones.) have not translated into differences in clinical outcomes (2,3) or length of therapy (4,5), leading one review to conclude that currently available evidence is insufficient to predict response to any one agent (6). Recent randomized clinical studies comparing escitalopram, the S-enantiomer of citalopram, with other SSRIs, however, suggests meaningful improvements in outcomes associated with the choice of escitalopram. In particular, several studies demonstrate that the onset of action of escitalopram is more rapid than that of alternative SSRIs (7,8). Conceptually, a more rapid response to treatment could be associated with higher rates of response and remission, a conclusion reached by some (8, 9,) but not all (11,12) studies that compare escitalopram with other SSRIs, including citalopram. I don’t think Moore is a good reference as this study showed a higher rate of response/remission, yet NOT associated with a more rapid onset of More rapid onset of antidepressant effect could also be associated with better adherence to treatment and longer duration on therapy. The study presented here examines antidepressant use patterns of patients who initiate therapy with SSRIs. The primary research objective is to compare (adjusting for selection factors) the percentage of patients initiating treatment on escitalopram and those initiating treatment on one of three generic antidepressants (fluoxetine, citalopram, and paroxetine) that continue on initial treatment for at least 2 and 6 months after starting antidepressant medication. We also examine the association of escitalopram with treatment patterns among patients who change treatment. We hypothesize that compared to patients initiating treatment on citalopram, fluoxetine, or paroxetine, patients initiating treatment with escitalopram are more likely to continue on their initial antidepressant after two and six months of treatment. We also hypothesize that patients initiating treatment with escitalopram who subsequently change treatment are less likely to switch antidepressants or add a second antidepressant, consistent with the notion that escitalopram acts faster than other generic antidepressants. Data & Sample
This study used administrative claims data from the HealthCore Integrated Research Database. As of July 2006, this Database included 13.8 million lives with linked medical, pharmacy, and eligibility files beginning in January 2000 from geographically dispersed United States health plans in the southeastern, mid-Atlantic, central, and western regions of the United States. Common insurance arrangements, including health maintenance, point of service, preferred provider organizations, and indemnity plans, were represented. Patients' identities were masked throughout the study in a limited data set format, in accordance with the 1996 Health Insurance Portability and Accountability Act (HIPAA). Institutional review board approval was not necessary because there was no patient intervention and the limited data was utilized in accordance with appropriate data use agreements with the covered entities as defined in HIPAA. Patients were included in the study sample if they had evidence of depression (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9] codes: 296.2x, 296.3x, 300.4, or 311), filled a prescription for a study antidepressant (escitalopram, citalopram, fluoxetine, or paroxetine), and had at least six months of eligibility in their health plan before and after the date of the first prescription fill of a study antidepressant (the index date). These eligibility criteria required a 6-month period prior to the index date during which there was no evidence of antidepressant treatment. Since escitalopram first became available in June 2002, only patients who initiated antidepressant therapy between June 2002 and April 2005 are included in the study sample. We selected citalopram, fluoxetine, and paroxetine as the comparator drugs because each was available in generic form during the study period. Patients are excluded from the study sample if they had evidence of bipolar or psychotic disorders (ICD-9 codes: 295.x, 296.0, 296.4x, 296.5x, 296.6x, 296.7x, or 296.8x), had pharmacy claims for lithium or antipsychotic medications, or had pharmacy claims for anticonvulsant medications without an indication of a seizure disorder (ICD-9 codes: 345.xx or 780.3x). Antidepressant Use Patterns
