improvement over the no-treatment group. Likewise, there was no significant placebo effect in studies in which objective outcomes (such as blood pressure) were measured by an independent observer. The placebo effect could be documented only in studies in which the outcomes (improvement or failure to improve) were reported by the subjects themselves. The authors concluded that the placebo effect does not have "powerful clinical effects," (objective effects) and that patient-reported improvements (subjective effects) in pain were small and could not be clearly distinguished from reporting bias. Other researchers (Wampold et al.) reanalyzed the same data from the 2001 meta-analysis and concluded that the placebo effects for objective symptom measures are comparable to placebo effects for subjective ones and that the placebo effect can exceed the effect of the active treatment by 20% for disorders amenable to the placebo effect,[83][84] a conclusion which Hróbjartsson & Gøtzsche described as "powerful spin".[85] Another group of researchers noted the dramatically different conclusions between these two sets of authors despite nearly identical meta-analytic results, and suggested that placebo effects are indeed significant but small in magnitude.[86]
Hróbjartsson and Gøtzsche's conclusion has been criticized on several grounds. Their meta-analysis covered studies into a highly mixed group of conditions. It has been reported that for measurements in peripheral organs the placebo effect seems to be more effective in achieving improvements in physical parameters (such as decreasing hypertension, improving FEV1 in asthma sufferers, or decreasing prostatic hyperplasia or anal fissure) than in improving biochemical parameters (such as cholesterol or cortisol) in various conditions such as venous leg ulcers, Crohn's disease, urinary tract infection, and chronic heart failure.[87] Placebos also do not work as strongly in clinical trials because the subjects do not know whether they might be getting a real treatment or a sham one. Where studies are made of placebos in which people think they are receiving actual treatment (rather than merely its possibility) the placebo effect has been observed.[88] Other writers have argued that the placebo effect can be reliably demonstrated under appropriate conditions.[89]
In another update by Hróbjartsson & Gøtzsche, published as a 2010 Cochrane systematic review which confirms and modifies their previous work, over 200 trials investigating 60 clinical conditions were included. Placebo interventions were again not found to have important clinical effects in general but may influence patient-reported outcomes in some situations, especially pain and nausea, although it was "difficult to distinguish patient-reported effects of placebo from response bias". The pooled relative risk they calculated for placebo was 0.93 (effect of only 7%) but significant. Effects were also found for phobia and asthma but were uncertain due to high risk of bias. In other conditions involving three or more trials, there was no statistically significant effect for smoking, dementia, depression, obesity, hypertension, insomnia and anxiety, although confidence intervals were wide. Several clinical (physical placebos, patient-involved outcomes, falsely informing patients there was no placebo) and methodological (small sample size, explicit aim of studying the placebo effect) factors were associated with higher effects of placebo. Despite low effects in general and the risk of bias, the authors acknowledged that large effects of placebo interventions may occur in certain situations.[90]
In 2013 Jeremy Howick and colleagues used Hróbjartsson & Gøtzsche's data to compare the magnitude of placebo effects with the magnitude of treatment effects.[91] They found a statistically significant difference between placebo and treatment effect sizes in trials with binary outcomes but not in trials with subjective outcomes.