Skip to main content

Welcome to a Grasp lesson

Create a free account to keep studying and create your own study path.

Lesson 6

Interpreting SAMSON Trial Comparisons and Nocebo Symptoms

Recommended resources

Welcome back. The previous lesson used blinded statin-versus-placebo trials to estimate the average excess of muscle symptoms attributable to statins. That comparison is essential, but it leaves another question: why might symptoms feel worse while taking a tablet even when its active ingredient is not the cause?

This lesson examines the SAMSON trial, which compared statin tablets, indistinguishable placebo tablets, and no tablets in people who had previously stopped statins because of symptoms. Its three-way design helps separate an effect of the statin ingredient from symptoms associated with taking a tablet and from symptoms present without tablets. It also shows why none of those comparisons can, by itself, diagnose the cause of one person’s symptoms.


Why SAMSON included a no-tablet condition

SAMSON enrolled 60 people who had previously abandoned statins because of intolerable symptoms. Each received a randomized sequence of twelve one-month periods: four with atorvastatin 20 mg, four with matching placebo tablets, and four with no tablets. Participants recorded symptom intensity daily on a scale from 0 to 100. They could stop tablets early in a month if symptoms became intolerable.

Read the study’s methods before interpreting its results.

Side Effect Patterns in a Crossover Trial of Statin, Placebo, and No Treatment

Read the SAMSON investigators’ description of who entered the trial and how the monthly conditions were assigned. The masking of statin versus placebo is crucial to interpreting the comparisons.

In “Methods,” begin with “Study design,” at the sentence beginning “This trial used a multiple-crossover.” Continue through “Participants,” “Randomization and masking,” and “Procedures and medications,” ending at the sentence about the placebo’s identical appearance. Read the trial design, focusing on prior statin intolerance, the randomized order of months, and which tablets could be distinguished.

The three comparisons answer different questions:

Comparison What it helps estimate
Statin versus placebo The added effect of the statin-containing tablet beyond taking a similar-looking tablet
Placebo versus no tablets The added symptoms associated with taking a tablet in this setting, beyond symptoms during no-tablet months
Statin versus no tablets The combined difference between statin-tablet months and no-tablet months

Participants could not tell statin from placebo by looking at the tablets. They necessarily knew when they were taking no tablet. Thus, the no-tablet condition supplies a useful baseline, but it is not blinded in the same way as the statin–placebo comparison. A higher placebo score than no-tablet score does not reveal one exact psychological or biological mechanism for every symptom.



What the average scores show

Read the results as symptom-intensity scores, not counts of people or episodes. The trial measured symptoms of any type; although muscle ache was the most common reason participants had stopped statins before enrollment, these results are not a muscle-pain-only measure.

Side Effect Patterns in a Crossover Trial of Statin, Placebo, and No Treatment

Now read the reported scores and the account of months stopped early. The results show both what happened on average and why “people stopped tablets” cannot automatically mean “the statin ingredient caused the symptoms.”

In “Symptom scores for statin, placebo, and no treatment,” start where the authors report that 49 of 60 participants completed all 12 months. Continue through “Early stopping of tablet months caused by intolerable symptoms,” ending with the sentence about participants who stopped only a statin or only a placebo month. Read these results. Focus on the three mean scores, the statin–placebo comparison, and early stopping in both tablet conditions.

The mean score was 8.0 during no-tablet months, 15.4 during placebo months, and 16.3 during statin months. In other words, scores were higher on either kind of tablet than with no tablet, while statin and placebo scores were close. The study did not detect a statistically clear difference between statin and placebo scores. That is not the same as proving that their true effects are exactly identical.

The investigators reported a nocebo ratio of 0.90. The idea becomes clearer by looking at the rounded mean scores. Relative to no-tablet months, the additional score was about on placebo and on statin. The placebo-associated increment was therefore roughly nine-tenths of the statin-associated increment. The published ratio was calculated from the study data, rather than from these rounded figures.

“90%” is not a percentage of participants, symptom episodes, or an individual’s pain. It describes a relationship between average increases in symptom scores above the no-tablet baseline in this trial. Nor does the small average statin–placebo difference rule out a pharmacological statin effect in a particular person.

Stopping tablets early tells a similar, more concrete story: participants stopped some statin months because of intolerable symptoms and some placebo months. Those stops represent real difficulties during the trial. Without knowing which tablet was taken, a stop alone would not establish what caused it.



Patterns hidden by an average

An average can conceal quite different experiences. The following SAMSON figure shows six illustrative participants. Each horizontal plot follows one person across monthly periods; the line traces daily symptom scores. Blue indicates placebo, red statin, and gray no tablets. These six were chosen to show patterns, not to estimate how common each pattern was among all 60 participants.

