| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
a Department of Preventive Medicine, University of Tennessee, Memphis
b College of Nursing, University of Tennessee, Memphis
Correspondence: Marshall J. Graney, Department of Preventive Medicine, University of Tennessee, Memphis, 66 North Pauline Street, Suite 633, Memphis, TN 38163. E-mail: mgraney{at}utmem.edu.
Decision Editor: Vernon L. Greene, PhD
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Nursing home • Functional health • Assessment
The Omnibus Budget Reconciliation Act (OBRA) of 1987 specified development of a uniform nursing home resident assessment instrument as part of a larger initiative to improve the quality of nursing home care. The Resident Assessment Instrument, including both the Minimum Data Set (MDS) for resident assessment and Resident Assessment Profiles (RAPs) for standardized care planning that are triggered by selected MDS items, were developed in response to that legislation. The MDS and its RAPs were implemented nationwide in 1989 for every nursing home resident in the United States receiving care reimbursed by Medicare or Medicaid. Since 1989, the uses of the MDS have expanded substantially beyond its original purpose of uniform resident assessment and care planning to include quality indicators and prospective reimbursement for care.
Concurrently, issues regarding implementation of the MDS in the clinical setting, psychometric properties of the MDS, and refinement of the MDS are being addressed both by the researchers who developed the MDS and by independent researchers who were not involved in the development of the MDS. As part of the current ongoing refinement of the MDS by independent researchers, the purpose of this study was to determine whether any single assessment of a resident's ADL performance may accurately evaluate the resident's performance compared with multiple and repeated assessments. Twenty-one repeated assessments within a 1-week period of time—including all three shifts—are recommended by the MDS directions, but research support for this recommendation was not documented by the MDS developers.
The MDS is a complex and often time-consuming instrument for staff to complete. The version of the MDS that is completed on admission contains approximately 450 items, and the MDS completed quarterly (every 3 months) contains approximately 250 items, with about half of the items used to trigger RAPs for care planning. These MDS data are supposed to be based on multiple assessments across time, ranging from 7 days to 90 days, and from multiple sources such as staff, consultants, family members, and residents.
It takes approximately 1–10 hours to complete the MDS (Teresi and Holmes 1992
), and the task may be delegated to nursing home staff directly involved in patient care (Teresi and Holmes 1992), to supervisory nursing home staff (Teresi and Holmes 1992), or to an "MDS nurse," whose sole responsibility is to perform MDS assessments (Ouslander 1997). The massive amounts of paperwork and staff time committed to the MDS are caused by the number of required MDS assessments throughout the year, the number of nursing home residents assessed, and the amount of time needed to complete each MDS assessment. Consequently, the MDS may often be viewed as unnecessary paperwork by nursing home staff (Ouslander 1994), as just another form to complete (Zulkowski 1999). "To the average floor nurse filling out the MDS it is just one task among many crying out to be done simultaneously" (Brooks 1996, p. 45).
An "MDS nurse" may take the responsibility for "paper compliance" with federal regulations to decrease the burden on staff (Crooks, Schnelle, Ouslander, and McNees 1995) and to help assure few or no citations by regulators for deficiencies of incomplete MDS assessments. In an article by Brooks 1996 on the use of the MDS in the clinical setting, researcher A. Morse asked, "Are you better off with fewer items with more validity or more items with no validity?" (p. 46) and computer systems manager C. Kowalski stated, "One of the biggest concerns is the value of all of this labor. We may be finding we could actually use 30% of data" (p. 46). Therefore, nursing home staff may be finding creative ways to minimize their time spent completing the MDS, such as completing all MDS items using data obtained at one point in time from one source (rather than from the multiple and possibly redundant and uninformative reassessments required by the MDS) or not scoring MDS items as positive that may trigger a RAP for care planning. Thus, accuracy of the MDS assessments, the appropriateness of subsequent care planning activities, and reimbursement for care may all be compromised. Burdensome assessments may compete for time with delivery of resident care, and assessments completed quickly instead of accurately may adversely affect quality of care, defeating the intent of OBRA.
Studies subsequent to the nationwide implementation of the MDS suggested that refinements of the MDS may be needed. For example, there was no reliable association found between research staff "wet checks" for incontinence and nursing home staff MDS incontinence assessments (Crooks et al. 1995). Some residents were observed by research staff as wearing glasses, but were coded as not wearing glasses on the MDS (Swanson 1995). Further, the number of dental visits has been found to be inversely related to MDS dental items (Thai, Shuman, and Davidson 1997), and Arvidson-Berjano, Blank, and Yellowitz (1996) suggested condensing two of the MDS oral hygiene items into a single item. Brandeis, Berlowitz, Hossain, and Morris 1995 found that only 14 of 22 MDS items triggering the pressure ulcer RAP were reliably associated with Stage II to IV pressure ulcers, whereas Zulkowski 1999 found that 18 other MDS items not identified as RAP triggers were correlated with pressure ulcers. Thus, some MDS data inaccurately represent a resident's status, and fewer MDS items may be needed to assess some domains and appropriately trigger RAPs. Revisions of the MDS to address these issues may result in a shorter but more clinically feasible resident assessment, decreasing nursing home staff burden while at the same time improving accuracy of the MDS and thereby enhancing care planning. Additional training of nursing home staff for MDS assessments (Lawton et al. 1998) and competency testing of MDS evaluators may also be necessary, such as that already required for use of the Functional Independence Measure (FIM), an established instrument used extensively in rehabilitation settings (Williams, Li, Fries, and Warren 1997).
