Introduction: Several versions of Folstein's minimental test (MMSE, Minimental State Examination) exist in Spanish. One of them is the 37-point MMSE (MMSE-37), which is adapted to people with a low cultural level. AIM. To study the interobserver and test-retest reliability of the MMSE-37.
Patients and methods: Interobserver reliability was studied in three professional observers who simultaneously administered the MMSE-37 to 40 patients from a community setting. Test-retest reliability was obtained in 35 patients by administering the MMSE-37 on two occasions 2-3 months apart. Statistics used: intraclass correlation coefficient (ICC), Bland-Altman difference plot method, and kappa index.
Results: Interobserver reliability study: multiple ICC of 0.99 (95% confidence interval, CI 95% = 0.99-0.99); kappa index of 0.84 (CI 95% = 0.72-0.97); the degree of agreement between the individual responses was 1 for 'pathological outcome', 0.79 for 'probably pathological outcome', 0.37 for 'probably normal outcome' and 0.89 for 'normal outcome'; ICC > 0.99 (CI 95% = 0.99-0.99) between all the pairs of observers. The systematic differences in the mean score between the pairs of observers did not exceed the score of 0.1, which represents a bias of 0.47% after removing the scale. Test-retest reliability study: ICC of 0.92 (CI 95% = 0.87-0.96), kappa index of 0.84 (CI 95% = 0.54-1.00) and mean difference in the magnitude of disagreement of 0.34 (CI 95% = -2.80-3.49), 1% of the percent error in the magnitude of the disagreement.
Conclusions: The MMSE-37 shows a high degree of test-retest reliability and reliability among professional observers, as in other versions of the MMSE.