Antidepressant therapy was categorized into four mutually-exclusive use patterns: continuation, discontinuation, switch, and augmentation based on counts of prescription refills in the first 2 and 6 months following the index prescription. We used these timeframes to examine patients’ early response and later response to therapy. We excluded patients with an index fill with more than 45 days supply (usually a 90 day supply) because there would be no anticipated refills in the first two months, in which case all patients would be classified as discontinuers. More than 96 percent of patients had an index fill with a days-supply of 45 days or less. Two-month use patterns are calculated for all study sample members using claims in the 60 days following the index date. Continuation at two months is defined as filling at least one additional prescription of the initial antidepressant with no fills of other study antidepressants. Therapy is augmented at two months if a patient had additional fills of the initial antidepressant and fills for another antidepressant. Switching at two months is defined as filling at least one prescription for a different antidepressant medication and having no additional fills of the initial antidepressant. Therapy is discontinued at two months if a patient had no antidepressant claims Six-month use patterns are measured in the 180 days after the index date for study sample members who continued therapy with their initial antidepressant in the first two months after initiation. Continuation at six months is defined as filling at least three additional prescriptions (index plus three fills) of the initial antidepressant with no fills of other antidepressants. Therapy is augmented at six months if a patient had at least three additional fills of the initial antidepressant and at least one concomittant fill for another antidepressant. Switching at six months is defined as filling at least one prescription for a different antidepressant medication and having no additional fills of the initial antidepressant after the first 60 days of therapy. Therapy is discontinued at six months if a patient had no more than two additional claims for the initial antidepressant and no other claims for another study antidepressant after the therapy index date Regression analysis
In any retrospective claims-based analysis of patient behavior, there is always the potential for selection bias. In this study, for example, patients who are more likely to continue therapy with any medication might simply be more likely to initiate therapy with escitalopram compared with other antidepressants. To account for potential selection bias, we used a two-stage procedure to examine the association between initial antidepressant choice (escitalopram versus other antidepressants) to subsequent use patterns two and six months after initiation. First, the propensity to initiate with escitalopram versus other antidepressants was estimated with a logistic regression. The predicted propensity to initiate with escitalopram versus a generically available antidepressant was then used as a weighting variable in the second stage of the model (13, 14). Two second-stage analyses were conducted to investigate the association between choice of initial antidepressant and subsequent use patterns. First, the likelihood of continuing therapy compared to changing treatment (discontinuing, augmenting, or switching) was estimated with a logistic regression. Second, among patients who did not continue therapy, the likelihood of therapy discontinuation compared with switching or augmentation was estimated, also with a Independent variables in the first stage model included demographic characteristics, indicators of disease severity, and information on patient utilization of medical services and prescription drugs. Demographic characteristics included age, gender, geographic region, year of therapy initiation, and type of insurance coverage. Disease severity variables included type of depression (major depression, single episode; major depression, recurrent episode; dysthymia; and depressive disorder, not otherwise specified), type of comorbid anxiety (when present), and measures of comorbid medical conditions prior to therapy initiation (Deyo-Charlson comorbidity index). Health care utilization was assessed in the six months before therapy initiation. These variables included a binary measure for having a psychiatric encounter (inpatient or outpatient), a count of the number of distinct medications filled in the six-month pre-period (unique number of generic product identifier codes), an indicator of having a drug claim for a benzodiazepine, and total patient cost-sharing for all psychotropic medications. Independent variables used for the second stage regressions included a binary variable indicating initiation with escitalopram or other antidepressants, the patient’s copayment at initiation, an indicator for having a physician office visit after initiation, and a variable indicating whether or not the patient had a psychiatric encounter (inpatient or outpatient) after initiation of index treatment. More than 500,000 patients filled a prescription for one of the study antidepressants and satisfied eligibility criteria before and after the index prescription. Among these patients, about 67,000 (I suggest the authors account for the reduction in sample from 500.000 to 67.000. As a reader you wonder why all theese patients did not qualify for the sample. (1st line result section)). also had (1) a diagnosis of depression, (2) no prescriptions in the six months prior to the index date of an antidepressant, antipsychotic, or mood stabilizer, and (3) no diagnosis of