Center to Improve Care of the Dying

INDICATORS OF MEDICAL CARE AT THE END OF LIFE AND SURVIVALTIME

Literature review

Decisions about life sustaining treatment ought to be based on both the patient’s likely outcomes of care and his or her informed preferences (President’s Commission 1983, Guidelines 1987, National Center for the State Courts 1991, American Thoracic Society 1991, Missouri Department of Health 1990). Over the course of an illness, there is usually a transition in the goals of care from the focus of extending and preserving life to that of maximizing the patient’s comfort. The timing of this transition, and the degree to which it is based on the informed preferences of the patient, are important outcome variables in this changing health care market. The current incentives are for maximal treatment unless the patient and/or family requests limitation of treatment. In the coming decade, these incentives may change such that less care is provided. The timing of these transitions plays an important role in the dying experience. Too often the transitions in goals of care occur late in the patient’s illness. The Study to Understand Prognoses and Preferences for Risks and Outcomes of Treatment (SUPPORT) found among a population of seriously ill adults that nearly 40% of patients died after 10 or more days in an ICU, on a ventilator, or in a coma. Over the five years of data collection, the length of ICU stays prior to death did decrease. However, this reduction was mainly due to a reduction in those with an ICU stay greater than one month.

The frequency and timing of DNR orders have been used as another marker of the aggressiveness of care. Numerous investigators have found that DNR orders are written within days of death (Gleeson 1990, Bedell 1984, Hakim 1996). Hakim and colleagues further found that the use of DNR orders, even after adjustment for patient preferences and prognoses, varied substantially based on the patient’s age, the subspeciality of the attending physician, and the hospital where the patient was being treated. Similarly, in a study of seriously ill adults, Wenger (manuscript in preparation) has found that the amount of time spent dying in an ICU, on a ventilator, or in a coma varied substantially among five study sites.

To the contrary, researchers have found a marked decline in aggressiveness of care based on a dramatic increase in hospitalization of nursing home residents (Holtzman 1994, Mor 1996). Of interest, Mor found that this reduction mainly occurred among those with severe cognitive impairment (Mor 1996). In a similar fashion, the locus of death and whether hospice referral was made is another proxy measure for aggressiveness of care. A British study showed that the majority of patients desired to die at home, yet many died at other locations. Further, consulted family members indicated that with more support, the patient’s wish to die at home could have been honored (Townsend 1990). Additionally, the rate of dying in the hospital has been found to vary based on geographic region of the country (Mor 1993) and hospital service area (Pritchard 1996, manuscript under review). This latter study found among the seriously ill that patient preferences did not predict location of death; however, ecological variables such as hospital utilization, reimbursement for hospice, and number of physicians predicted dying in the hospital.

For any given patient, a cogent argument could be made for the aggressiveness of medical care. In reviewing the literature, we did not find any instrument that addressed the appropriateness of “aggressive care” via a delphi panel or that provided guidelines recognized by a professional society. Several professional organizations (e.g., the Society of Critical Medicine, the American Medical Association) have endorsed the concept of futility. One study showed that a prognosis based futility guideline would not impact dramatically on patients with one of nine life threatening illnesses (Teno 1995). However, I would suggest that the measures below be examined from a population perspective and that such measures (e.g., the time spent in an ICU prior to death) be considered as screening instruments to characterize how a health care system cares for dying patients.

Measures of aggressiveness of care

• Timing of DNR orders (chart based)
• Timing of important decisions about life sustaining treatment (chart based)
• Actual and preferred site of death (depending on the medical records system, will most likely need to be interviewer administered to ascertain whether locus of dying was the preferred site)
• Time spent using aggressive treatment prior to death (chart based)
• Measures of patient’s preferred approach to care and the degree to which that approach is being followed (interviewer administered)
• Medical care in the last 48 hours of life (chart based and interviewer survey)

Finally, it is important that consideration be given to the measure of survival time. Mortality rates have been utilized for decades as a measure of quality of medical care. A greatly shortened survival time is cause for concern and for further clarification of whether such reduction is clinically appropriate.

