SAT-5 Lilly: Advancing CNS research into the twenty-first centu~' 257

Design and Methodology Patients assessed as depressed by their Dr, mary care physician and who were candidates for a new antidepressant prescription were presented with the option of participating in the study. A study coordinator assessed patients for eligibihty and completed a base line assessment of depression severity and functional ~mpairment. Patients were stratified according to the presence/aosence of current DSM-IIIR ma- jor depression by structured interwew and randomized to initial treatment with imi, dml or fix using computer-generated random numbers. All patients received a standard patient education package concerning antidepressant treatment at the t ime of the initial prescnption fill. Data collection for each patient will continue for two years from the t ime of randomization. The study will conclude in June, 1996 Naturalistpc follow-up care has been pro- vided by the referring physician who has made all decisions concermng antidepressant dosage, duration of treatment, change in antidepressant drug and referral for specialty care

A research assistant has conducted or will conduct follow-up evaluations by telephone at intervals of 1,3, 6.9, 12, 18 and 24 months. A t the time ol study entry, study participants were informed that follow-up assessments were completely independent of clinical management and that assessment would continue for two years regardless of subsequent antidepressant use Follow-up evaluations consists of the SCID depresspon module, the HAMD 17, the SCL-90 depression and anxiet), subscales, the SF-36. an assessment of days of disability and restricted acbvity, an antidepressant quest~onnaqre to assess current antidepressant use/reasons for discontinuation/s~de effect severity, a utilization questionnaire to assess non-managed health care service utilization and an assessment of global improvement. Additlona! data available via automated data systems includes medication use, medlcai service utilization, and direct medical costs

Outcomes. The primary clinical outcome is the HAMDI7. The pnmar~,. functional outcome is the SF-36 Emotional Functioning score with sec ondary functional outcomes to include days of d~sability and days of re stricted activity The primary cost outcome is the total direct medical costs with additional analyses of indirect costs and total direct and indirect costs Utilization outcomes include the total number of outpatient visits menta health/substance abuse wslts, total number of inpatient days, and ores ence/absence of any inpatient stay. Intermediate outcomes include the severity of antidepressant treatment and treatment adherence.

The presentation will focus on the design of the study ¢ollowee by pre liminary results of 6 month data

References Broadhead WE, Blazer DG. George LKr Tse C 1990 Depression disabdity days and days

lost from work in a prospect ve epidemioioglc suwey JAMA 2642524 2528 Katon W Von Korff M, Lin E, Bush T, Ormel J 1992 Adequacy and duration of ant~de

pressant treatment in primary care Meal Care 30(1 ):67-76 Shapiro S, Skinner EA, Kessler LG, Von Korff M, German PS, Tlschler GL et al 1984

Utilization of health and mental health services three epldemlologic catchment area sites Arch Gen Psychiat 41 971 978

Simon G, Von Korff M, Wagner EH, Barrow W 1993 Patterns of antidepressant bse ir" commumtv practice Gen Hosp Psychiatry 15:399-408

Wells KB, Stewart A, Harys RD, Burnan MA, Rogers W, Danials Me t a! 1989 The fJnc tioning and well-being of depressed patients Results from the Medical Outcome Study JAMA 262:914-19

SAT-5-7 1 Biological enrichment in the study, of drug: I

placebo differences of patients with affective disorders

StevenJames, John Heil igenstein Douglas ~anes, GarvTollefson ZJ//y Research Laboratories Lilly Corporate Center Build,'ng 31/2 Drop Code 2128 Indianapolis, IN. 46285

Disruptions in the sleep-wake :ycle are frequently heard complaints re ported by depressed patients To better understand the significance of these problems, an impressive body of knowledge has oeen developed over the past decade examining the association between sleep parameters and clinical presentation of affectlve disorders

Despite this heighten recognition that sleep disturbances offer ~m13ortant clues in the identification of underlying affectwe illness, sleep laooratory studies rarely are included in the evaluation of depressed patients Ques l ions still remain concerning the clinical significance of disturbed sleep *o" 'normal' sleep in depressed patients) in art r ing at the diagnosis of affectlve disorder Also unanswered ~s what p,ognostlc value disturbed sleep car" provide in predicting treatment ou:come with antidepressant medication o" placebo.