bipolar disorder or a psychotic disorder. Of the remaining patients, the 43,921 with an index date on or after June 1, 2002 (escitalopram’s launch date) to April 2005 made up the final research sample. On average, sample members are about 40 years old and slightly more than 30 percent were male (Table 1). The predominant insurance types are HMO (57.8 percent) and PPO (28.1 percent) and nearly half of the sample came from one plan in a western state. The most common type of depression diagnosis is depressive disorder not otherwise specified and about a quarter of the sample also had evidence of an anxiety disorder. About 20 percent of the sample had claims for a benzodiazepine before antidepressant initiation while the average number of unique prescription drugs used before initiation of antidepressant therapy was three. Forty-five percent of the sample initiated antidepressant therapy with escitalopram and 55 percent initiated with a generically available antidepressant. Does this ratio (45%ESC) vs 55% generic CIT, PAR, FLU) correspond to the US market ; in order to evaluate if this sample is representative ? Among escitalopram initiators, 66.1percent continued escitalopram therapy within two months of initiation compared with 61.9 percent of patients initiating with other antidepressants (Table 2). Sample members who did not initiate therapy with escitalopram are more likely to discontinue, switch, or augment antidepressant therapy than escitalopram users in the two months after initiation (38.1 percent compared with 33.9 percent). This is the same as the prior sentence flipped. Why not differentiate between switch+augment therapy vs discontinue therapy as done below for the 6 months analysis? Antidepressant use patterns in the six months after initiation are similar to two-month use patterns. Patients initiating with escitalopram are more likely to continue therapy after six months than patients initiating with other antidepressants (71.2 percent versus 66.3 percent). Compared with those initiating treatment on other antidepressants, escitalopram initiators are less likely to switch or augment therapy by six months after index (7.1 percent versus 11.0 percent), while about the same proportion discontinue therapy by six months (21.8 percent versus 22.7 Propensity Score Model
To account for potential selection effects based on the choice of initial antidepressant, we estimated a logit model of the probability of initiating therapy with escitalopram or other antidepressants. These estimated probabilities are used as weights in the second stage logit regressions. Several observable variables were statistically significant predictors of being prescribed escitalopram, including age (36 to 55, 56 to 64), index year, having a psychiatric office visit in the six months prior to therapy initiation, some types of depression and anxiety, geographic region, health plan line of business (WHAT IS THIS?),, use of benzodiazepines before therapy initiation, and total psychotropic drug costs before therapy initiation (data not Continuation of Initial Antidepressant Therapy
In both the two- and six-month use pattern analyses, patients who initiate antidepressant therapy with escitalopram are more likely to continue initial antidepressant treatment (Table 3). In the two-month analysis, patients initiating with escitalopram are 30 percent more likely to continue. This figure is based on the Odds ratio, which can be questioned here: continuation rates were globally high in both groups; Odds ratio is clearly (too much ??) favouring escitalopram as compared to a RR ( “relative likelihood); is this not what overselling ? In the last sentence the OR=1.30 is interpreted as a relative risk of 1.30. The OR and RR are almost identical for rare events but here over 60% continue therapy. So it may be more intuitive for non-statisticians, but in this case misleading, to present the OR’s in the text as relative risks (this interpretation is continued below throughout the paper).Patients are 5 percent (p = 0.035) more likely to continue with their initial antidepressant therapy if they have an office visit after their index date. Patient cost-sharing for their initial antidepressant has a small, but significant, influence on their subsequent use pattern. The larger their cost-share, the less likely patients are to continue on their initial antidepressant (p < 0.01). (Though the estimated odds ratio was 1.00, the marginal effect of an increase in copayment was statistically significant and negative.) In the analysis of six-month use patterns, patients initiating with escitalopram are 42 percent (p < 0.01) more likely to continue with their initial therapy. Sample members who had a psychiatric encounter (either inpatient or outpatient) in the six months after therapy initiation were 10 percent less likely (p < 0.01) to continue with their initial therapy. Patient cost-sharing for their initial antidepressant again has a small but statistically significant influence on 6-month use pattern. The larger their cost-share, the less likely patients are to continue on their initial Changes of Initial Antidepressant Therapy
