The conduct of this study was approved by the Institutional Review Board at Hebrew Senior Life.
Participant facilities were required to have greater than 60 beds, and be located within a 60-mile radius of metropolitan Boston. Participant facilities represented a convenience sample of NHs meeting these criteria. Federal Wide Assurance numbers are needed to conduct research in NHs and were obtained for all facilities.
Based on the literature,5,13–15
facility variables potentially associated with the quality of end-of-life care were obtained, a priori, from the On-Line Survey of Certification of Automated Records (OSCAR),15
and the Center for Medicare and Medicaid (CMS) Nursing Home Compare website.16
For comparison purposes, these data were also ascertained for all NHs across the US. Demographic data included: the number of beds, racial/ethnic profile (% non-whites), sex (% female), whether the proportion of Medicaid beds exceeded 85%, and whether the facility had a special care dementia unit. Ownership variables included for-profit status and whether the facility was part of a corporate chain. Staffing variables included the number of licensed nurses and nursing assistants per resident per hour per day. Finally, quality indicators included the proportion of high-risk residents with pressure sores, the use of trunk or limb restraints, and the number of deficiencies on the most recent state inspection.16
Quality indicators were not available from one facility because it was licensed as a hospital rather than a NH.
The study population was comprised of 2 groups of subjects (ie, dyads): (1) NH residents with advanced dementia and (2) their health care proxies (HCPs).
Eligibility criteria for NH residents with advanced dementia included the following: (1) age ≥65 years, (2) length of NH stay >30 days, (3) Cognitive Performance Score (CPS) equal to 5 or 6,17,18
(4) cognitive impairment due to dementia (any type), (5) Global Deterioration Scale (GDS) equal to 7,19
and (6) a formally or informally appointed HCP was available who could communicate in English. HCPs had to communicate in English because the instruments included in their interviews had not been translated into other languages. Residents in short-term rehabilitation units and those in a coma were excluded. Finally, residents with other causes of severe cognitive impairment (ie, stable major stroke, traumatic brain injury, tumor, or chronic psychiatric conditions) were excluded because their end-of-life course would differ from those with a progressive dementia.
The CPS scale uses 5 variables from the Minimum Data Set (MDS)20
to group residents into the following 7 hierarchical cognitive performance categories; 0=intact, 1=borderline intact, 2=mild impairment, 3=moderate impairment, 4=moderately severe impairment, 5=severe impairment, and 6=very severe impairment. 17,18
The CPS was chosen to screen large NH populations for severe cognitive impairment for the following reasons: (1) the MDS is a federally mandated, standardized assessment instrument, completed on all residents living in licensed US NHs, (2) according to the CPS algorithm, residents are categorized as having a score of 5 or 6 based on single MDS variable; cognitive skills for daily decision-making must be equal to 3 (range 0 to 3, 3 indicates severe impairment), and (3) CPS score is a validated measure of cognitive disability. A CPS score of 5 corresponds to mean Mini-Mental State Examination score of 5.1±(SD)5.3.18
In addition to the CPS score, subjects were also required to be at GDS stage 7. The GDS classifies dementia into 7 stages (1 to 7) based on broad descriptions of the cognitive and functional deficits that typify each stage.19
Although there has been limited validation of the GDS, it is a widely recognized instrument classifying the late stages of dementia. Stage 7 of the GDS is distinguished by the following features: very severe cognitive decline with minimal to no verbal communication, assistance needed to eat and toilet, incontinence of urine and stool, and loss of basic psychomotor skills (eg, may have lost ability to walk).
A 2-stage screening protocol was used to identify eligible subjects. To conduct this screening procedure, a waiver of individual authorization for disclosure of personal health information was granted by the Institutional Review Board at Hebrew Senior Life.
As a first step, computerized MDS data provided by the facility identified residents with CPS scores of 5 or 6 on their most recent assessment. If the facility’s MDS data were not computerized, nurses on all units were asked to identify residents with a CPS score of 5 or 6. In the second step, a research assistant (RA) reviewed the charts of residents with a CPS score of 5 or 6 for eligibility criteria, including dementia as a cause of the cognitive impairment. If the diagnosis of dementia was ambiguous, the residents’ primary care physicians were contacted for their opinions.
