IAGS 38:621-627, 1990
CLINICAL INVESTIGATION
Are Elderly People Less Responsive to Intensive Care? Albert W.Wu,MD,* Haya R. Rubin, MD, PhD,* and MarkJ. Rosen, M D t
Older patients may be excluded from intensive care units because of the perception that they will benefit less than younger patients. To determine if advanced age is associated with increased mortality independent of severity of illness, we compared older and middle-aged patients admitted to a medical intensive care unit. W e reviewed the charts of 130 patients age 75 years or older and 135 patients age 55 to 65 admitted over a 30-month period. W e controlled for severity of illness using the Acute Physiology Assessment and Chronic Health Evaluation (APACHE 11) system without including points for age (APACHE 11M).The groups were similar with regard to gender, whether or not they had a private attending physician, mean APACHE 11M score, and diagnoses, except that older patients had more chronic obstructive pulmonary disease. Hospital stay was slightly longer in the older group (37 us. 39 days, rank sum, P < .02).Hospital mortality was significantly greater in the older group (39% us. 51 %, Chi-square P < .05) with a crude relative risk of
1.32 (95% confidence interval [Cl]: 1.01, 1.73). However, the relation of age group to mortality differed for patients with different diagnoses. When we used logistic regression to adjust for APACHE UM,whether the patient had a private attending physician, primary admitting diagnosis, or presence of cancer, older patients did not have a significantly greater risk of dying (adjusted relative risk, 1.05; 95% Cl: 0.97, 1.12). When pulmonary artery catheterization was added to the model, it independently predicted mortality (adjusted relative risk, 1.47; 95% Cl: 1.05, 2.06). APACHE 11M (calculated without the inclusion of age) was an excellent predictor of mortality. Older age did not predict mortality once severity of illness, admitting diagnosis, and the presence of underlying malignancy were taken into consideration. Further studies should explore the effects of age separately for patients with different diagnoses as well as control for severity of illness. J Am
I
tients has been examined previouslyI1-l2few studies have compared the outcome of intensive care in elderly patientswith that of younger patients with similar severity of illness. The purpose of this study is to determine if advanced age is associated with increased mortality in a general medical intensive care unit, after accounting for severity of illness. The APACHE I1 (Acute Physiology Assessment and Chronic Health Evaluation)system provides a quantitative estimate of severity of illness and accurately indicates prognosis in the intensive care unit (ICU)setting when used in prospective and retrospective eva1uations.l3-l* The APACHE I1 score includes points for age group, acute physiologic derangements, and chronic health status. The original validation studies refer to age as "a well-documented risk factor for death from acute illness, that is independent of the severity of disease," n and older age groups were found to have a higher mortality rate independent of the acute physiology and chronic health scores. However, subsequent studies of the effect of age and severity of illness on mortality have yielded conflicting r e s ~ l t s . ~ . 9 . ~ ~ We conducted a retrospective study to determine the
s a patient's age an appropriate criterion to use in making decisions about intensivecare? In most hospitals, the limited number of intensive care beds requires physicians to decide who will be admitted to them. Patients considered to have an extremely poor prognosis may not be offered admission because of limited beds and the low likelihood of benefit. Many feel that elderly patients, because of chronic illnesses and impaired physiologic reserve, have a worse prognosis then younger individuals, and there is evidence to suggest that physicians may be less aggressive in treating elderly patients.' Although intensive care in elderly pa-
From the *Department of Medicine, University of California, San Francisco, San Francisco, California and tDeparhnent of Medicine, The Mount Sinai Medical Center, New York, New York. Supported in part by a grant from the Catherine and Henry Gaisman Foundation and the Veterans Administration/Robert Wood JohnsonClinical ScholarsProgram. Presented at the society of Critical Care Medicine Annual Educational and Scientiiic Symposium, Anaheim, California, May 26-29,1987. Address correspondenceand reprintrequests to Albert W.Wu,MD, Robert Wood JohnsonClinical Scholars Program, 350 Pamassus Avenue, Room 407, San Francisco, CA 94117. 0 1990 by the American Geriatrics Society
Geriatr SOC38:621-627,1990
0002-8614/90/$3.50
622 W U ET AL
IAGS-JUNE 1990-VOL 38, NO. 6
relation between age and mortality among individuals in a general medical ICU with comparable severities of illness, as quantified by APACHE I1 scores calculated without including age.
