Novel approach to substantiate cause-and-effect relationships: combining study ranking and statistical modelling,...

The present study aims at substantiation of cause-and-effect relationships via a combination of ranking studies and statistical modelling of the highest scoring studies.

Due to the fact that independent datasets of studies identified in the ranking part of the study were difficult to obtain the second part was performed using a different aim. The ranking part concentrated around studies on the association between vitamin D status and severity (morbidity/mortality) of COVID-19 infection in hospitalized patients with criteria focussing on physiological and statistical relevance. The topic of the second part of our study changed into the impact of postbiotic consumption on infectious episodes in children to accommodate testing the statistical modelling approach. The latter consisted of the construction of a multivariate confounding model based on the data of one study and subsequent validation of the achieved model via data of an independent study with a similar experimental design. Sensitivity and specificity were assessed in both studies applying discriminant analysis.

The first part of our project consisted of five statistically-based criteria to rank the various studies resulting in the highest score obtained by Hernandez et al. (J Clin Endocrinol Metab 106, e1343-e1353. 10.1210/clinem/dgaa733, 2021). Subsequently, the second part using the datasets of the postbiotic studies yielded a significant model showing impact by consumption of the postbiotic on the reduction of infectious episodes. The sensitivity and specificity outcome ranged between 68 to 94%. Importantly changing the datasets did not affect the significant impact of the postbiotic.

The combination of a ranking and a statistical modelling approach supports the validation of causal-effect relationships via objective criteria.