Overview of significant changes in the Minimum Data Set for nursing homes version 3.0.

The Minimum Data Set (MDS) is a standardized assessment that is completed on all residents admitted to Medicare certified nursing homes in the US. It is also completed on all residents admitted to Veteran Health Administration Community Living Centers. Its content addresses multiple domains of resident health and function and is intended to facilitate better recognition of each resident's needs. A new version of the MDS, MDS 3.0, was implemented in October, 2010. This article highlights significant clinical changes found in the MDS 3.0, including new structured resident interviews to assess mood, preferences, pain and cognition; inclusion of the Confusion Assessment Method to screen for delirium; revised psychosis and behavior items; revised balance and falls sections; revised bladder and bowel assessment items; revised pressure ulcer assessment items; revisions to the nutrition items; items reporting on resident expectations for return to the community; and changes to race/ethnicity item and language report. These changes aim to improve the clinical utility of these assessment items.

Making the investment count: revision of the Minimum Data Set for nursing homes, MDS 3.0.

BACKGROUND: The Minimum Data Set (MDS) is a potentially powerful tool for implementing standardized assessment in nursing homes (NHs). Its content has implications for residents, families, providers, researchers, and policymakers, all of whom have expressed concerns about the reliability, validity, and relevance of MDS 2.0. Some argue that because MDS 2.0 fails to include items that rely on direct resident interview, it fails to obtain critical information and effectively disenfranchises many residents from the assessment process. PURPOSE: Design a major revision of the MDS, MDS 3.0, and evaluate whether the revision improves reliability, validity, resident input, clinical utility, and decreases collection burden. DESIGN AND METHODS: In the form design phase, we gathered information from a wide range of experts, synthesized existing literature, worked with a national consortium of VA researchers to revise and test eight sections, pilot tested a draft MDS 3.0 and revised the draft based on results from the pilot. In the national validation and evaluation phase, we tested MDS 3.0 in 71 community NHs and 19 VHA NHs, regionally distributed throughout the United States. The sample was selected based on scheduled MDS 2.0 assessments. Comatose residents were excluded. A total 3822 residents of community NHs in eight states were included. The evaluation was designed to test and analyze inter-rater agreement (reliability) between research nurses and between facility staff and research nurses, validity of key sections, response rates for interview items, anonymous feedback on changes from participating nurses, and time to complete the MDS assessment. RESULTS: The reliability for research nurse to research nurse and for research nurse to facility staff was good or excellent for most items. Response rates for the resident interview sections were high: 90% for cognitive, 86% for mood, 85% for preferences, and 87% for pain. Staff survey responses showed increased satisfaction with clinical relevance, validity and clarity compared with MDS 2.0. The test version of the MDS 3.0 took 45% less time for facilities to complete. IMPLICATIONS: Improving the reliability, accuracy, and usefulness of the MDS has profound implications for NH care and public policy. Enhanced accuracy supports the primary legislative intent that MDS be a tool to improve clinical assessment and supports the credibility of programs that rely on MDS.

Testing the PHQ-9 interview and observational versions (PHQ-9 OV) for MDS 3.0.

