Bringing more definition to end-of-life decisions

Most nursing home residents will eventually die there. Yet only 3.1% of residents at the end of life (i.e., with fewer than six months to live) are receiving hospice care. The other 96.9% die without some or all of the specialized care that might ease their final months, weeks, and days. This article explores the challenges of identifying the resident at the end of life using a recent study by LTCQ, Inc., on 630 geographically diverse facilities in the United States.


Nursing home staff members are often surprised when a long-term resident dies. It seems odd that the clinicians providing daily care to a resident would not notice that a resident is dying. Perhaps it is because they are caring for the resident daily; the slow decline can be difficult to perceive. Another reason might be the emphasis on providing the medical or rehabilitative model of care rather than a palliative (or comfort care) one. Clinicians are trained to heal—but there is no cure for old age. Caring for actively dying residents requires specialized education, which often is not provided even though hospice care was introduced to the United States in the mid-1970s.

It is only logical that if the clinical staff are surprised when a resident dies, the family and even the resident are not prepared either. In fact, 58.1% of residents at the end of life in LTCQ's study did not have signed advance directives such as those labeled “do not resuscitate” (DNR) or “do not hospitalize” (DNH), which indicates that those residents were probably receiving unnecessary care. The LTCQ study found that 85.5% of residents at the end of life were hospitalized or sent out for emergency room visits and 40.6% had new feeding tubes inserted.

How to identify end-of-life residents

Morris et al1 developed the Personal Severity Index (PSI), derived from 25 MDS items statistically related to the likelihood of death within six months. In this scale, a resident with a score of 9 or higher would be considered to be at higher risk of death. The table shows the 25 MDS items used in the PSITable 1. It is important to note that this is a predictive model based on severity scores in these categories. Listing on this table does not necessarily mean that residents with all of these issues are at the end of life.

Table The 25 MDS Items

Items Used in Morris et al's PSI Items on Full and Quarterly MDS

Age (90+)

Cognitive Decision Making (B4 = 3)

Delirium (B5e = 2)

Ability to Understand (C6 = 2,3)

Transfer (G1bA = 3,4,8)

Locomotion (G1eA = 3,4,8)

Eating (G1hA = 3,4,8)

Personal Hygiene (G1jA=3,4,8)

Sad Mood, Repetitive Verbalizations (E1c = 2)

Sad Mood, Something Terrible About to Happen (E1g = 2)

Acute Episode (J5b)

Unstable (J5a)

Change in Care Needs (Q2 = 2)

End Stage Disease (J5c)

Bowel (H1a=2,3,4)

Weight Loss (K3a)

Pressure Ulcer (M2a = 1,2,3,4)

Stasis Ulcers (M2b = 1,2,3,4)

Items on Full MDS But Not on Quarterly MDS

Inability to Lie Flat Due to Shortness of Breath (J1b)

Oxygen Therapy (P1g)

Problem Swallowing (K1b)

Time Awake Afternoon (N1b=0)

Cardiac Dysrhythmias (I1e)

Congestive Heart Failure (I1f)

Skin Tears or Cuts (M4f)

For an example of how the PSI is scored, consider a resident who is 90 years old with severely impaired cognitive skills for daily decision making. She is dependent for eating and requires extensive assistance with transfer, locomotion, and personal hygiene. She has experienced recent weight loss, has frequent bowel incontinence, and has a stage II pressure ulcer. By using a straight count of these MDS items we see that this resident's PSI score is 9, the beginning of high risk.

The benefits of knowing a resident's PSI score are tremendous. It enables providers to set families' expectations and better avert the emotional decision making that often leads to inappropriate care. Family members will understand that nothing can be done to “save” their loved one, and they can focus on saying goodbye and making any necessary arrangements. Clinical staff can accurately assess the care the resident is receiving and ensure that it has an appropriate emphasis on pain management and depression management, and that it supports quality of life.

Typical Case

A male resident has diagnoses including congestive heart failure, cardiac dysrhythmia, and peripheral vascular disease. He has been hospitalized three times in the past year and a half and, since his most recent readmission, his PSI score has remained at 11 (high risk). He has a history of stage IV stasis ulcer and has been losing weight since the previous two hospitalizations. He is now assessed as having periods of lethargy, has pain daily, is receiving 20 medications, and has a trunk restraint used daily.

None of his MDS assessments are coded with end-stage disease (J5c), and he is not receiving hospice services (P1ao). The requirements for coding both of these items are stringent: A written physician's certification that the resident has six or fewer months to live must be obtained before coding J5c. The resident did not have a DNR order in place until after his most recent hospitalization.

The pattern of decline in residents with chronic organ failure (such as with this resident) is pointed out by Lunney et al2 as a distinct “trajectory of dying” characterized by gradual decline, with periodic exacerbations of the disease. While these exacerbations may be resolved, the resident's overall functioning continues to worsen.

In this case it appears that the decline was so gradual that it was not noticed. As a result, inappropriate interventions such as high-intensity rehabilitation and restraints were part of this resident's care plan. Lunney et al's study found that, in general, end-of-life residents were restrained more than twice as often as nonterminal residents. Because of the gradual decline, this particular resident's family will probably not be adequately prepared for the resident's death—an important risk-management issue for the facility.

More Findings

A positive outcome in LTCQ's analysis was that 98.4% of facilities having the highest 10th percentile of end-of-life residents reduced the total number of unnecessary medications given, while new medications for comfort, such as pain medications or sedatives, were added.

Although the vast majority of nursing homes do their best to provide the care that their residents need, LTCQ's study shows that providers need to more accurately identify end-of-life residents so that the needs of the resident, family, and facility staff can be fulfilled. Each group needs timely, accurate information to help them make appropriate decisions.

Jennifer K. Gross, RN, BSN, BA, and Steven B. Littlehale, MS, APRN, BC, are consultants, and Barbara Shea is Director of Marketing, for LTCQ, Inc., Lexington, Mass.

For further information, phone (781) 457-5916 or visit To send your comments to the authors and editors, e-mail


  1. Morris JN, Jones R, Morris S, Fries BE. Attachment 13: Proximity to death, a modeling tool for use in nursing homes. Hebrew Rehabilitation Center for Aged Research and Training Institute; CMS, Contract No. 500-95-0062; and interRAI.
  2. Lunney JR, Lynn J, Hogan C. Profiles of older medicare decedents. Journal of the American Geriatrics Society 2002; 50:1108-12.

Topics: Articles , Clinical