Among patients who discontinue, switch, or augment their initial therapy, patients initiating with escitalopram are more likely to discontinue but less likely to switch or augment when compared to those who initiate treatment on other antidepressants (Table 4). At two months, patients initiating with escitalopram are 60 percent more likely to discontinue therapy (p<0.01) and at six months they are 79 percent more likely to discontinue (p<0.01). these figures ( likely correct if derived from the Odds ratio) do give a false impression of very important difference. Cfr remark hereabove: is it relevant to use the Odds ratio ?Other factors also influence the likelihood of discontinuing treatment or switching/augmenting at both two and six months among those who did not continue with their initial therapy. Having a physician’s office visit is associated with a 50 percent reduction in the likelihood of discontinuation compared with switching or augmenting antidepressant therapy (p<0.01). Higher patient cost sharing for their initial antidepressant is also associated with a decreased likelihood of discontinuation (p<0.01). (Though the estimated odds ratio was 1.00, the marginal effect of an increase in copayment was We conducted a number of sensitivity analyses to examine the robustness of the primary findings, such as excluding all patients with initial fills in 2002 since this was the launch year of escitalopram and the proportion of escitalopram initiators in that year was much larger than other antidepressant initiators. Findings did not differ qualitatively or quantitatively when we excluded patients who did not initiate antidepressant therapy until 2003 or later (in any analysis). DISCUSSION
There are several possible reasons why patients prematurely discontinue antidepressant treatment, including inadequate response, side effects, and resolution of symptoms. Moreover, whether operating through cognitive deficits, feelings of hopelessness, or other factors, depression itself directly affects the likelihood of adherence to treatment of all types (15), and it follows that treatments associated with faster resolution of symptoms would also be associated with better adherence. Our results are consistent with this hypothesis, finding that escitalopram, an antidepressant associated with faster response rates among patients with depression, is also associated with higher rates of continuation. When compared to those receiving alternative SSRIs, study sample members who initiated with escitalopram were 30 percent more likely to continue with their initial therapy after two months, and they were 42 percent more likely to There are other possible explanations for our findings. In several studies, 35 percent to 55 percent of patients who prematurely discontinue antidepressants reported doing so because their symptoms are resolved or they “feel better” (16, 17), and in one study, feeling better was confirmed by improvement in functioning (16). Thus, it is possible that higher rates of discontinuation associated with use of alternative antidepressants were the result of symptom resolution. This conclusion would not be consistent with the existing pharmacodynamic studies, but more research is needed to clarify the reasons for continuation and discontinuation. Moreover, continuation of treatment has benefits beyond the point of symptom resolution, especially reductions in the probability of relapse or recurrence (18-20). Our models of use patterns among those who changed treatment (i.e., those who discontinued, switched, or augmented) are also consistent with the hypothesis that escitalopram is associated with a more rapid antidepressant response and symptom resolution. The appropriate clinical response to continuing symptoms or side effects is a change in treatment, either switching antidepressants in the case of side effects, or switching or augmenting in the case of treatment failure. We hypothesize that escitalopram users are more likely to discontinue and less likely to switch or augment because they have higher response rates and fewer side effects when compared with the alternative antidepressants studied here. Comparative studies will be required to determine whether this hypothesis is correct. We also find that factors other than choice of antidepressant have a significant influence on use pattern. Sample members who had any office visit in the two months after initiation were also more likely to remain on their initial therapy, suggesting that follow-up consultations with a physician are an important part of depression therapy management. Consistent with recent research in other therapeutic areas (21-23), we also find that the level of cost sharing to the patient was also a driver of antidepressant use patterns. In general, patients with higher initial copayments were slightly less likely to continue use of their initial antidepressant. We also found, perhaps counter-intuitively, in the six-month