OBJECTIVES: To test the feasibility and validity of the Patient Health Questionnaire-9 item interview (PHQ-9) and the newly developed Patient Health Questionnaire Observational Version (PHQ-9 OV) for screening for mood disorder in nursing home populations. METHODS: The PHQ-9 was tested as part of the national Minimum Data Set 3.0 (MDS 3.0) evaluation study among 3822 residents scheduled for MDS 2.0 assessments. Residents from 71 community nursing homes (NHs) in eight states were randomly included in a feasibility sample (n = 3258) and a validation sample (n = 418). Each resident's ability to communicate determined whether the PHQ-9 interview or the PHQ-9 OV was initially attempted. In the validation sample, trained research nurses administered the instruments. For residents in the validation sample without severe cognitive impairment (3 MS ≥30) agreement between PHQ-9 and the modified Schedule for Affective Disorders and Schizophrenia (m-SADS) was measured with weighted kappas (κ). For residents with severe cognitive impairment (3MS <30), agreement between PHQ-9 interview or PHQ-9 OV and the Cornell Scale for Depression in Dementia (Cornell Scale) was measured using correlation coefficients. Staff impressions were obtained from an anonymous survey mailed to all MDS assessors. RESULTS: The PHQ-9 was completed in 86% of the 3258 residents in the feasibility sample. In the validation sample, the agreement between PHQ-9 and m-SADS was very good (weighted κ = 0.69, 95% CI = 0.61-0.76), whereas agreement between MDS 2.0 and m-SADS was poor (weighted κ = 0.15, 95% CI = 0.06-0.25). Likewise, in residents with severe cognitive impairment, PHQ correlations with the criterion standard Cornell Scale were superior to the MDS 2.0 for both the PHQ-9 (0.63 vs 0.34) and the PHQ-9 OV (0.84 vs 0.28). Eighty-six percent of survey respondents reported that the PHQ-9 provided new insight into residents' mood. The average time for completing the PHQ-9 interview was 4 minutes. DISCUSSION: Compared with the MDS 2.0 observational items, the PHQ-9 interview had greater agreement with criterion standard diagnostic assessments. For residents who could not complete the interview, the PHQ-9 OV also had greater agreement with a criterion measure for depression than did the MDS 2.0 observational items. Moreover, the majority of NH residents were able to complete the PHQ-9, and most surveyed staff reported improved assessments with the new approach.

MDS 3.0: brief interview for mental status.

OBJECTIVES: To test the feasibility and validity of the Brief Interview for Mental Status (BIMS) as a performance-based cognitive screener that could be easily completed by nursing home staff. The current study examines the performance of the BIMS as part of the national testing of the Minimum Data Set 3.0 (MDS 3.0) for Nursing Homes. METHODS: The BIMS was tested as part of the national MDS 3.0 evaluation study among 3822 residents scheduled for MDS 2.0 assessments. Residents were from 71 community nursing homes (NHs) in eight states. Residents were randomly included in a feasibility sample (n = 3258) and a validation sample (n = 418). Cognition was assessed with three instruments: the Brief Interview for Mental Status (BIMS), the MDS 2.0 Cognitive Performance Scale (CPS), and the Modified Mini-Mental State Examination (3MS). Trained research nurses administered the 3MS and BIMS to all subjects in the validation study. The CPS score was determined based on the MDS 2.0 completed by nursing home staff who had undergone additional training on cognitive testing. Standard cutoff scores on the 100-point 3MS were used as the gold standard for any cognitive impairment (<78) and for severe impairment (<48). Staff impressions were obtained from anonymous surveys. RESULTS: The BIMS was attempted and completed in 90% of the 3258 residents in the feasiblity sample. BIMS scores covered the full instrument range (0-15). In the validation sample, correlation with the criterion measure (3MS) was higher for BIMS (0.906, P < .0001) than for CPS (-0.739, P < .0001); P < .01 for difference. For identifying any impairment, a BIMS score of 12 had sensitivity = 0.83 and specificity = 0.91; for severe impairment, a BIMS score of 7 had sensitivity = 0.83 and specificity = 0.92. The area under the receiver operator characteristics curve, a measure of test accuracy, was higher for BIMS than for CPS for identifying any impairment (AUC = 0.930 and 0.824, respectively) and for identifying severe impairment (AUC = 0.960 and 0.857, respectively). Eighty-eight percent of survey respondents reported that the BIMS provided new insight into residents' cognitive abilities. The average time for completing the BIMS was 3.2 minutes. DISCUSSION: The BIMS, a short performance-based cognitive screener expressly designed to facilitate cognitive screening in MDS assessments, was completed in the majority of NH residents scheduled for MDS assessments in a large sample of NHs, demonstrating its feasibility. Compared with MDS 2.0 observational items, the BIMS performance-based assessment approach was more highly correlated with a criterion cognitive screening test and demonstrated greater accuracy. The majority of surveyed staff reported improved assessments with the new approach.