MATERIALS AND METHODS Hospital and Intensive Care Unit The Mount Sinai Medical Center is a 1,212-bed tertiary care hospital in New York City. The Medical Special Care Unit consists of eight beds used for general medical intensive care. Coronary care and surgical patients are usually treated in other units. Decisions regarding admission to the unit are made by a resident in internal medicine, with consultation from a fellow or from the director of the unit in cases of uncertainty or objections by other medical staff. Decisions to discharge patients are usually made jointly by the attending physician and house staff rotating through the unit. Patients frequently are discharged because a bed is needed for another patient who may be more critically ill or more “salvageable.” Study Design Patients admitted to the ICU between January 1, 1982 and June 30, 1984 were eligible for study. All available charts of patients 75 years or older were analyzed and compared with those of patients between the ages of 55 and 65. These age groups were selected to compare elderly patients with a group with comparable diagnoses whose members would not have been refused ICU admission on the basis of age alone. Measures Independent variables included those hypothesized to affect mortality: age, gender, whether or not the patient had a private attending physician (private/service), primary ICU admitting diagnosis, severity of illness, and major diagnostic and therapeutic interventions. The primary admitting diagnosis was determined from the resident’s admission note. We classified the most common admitting diagnoses as having high or low risk; cardiopulmonary arrest, septic shock, and pneumonia were considered ”high risk’ whereas gastrointestinal hemorrhage, chronic obstructive pulmonary disease, congestive heart failure, and neuromuscular disease were classified as “low risk.” Less frequent diagnoses included asthma, adult respiratory distress syndrome, pulmonary embolism, restrictive lung dis-
ease, upper airway obstruction, hemoptysis, pneumothorax, hypertensive crisis, arrhythmia, hypovolemia, coronary artery disease, seizure, intracranial hemorrhage, stroke, acute renal failure, anemia, diabetic ketoacidosis, drug overdose, and metabolic imbalance. Major interventions were defined as intubation and mechanical ventilation, pulmonary or peripheral arterial catheters, peritoneal or hemodialysis, cardiac pacemakers, or tube thoracostomy. The APACHE I1 system point score is based on 11 routine physiologic measurements, the Glasgow coma score, age, and previous health status.’ To measure severity of illness, a score was calculated for each patient without including age (APACHE IIM), based on information recorded in the chart during the first 24 hours of ICU hospitalization. Physiologic measurements were abstracted from detailed nursing flow sheets. Glasgow coma score was determined using medical and neurologic data in the ICU chart. Outcomes recorded were length of ICU and hospital stay, readmission to the ICU, and ICU and hospital survival.
Statistical Analysis For continuous variables, differences between group means were determined using the unpaired two-tailed Student’s t test. Where distributions were not normal, the Wilcoxon rank-sum test was used. The Chi-square test was employed to compare proportions for different groups. Logistic regression analysis was used to examine the independent relation to mortality of age, severity of illness, gender, private/ service status, and treatment. We calculated adjusted relative risk for each independent variable using the most typical values for each of the other variables. RESULTS During the 30-month study period there were 895 admissions to the ICU. One hundred seventeen of these were readmissions; only first ICU admissions were considered to ensure that all observations were independent. The mean age for the 778 patients was 57 years, witharangeof 15 to97; 181(23%)wereage55 to65and 164 (21%) were age 75 or older. Eighty of these charts were incomplete or unavailable for review (46 in the younger and 34 in the older group). The age and gender distribution for those patients whose charts were avail-
TABLE 1. STUDY GROUP CHARACTERISTICS (55 to 65 Years)
Group I1 (275 Years)
135 63 (47) 90 (67) 15.8 f 9.5
130 54 (42) 99 (76) 16.1 & 9.1
Group I Number Male (%) Number with private attending physician (96) Mean APACHE IIM score & SD (does not include age) There were no significant differences between groups I and II.
IAGS-JUNE 1990-VOL 38, NO. 6
MEDICAL INTENSIVE CARE IN ELDERLY PEOPLE
623
TABLE 2. PRIMARY ADMITTING DIAGNOSES Group I
(55 to 65 Years)
High-risk diagnoses Cardiopulmonaryarrest (%) Septic shock (%) Pneumonia (%) Low-risk diagnoses Gastrointestinalhemorrhage (%) Chronicobstructive lung disease (%)* Congestive heart failure (%) Neuromuscular disease (%) Other diagnoses (%)t Total
Group I1 (2:75 Years)
19(14) 15(11) 16(12)
21(16) 18(14) 16(12)
Chi-square, P < .005. t Chi-square, P < .02.
able did not differ significantly from those whose charts were missing.
Sample Characteristics Charts were available for 135 patients aged 55 to 65 years (mean, 60.2) and 130 aged 75 years or older (mean, 80.5). We found no differ-
ences between the groups in gender, private/service status, and mean APACHE IIM score (Table 1).
