Item-Level Psychometric Properties of the Brief Interview for Mental Status for Patients Discharged to Skilled Nursing Facilities

PURPOSE: The Brief Interview for Mental Status (BIMS) is among the most commonly used observational cognitive assessment for patients with chronic cognitive impairments. As a performance-based screening tool, the BIMS assesses basic and common cognitive functions including repetition of words, temporal orientation, and item recall (Saliba et al., 2012; Tariq et al., 2006; White & Davis, 1990). In addition, the BIMS was designed to assess the needs of residents in skilled nursing facilities (SNFs) and identify whether they had delirium behaviors.

However, a previous study found that the BIMS did not differentiate between residents with normal cognition and those with mild cognitive impairment or between mild and moderate dementia on the basis of sensitivity testing (Mansbach et al., 2014). This previous finding leads to a concern about whether the BIMS can provide reliable information to assess patients’ cognitive function. Understanding the item-level psychometric properties of the BIMS could help identify the specific areas in which the BIMS is performing poorly or the ways in which it can be improved. Occupational therapy practitioners in SNFs need to know whether the BIMS can generate reliable and useful assessment results. This study investigated the item-level psychometric properties of the BIMS based on Rasch analysis for patients admitted to SNFs using Medicare beneficiary data.

METHOD: We conducted a secondary data analysis using 100% Medicare data of patients admitted to SNFs after hospital discharge from January 2013 through December 2014. We included only residents who completed the BIMS based on the screening question “Should Brief Interview for Mental Status be conducted? (yes/no).” We excluded residents who were rarely or never understood or could not respond to others either verbally or in writing at the time the BIMS was administered. We selected the first SNF stay for each beneficiary within the 5-day readmission of the Minimum Data Set (MDS) assessment because the BIMS is part of the MDS assessment. We included only respondents aged 65 or older at SNF admission.

The BIMS has seven items and three response rating scales with a total score ranging from 0 (lowest function/no item answered correctly) to 15 (highest function/all items answered correctly). We examined the factor structure using three model fit indices of confirmatory factor analysis (CFA): comparative fit index (CFI; >.95 for good fit), Tucker-Lewis Index (TLI; >.95 for good fit), and root mean square error of approximation (RMSEA; <.06 for good fit). We assumed that all BIMS items measure the same latent trait: basic cognitive function. The Rasch residual principle component analysis (PCA) was used to examine the underlying dimensionality of the BIMS. After the one-factor model was validated, we used Rasch analysis with the partial credit model to examine unidimensionality.

RESULTS: A total of 1,048,575 participants discharged to SNFs were included. Patients had a mean age of 78.5 (SD = 11.0; range = 65–119). The majority of patients were female (62.5%), widowed (40.1%), and White (81.7%). Two CFA model fit indexes met the criteria (CFI = .99, TLI = .98, RMSEA = .08). The Rasch PCA showed that the Rasch dimension explained 61.4% of the variance in the data, above the threshold of 40%. The eigenvalue of first contrast was 1.57 (<2.0), indicating that no meaningful residuals were left after extracting the Rasch dimension, implying unidimensionality in the BIMS. Six of seven items met the fit statistics criteria of 0.9–1.1 after adjusting for sample size. One item, repetition of three words, was the easiest item to answer and the only misfit item (infit = 1.17); around 40% of the sample had the maximum score, indicating a ceiling effect. Person strata was 1.53 based on the person separation index of 0.90.

CONCLUSION: Our findings partially support the unidimensionality of the BIMS, suggesting the BIMS can be used to measure basic cognitive function for SNF residents. However, the BIMS was too easy for almost 40% of respondents, implying the need to add more difficult items to the scale. In addition, the BIMS could barely differentiate between two cognitive levels (i.e., with or without cognitive impairment), which may limit the feasibility of the instrument. Because the BIMS was originally designed as a screening tool for patients admitted to SNFs, adding more challenging cognitive items could improve its feasibility and increase the meaningfulness of outcome interpretation. We suggest that rehabilitation practitioners, including occupational therapists, used the BIMS with cautious interpretation.

References

Mansbach, W. E., Mace, R. A., & Clark, K. M. (2014). Differentiating levels of cognitive functioning: A comparison of the Brief Interview for Mental Status (BIMS) and the Brief Cognitive Assessment Tool (BCAT) in a nursing home sample. Aging and Mental Health, 18, 921–928. https://doi.org/10.1080/13607863.2014.899971

Saliba, D., Buchanan, J., Edelen, M. O., Streim, J., Ouslander, J., Berlowitz, D., & Chodosh, J. (2012). MDS 3.0: Brief Interview for Mental Status. Journal of the American Medical Directors Association, 13,611–617. https://doi.org/10.1016/j.jamda.2012.06.004

Tariq, S. H., Tumosa, N., Chibnall, J. T., Perry, M. H., & Morley, J. E. (2006). Comparison of the Saint Louis University Mental Status Examination and the Mini-Mental State Examination for detecting dementia and mild neurocognitive disorder: A pilot study. American Journal of Geriatric Psychiatry, 14, 900–910. https://doi.org/10.1097/01.JGP.0000221510.33817.86

White, H., & Davis, P. B. (1990). Cognitive screening tests.Journal of General Internal Medicine, 5, 438–445. https://doi.org/10.1007/BF02599435