analysis, that sample members with a psychiatric office visit after therapy initiation were less likely to continue with their initial antidepressant therapy. However, there are several possible explanations for this phenomenon. Most important among these, those who seek psychiatric specialty care are more likely to do so because of failure to respond to initial treatment or side effects, both of which are indications for As in any comparison of treatment effects that relies on observational data, our study provides only limited opportunity to make causal inferences. While we use propensity scores to mitigate this limitation, unobserved characteristics could still result in an imperfect balance between the escitalopram and other antidepressant cohorts. An additional limitation is that the diagnosis of a depressive or anxiety disorder on insurance claims is imperfect at best; patients with these disorders that are not coded in claims data but are treated with antidepressants were excluded from our research sample. Despite these limitations, our study contributes to the literature in several ways. The mix of available antidepressants has changed considerably, with new branded drugs coming available while older SSRIs becoming available in generic form. Our study suggests that there is continued clinical value in these new antidepressants, even when those new drugs are very similar to existing drugs. However, there is much to be learned about the reasons for the differences in use patterns observed here. This study also adds to our understanding of comparative effective methods using observational data from insurance claims. To our knowledge, this is the first such study that uses the propensity score as weight rather than a Our use of changes in treatment at two and six months is also a novel approach to the more static measures of use pattern that have used in past studies. Our goals were to more accurately reflect current treatment standards that recommend changes in treatment after 6 to 8 weeks when there is an inadequate response to treatment and provide information corresponding to current National Center for Quality Assurance criteria for appropriate depression care requiring 6 months of continuous care. We also wanted to recognize the fact that side effect profiles differ over time (e.g., nausea more likely early in treatment while sexual side effects are more noticeable later in the course of treatment), and that these differences may differentially influence the timing of premature discontinuation. Although we did not observe different switching patterns between the early and late periods, we find this new method conceptually appealing and straightforward to implement. CONCLUSION
Patients who initiated drug therapy with a branded antidepressant (escitalopram) are more likely to continue a monotherapy regimen than patients who initiated with an antidepressant available in generic form (fluoxetine, citalopram, or paroxetine). Among patients who did not continue on monotherapy with their initial antidepressant, escitalopram initiators are more likely to discontinue than switch or augment their therapy than patients who initiated with a generic REFERENCES
1. Preskorn SH. Clinically relevant pharmacology of selective serotonin reuptake inhibitors. An overview with emphasis on pharmacokinetics and effects of oxidative drug metabolism. Clinical Pharmacokinetics. 32 supplement 1: 1-21, 1998 2. Croghan TW, Sherbourne CD, Schoenbaum M, Koegel P. A Framework to Improve the Quality of Care for Depression in Primary Care. Psychiatric Services 57(5): 623-631, 3. Simon GE. Choosing a First-Line Antidepressant. JAMA 286: 3003-3004, 2001 4. Tai-Seale M, Croghan TW, and Obenchain RL. Determinants for Antidepressant Treatment Compliance: Implications for Policy. Medical Care Research and Review 5. Polsky D, Onesirosan P, Bauer MS, Glick HA. Duration of therapy and health care costs of fluoxetine, paroxetine, and sertraline in 6 health plans. Journal of Clinical Psychiatry 6. Gartlehner G, Hansen RA, Thieda P, DeVeaugh-Geiss AM, Gaynes BN, Krebs EE, Lux LJ, Morgan LC, Shumate JA, Monroe LG, Lohr KN. Comparative Effectiveness of Second-Generation Antidepressants in Pharmacologic Treatment of Adult Depression. Comparative Effectiveness Review No. 7. (Prepared by RTI International-University of North Carolina Evidence-based Practice Center under Contract No. 290-02-0016.) Rockville, MS: Agency for Healthcare Research and Quality. January 2007. Available at www.effectivehealthcare.ahrq.gov/reports/final.cfm. 7. Kasper S, Spadone C, Verpillat P, Angst J. Onset of action of escitalopram compared with other antidepressants: results of a pooled analysis. International Clinical 8. Wade A, Friis Andersen H. The onset of effect for escitalopram and its relevance for the clinical management of depression. Current Medical Opinion 22(11):2101-2010, 2006. 