The sleep laboratory can prov,de obiect ve n;ormation about the physl ological measures of human s leep Al:bougn there,s htt leagreeme~t ore"

what role these sleep-EEG measures from the sleep laboratory play in af- festive disorders, a critical need remains to develop objective methods to ~dentify depressed patients likely to respond to pharmacotherapy. In addi- tion, it is equally important to screen out patients complaining of depression from causes other than affective disorders and to recognize patients likely to respond to placebo To address these needs we have incorporated sleep laboratory assessments into our clinical trials of depressed patients prior to randomization onto recognized antidebressants or investigational com- oounds.

In developing our protocol for sleep laboratory evaluation we sought to utihze sleep EEG parameters to enrich our population of depressed patients under study We anticipated that we would identify a group of patients with changes in EEG-sleep parameters associated with affective disorders. By segregatqng these patients with reported biological markers in sleep for malor depression we hoped to recruit a more 'homogeneous' population of depressed patients.

We also predicted that patients with changes in sleep revealed through ;)olysomnographic techniques would have a more robust response to an- tidepressant treatment. Previous studies had already reported this observa- tion with tricyclic antidepressants but the applicability of that earlier work to treatment with newer classes of antidepressants was unknown.

Finally, we wanted to investigate whether the presence (or absence) of cnanges in EEG monitored sleep would permit selection of patient popu- lations less likely to respond to placebo. We anticipated that patients with more disturbance in sleep and possible greater severity of overall symp- toms would ~mprove with antidepressants and be less likely to respond to placebo

To address tnese issues, we enrolled 160 0argents in a clinical trials program evaluating a novel antidepressant compound. All subjects were screened by a physician experienced in sleep disorders. After one week of maintaining strict sleep hygiene guidelines, all batients were studied for two or three nights in a sleep laboratory. During the first night a complete polysomnogram (PSG) was performed with careful attention to evaluating respiration, motor activity and arousals for evidence of disturbed sleep as a result of sleep apnea, nocturnal myoclonus or other primary sleep disorders. The second and third night were a modified PSG and obtained to evaluate sleep staging only.

After the laboratory analysis 105 patients were found to have complaints of depression while maintaining strict sleep hygiene and in the absence of a primary sleep disorden Fifty three patients were identified with REM latencies less than 65 minutes and were grouped in the short REM latency group. The remaining 52 patients all had REM latencies greater than 65 minutes and were grouped into the normal REM latency group. Patients w~thin each group were randomized onto study drug or placebo and results determined by the Hamilton Depression Rating Scale {HDRS).

We present the results of this study and compare with earlier work studying the association between sleep parameters and antidepressant response We also will discuss the role of sleep evaluation on identification of individuals likely to respond to placebo condit ions

References Nelson, JC and Charney. DS (1980) Primary Affective Disorder criteria and the

endogenous reactive distinction Arch Gen Psychiatry 37,787 793 Giles, D G, Roffwarg, HP, Schlesser, MA. and Rush, A J {1986}Which endogenous de-

pressive symptoms relate to REM latency reduction? Biological Psychiatry 21,473- 482

Spiker. D G . Coble, P, Cofsky, J, Foster, FG and Kupfer, D J (1978) EEG sleep and sever- ity of depresskon. Biological Psychiatry 13, 485~188

I SAT-5-8 I Adaptive randomization in clinical trials: a i

placebo compromise

A Wood 1, D Faries 2, j . Heiligenstein 2 1 L/fly Research Centre Ltd, Er/ Wood Manor, Wind/esham, Surrey, GU20 6PH, 2 UK," Eli DTly and Company, L/fly Corporate Centre, Indianapohs, Indiana 46285, LZS.A.

A number of innovative strategies have been explored in order to minimize the exposure of subjects included in clinical trials to either placebo or in- effectwe novel drug candidates. These strategies are particularly important m the area of psychopharmacology where the morbidity associated with a failure of treatment can be particularly severe

T~e so-called adaptive study design, known colloquially as "playing the winner", ~s based on a Bayesian statistical argument and progressively modifies trle randomization schedule based on outcomes from foregoing 0argents ~n the study in order to maximize the chance of subsequent sub- lects being randomized to efficacious treatments.