Why not just ask the resident?

This research evaluated a draft preference assessment tool (draft-PAT) designed to replace the current Customary Routine section of the Minimum Data Set (MDS) for nursing homes. The draft-PAT was tested with a sample of nursing home residents to evaluate survey-level administration time and noncompletion rates, as well as item-level nonresponse rates, response distributions, and test-retest reliability. Modifications to the draft-PAT were then retested with a subsample of residents. Completion times were brief (generally less than 10 minutes), and only a small percentage of residents were unable to complete the interview. Item-level nonresponse rates were low for the draft-PAT (0% to 8%) and even lower during retesting for items advanced to the national field trial (0% to 4%). Item response distributions indicated reasonable use of all options across both testing occasions, and item-level test-retest reliability was high. This study found that nursing home residents can reliably report their preferences. Eighteen items from the modified draft-PAT were advanced to the national field trial of the MDS 3.0. Inclusion of the PAT in the MDS revision underscores increased emphasis on including residents' voice in the assessment process.

Nursing home assessment of cognitive impairment: development and testing of a brief instrument of mental status.

OBJECTIVES: To test the accuracy of a brief cognitive assessment of nursing home (NH) residents and to determine whether facility nurses can reliably perform this assessment. DESIGN: Cross-sectional, independent cognitive screening tests with NH residents. SETTING: Six Department of Veteran Affairs nursing facilities. PARTICIPANTS: Three hundred seventy-four residents from six regionally distributed Veteran Affairs NHs. MEASUREMENTS: Three cognitive assessment instruments: the Brief Interview of Mental Status (BIMS), created for this study; the Minimum Data Set (MDS) 2.0 Cognitive Performance Scale (CPS), and the Modified Mini-Mental State Examination (3MS) as the criterion standard. The 15-point BIMS tests memory and orientation and includes free and cued recall items. Research assistants administered the 3MS and BIMS to all subjects. Facility nurses administered the same BIMS to a subsample. RESULTS: Three hundred seventy-four of 417 (89.7%) residents approached completed the 3MS and research assistant-administered BIMS (BIMS-R); 212 residents also received a facility nurse-administered BIMS (BIMS-N). The BIMS-R was more highly correlated with the 3MS than was the CPS (Pearson correlation coefficient (r)=0.79 vs 0.62; P<.01 for difference). For the subset who received facility assessments, the BIMS-N was also more highly correlated with the 3MS (Pearson r=0.74 vs 0.65; P<.01 for difference). For any impairment (3MS<78), the area under the receiver operator characteristic curve (AUC) was 0.86 for the BIMS, versus 0.77 for the CPS. For severe impairment (3MS<48) the AUC was 0.94, versus 0.85 for the CPS. CONCLUSION: In this population, a brief cognitive test is a more accurate approach to cognitive assessment than the current observational methods employed using the MDS 2.0.

What the resident meant to say: use of cognitive interviewing techniques to develop questionnaires for nursing home residents.

PURPOSE: Emphasis on consumer-centered care for frail and institutionalized older adults has increased the development and adaptation of surveys for this population. Conventional methods used to pretest survey items fail to investigate underlying sources of measurement error. However, the use of the cognitive interview (CI), a method for studying how respondents answer survey items, is not well established or documented in this population. This study demonstrates how CIs can be used to improve questionnaires intended for nursing home residents. DESIGN AND METHODS: CIs were conducted with 29 nursing home residents in order to identify potential problems with prospective survey items. We used scripted probes to standardize the interviews and adapted the Question Appraisal System to enumerate and classify the problems discovered. RESULTS: We fielded between one and five versions of each item in an iterative process that identified 61 item-specific problems. Additionally, residents' cognitive responses suggested that some screened their answers on the basis of perceived physical and environmental limitations, and some had difficulty answering items about preferences that fluctuate day to day. These findings led us to modify the items and response set to simplify the respondents' cognitive task. IMPLICATIONS: This study illustrates how CI techniques can be used to understand residents' comprehension of and response to survey items.

