Development of a minimum data set-based depression rating scale for use in nursing homes

BACKGROUND

depression is common but under-diagnosed in nursing-home residents. There is a need for a standardized screening instrument which incorporates daily observations of nursing-home staff.

AIM

to develop and validate a screening instrument for depression using items from the Minimum Data Set of the Resident Assessment Instrument.

METHODS

we conducted semi-structured interviews with 108 residents from two nursing homes to obtain depression ratings using the 17-item Hamilton Depression Rating Scale and the Cornell Scale for Depression in Dementia. Nursing staff completed Minimum Data Set assessments. In a randomly assigned derivation sample (n = 81), we identified Minimum Data Set mood items that were correlated (P < 0.05) with Hamilton and Cornell ratings. These items were factored using an oblique rotation to yield five conceptually distinct factors. Using linear regression, each set of factored items was regressed against Hamilton and Cornell ratings to identify a core set of seven Minimum Data Set mood items which comprise the Minimum Data Set Depression Rating Scale. We then tested the performance of the Minimum Data Set Depression Rating Scale against accepted cut-offs and psychiatric diagnoses.

RESULTS

a cutpoint score of 3 on the Minimum Data Set Depression Rating Scale maximized sensitivity (94% for Hamilton, 78% for Cornell) with minimal loss of specificity (72% for Hamilton, 77% for Cornell) when tested against cut-offs for mild to moderate depression in the derivation sample. Results were similar in the validation sample. When tested against diagnoses of major or non-major depression in a subset of 82 subjects, sensitivity was 91% and specificity was 69%. Performance compared favourably with the 15-item Geriatric Depression Scale.

CONCLUSION

items from the Minimum Data Set can be organized to screen for depression in nursing-home residents. Further testing of the instrument is now needed.

Identifying nursing home residents at risk for falling

OBJECTIVES

To develop a fall risk model that can be used to identify prospectively nursing home residents at risk for falling. The secondary objective was to determine whether the nursing home environment independently influenced the development of falls.

DESIGN

A prospective study involving 1 year of follow-up.

SETTING

Two hundred seventy-two nursing homes in the state of Washington.

PARTICIPANTS

A total of 18,855 residents who had a baseline assessment in 1991 and a follow-up assessment within the subsequent year.

MEASUREMENTS

Baseline Minimum Data Set items that could be potential risk factors for falling were considered as independent variables. The dependent variable was whether the resident fell as reported at the follow-up assessment. We estimated the extrinsic risk attributable to particular nursing home environments by calculating the annual fall rate in each nursing home and grouping them into tertiles of fall risk according to these rates.

RESULTS

Factors associated independently with falling were fall history, wandering behavior, use of a cane or walker, deterioration of activities of daily living performance, age greater than 87 years, unsteady gait, transfer independence, wheelchair independence, and male gender. Nursing home residents with a fall history were more than three times as likely to fall during the follow-up period than residents without a fall history. Residents in homes with the highest tertile of fall rates were more than twice as likely to fall compared with residents of homes in the lowest tertile, independent of resident-specific risk factors.

CONCLUSIONS

Fall history was identified as the strongest risk factor associated with subsequent falls and accounted for the vast majority of the predictive strength of the model. We recommend that fall history be used as an initial screener for determining eligibility for fall intervention efforts. Studies are needed to determine the facility characteristics that contribute to fall risk, independent of resident-specific risk factors.

Depression in a long-term care facility: clinical features and discordance between nursing assessment and patient interviews

OBJECTIVE

Nurses commonly observe more depression than is diagnosed and treated in nursing homes. Accordingly, we aimed to describe the clinical features of untreated nursing home residents whom nurses identify as depressed and to compare nurse ratings of depressed nursing home residents with ratings from direct interviews and patient self-reports.

DESIGN

Cross-sectional survey followed by semi-structured diagnostic interviews of depressed patients and their nurses.

SETTING

A large academic, multi-level, long-term care facility.

PARTICIPANTS

Thirty-seven patients aged 74-99 (mean age 88.4) whom nurses identified as having daily symptoms of depression. Subjects had Mini-Mental State Exam (MMSE) scores > 10 (mean score 21.2), were not acutely or terminally ill, and were able to participate in an interview.

MEASUREMENTS

DSM-III-R mood diagnoses and separate ratings of interviews with nurses and patients using the Cornell Scale for Depression.

RESULTS

Nurses observed daily symptoms of depression in 110 of 495 (22%) long-term care residents on units not reserved for advanced dementia. Of these 110 patients, 58 (53%) were not receiving antidepressants. Of 37 patients eligible for interviews, nine met criteria for major depression, 20 met criteria for another non-major depression diagnosis, and eight did not have a diagnosable mood disorder. Cornell scale ratings derived exclusively from interviews of nurses were similar across the three diagnostic groups (12.5, 9.9, and 9.5, respectively; P = .31; mean 10.5), whereas Cornell scale ratings from patient interviews differed among groups (15.9, 6.9, and 4.1, respectively; P < .001; mean 8.4). Correlation between nurse Cornell ratings and patient Cornell ratings was poor (r = .27), especially for patients with non-major forms of depression (r = -.20). MMSE and Cumulative Illness Rating Scale (CIRS-G) scores were similar in the three groups.

CONCLUSIONS

Nurses frequently observed symptoms of depression in a long-term care setting, and many symptomatic patients were not being treated with antidepressants. In these patients, nurse-derived symptom ratings did not vary across DSM-III-R diagnostic categories and correlated poorly with ratings from direct patient interviews. These findings suggest that nurse caregivers may contribute important diagnostic information about non-major depression and raise questions about the application of standard diagnostic categories to late-life depression in the nursing home.