9. Colonna L, Andersen HF, Reines EH. A randomized, double blind, 24-week study of escitalopram (10 mg/day) versus citalopram (20 mg/day) in primary care patients with major depressive disorder. Current Medical Research Opinion 21(10): 1659-1568 10. Moore N, Verdoux H, Fantino B. Prospective, multicentre, randomized, double-blind study of the efficacy of escitalopram versus citalopram in outpatient treatment of major depressive disorder. International Clinical Psychopharmacology 11(2): 129-136, 2005. 11. Burke WJ, Gergel I, Bose A. Fixed-dose trial of the single isomer SSRI escitalopram in depressed outpatients. Journal of Clinical Psychiatry 63(4): 331-336, 2002 12. Kasper S, de Swart H, Friis Andersen H. Escitalopram in the treatment of depressed elderly patients. American Journal of Geriatric Psychiatry 13(10): 884-891, 2005 13. Hirano K, Imbens G. Estimation of causal effects using propensity score weighting: An application to data on right heart catheterization. Health Services and Outcomes Research 14. Rosenbaum P. Model-based direct adjustment. Journal of the American Statistical 15. DiMatteo MR, Lepper HS, Croghan TW: Depression is a risk factor for noncompliance with medical treatment: results of a meta-analysis of effects of anxiety and depression on patient adherence. Archives of Internal Medicine 160:2101–2107, 2000 16. Demyttenaere K, Enzlin P, Dewe W, et al: Compliance with antidepressants in a primary care setting: 2. the influence of gender and type of impairment. Journal of Clinical 17. Linden MH, Gothe RW, Dittman B, et al: Early termination of antidepressant drug treatment. Journal of Clinical Psychopharmacology 20:523–530, 2000 18. Melfi C.A., A. Chawla, T.W. Croghan, M.P. Hanna, S. Kennedy, and K. Sredl. Effect of Adherence to Guidelines for Depression Treatment on Relapse and Recurrence in a Medicaid Population. Archives of General Psychiatry 55:1128-1132, 1998 19. Reimherr FW, Amsterdam JD, Quitkin FM, Rosenbaum JF, Fava M, Zajecka J, Beasley CM Jr, Michelson D, Roback P, Sundell K. Optimal length of continuation therapy in depression: a prospective assessment during long-term fluoxetine treatment. American Journal of Psychiatry 155(9):1247-53, 1998 20. Sood N, M. Treglia, R.L. Obenchain, B. Dulisse, C.A. Melfi, and T.W. Croghan. Determinants of Antidepressant Treatment Outcome. American Journal of Managed Care 21. Joyce, G.F., J.J. Escarce, M.D. Solomon, D.P. Goldman. Employer drug benefit plans and spending on prescription drugs. JAMA 288 (14): 1733-1739, 2002 22. Huskamp, HA, P.A., Deverka, A.M. Epstein, R.S. Epstein, K.A. McGuigan, R.G. Frank. The effect of incentive-based formularies on prescription-drug utilization and spending. New England Journal of Medicine 349(23): 2224-2232, 2003 23. Gibson T.B., T.L. Mark, K. Axelsen, O. Baser, D.A. Rublee, K.A. McGuigan. Impact of statin copayments on adherence and medical care utilization and expenditures. American Journal of Managed Care 12: SP11-SP19, 2006 (Percentages, Unless otherwise specified) Initiate with
Initiate with
Escitalopram
Other Total
Psychotropic Drug Out-of-Pocket Costs*** Number of Patients
Percent of Patients
1 Other includes Indemnity, Traditional, Fee-for-service, and other types of insurance. 2 Measured in the six-month pre-index period. ** p<0.05; ***p<0.01 ANTIDEPRESSANT USE PATTERNS IN THE TWO AND SIX MONTHS AFTER THERAPY INITIATION Two Months After Initiation***
Six Months After Initiation***
Initiated
with: Escitalopram Other
Total Escitalopram Other
Total 19,755
43,921 13,061 14,957 28,018
Percent 45.0
100.0 46.6 53.4 100.0
Note: Other antidepressants in the final research sample included fluoxetine, citalopram, and paroxetine. *** Distribution of use patterns between escitalopram initiators and other antidepressant initiators is statistically different from one another at the .01 level, chi-squared test. CONTINUATION OF ANTIDEPRESSANT THERAPY, REGRESSION ANALYSIS Odds Ratio
Confidence Interval
Explanatory Variables
Two-Month Use Pattern
Six-Month Use Pattern
Note: Dependent variable in logistic regression analysis was whether or not the sample member continued therapy with initial antidepressant within the two- or six-month period after therapy initiation. All regression analyses use propensity score weights. 1 The marginal effects on the index copayment variable are –0.0013 for the two-month period and –0.0026 for the six-month period. DISCONTINUATION OF ANTIDEPRESSANT THERAPY, REGRESSION ANALYSIS Odds Ratio
Confidence Interval
Explanatory Variables
Two-Month Use Pattern
Six-Month Use Pattern
Note: Research sample for these regressions was limited to those patients who discontinued, switched, or augmented initial drug therapy in either the two- or six-month followup period. Dependent variable in logistic regression analysis was whether or not the sample member discontinued therapy with initial antidepressant within the two- or six-month period after therapy initiation. All regression analyses use propensity score weights. 1 The marginal effects on the index copayment variable are –0.0015 for the two-month period and –0.0034 for the six-month period. P value ??

Source: http://www18.georgetown.edu/data/people/stotom/publication-26108.pdf