The development of a CAHPS instrument for Nursing Home Residents (NHCAHPS).

We report on a federal initiative to develop a CAHPS (The Consumer Assessment of Healthcare Providers and Systems) survey to measure residents' experiences with quality-of-care and quality-of-life in nursing homes (known as NHCAHPS). We focus on how we created and tested questions for inclusion in the instrument and tested a possible cognitive screener to determine which residents could participate in a NHCAHPS interview. The major lessons learned were: (1) In contrast to other CAHPS surveys, ratings were more useful than reports because of the difficulty that residents had with summarizing over time and people; (2) consistent with other CAHPS surveys, the 0 to 10 response scale appeared to work well with nursing home residents for many of the quality-of-care questions; however, a different response scale was needed for many of the quality-of-life items; and (3) in contrast with typical survey methodology and other CAHPS surveys where explicit time reference periods are used, a non-specific present reference period in questions seemed to work best.

Derivation and confirmation of scales measuring medical directors' attitudes about the hospitalization of nursing home residents.

OBJECTIVE: To derive and confirm scales measuring medical director's attitudes about hospitalization of nursing home residents. METHOD: The authors surveyed nursing facility medical directors about the necessity of hospitalizing residents for eight clinical conditions and compared the ratings to those obtained from an expert panel to derive a relative hospitalization score. They also asked about factors that might influence hospitalization decisions. They performed a factor analysis to derive scales that measure attitudinal determinants of hospitalization and used the relative hospitalization score to confirm the scales. RESULTS: The survey had a 79% response rate. The relative hospitalization score demonstrated that medical directors were slightly less likely to recommend hospitalization than expert panel physicians. Factor analyses yielded 10 scales focusing on nursing home functioning, economics, resident specific considerations, and physician attitudes. Eight of the 10 scales had significant bivariable associations with the relative hospitalization score, and 6 had significant multivariable associations. DISCUSSION: Medical directors identify multiple determinants of hospitalization for nursing facility residents across several domains. Hospitalization decisions for nursing facility residents are complex and involve clinical and nonclinical factors.

Nursing home capabilities and decisions to hospitalize: a survey of medical directors and directors of nursing.

OBJECTIVES: To obtain information from decision makers about attitudes toward hospitalization and the factors that influence their decisions to hospitalize nursing home residents. DESIGN: Cross-sectional survey. SETTING: Four hundred forty-eight nursing homes, 76% of which were nonprofit, from 25 states. PARTICIPANTS: Medical directors and directors of nursing (DONs). MEASUREMENTS: Participants were surveyed about resource availability, determinants of hospitalization, causes of overhospitalization, and nursing home practice. RESULTS: The survey response rate was 81%, with at least one survey from 93% of the facilities. Medical directors and DONs agreed that resident preference was the most important determinant in the decision to hospitalize, followed by quality of life. Although both groups ranked on-site doctor/nurse practitioner evaluation within 4 hours as the least accessible resource, they did not rank doctors not being quickly available as an important cause of overhospitalization. Rather, medical directors perceived the lack of information and support to residents and families around end-of-life care and the lack of familiarity with residents by covering doctors as the most important causes of overhospitalization. DONs agreed but reversed the order. Medical directors and DONs expressed confidence in provider and staff ability, although DONs were significantly more positive. CONCLUSION: Medical directors and DONs agree about most factors that influence decisions to hospitalize nursing home residents. Patient-centered factors play the largest roles, and the most important causes of overhospitalization are potentially modifiable.

Function and response of nursing facilities during community disaster.