Assessment of ADLs is of critical importance in the nursing home setting, because functional abilities of residents are directly related to cost of care provided by nursing assistants (Smith, Hogan, and Rohrer 1987). Research evaluating a MDS composite indice of ADLs, using data collected by nursing home staff rather than by research staff, indicated that the validity of ADL assessments was good (Lawton et al. 1998). However, it is unknown whether staff actually used the required number of 21 observations to obtain ADL data for their MDS assessments, because nursing home staff may not follow MDS protocols (Morris et al. 1994). We believe that, because of time constraints, nursing home staff are generally using fewer than 21 observations.
We were unable to identify any studies related to the development of the MDS that supported using either 7 days of assessments or observations during all three shifts each day. Generally, the MDS and FIM are used to measure changes in status over longer periods of time, such as 3 months. Theoretically, it might be necessary to assess ADLs for 7 days because a resident's status may change over time, and it might be necessary to assess ADLs across all three shifts because residents may be fatigued at the end of the day. However, previous research on stability of 55 White residents' ADLs on the day shift documented stability of 10 ADLs during a 4-day period of time following nursing home admission using Scaled Outcome Criteria (Engle 1985). The day shift was chosen for study because the majority of ADLs occur during the day shift. In another study of 647 Black and White nursing home residents, hygiene, grooming, dressing, and transferring improved during a 2-week period, whereas feeding, ambulation, urination, and defecation were stable across three assessments (Day 1, Day 8, Day 15) following admission (Engle and Graney 1993). As regards the stability of ADLs across all three shifts, Ouslander, Schnelle, Simmons, Bates-Jensen, and Zeitlin 1993 found that residents were consistently incontinent day and night. Thus, several studies have established that most resident ADLs may be stable during 1- and 2-week periods when assessed on different shifts.
This prospective longitudinal study evaluated the equivalence of multiple assessments of ADL performance over a 7-day period within the required 14 days following nursing home admission using the MDS, with data collected during the day shift on 3 different days. The purpose of the study was to determine if any single assessment may evaluate ADL performance as accurately as multiple, repeated assessments.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
), nursing assistants have the most accurate resident ADL data (Morris et al. 1990), and nursing staff complete the MDS by obtaining data from nursing assistants. Nursing assistants were blind to the purpose of the study.
Site
Residents enrolled in the study were newly admitted or readmitted to either of two large county-financed nursing homes, managed by a for-profit corporation, that have historically provided care to indigent Black and White older adults. The nursing homes were dually certified for skilled and intermediate care and had 450 and 314 beds, respectively.
Sample
Residents enrolled in the study met the criteria of (a) being a new admission or readmitted to the nursing home, (b) remaining in the nursing home during the first 14 days following admission, and (c) not refusing to participate in the study. A waiver for consent was granted by the University of Tennessee Health Science Center's Institutional Review Board because this study's data were similar to data routinely collected in a nursing home for planning care. Both residents able and unable to give consent were asked to participate. Residents were not included in the study if they and/or their family expressly declined participation.
Residents in the first half of the sample were enrolled in the study as sequentially admitted to either of the two participating nursing homes. However, the three assessments placed a burden on the nursing staff. Consequently, to not jeopardize conduct of the larger study, residents in the second half of the sample were selected ad libitum from the series of later admissions.
Sample Size
A power analysis was accomplished at the time of study design to estimate the number of participants needed in a repeated measures study to obtain statistically significant evidence of the average within-subject ADL effect sizes previously documented in studies using non-MDS data. That statistical power analysis determined that a sample size of 54 or more residents would provide power of 0.80 or better to reject a false null hypothesis of no within-subject difference between measurements. This estimate assumed use of a two-tailed test at the .05 level of statistical significance, and was based on the average two-week within-subject effect size of 0.24 documented in non-MDS ADL measures showing significant change in previous research (Engle and Graney 1993). Because of the burden to nursing assistants, data collection was ended when a sample size of 42 completed data collection protocols was attained. This shortfall in recruitment numbers reduced the power of the study to 0.70 under the original set of assumptions.
Measurement
Demographics.
Demographic data were obtained from the MDS for gender (MDS question AA2), birthdate (MDS question AA3), and race/ethnicity (MDS question AA4).
ADL Performance.