An information sheet describing the study was mailed to the HCPs of eligible residents. Unless HCPs requested no further contact, they were telephoned 1 week later to obtain informed consent. HCPs provided consent for themselves and for the residents with advanced dementia.
Screening at each facility was conducted in the aforementioned fashion at the time of initial NH recruitment, and on a quarterly basis thereafter. Entry of individual facilities into the study was staggered because of the intense effort required at the time of initial screening.
Data Collection Protocol
Resident assessments were conducted at baseline and quarterly thereafter for up to 18 months or until death (). If the resident died, a death assessment was completed within 14 days. Resident assessments involved a chart review, a nursing interview, and a brief clinical examination. The follow-up period of 18 months was chosen as a reasonable time to ensure an adequate number of clinical events and deaths would occur to analyze study outcomes.
FIGURE 1 The CASCADE study data collection protocol. Assessments of residents with advanced dementia were conducted at baseline and quarterly intervals thereafter for up to 18 months or until death. If the resident died, a death assessment was completed within (more ...)
HCP assessments were conducted by telephone interviews at baseline and on a quarterly basis thereafter for a maximum of 18 months. Every effort was made to conduct HCP interviews within 14 days of the corresponding resident assessment. If the resident died, HCPs were interviewed 2 and 7 month postdeath. Two months was chosen for the first postloss interview to allow enough time to pass after death to maximize the HCP participation, yet proximal enough to death to allow them to remember the circumstances around the dying process. The timing of the 7-month postloss assessment was chosen because a diagnosis of complicated grief requires at least 6 months of symptomatic distress.
HCPs were dropped from the study if they could not be contacted for 2 consecutive assessment periods. If HCPs were lost to follow-up or withdrew, residents within that dyad remained in the study unless the HCPs specifically requested their withdrawal.
Data were collected by 3 trained RAs. In most cases, an individual RA was assigned to specific facilities and collected data from both the resident and HCP comprising a single dyad. These assignments were intended to enhance the establishment of an ongoing relationship between an individual RA with facility staff and HCPs.
Data Collection Elements
Valid and reliable instruments were used to measure data elements whenever possible (), and measures specifically designed for use in advanced dementia were used if available. The variables needed to achieve the specific aims of the CASCADE study were selected based on the literature,2–14,21–30
and clinical experience. All items had closed-ended response options.
CASCADE Data Collection Elements
Resident data abstracted from the NH medical records included: demographic information, comorbid health conditions, medications, advance care planning, interventions, health care utilization, sentinel events, discomfort, and nutritional markers. Nurse interviews provided data regarding: functional status, physical comfort, and eating patterns. Direct clinical examination of the residents was used to assess cognition, and to obtain the anthropomorphic measures needed to calculate body mass index.
Demographic data included: age, sex, race, ethnicity, religious affiliation, education, primary language, marital status, length of NH stay, and prior living situation. All comorbid health conditions were identified at baseline. Daily medications were determined from the Medical Administration Record at each assessment. Detailed information was gathered regarding the specific administration of antibiotics, analgesics, and antipsychotic medications.
The following advance directives were ascertained at all assessments: do-not-resuscitate, do-not-hospitalize, do-not-intubate, and withholding feeding tubes, parenteral therapy, or antibiotics. The presence of a living will and the directives described within was also determined. Use of interventions included: feeding tubes, parenteral therapy, venipunctures, artificial ventilation, trunk or limb restraints, and blood transfusions.
Sentinel events, defined as new medical conditions that had the potential to lead to a significant change in health status and a shift in the goals of care (eg, pneumonia, hip fracture, etc.), occurring between follow- up assessments were identified and details regarding the management and outcome of these events were ascertained. The chart was also abstracted for the use of the following health services: hospitalizations, emergency room visits, intensive care admissions, hospice referrals, involvement of nurse practitioners and physician assistants, visits by primary care providers, and whether the resident resided in a special dementia care unit. To assess nutritional status, measures of body weight, serum albumin, and hematocrit recorded in the medical chart within 30 days of each assessment were obtained.
