Table 2 Spin classification for non-randomized studies assessing therapeutic interventions. Examples provided are from manuscripts and published abstract and full-texts from our sample

Misleading reporting

Not reporting adverse events or lack of focus on harm

Results are reported without warnings on important or relevant safety issue.

“This trial showed that overnight switch from oral ropinirole to transdermal rotigotine, with a dose conversion ratio of 1.5:1, was well tolerated in Korean patients with no apparent loss of efficacy.” 45/116 experienced adverse events and 13 stopped treatments because of adverse events.

Selective reporting

Only a subset of the original outcomes or analysis planned in a study is fully reported.

“The results suggest that [Internet-based cognitive therapy] ICBT with therapist support has the potential to reduce [obsessive-compulsive disorders] OCD symptoms, depressive symptoms and general functioning”. No report of the lack of improvement in quality of life.

Misleading description of study design

Study design is presented as more robust than it is actually.

“Based on this prospective case control study tranexamic acid seems not to have a benefit in posterior lumbar spine surgery.” It was a retrospective study involving 97 patients and nothing was prospective in this study.

Use of linguistic spin

Any word or expression emphasizing the beneficial effect of the therapeutic intervention

“Both operative methods could be performed safely in the early stages after the introduction of surgery and could consistently obtain excellent surgical performance.”

No consideration of the limitations

Important limitations are not taken into account in the interpretation of the results.

“The TABADO program, targeting teenagers in vocational schools, was effective in producing a higher 12-month abstinence rate among all smokers in the intervention group.” No adjustment was made on major confounding variables.

Selective citation of other studies

Only previous studies concordant with the current study findings are acknowledged or other important studies in the field are not reported.

“It would be interesting to know its efficacy and safety in correcting high myopic astigmatism and how it changes the shape of the cornea.” Several publications already exist in this field.

Inadequate interpretation

Claim an effect for non-statistically significant results

Therapeutic intervention is presented as effective despite a non-statistically significant result.

“The use of [Automated CardioPulmonary Resuscitation] A-CPR resulted in a higher rate of survival to hospital compared with [Conventional CardioPulmonary Resuscitation] CPR” in a restrospective study involving 66 patients where the propensity score adjusted Odds Ratio was 1.69 [0.79; 3.63].

Claim an equivalence for non-statistically significant results despite a wide confidence interval

Therapeutic intervention and comparator are presented as equivalent when a comparison test is not statistically significant with a large confidence interval.

“The authors concluded that: 1) mortality during follow-up was statistically similar for both groups; (…)” In this retrospective cohort study involving 352 patients, the survival rate at 20 year was 69.3 % with mechanical mitral valve substitutes and 56.6 % with biological substitutes. The hazard ratio was not statistically significant but had a wide 95 % confidence interval (HR = 1.21 [0.79; 1.86], p = 0.386).

Ruling out safety for non-statistically significant results

Therapeutic intervention is presented as safe based on a non-statistically significant comparison test, despite a large confidence interval.

“Long-term treatment with esomeprazole (20 mg once daily) is well tolerated and efficacious” despite 16.29 % of patients (n = 22) experiencing adverse events judged to be possibly related to treatment with esomeprazole (mostly mild and transient).

Causal language or causal claim

Results are presented with a sentence implying a cause-and-effect link between the intervention and the outcome

“Treatment with oral valacyclovir as the sole antiviral therapy resulted in complete resolution of retinitis.” This was a before–after study involving 10 patients.

Claim of any significant difference despite lack of statistical test

Therapeutic intervention and comparator are compared despite no proper statistical test reported.

“pH and HCO³⁻ significantly increased after hemodialysis sessions with both aspirin only and aspirin + dipyramidol, the increase of pH and HCO³⁻with aspirin only was significantly larger than aspirin + dipyramidol.” No statistical test was reported.

Focus on statistical significance instead of clinical relevance

Results are presented by their statistical significance without considering the clinical relevance of the effect size.

“While the [Clinical Global Impression-Schizophrenia] CGI-SCH overall score improved in both groups after switching, there was a significantly greater change in those who switched from olanzapine (difference of 0.29 points, p = 0.013)”. The CGI-SCH scale range from 0 to 7.

Inadequate extrapolation

Inadequate extrapolation to larger population, intervention or outcome

Results are generalized to another population, intervention or outcome than those of the study (such as surrogate outcomes)

“This intervention approach has the potential to impact on the progression of colorectal cancers and other cancers or chronic diseases.” The intervention focused on colorectal cancers only.

Inadequate implication for clinical practice

Authors recommend the use of the therapeutic intervention for clinical practice.

“One should not hesitate to perform an osteotomy in difficult cases.” in a prospective cohort study involving 109 patients.