Skip to Content
Interactive Textbook on Clinical Symptom Research Logo


Home button

Trial Design: Pain Sections
Author Bio
Introduction
Currently selected section: Placebo Effects
Single Dose Trials
Repeated Dose Trials
Explanatory Versus Pragmatic
Dose-Response
Parallel Group Versus Crossover
Conclusion
 

Commentary of Eugene Laska, Statistician

Chapter 1: Clinical Trials of Pain Treatment: Placebo Effects and Their Implications for Pain Studies
        
Photo of Pills
Compared to trials of treatments for structural disease, symptom treatment trials are susceptible to both false positive and false negative results. A major source of both false positives and negatives is the placebo effect, which in analgesic trials is often substantial and may have a duration of weeks or months (Turner et al., 1994), but appears to have minimal effects on measures of established structural disease. The frequency of placebo effects and their implications for pain studies effects varies among analgesic studies from zero to more than 60% (Turner et al., 1994; Moore et al.,1998). Efforts to identify and exclude "placebo responders" from analgesic trials have not succeeded; given repeated opportunities, most people will manifest a placebo response. (Investigators in other fields, such as affective disorders, often try to identify placebo responders to a placebo run-in period and exclude them from further study.)

Several factors are recognized to increase the magnitude of the placebo analgesic response, including patients’ expectations that they would get relief, clinicians’ warmth, prestige, and positive attitude, and the invasiveness of the intervention. Price (1988) has shown that type of placebo nerve block, saline injections of the stellate ganglion, produced an average of 19 hours of near-complete pain relief in patients with complex regional pain syndromes. Full scale surgical procedures produce even more dramatic placebo responses. A variety of operations that later proved to be useless, including gastric freezing for duodenal ulcers and actual or sham internal mammary artery ligation for angina pectoris, were initially reported to improve or eliminate the pain of 60-100% of patients for a year after surgery (Turner et al., 1994).

If one is to avoid false positives, one must strive to maximize the effectiveness of blinding procedures, and check to see if patients can guess their study assignment by the appearance, taste, or side effects of the treatments (Moscucci et al., 1987). In studies of drugs that have unmistakeable side effects, some investigators use "active placebos" that mimic the side effects of the analgesic (Max, 1991; Greenberg and Fisher, 1994) (Figure 2.1 ; from Max et al., 1988).



Click on image to enlarge

Figure 2.1: Side Effects Can Bias Patients Towards Spurious Reports of Efficacy: Pain intensity vs. time in a parallel group comparison of amitriptyline (mean dose 65 mg/day), lorazepam (2.4 mg/day) and lactose placebo in patients with post-herpetic neuralgia. Both active drugs but not lactose placebo produced moderate to severe sedation in all patients. Amitriptyline was superior to the inert placebo, reaching statistical significance in week 6. Patients taking lorazepam initially reported pain reduction, during the time in which sedation was most pronounced, but this effect dissipated after the first few weeks. It is possible that patients who noted sedation thought they were on a strong analgesic and that this belief biased them towards reporting pain relief. To improve blinding and reduce this potential bias in subsequent studies, we have used small doses of lorazepam and benzotropine, other drugs to mimic sedative and anticholinergic side effects of experimental medications ( From Max et al., 1988a).

        

Page 5 of 52
        
Previous Section