JOURNALOFGENERALINTERNALMEDICINE, Volume 7 (January/February), i992

from a systematic bias. Extraordinary instances of prolonged life support are used to infer a standard of practice for the general population. What we need to know before patients and physicians spend their health discussing death, is the probability that such discussion is relevant for an individual. What patient profiles mandate this discussion? z Besides patients with terminal cancer, what other non-hospitalized patients w o u l d benefit from ACDs? Cardiopulmonary resuscitation and intensive care are expensive and, at times, undignified, but the execution of ACDs by the majority of the population will not be inexpensive. What is the c o s t - b e n e f i t ratio for ACDs as a public policy? Any benefit of ACDs w o u l d be negated if previously stated preferences are regularly reversed whenever death is imminent. What is the effect of "framing" on the expression of patient preferences? 3 What are the conditional and temporal limitations to any previously expressed preference? It may be more ethical, more economical, and more relevant for patients and physicians to negotiate the intensity and duration of medical care on a day-to-day basis than by prior directive. Perhaps ACDs have an appropriate, but not ecumenical, role in ambulatory care medicine.--DOMINIC J. B ~ , MD,

Veterans Affairs Regional and Medical O~ce Center, White River Junction, VT05009-0001 References 1. Brunetti LL, Carperos SD, Westlund RE. Physicians' attitudes towards living wills and cardiopulmonary resuscitation. J Gen Intern Med. 1991;6:323-9. 2. MossAH. Informing the patient about cardiopulmonaryresuscitation: when the risks outweigh the benefits. J Gen Intern Med. 1989;4:349-55. 3- TverskyA, KahnemanD. The framingofdecisionsandthepsychology of choice. Science. 1981 ;211:453-8.

In reply:--We agree with Dr. Balestra that, within the time limits of an office visit, patients and physicians must communicate about many health care matters. Although we scored responses on a scale from 1 to 5, we were not implying that a score of " 5 " was expected. Indeed, it is unrealistic to expect that physicians address cardiopulmonary resuscitation (CPR) wishes during each office visit or to the exclusion of more emergent matters. However, we believe that a patient's wishes regarding extraordinary care are more than merely a "theoretical contingency." After all, Karen Ann Quinlin and Nancy Cruzan were young w o m e n in good health before meeting their tragic fates. The probability that a young, healthy person will suffer a catastrophic illness, such as severe trauma, is small; however, in this age group, it is the leading cause of death. Because primary care physicians provide longitudinal care, they are in the best position to question patients (of whatever age), in a nonemergent setting, about their health care p r e f e r e n c e s - in much the same way as other " p r e v e n t i v e " medical issues are discussed. Additional motivation for documenting patient wishes comes from the Cruzan case. In Cruzan, the Supreme Court ruled that a competent person's decision to forgo extraordinary care should be honored; however, the Court did not guarantee that the unexpressed wishes of an incompetent person should bear equal weight and be followed. The clear implication of Cruzan is, if a person has chosen not to receive extraordinary care, then these wishes must be made known. Execution of advance care directives (ACDs) should not be expensive, as they are readily available free, or at nominal cost from state medical societies, bar associations, and special

1 ~.1

interest groups. Unlike last wills and testaments, an attorney is not needed to execute one! Relative to the cost of unwanted medical care, the cost/benefit ratio of ACD use, although unknown, should be very low. The benefit of ACDs could be negated if preferences are consistently reversed. However, there are no data to suggest that such reversals occur with any great regularity. We believe it is far more paternalistic to avoid initiating discussion about CPR preferences just because you believe the patient's wishes will change. If physicians communicate with and educate their patients about the benefits of ACDs, perhaps the old saying from medical school of "see one, do one, teach o n e " will translate into a greater chance that extraordinary care issues will be readily discussed by both doctors and their p a t i e n t s . - Louis L. BRLrNETrl, MD, JD, Carolinas Medical Center, Charlotte, NC 28232-2861; and STEPHANIE CARPEROS, MD, The Davidson Clinic, Davidson, NC 28036

Ultrasonography for Abdominal Aortic Aneurysm To the Editors:-- In their recent review, ~ Reuler and Kumar cite data from an analysis by Quill et al. 2 to support their conclusion that ultrasound screening for abdominal aortic aneurysm (AAA) is not cost-effective. This conclusion can be challenged on several grounds. The cost per life saved for one-time (or every-ten-year) ultrasound screening of $78,000 compares favorably with the costs per life saved of many accepted medical practices and was considered to be cost-effective by Quill et al. in the original article. The $78,000 consisted almost entirely of the cost of treatment, which in this model included operation on all AAAsdetected, even those less than 4 cm. A screening program that employed a more selective use of surgery, as described elsewhere in the review, would considerably reduce the cost per life saved. Reducing the prevalence fourfold (from 10% to 2.5%) raised the cost per life saved by less than 40%. The figures of $150,000 per diagnosis and $600,000 per life saved quoted in the review refer to annual retesting, w h i c h is clearly too short an interval and, to my knowledge, has never been recommended. Reuler and Kumar also consider abdominal palpation to screen for AAA to be unsupportable because of low positive predictive value and sensitivity. The only consequence of a positive predictive value of 22% is that four confirmatory ultrasound tests will be done per diagnosis, still a very high yield. The 50% sensitivity of abdominal palpation makes it less desirable than ultrasonography for screening since fewer AAAs are detected, but the costs per AAA detected are similar (about $1,500, assuming 10% prevalence, $150 for ultrasonography, and $ 25 for abdominal palpation) and the largest, most dangerous AAAsare preferentially detected, 3 so it is better than nothing. The authors have provided internists with much useful information about abdominal aortic aneurysms. It would be unfortunate if, in the process, the internist's crucial role as diagnostician were d i m i n i s h e d . - FRANKA. LEDERLE,MD, D# vision of General Internal Medicine, Department of Medicine, Minneapolis VA Medical Center, Minneapolis, MN 55417 References 1. Reuler JB, Kumar KL. Abdominal aortic aneurysm. J Gen Intern Med. 1991 ;6:360-6. 2. Quill DS, Colgan MP, Sumner DS. Ultrasonic screening for the