Residents' performance of ADLs was evaluated using MDS questions from "Section G. Physical Functioning and Structural Problems" (G1a to G1j, G2) and "Section H. Continence in [the] Last 14 Days" (H1a, H1b). Included in this set of assessments were two measures of continence—bowel and bladder; six measures of mobility—bed mobility, mobility off unit, mobility on unit, transferring, walking in corridor, and walking in room; and five measures of personal care—bathing, dressing, eating, hygiene, and toileting. MDS ADL performance is assigned scores ranging from independent (scored 0) to total dependence (scored 4) for MDS questions G1a to G1j, G2, H1a, and H1b. If the activity did not occur during the entire 7 days, the activity was scored 8. We recoded this lowest level of performance from "8" to "5," so ADL performance was scored on a 0 to 5 scale. This recoding affects only the descriptive statistics presented; it has no effect on the inferential statistics presented in this study.
Statistical Analysis
Resident scores were concentrated at the most impaired performance levels for most ADLs measured. Consequently, a nonparametric data analysis strategy was adopted because data conformed neither to assumptions of scale continuity nor to those of normal distributions. Statistical analysis of clinical series data was accomplished using the Friedman two-way analysis of variance for ranks (Siegel 1956), a nonparametric repeated measures analysis of variance in which several measurements of an ADL for each participant were evaluated for evidence of within-subject difference. This analysis determined if there was reliable evidence to say that there was at least one significant difference among the several within-subject measurements. Computations of test statistics were accomplished in accordance with Siegel 1956, and p values were obtained using SAS for Windows version 6.12 (SAS Institute, Inc 1990). A p value of .05 or less was the criterion for rejection of a null hypothesis.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
; Engle and Graney 1993).
Some ADLs showed changes in previous research, but there are two reasons why these changes were not detected using MDS data in the present study. First, earlier research (Engle and Graney 1993) used a 2-week period, twice as long as the stipulated 7-day MDS observation period, allowing more time for change to take place. Second, MDS ADL data were less sensitive to change because of their psychometric properties. Although the MDS (current study) and the Scaled Outcome Criteria (SOC; Engle 1985; Engle and Graney 1993, Engle and Graney 1995) both use 6-point scales, MDS calibrations are the same for all ADLs, whereas SOC calibrations are specific to each ADL assessed. The SOC calibrations also make finer distinctions of performance among more impaired residents, whereas the MDS calibrations are coarser in that they group impaired residents together in fewer scale values representing the worst ADL performance. Consequently, the MDS ADL data for nursing home residents were highly skewed.
Time demands placed on nursing home staff by only three assessments in this study reflected just part of the far greater demands required by following MDS directions for 21 ADL assessments. To comply with these greater demands, nursing home staff may disregard MDS directions, or they may document the 21 ADL assessments with flow sheets. However, nursing assistants often complete flow sheets hastily at the end of the shift, with emphasis on speed rather than on accuracy. Thus, nursing home staff may emphasize completeness, to avoid citations, instead of accuracy and may also avoid scoring MDS items that would trigger RAPs. Requirements for redundant assessments may focus staff efforts on paperwork, whereas fewer required assessments may allow staff to focus on accuracy, quality of care, and completeness of assessments. Additional research is needed to evaluate accuracy versus completeness of MDS assessments and to consider the cost of staff time required for completing MDS assessments relative to resident outcomes and quality of care.
The study sample was comprised of newly admitted or newly readmitted residents, and therefore younger residents, male residents, and Black residents were overrepresented in comparison to the general long-term care population. Research is needed to identify possible cognitive impairment (Phillips, Chu, Morris, and Hawes 1993), gender, race, and age differences in the stability of ADLs. In addition to research to identify characteristics of residents whose ADLs are unstable, and who will require more than three assessments, additional research is needed to document stability or change in ADLs across three shifts and to document the impact of multiple assessments on RAPs and care planning.
In summary, this study documented stability of three assessments of MDS ADLs during the day shift over a 7-day period. According to this study's results, fewer than the required 21 assessments may be used for accurate evaluation of residents' ADL performance using the MDS.
|
Acknowledgments |
|---|
Received for publication February 8, 1999. Accepted for publication February 22, 2000.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  B. M. Bates-Jensen, S. F. Simmons, J. F. Schnelle, and C. Alessi Evaluating the Accuracy of Minimum Data Set Bed-Mobility Ratings Against Independent Performance Assessments: Systematic Error and Directions for Improvement Gerontologist, December 1, 2005; 45(6): 731 - 738. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Mirotznik Does Cognitive Status Moderate the Health Effects of Single-Person Room Transfers on Nursing Home Residents? Gerontologist, October 1, 2002; 42(5): 634 - 642. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. E. Tinetti, D. Baker, W. T. Gallo, A. Nanda, P. Charpentier, and J. O'Leary Evaluation of Restorative Care vs Usual Care for Older Adults Receiving an Acute Episode of Home Care JAMA, April 24, 2002; 287(16): 2098 - 2105. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. F. Engle, M. J. Graney, and A. Chan Accuracy and Bias of Licensed Practical Nurse and Nursing Assistant Ratings of Nursing Home Residents' Pain J. Gerontol. A Biol. Sci. Med. Sci., July 1, 2001; 56(7): M405 - 411. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