Physical comfort during the prior 90 days was measured by nursing interview using the Symptom Management at the End-of-Life in Dementia scale at baseline and all follow-up assessments.31
At the death assessment the Comfort Assessment in Dying with Dementia scale was used to describe symptoms during last 7 days of life.31
These scales were specifically designed to quantify physical discomfort in advanced dementia. The Quality of Life in Late-Stage Dementia was administered to nurses at all time periods to assess overall quality of life.32
Commonly used scales to measure functional status (eg, Katz ADL) are not sensitive to change in advanced disease. Therefore, the Bedford Alzheimer’s Nursing Severity-Subscale (BANS-S) (range 7 to 28; higher scores indicate more functional disability) was used to measure functional disability at all assessments.28
The Functional Assessment Staging (FAST) was also collected.33
The FAST consists of 7 major stages, with stage 7 representing the most advanced phase of dementia. Nurses also provided detailed information regarding feeding problems, oral intake at each meal, pressure ulcer risk (Braden Scale),34
and the existence of any pressure ulcers (categorized as stages 1 to 4).35
Cognitive assessments were conducted by direct clinical examination of the resident at each time period using the Test for Severe Impairment (range 0 to 24, lower scores indicate greater cognitive impairment),36
a validated and reliable measure of cognition in persons with advanced disability. As most residents were bedfast, knee height was used to estimate the resident’s height to calculate the body mass index.37
Finally, if the resident died, the primary and contributing causes of death recorded on the death certificate were ascertained. Death certificates were obtained from municipal offices if a copy was not in the resident’s chart.
Data gathered during the HCP interviews included: demographic information, advance care planning, communication, satisfaction with care, decision-making, health status, mood, and bereavement adjustment.
Demographic data included: age, sex, race, ethnicity, primary language, education, household income, religious affiliation, religiousity (importance of faith, frequency of religious services attendance), marital status, work status, relation to the resident, the number of years as HCP, and proximity of the HCPs’ primary residence to the NH.
At the baseline and each quarterly follow-up interview, HCPs were asked about the following advance care planning issues: primary goal of care (life-prolongation vs. comfort), their understanding of the health problems expected in advanced dementia, whether they perceived dementia as a terminal condition, and how close they believed the resident was to the end of life. HCPs were also asked whether the residents previously specified their care preferences to them (verbally or in a living will). At each quarterly follow-up, HCPs stated whether any of the residents’ advance directives (eg, DNR) had changed since the prior interview.
In terms of communication, HCPs were asked with whom (ie, physician, social worker) and for how long they discussed the residents’ advance directives at the time NH admission (baseline interview), and at the time of any changes (quarterly interviews). HCPs were also asked whether a NH physician did any of the following: sought the HCP’s opinion regarding goals of care, counseled the HCP about the expected health problems in advanced dementia, explained comfort and life-prolonging care, and provided prognostic counseling.
Satisfaction with overall end-of-life care was assessed at each time period using The Satisfaction with Care at the End-of-Life in Dementia Scale (SWCEOLD). 31
In addition, The After-Death Family Interview was administered at the 2 months postdeath interview. 27,30
The After-Death Family Interview specifically focused on care provided during the dying process in the following domains: physical comfort, shared decision-making, communication, treatment of the resident with respect, and tending to the emotional family needs.
Decision-making was assessed at each follow-up interview. HCPs were asked whether they had made any major health care decisions for the resident since the prior interview and if so, the nature of that decision (ie, pneumonia treatment, hospitalization, feeding problem). If HCPs’ had made a decision, their satisfaction with the decision-making process was measured using the Decision Satisfaction Inventory.38
The well-validated SF-1239
was used to measure the HCPs’ health status at each interview. In addition, HCPs were asked whether they experienced any hospitalizations, emergency room visits, or physician visits since the prior assessment. Mood was assessed with the K6, a 6-item instrument assessing symptoms of psychological distress used in the National Health Interview Survey, the World Health Organization’s World Mental Health (WMH) Initiative, and elsewhere.40
Finally, The Inventory of Complicated Grief (ICG),41,42
was administered to HCPs 2 and 7 months after the residents’ death. The ICG assesses symptoms of complicated mourning (eg, separation and traumatic distress). A related instrument, the ICG-Short Form: Pre-loss was also administered to the HCP during the baseline interview to test the hypothesis that HCPs who have complicated grief before the residents’ death are more likely to experience this condition after their death.