Advance Directives and Medical Treatment at the End of Life

01/01/2004
Summary of working paper 9955
Featured in print Digest

Laws dealing with advance directives and surrogates do not bring "any savings in medical expenditures at the end of life."

Many Americans have decided to prepare written instructions on their preferences for specific medical treatments in the event of a loss of mental or physical competence. This trend toward preparing "advance medical directives" has been debated extensively by physicians, philosophers, and social scientists.

One issue is whether these end-of-life instructions would reduce the substantial health care resources devoted to patients near death. In 1990, the 6.6 percent of Medicare recipients who died accounted for 22 percent of program expenditures, a pattern that has changed little over time.

Another issue is whether patient autonomy and overall well-being would be enhanced by the use of advance directives. Although society has reached a consensus that treatment decisions should reflect a patient's informed preferences, this ideal is often not implemented in practice. Traditionally, physicians have made such treatment decisions in consultation with a patient's family members. In addition, physicians may disregard a patient's earlier stated preferences, believing that terminal patients' preferences change as their illnesses progress.

In Advance Directives and Medical Treatment at the End of Life (NBER Working Paper No. 9955), NBER Research Associate Daniel Kessler and Dr. Mark McClellan provide new evidence on the consequences of states' laws governing advance medical directives for treatment at the end of life. They use a randomly selected sample of elderly Medicare beneficiaries who died between 1985 and 1995 and estimate the effects of two types of laws: laws enhancing incentives for compliance by physicians and hospitals with advance directives; and laws requiring the appointment of a health care surrogate -- such as a family member or guardian -- in the absence of an advance directive.

The authors report three key findings. First, laws enhancing incentives for compliance reduce the probability of dying in an acute care hospital by .76 percentage points. Second, laws requiring the appointment of a surrogate to decide on treatment change the mix of treatment in the last month of life, increasing the probability of receiving acute care in the last month of life by .98 percentage points and decreasing the probability of receiving non-acute care by 1.79 percentage points. Third, neither type of law brings any savings in medical expenditures at the end of life.

In particular, laws enhancing incentives for compliance with advance directives increase medical expenditures in the last month of life by $335; laws requiring delegation increase expenditures in the last month of life by $379. On a base of average expenditures in the last month of life of $14,122 (1995), this amounts to a 2.4 and a 2.7 percent increase, respectively.

The authors also find that laws governing end-of-life care have different effects on different types of patients. For example, among patients who died from cancer, laws enhancing compliance with advance directives reduced the long-run probability of dying in an acute care hospital by almost twice as much - 1.38 percentage points - as for the entire sample of decedents. This is consistent with laws having a larger causal effect for patients for whom end-of-life care is particularly important. In addition, laws enhancing compliance with advance directives have a greater impact on less-educated patients, but laws requiring delegation in the absence of an advance directive have a greater impact on more-educated patients.

The authors point out that their findings are consistent with previous clinical research on the effects of advance directives. Advance directives are not simply a means for refusing treatment, although this is the most common request. In addition, previous research has also suggested that surrogates systematically opt for more intensive treatment than patients would prefer.

The authors conclude with several suggestions for further research. Unless patients receive too little acute and too much non-acute care at the end of life, laws requiring delegation do not improve the alignment of end-of-life treatment with patient preferences, particularly for more-educated patients. The authors hypothesize that this may be because educated patients also have educated surrogates, who are better able to persuade medical care providers of the patient's perceived wishes. If this is the case, then programs to encourage surrogates to communicate better with their patients could enhance patient autonomy and conserve health care resources.

-- David R. Francis