12Z

LETTERS It would be helpful if the authors could cite earlier work or their own experience on the validity of these vision tests in d e m e n t i a . - CATHYA. ALESSI,MD, Department o f Medicine,

detection of abdominal aortic aneurysms. Surg Clin North Am. 1989;69:713-20. 3. Lederle FA, WalkerJM, Reinke DB. Selectivescreening foralxtominal aortic aneurysms with physical examination and ultrasound. Arch Intern Med. 1988;148:1753-6.

University of California, Los Angeles, Sepulveda VA Medical Center, Sepulveda, CA 91343 Reference

Visual Impairment and Alzheimer's Disease

1. Uhlmann RF, Larson EB, Koepsell TD, Rees TS, Duckert LG. Visual impairment and cognitive dysfunction in Alzheimer's disease. J Gen Intern Med. 1991;6:126-32.

To the Editor: - - I read with much interest the work by Uhlmann et al. on the association between visual impairment and cognitive dysfunction in Alzheimer's disease) However, I found it difficult to interpret the results of the study without information about the use in demented individuals of the Snellen and Rosenbaum methods for testing far and near vision. It seems likely that the validities of vision testing differ between demented and nondemented individuals. This could lead to bias due to differential misclassification of visual impairment, i.e., a difference in classifications of visual impairment between cases and controls. The potential importance of differential misclassification bias in this type of study is illustrated by the following extreme example (Table 1). Assume that in the 87 matched pairs the true prevalence of visual impairment in both cases and controls is 37% (equal to the prevalence in controls in Table 1 of the Uhlmann article), and the true odds ratio is 1.0 (i.e., no effect). Suppose 20% of demented subjects with normal vision are misclassified as visually impaired. Then the observed odds ratio w o u l d be 2.4 (95% CI = 1.05, 5.48). If this bias away from the null value is not recognized, the investigator w o u l d erroneously interpret an effect of visual impairment on dementia.

In reply: - - Dr. Alessi raises a concern regarding possible bias due to misclassification of visual impairment. In our study, we described how we tested patients for visual impairment using standard methods that are considered "gold standards." We acknowledge that there are legitimate concerns regarding vision testing in demented individuals. However, we are not aware of any evidence that differential misclassification might have occurred. We also note that one can easily make an association "go away" by developing hypothetical data, as Dr. Alessi has done. We based our conclusions on the actual data observed, the biologic plausibility of the association, and the consistency of the association with the effect of other sensory impairment (hearing loss) on cognitive dysfunction in Alzheimer's disease. We respectfully submit that Dr. Alessi's letter provides no new information that w o u l d warrant reconsideration of our study.--RICHARD F. UHLMANN,MD, MPH, ERICB. LARSON,MD, MPH, THOMASD. KOEPSELL,MD, MPH, THOMASS. REES, PhD, and LARRYG. DuCKERT, MD, PhD, University of Washington and Harborview Medical Center, Seattle, WA 98104

TABLE 1

Hypothetical Example:Differential MisclassificationBias ( N = 87 Pairs)* a) "'True" cell frequencies

Demented Patients impaired Vision Normal Vision

Total (%)

Impaired Vision

22

10

32 (37%)

Normal Vision

10

45

55 (63%)

32 (37%)

55 (63%)

87 (100%)

NondementedPatients

Total (%) *Odds ratio of discordant pairs. 10/10 = 1.0 (95% CI = 0,4, 2.4).

b) Observed cell frequencies if 20% of demented subjects with normal vision in the above table were misclassified as vision impaired (i.e., differen~'al misclassificabbn)

Demented Patients Impaired Vision Normal Vision

Total (%)

impaired Vision

24

8

32 (37%)

Normal Vision

19

36

55 (63%)

43 (50%)

44 (51 Olo)

87 (100%)

NondementedPatients

Total(%) *Odds ratio of discordant pairs, 19/8 = 2.4 (95% CI = 1.1,5.5).

Ultrasonography for abdominal aortic aneurysm.

JOURNALOFGENERALINTERNALMEDICINE, Volume 7 (January/February), i992 from a systematic bias. Extraordinary instances of prolonged life support are use...
216KB Sizes 0 Downloads 0 Views