OBJECTIVES: We sought to describe the role and function of nursing facilities after disaster. METHODS: We surveyed administrators at 144 widely dispersed nursing facilities after the Los Angeles Northridge earthquake. RESULTS: Of the 113 (78%) nursing facilities that responded (11 365 beds), 23 sustained severe damage, 5 closed (625 beds), and 72 lost vital services. Of 87 nursing facilities implementing disaster plans, 56 cited problems that plans did not adequately address, including absent staff, communication problems, and insufficient water and generator fuel. Fifty-nine (52%) reported disaster-related admissions from hospitals, nursing facilities, and community residences. Nursing facilities received limited postdisaster assistance. Five months after the earthquake, only half of inadequate nursing facility disaster plans had been revised. CONCLUSIONS: Despite considerable disaster-related stresses, nursing facilities met important community needs. To optimize disaster response, community-wide disaster plans should incorporate nursing facilities.

Tube-feeding versus hand-feeding nursing home residents with advanced dementia: a cost comparison.

OBJECTIVE: To compare the costs associated with caring for severely demented residents nursing homes with and without feeding tubes. DESIGN: Retrospective cohort study. SETTING: A 700-bed long-term care facility in Boston. PARTICIPANTS: Nursing home residents aged 65 years and over with advanced dementia and eating problems for whom long-term feeding tube had been discussed as a treatment option. MEASUREMENTS: Costs were compared over the 6 months that followed the tube-feeding decision for those residents who did and did not undergo feeding tube placement for the following items: nursing time, physician assessments, food, hospitalizations, emergency room visits, diagnostic tests, treatment with antibiotics and parenteral hydration, and feeding tube insertion. RESULTS: Twenty-two subjects were included, 11 were tube-fed (mean age 84.3 years +/- 6.0) and 11 were hand-fed (mean age 90.2 years +/- 9.1). The daily costs of nursing home care were higher for the residents without feeding tubes compared with residents with tubes ($4219 +/- 1546 vs $2379 +/- 1032, P = 0.006). Nonetheless, Medicaid reimbursement to nursing homes in at least 26 states is higher for demented residents who are tube-fed than for residents with similar deficits who are not tube-fed. Costs typically billed to Medicare were greater for the tube-fed patients ($6994 +/- 5790 vs. $959 +/- 591, P < 0.001), primarily because of the high costs associated with initial feeding tube placement and hospitalizations or emergency rooms visits for the management of complications of tube-feeding. CONCLUSIONS: Nursing homes are faced with a potential fiscal incentive to tube-feed residents with advanced dementia: tube-fed residents generate a higher daily reimbursement rate from Medicaid, yet require less expensive nursing home care. From a Medicare perspective, tube-fed patients are expensive due to the high costs associated with feeding tube placement and acute management of complications. Further work is needed to determine whether these potential financial incentives influence tube-feeding decisions in practice.

Evaluating the planned substitution of the minimum data set-post acute care for use in the rehabilitation hospital prospective payment system.

OBJECTIVE: The objective of this study was to evaluate the payment implications of substituting the Minimum Data Set-Post Acute Care (MDS-PAC) for the FIM trade mark instrument for use in the planned prospective payment system (PPS) for inpatient rehabilitation hospitals. FIM trade mark is a trademark of the Uniform Data System for Medical Rehabilitation, a division of UB Foundation Activites, Inc. RESEARCH DESIGN: We used a prospective cross-sectional design using consecutive sampling. SUBJECTS: We studied all Medicare admissions with stays of 3 days or more over a 2-month period to 50 inpatient rehabilitation hospitals in 22 states. MEASUREMENTS AND METHODS: Each participating institution completed both the FIM and the MDS-PAC assessments on all participants. Items from the MDS-PAC were combined and translated to create "FIM-like" items. We assessed agreement of classification into prospective payment cells using FIM assessment data and also using MDS-PAC data. Statistical adjustments were applied to improve the level of agreement. RESULTS: The mean differences between the FIM motor and cognitive scales and their MDS-PAC translations were 2.4 (mean = 45) and 0.0 (mean = 28), respectively, with scale correlations of.85 and.84. Weighted kappas on individual items ranged from.32 to.64. There were substantial hospital-specific differences in scoring. Payment cell classification using FIM data agreed with that using MDS-PAC data only 56% of the time. Twenty percent of the facilities experienced revenue shifts larger than 10%. CONCLUSION: Despite better item-level agreement than previously observed, poor payment cell agreement and substantial revenue shifts indicated that the MDS-PAC should not be substituted for the FIM trade mark instrument in the rehabilitation hospital PPS.

