Arthur Shapiro and colleagues published a randomized, double-blinded study in the 1983 Journal of Psychiatric Research entitled Diazepam: how much better than placebo? assessing the effectiveness of diazepam versus placebo and weekly psychotherapy in the treatment of “neurotic” anxiety. At the time of publication, there were 70 million diazepam prescriptions in the U.S., with world-wide sales exceeding 650 million dollars each year. The authors note that “these impressive figures imply impressive effects, ” yet their “anti-anxiety effects, for which these drugs are most used, are unclear and certainly not impressive.”  


Of the 2805 psychiatric patients who were assessed for outpatient treatment between 1972-75, 224 met criteria for the study and 139 completed all 6 visits. Patient underwent psychiatric assessment with standardized testing at the initial visit. Patients who met the following criteria were considered for the study:

  1. Age 17-65
  2. 6th grade education and above
  3. Symptoms of anxiety not related to psychosis, organic brain disease, dementia, alcoholism, drug addiction, or medical illness
  4. Not in psychiatric treatment in the year prior to the study
  5. Not on “clinically meaningful” doses of antipsychotics or antidepressants 6 months prior to the study
  6. Not on anti-anxiety drugs in the past month
  7. Rated by psychiatrist as more anxious than depressed
  8. Having sufficient anxiety to interfere with function
  9. Likeliness to respond to 6 weeks of treatment with medication and psychotherapy

Patients were randomly assigned to treatment with diazepam or matching placebo tablets.

Patient were assessed over a period of 6 weeks with weekly return visits by standardized testing and interview with a psychiatrist. Each patient received 15-20 minutes of psychotherapy. The authors note that DSM-III criteria were not in effect at the time of the study, but 161 of the 224 patients in the study met criteria for generalized anxiety disorder (GAD) based on DSM-III definitions. The sample population was in general young (mean age 28.9), Caucasian (88%), female (63%), and middle social class with greater than high school education (64%).  

Psychiatrists rated patients on global improvement at each weekly visit with a 7 point scale ranging from “markedly better to markedly worse.” Patients rated their own global improvement using the same scale. Likewise, both psychiatrist and patient rated the anxiety level at each visit.   These 4 dependent variables were used to measure response to treatment.

Nonspecific variables were also assessed, such as favorable reaction to placebo, patient preference for drug therapy, and positive attitudes to doctors by patients (and vice versa).

The dosage for the first week of treatment was 10 mg for both diazepam and placebo. Doses were subsequently “clinically titrated and resulted in an average dose throughout the 6 visits of 21.4 mg”.


Total sample (N=224): Patients treated with diazepam rated themselves as significantly more improved on a global scale than placebo patients only at the first week revisit. Psychiatrists also rated their patient as improved only at the first revisit.

Completed sample (N=139): Psychiatrists, but not patients, rated the diazepam group as significantly more improved only at the first revisit.

Patient-psychiatrist agreement sample (patients were excluded if the two disagreed in their assessment): There was no significant difference between the diazepam and placebo groups.

Anxiety assessment: Similar results were obtained using the anxiety assessment scales in the total and completed patient samples.

Nonspecific variables: positive placebo reaction, patient pre-treatment attitude to physician, and patient attitude toward physician were significant factors in patient improvement at both the first revisit and end of study.


This study did not provide evidence that diazepam is effective in the long-term treatment of anxiety. In fact, in this population sample, it was shown to only be effective in the first week of treatment. Nonspecific factors played a substantial role in predicting treatment outcome, raising the question that the “success” of diazepam could be in part due to these non drug-related factors.

The authors did note a few study limitations, including:

  1. Results only apply to the demographic in the sample population
  2. Patients’ diagnosis of anxiety was based on pre-DSM-III criteria
  3. Drug compliance was assessed by counting tablets instead of measuring blood diazepam levels

Based on the study results, the authors conclude that “most patients can probably be successfully treated without diazepam, and possibly with brief, supportive psychotherapy.” They also mention that while these results are limited to diazepam, they could be generalized to all benzodiazepines. For this reason, the long term use of benzodiazepines is “not supported by the data of this study and the accumulated data in the literature”.