ResearchResearch summary
Nocebo and adverse events in placebo groups: 2018 overview
Across 20 reviews, the median adverse-event rate in placebo groups was 49.1%; a separate 15-trial comparison with no treatment was not significant.
In this pathway
- Current topic
- Nocebo and adverse events in placebo groups: 2018 overviewCurrent

Bottom line
Across trials for many conditions, adverse events were also reported in placebo groups. The cross-review median and separate 15-trial comparison cannot establish what caused any individual's symptom, and Candida or antifungal treatment was not studied.
Why this source matters
It gives context for how often placebo-group participants reported adverse events across prior reviews, while showing why an untreated comparator matters. It does not prove that placebo or negative expectations caused symptoms.
Study and population
This rapid overview included 20 systematic reviews of randomized trials, with 1,271 trials and 250,726 placebo recipients. Most included reviews concerned pain-related conditions; others covered neurologic, psychiatric, cardiovascular, and other conditions.
What was tested
The authors summarized adverse-event prevalence and dropout attributed to adverse events in placebo groups. They separately identified 15 individual trials with untreated, placebo, and active-treatment groups; this comparison was not a pooled estimate of all 1,271 trials. The underlying trials supplied their own adverse-event definitions and assessment times.
Results
The median prevalence of at least one reported adverse event in placebo groups was 49.1% (IQR 25.7–64.4% in the abstract and main text; Table 1 prints 25.8–64.5%). Median adverse-event-related dropout was 5.0% (IQR 2.28–8.4%; Table 1 prints 5% 2.3–8.4%). These are cross-review medians, not personal risk estimates. Separately, the 15-trial comparison reported adverse events in 29/682 participants without treatment (4.25%), 40/614 with placebo (6.51%), and 48/809 with active treatment (5.93%). The placebo-versus-untreated difference was not statistically significant (P=0.07). Serious harms and deaths were not reported as pooled outcomes, so these adverse-event rates do not estimate serious-harm or mortality risk.
Applicability
These results describe reporting patterns in selected trial reviews across different conditions and interventions. They do not estimate the chance that a person will have an adverse event from a specific treatment, supplement, cleanse, restrictive diet, or Candida-labelled protocol.
Limitations
The overview had a narrow PubMed search ending in November 2017. Included conditions, interventions, event definitions, ascertainment, follow-up, and denominators varied; review quality also limited interpretation. The 15-trial comparison was small, had sparse events, and provided no pooled effect estimate or confidence interval. The review cannot distinguish natural symptoms, expectations, and misattribution, and it did not study Candida or antifungal interventions.
Contribution to the evidence
Contextual. The overview documents adverse-event reports in placebo groups and a separate untreated comparison, but it does not prove a nocebo effect or support a Candida-specific claim.
Practical change
No practical change. The overview cannot explain an individual's symptom or establish a Candida mechanism, and it does not change the existing advice not to treat worsening as proof that an intervention is working.
Funding and conflicts
The authors reported no external funding and declared no competing interests.
Source
Read the original publication (published 11 December 2018).
Study type, population, and sample scale
- Study type
- Rapid overview of systematic reviews with a separate descriptive trial comparison
- Population
- Participants in placebo-controlled randomized trials across clinical conditions; included reviews mostly covered pain-related conditions.
- Sample scale
- 20 systematic reviews, 1,271 randomized trials, and 250,726 placebo recipients; a separate untreated comparison covered 15 trials
Results
- Across 20 systematic reviews covering 1,271 randomized trials and 250,726 placebo recipients, the median prevalence of at least one reported adverse event in placebo groups was 49.1% (IQR 25.7–64.4% in the abstract and main text; Table 1 prints 25.8–64.5%). This is a median across heterogeneous reviews, not an estimate of an individual's risk.
- The median rate of dropout attributed to adverse events was 5.0% (IQR 2.28–8.4% in the abstract and main text; Table 1 prints 5% [2.3–8.4%]).
- In a separate, non-pooled comparison of 15 trials with untreated, placebo, and active-treatment groups, adverse events were reported by 29/682 without treatment (4.25%), 40/614 with placebo (6.51%), and 48/809 with active treatment (5.93%). Placebo versus no treatment was not statistically significant (P=0.07).
- The overview did not report serious harms or deaths as pooled outcomes, so the adverse-event rates do not estimate serious-harm or mortality risk.
Limitations
- Cross-review median prevalence and dropout rates summarize placebo groups across the included reviews.
- Conditions, interventions, definitions, ascertainment methods, follow-up times, and denominators varied; the medians are not individual risk estimates.
- The placebo-versus-untreated comparison drew on 15 trials with three groups.
- It involved sparse events across mixed conditions and interventions, was not pooled, and had no pooled effect estimate or confidence interval; the placebo-versus-untreated difference was not statistically significant (P=0.07).
- Adverse events were drawn from the underlying reviews and trials.
- There was no common event definition or follow-up time, and the overview could not determine whether a report reflected natural history, negative expectations, or misattribution.
- The review reported adverse events and dropout attributed to adverse events.
- Serious harms, deaths, Candida outcomes, and intervention-specific harms were not reported as pooled outcomes; the adverse-event rates do not estimate those risks.
- The overview included trials across multiple conditions.
- It did not study Candida, fungal infection, antifungals, supplements, cleanses, or restrictive diets and cannot establish a Candida-specific mechanism.
- The authors conducted a rapid review.
- The search was narrow, using PubMed on 22 November 2017 with the title query nocebo[TI] AND Review[ptyp], plus expert contact and reference searching; only 8 of 20 reviews met both abbreviated AMSTAR criteria.
Applies to
- How adverse events were reported in placebo arms across the conditions and reviews included in this overview.
- General context about adverse-event reporting when interpreting trial findings.
Does not apply to
- Estimating one person's chance of symptoms or determining the cause of an individual's adverse event.
- Candida or fungal infection, antifungal treatment, supplements, cleanses, or restrictive diets; these were not studied.
Contribution to the current conclusion
Contextual
The overview adds broad context about adverse-event reporting in placebo groups. Its heterogeneous cross-review results and a small, nonsignificant placebo-versus-untreated comparison cannot identify the cause of an individual's symptoms, establish a nocebo mechanism, or be transferred to Candida interventions.
Related evidence mapPractical change or no change
No change to Candida guidance. The overview provides general context about adverse-event reporting in placebo groups but cannot explain an individual's symptoms or establish a Candida-specific mechanism.
Funding and conflicts
- Funding
- The authors reported no external funding.
- Conflicts
- The authors declared no competing interests.
- Paper date
- 2018-12-11
Sources
View all 1 source
- Rapid overview of systematic reviews of nocebo effects reported by patients taking placebos in clinical trials
- Source type
- Systematic review
- Published
- 2018-12-11