A two-phased screening paradigm for evaluating candidate medications for cocaine cessation or relapse prevention:...

Cocaine pharmacotherapy trials are often confounded by considerable variability in baseline cocaine-use levels, obscuring possible medication efficacy. Testing the feasibility of using a prerandomization, abstinence-induction protocol, we screened three candidate medications to explore treatment response in patients who did, or did not, achieve abstinence during an extended baseline phase.

Eligible treatment-seeking, cocaine-dependent subjects entered a 4-week baseline period (Phase I) with high-value abstinence contingent vouchers and two motivational interviewing sessions, followed by a 12-week medication trial (Phase II) with random assignment stratified on Phase I abstinence status to (1) modafinil (400mg/d), (2) levodopa/carbidopa (800/200mg/d), (3) naltrexone (50mg/d), or (4) placebo. Treatment consisted of thrice-weekly clinic visits for urine benzoylecgonine testing and weekly cognitive behavioral therapy with contingency management targeting medication compliance.

Of the 118 subjects enrolled, 81 (80%) completed Phase I, with 33 (41%) achieving abstinence, defined a priori as 6 consecutive cocaine-negative urines. Tests of the interaction of each medication (active versus placebo) by baseline status (abstinent versus nonabstinent) permitted moderator effect analysis. Overall, baseline abstinence predicted better outcome. Cocaine-use outcomes for levodopa and naltrexone treatment differed as a function of Phase I abstinence status, with both medications producing benefit in nonabstinent but not baseline-abstinent subjects. There was no evidence of a moderator effect for modafinil.

The two-phase screening trial demonstrated that subgrouping of patients with respect to baseline abstinence status is feasible and clinically useful for exploring cocaine cessation and relapse-prevention effects of candidate medications.