The value of predictive instruments in the screening of acute stroke: an umbrella review on previous systematic reviews

Alireza Baratloo; Mobin Mohamadi; Mohammad Mohammadi; Amirmohammad Toloui; Arian  Madani Neishaboori; Seyedeh Niloufar  Rafiei Alavi; Ali Nahiyeh; Mahmoud Yousefifard

doi:10.18502/fem.v6i3.9400

Alireza Baratloo Research Center for Trauma in Police Operation, Directorate of Health, Rescue and Treatment, Police Headquarter, Tehran, Iran
Mobin Mohamadi Student Research Committee, Iran University of Medical Sciences, Tehran, Iran.
Mohammad Mohammadi Student Research Committee, Iran University of Medical Sciences, Tehran, Iran.
Amirmohammad Toloui Physiology Research Center, Iran University of Medical Sciences, Tehran, Iran.
Arian Madani Neishaboori Physiology Research Center, Iran University of Medical Sciences, Tehran, Iran.
Seyedeh Niloufar Rafiei Alavi Physiology Research Center, Iran University of Medical Sciences, Tehran, Iran.
Ali Nahiyeh Faculty of Medicine, University of Central Lancashire, Preston, United kingdom.
Mahmoud Yousefifard Physiology Research Center, Iran University of Medical Sciences, Tehran, Iran.

DOI: https://doi.org/10.18502/fem.v6i3.9400

Keywords: Decision Support Techniques; Diagnosis; Emergency Medical Service; Stroke

Abstract

Objective: Although various predictive instruments have been introduced for early stroke diagnosis, there is no consensus on their performance. Therefore, we decided to assess the value of predictive instruments in the detection of stroke by conducting an umbrella review.

Methods: A search was performed in the Medline, Embase, Scopus and Web of Science databases by the end of August 2021 for systematic reviews and meta-analyses. Original articles included in the systematic reviews were retrieved, summarized and pooled sensitivity, specificity and diagnostic odds ratio were calculated. The level of evidence was divided into five groups: convincing (class I), highly suggestive (class II), suggestive (class III), weak (class IV) and non-significant.

Results: The value of 33 predictive instruments was evaluated. The sample size included in these scoring systems’ assessments varied between 182 and 47072 patients. The level of evidence was class I in one tool, class II in 18 tools, class III in 2 tools, class IV in 11 tools, and non-significant in one tool. Apart from Med PACS, which had a low diagnostic value, other tools appeared to be able to detect a stroke. The optimum performance for diagnosis of stroke was for ROSIER, NIHSS, PASS, FAST, LAMS, RACE and CPSS.

Conclusion: Convincing to suggestive evidence shows that ROSIER, NIHSS, PASS, FAST, LAMS, RACE and CPSS have the optimum performance in identifying stroke. Since ROSIER’s calculation is simple and has the highest sensitivity and specificity among those predictive instruments, it is recommended for stroke diagnosis in pre-hospital and in-hospital settings.