Daily symptom scores for six illustrative SAMSON participants across statin months (red), placebo months (blue), and no-tablet months (gray). The examples show persistent background symptoms, few symptoms on rechallenge, and symptoms during both kinds of tablet month.

In the top pair, substantial symptoms appear even during no-tablet months: a statin month is not needed to explain their continuing symptom burden. The middle pair reported very little during this particular year despite having previously stopped statins because of symptoms. In the bottom pair, symptoms appear during both statin and placebo months, with quieter stretches during no-tablet months. That pattern is consistent with a tablet-associated nocebo effect, rather than an effect unique to the statin ingredient.

Read the investigators’ explanation of why the placebo and no-tablet conditions are both needed.

Side Effect Patterns in a Crossover Trial of Statin, Placebo, and No Treatment

The investigators use these individual patterns to explain what an ordinary stop-and-restart history can miss. Read their interpretation alongside the figure, while remembering that an illustrative graph is not a diagnosis.

Read “The need for a planned schedule and symptom tracking,” followed by “The need for a placebo” and “The need for no-tablet periods.” Start at the sentence beginning “Because SAMSON was an exercise in personalized medicine” and continue to the end of the last subsection, where the authors compare no-treatment and placebo months. Read the explanation. Notice how no-tablet months distinguish persistent background symptoms from symptoms associated with taking either tablet.

This design also exposes a trap in informal stop-and-restart observations. A person may stop a tablet when symptoms are particularly bad and restart it once feeling better. Symptoms that fluctuate for other reasons can then improve after stopping and worsen after restarting. The timing feels persuasive, but without a planned, blinded comparison it cannot isolate the statin ingredient.

SAMSON examined timing directly. Read the short results section on symptom onset and relief.

Side Effect Patterns in a Crossover Trial of Statin, Placebo, and No Treatment

This section tests whether prompt worsening on starting a tablet, or prompt relief on stopping it, distinguished statin from placebo in the trial.

In “Time-course of onset and offset of symptoms,” begin at “We studied whether the intensity of symptom onset” and read through the end of the subsection. Read the timing analysis and the immediately following sentence about relief within three days. Focus on the comparisons between the two tablet types, not just on how quickly scores changed.

In this group, neither the intensity of symptoms on starting nor the degree of relief on stopping reliably distinguished statin from placebo. Rapid change was possible with either. Timing remains useful information to describe to a clinician; it simply is not, on its own, proof of a statin-specific effect.



“Nocebo” describes a possible cause, not imaginary pain

A nocebo effect is a genuine adverse experience influenced by expectations or the context of an intervention, rather than necessarily by its active ingredient. The word does not mean that someone invented a symptom or could simply choose not to feel it. SAMSON’s placebo months matter precisely because participants could have substantial symptoms while taking tablets with no statin in them.

The following short discussion places nocebo effects beside other possible sources of symptoms, including background conditions and true pharmacological effects.

Why Do I Get Side Effects? Personalized (N-of-1) Trials for Statin Intolerance and the Nocebo Effect

Read this explanation to keep the interpretation humane and precise: genuine symptoms can have different causes, and more than one cause may be relevant to a person.

Under “3.2. Medical Implications,” read subsections “3.2.1. Source 1: ‘Ever Present’ Chronic Background Symptoms” through “3.2.4. Source 4: True Pharmacological Side Effect.” Start at “When patients are asked about symptoms on statins” and finish at the end of Source 4, with the sentence ending “may be an example of this.” Read the four sources. Pay particular attention to the explicit distinction between a nocebo effect and an imaginary symptom.

There are limits to what SAMSON establishes. It was a relatively small study of people who had already stopped statins because of symptoms; 49 of 60 completed the full year. It studied one statin and dose in one-month periods, and excluded people with dangerous previous effects such as rhabdomyolysis. The result should not be read as “statins cannot cause symptoms,” as a safety finding about rare serious injury, or as a way to classify every person’s experience from a graph.

Instead, SAMSON supplies a more careful interpretation: in this selected group, symptoms were markedly higher while taking either kind of tablet than while taking no tablet, but were similar on statin and placebo on average. An individual’s symptoms deserve attention regardless of their eventual explanation. If symptoms occur in clinical care, their history and any appropriate assessment belong in a clinician conversation rather than an independently arranged medication change or an improvised stop-and-restart test.



Takeaway

The statin–placebo comparison asks whether the active ingredient adds symptoms; the placebo–no-tablet comparison shows how much symptom burden accompanies tablet-taking in this trial; and the no-tablet months reveal background symptoms. SAMSON’s average scores and individual patterns make nocebo effects intelligible without dismissing pain as imaginary. They also show why symptom timing alone cannot identify the cause for one person.

The next lesson turns to another potential harm—newly diagnosed diabetes—and uses trial results to examine how its absolute risk varies with treatment intensity and baseline blood glucose.