In the ensuing sections, proposed measures for each of these variables will be presented with potential alternate versions based on a review of the literature. Next, a listing of available tools to adjust for “case-mix” are reviewed. Finally, I will end with recommendations for future research.

a. Timing of DNR orders (chart based) – Resuscitation is a paradigmatic decision for examination of decision making at the end of life. Several researchers have examined the timing of DNR order from hospital admission and death (Bedell 1984, Gleeson 1990, Zimmerman 1986, Hakim 1996). The measure is operationalized as the first “official documentation” of discussion and an order to forego resuscitation.

b. Timing of important decisions about life sustaining treatment (chart based) – Each institution can decide what the most important life sustaining treatments are in the care of its patients. Based on work done in the SUPPORT project and in an ongoing project to examine end of life care in an HMO, we have utilized the enclosed instrument to examine decisions about use of mechanical ventilators, vasopressors, feeding tubes, dialysis, and antibiotics.

c. Site of death and whether this was the preferred site of death – The toolkit should examine this domain in terms of the timing of referral to hospice (discussion and the timing of transfer) and site of death based on review of the medical record. I further recommend that an interview be conducted after death with a close family member to ascertain where the patient died and the degree to which this was the patients’ preferred site.

Where did (his/her) death take place?

Was inpatient Hospice involved in the care of (PATIENT), so that (he/she) stayed in a hospice facility?

For about how long did (PATIENT) get inpatient hospice care before (his/her) death?

Was outpatient Hospice involved in the care of (PATIENT), so that a hospice worker cared for (him/her) in the home?

For about how long did (PATIENT) get outpatient hospice care before (his/her) death?

Do you think that (PLACE OF DEATH) was where (PATIENT) would have most wanted to die?

What would have allowed (PATIENT) to die at (his/her) preferred place of death? (open ended)

There are alternate versions that have been proposed to get as this construct:

i. Townsend (1990) and colleagues examined site of death among a cohort of patients.
ii. McCormick (1991) examined site of death among AIDS patients.
iii. McCusker (1984) developed an attitudinal survey to examine patients preferences for dying at home.
v. Mor and colleagues created an instrument to measure preferences for dying at home among patients with AIDS.

d. Time spent prior to dying using aggressive treatment (chart based) - I would propose that a simple count of the number of days spent in an ICU or on a ventilator be done among those who died in the hospital.

e. Measures of patient’s preferred approach to care and the degree to which that approach is being followed (interviewer administered) - The aim of these two questions is to ascertain the degree to which patient preferences are not followed.

This question has adequate test-retest performance (62% agreement with the majority of change occurring from persons who stated “don’t know” now stating a preference) and a modification of this question has been used in a national study of dying persons.

During the last week of (PATIENT’S) life, did (he/she) prefer a course of treatment that focused on extending life as much as possible, even if it meant more pain and discomfort, or on a plan of care that focused on relieving pain and discomfort as much as possible, even if that meant not living as long?

To what extent were these wishes or preferences followed in the medical treatment (he/she) received during the last month of life? Were they followed.…

f. Medical care in the last 48 hours of life (chart based and interviewer survey). There are four chart based instruments which should be considered under this domain:

i. Support Team Assessment Scedule (Higginson 1992) - This is a comprehensive instrument which is completed by medical staff and patient. It allows the tracking of a number of domains. Reliability is good among a trained staff.
ii. Audit of palliative care – The Royal College of Physicians (1991) has identified performance indicators that they adapted for review of the medical record in England.
iii. Performance indicators of quality of dying in nursing home - Based on an exhaustive review of the literature and input from nursing home medical directors, Keay and colleagues (1994) have outlined performance criteria for examining medical care among persons who die in nursing homes.
iv. Modifications of SUPPORT Instruments (Teno, in preparation) and Dying Experience at Dartmouth Study (Goodlin, in preparation) - Based on instruments used in SUPPORT, Lynn, Teno, and Goodlin have created two chart review instruments to examine the dying experience of patients who died in the hospital.