An assessment tool translation study.

Policymakers hoped to substitute a new, multi-purpose, functional assessment instrument, the minimum data set post-acute care (MDS-PAC), into the planned prospective payment system (PPS) for inpatient rehabilitation hospitals. PPS design requires a large database linking treatment costs with measures of the need for care, so the PPS was designed using the functional independence measure (FIM) database linked to Medicare hospital claims. An accurate translation from the MDS-PAC items to FIM--like items was needed to ensure payment equity under the substitution. This article describes the translation efforts and some of the problems that led policymakers to abandon the effort.

Tube-feeding versus hand-feeding nursing home residents with advanced dementia: a cost comparison.

OBJECTIVE: To compare the costs associated with caring for severely demented residents nursing homes with and without feeding tubes. DESIGN: Retrospective cohort study. SETTING: A 700-bed long-term care facility in Boston Participants: Nursing home residents aged 65 years and over with advanced dementia and eating problems for whom long-term feeding tube had been discussed as a treatment option. MEASUREMENTS: Costs were compared over the 6 months that followed the tube-feeding decision for those residents who did and did not undergo feeding tube placement for the following items: nursing time, physician assessments, food, hospitalizations, emergency room visits, diagnostic tests, treatment with antibiotics and parenteral hydration, and feeding tube insertion. RESULTS: Twenty-two subjects were included, 11 were tube-fed (mean age 84.3 years +/- 6.0) and 11 were hand-fed (mean age 90.2 years +/- 9.1). The daily costs of nursing home care were higher for the residents without feeding tubes compared with residents with tubes ($4219 +/- 1546 vs $2379 +/- 1032, P = 0.006). Nonetheless, Medicaid reimbursement to nursing homes in at least 26 states is higher for demented residents who are tube-fed than for residents with similar deficits who are not tube-fed. Costs typically billed to Medicare were greater for the tube-fed patients ($6994 +/- 5790 vs. $959 +/- 591, P < 0.001), primarily because of the high costs associated with initial feeding tube placement and hospitalizations or emergency rooms visits for the management of complications of tube-feeding. CONCLUSIONS: Nursing homes are faced with a potential fiscal incentive to tube-feed residents with advanced dementia: tube-fed residents generate a higher daily reimbursement rate from Medicaid, yet require less expensive nursing home care. From a Medicare perspective, tube-fed patients are expensive due to the high costs associated with feeding tube placement and acute management of complications. Further work is needed to determine whether these potential financial incentives influence tube-feeding decisions in practice.

Transmission of financial incentives to physicians by intermediary organizations in California.

Many U.S. physicians participate in provider-sponsored organizations that act as their intermediaries in contracting with managed care plans, particularly where capitation contracts are used. Examining a survey of 153 intermediary entities in California, we trace the cascade of financial incentives from health plans through physician organizations to primary care physicians. Although the physician organizations received the vast majority (84 percent) of their revenues through capitation contracts, most of the financial risk related to utilization and costs was retained at the group level. Capitation of primary care physicians was common in independent practice associations (IPAs), but payments typically were restricted to primary care services. Thirteen percent of medical groups and 19 percent of IPAs provided bonuses or withholds based on utilization or cost performance, which averaged 10 percent of base compensation.