A systematic review and meta-analysis of the Test of Memory Malingering in adults: Two decades of deception detection

Abstract Objective: The present study, adhering to Preferred Reporting Items for Systematic review and Meta-Analyses (PRISMA) guidelines, is the first systematic review and meta-analysis of the Test of Memory Malingering (TOMM) to examine traditional and alternative cutoffs across Trial 1, Trial 2, and Retention. Method: Search criteria identified 539 articles published from 1997 to 2017. After application of selection criteria, 60 articles were retained for meta-analysis. Classification accuracy statistics were calculated using fixed- and random-effects models. Results: For Trial 1, a cutoff of <42 was found to result in the highest sensitivity value (0.59–0.70) when maintaining specificity at ≥0.90. Traditional cutoffs for Trial 2 and Retention were highly specific (0.96–0.98) and moderately sensitive (0.46–0.56) when considering all available studies and only neurocognitive/psychiatric samples classified by known-groups design. For both trials, a modified cutoff of <49 allowed for improved sensitivity (0.59–0.70) while maintaining adequate specificity (0.91–0.97). A supplementary review revealed that traditional TOMM cutoffs produced >0.90 specificity across most samples of examinees for whom English is not the primary language, but well-below acceptable levels in individuals with dementia. Conclusions: The TOMM is highly specific when interpreted per traditional cutoffs. In individuals not suspected of significant impairment, findings indicate that a less conservative TOMM Trial 2 or Retention cutoff of <49 can be interpreted as invalid, especially in settings associated with higher base rates of invalidity and, thus, higher positive predictive power. A cutoff of <42 on Trial 1 can also be interpreted as invalid in most settings.