II. Examination of Survival Time

As noted previously, survival time is one of oldest measures of medical care outcomes. Comparison of unadjusted mortality rates or survival time can be misleading because different institutions may have patients with different disease severity and socio- economic status. Listed below are instruments that one might consider to adjust for such differences. We have listed only those instruments which have been independently validated on a new sample.

a. Apache III Prognostic System (Knaus 1991) - This is the most widely used prognostication system for patients hospitalized in an ICU. Chart abstraction has been shown to be reliable. The model estimate (i.e., surviving this hospitalization) has been independently validated. Instruments and computerized platform for APACHE III are available from Apache Medical Systems.
b. Simplified Acute Physiology Score II (Le Gall 1993) – A prognostic model that predicts hospital survival for patients hospitalized in an ICU. The scoring sheet is published in JAMA, 1993, pages 2960-61.
c. SUPPORT Prognostic Model (Knaus 1996) - One of the first models to predict survival time among seriously ill adults. This model is only applicable if institutions choose to utilize the same strict entrance criteria.

IV. Priorities for Future Research

a. We need better regression models that would adjust for confounders in predicting the timing of discussion and decision making about the use of CPR and life sustaining treatment. If adequate models were developed, health care institutions could compare their performances to a “best practice” or national standard.
b. From a clinical standpoint, the development of instruments which could track palliative care management and graphically display the results over time would provide important feedback to improve and enhance patient care.
c. Consideration should be given to developing performance standards (or guidelines) for process variables in the care of the dying. This work should consider building off the audit system outlined by the Royal College of Physicians in London (1991).

References

American Thoracic Society Bioethics Task Force. Withholding and withdrawing life-sustaining therapy. Ann Intern Med 1991;115:478-485.

Bedell SE, Delbanco TL. Choices about cardiopulmonary resuscitation in the hospital. When do physicians talk with patients? N Engl J Med 1984;310:1089-1093.

Gleeson K, Wise S. The do-not-resuscitate order. Still too little too late. Arch Intern Med 1990;150:1057-1060.

Guidelines on the termination of life-sustaining treatment and the care of the dying. Indianapolis: Indiana Univ. Pr., 1987.

Hakim RB, Teno JM, Harrell FE. Factors associated with the Do-Not-Resuscitate Orders: Patients’ preferences, prognoses, and physicians’ judgments. Annals of Internal Medicine 1996;125:284-293.

Higginson IJ, Wade AM, McCarthy M. Effectiveness of two palliative care support teams. J Public Health Med 1992;14:50-6.

Holtzman J, Pheley AM, Lurie N. Changes in orders limiting care and the use of less aggressive care in a nursing home population. JAGS 1994;42:275-279.

Keay TJ, Fredman L, Tlaker GA, et al. Indicators of Quality Medical Care for the Terminally Ill in Nursing Homes. JAGS 1994;42:853-860.

Knaus WA, Harrell FE Jr., Lynn J, et.al. The SUPPORT prognostic model. Objective estimates of survival for seriously ill hospitalized adults. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments. Ann Intern Med 1996;122:191-203.

________, Wagner DP, Draper EA. The Apache III Prognostic System: Risk prediction of hospital mortality for critically ill hospitalized adults. Chest 1991;100:1619-1636.

Le Gall JR, Lemeshow S, Saulnier F. A new simplified acute physiology score (SAPS II) based on a European/North American multicenter study. JAMA 1993;270: 2957-2963.

McCusker J. Development of scales to measure satisfaction and preferences regarding long term and terminal care. Med Care 1984;22:476-493.

Missouri Department of Health. Cruzan v. Director 1990; 110 S.Ct. 284:2855-2856.

National Center for the State Courts. The Coordinating Council on Life-Sustaining Medical Treatment Decisions by the Court. Guidelines for State Court Decision Making in Authorizing or Withholding Life-Sustaining Medical Treatment. Williamsburg, VA: West Publishing, 1991.

President’s Commission for the Study of Ethical Problems in Medicine and Biomedical and Behavioral Research, deciding to forego life-sustaining treatment: A report on the ethical, medical, and legal issues in treatment decisions. Washington, DC: Government Printing Office, 1983.

Royal College of Physicians. Palliative care: Guidelines for good practice and audit measures. J of the Royal College of Physicians of London;25:325-328.

Townsend J, Frank AO, Fermont D, et.al. Terminal cancer care and patient’s preference for place of death: a prospective study. Br Med J 1990;301:415-417.

Zimmerman JE, Knaus WA, Sharpe SM, et.al. The use and implications of do not resuscitate orders in the intensive care units. JAMA 1986;255:351-356.

Section prepared by Joan Teno