Admission Diagnoses Table 2 lists the primary diagnoses at the time of ICU admission in both age groups. The distribution of diseases was similar in the two study groups except that more elderly patients had chronic obstructive pulmonary disease whereas fewer had "other" diagnoses. Hospital Course and Major Interventions Two hundred eleven patients (8096) had invasive interventions, including mechanical ventilation, dialysis, chest
tube placement, or invasive hemodynamic monitoring (Table 3). There was no significant difference in the proportion of younger and older patients who had an invasive intervention or in the mean number of procedures performed per patient. However, pulmonary arterial catheterization was performed more frequently in the older patients (76% vs. 62Y0, P 5.02). There were no significant differencesbetween groups in length of ICU stay or ICU readmission. However, once discharged from the ICU, older patients stayed longer in the hospital (Table 3).
Relation of Age to Mortality For both age groups, the APACHE IIM score correlated well with ICU and hospital death rates (Figure 1). Mean APACHE IIM score was 12.3 f 7 for survivors compared with 20.5 f 10 for those who died (t test, P < .0001). Overall ICU mortality was 21Yo and total hospital
TABLE 3. PROCEDURES, INTENSIVE CARE UNIT (ICU), AND HOSPITAL COURSE Major intervention (%) No.of interventions/patient Pulmonary artery catheterization (%y Endatracheal intubation (%) Mean ICU days Mean hospital days after ICU discharget Total hospital dayst ICU readmission (%) Mortality (%)$ ICU mortality (%) Hospital mortality after ICU discharge (%)€j Chi-square, P 5 . 0 2
t Runk sum, P 5.02.
$ Chi-square, P 5.05. 5 Chi-square, P 5.001.
55 to 65 Years
2 7 5 Years
Total
105(78) 1.8 84(62) 35(26) 7.4 21.8 37.4 12 (9) 52(39) 34(25) 18(13)
106(82) 2.0 99(76) 42(32) 6.3 25.3 39.1 10 ( 8 ) 66(51) 22(17) 44(34)
211(80) 183(69) 77(29)
22 (8) 118(45) 56(21) 62(23)
624 W U ET AL
IAGS-IUNE 1990-VOL 38, NO. 6
100
Percent Died
6 2
80 60
FIGURE 1. Hospital mortality by APACHE IIM score (without age) comparing patients aged 55 to 65 years with those 75 years of age and older. Numbers above each bar indicate numbers of patients.
40 20 0
0-4
5-9
10-14
15-19
20-24
25-29
30-34
35*
Apache II Score Without Age 55-65
mortality was 45% (Table 3). For younger and older groups, the difference between rates of ICU mortality was not significant. However, a sigdicantly greater proportion of elderly patients died in the hospital after discharge from the ICU, so elderly patients had a significantly greater total hospital mortality rate (39%vs. 51%, P 5 .05). Table 4 lists crude relative risks of mortality depending on the patient's age group, gender, private/service status, diagnostic risk group (as classified in Table 2), presence or absence of underlying diagnosis of malig-
75 and older
nancy, and whether or not the patient received pulmonary arterial catheterization. All factors (except gender) were significantly associated with mortality. Logistic regression was performed using admission characteristics: APACHE IIM, age group, gender, private/service status, presence of cancer, and presence of a high-risk admission diagnosis as independent variables (Table 4). Older patients did not have a sigdicantly greater risk of dying controlling for other variables. Patients with higher APACHE IIM scores were more likely to die as were those with underlying cancer.
TABLE 4. RELATION OF PATIENT DEMOGRAPHIC AND CLINICAL VARIABLES TO MORTALITY Independent Variab1e Age group 55 to 65 years 2 75 years Gender Male Female Status Service Private Primary diagnosis Low risk High risk Malignancy Absent Present Pulmonary artery catheter No Yes Intuba tion
No Yes
Died
(%)
Crude Relative Risk (95%CI)
52/135 66/130
(39) (51)
Old/young = 1.32 (1.01, 1.73)
59/117 59/148
(50) (40)
Male/female = 1.265 (0.97, 1.66)
26/76 92/189
(34) (49)
Private/service (1.03, 1.96)
26/82 69/105
(32) (66)
High/low risk = 2.07 (1.52, 2.82)
89/224 29/41
(40) (71)
Present/absent = 1.78 (1.31, 2.42)
18/82 100/183
(22) (55)
Yes/no = 2.49 (1.73, 3.57)
57/188 61/77
(30) (79)
Yes/no = 2.61 (2.02, 3.39)
= 1.42
JAGS-IUNE 1990-VOL.38, NO. 6
APACHE IIM score alone classified 69% of cases correctly; APACHE IIM for patients with cancer and a high-risk admission diagnosis classified 74% of cases correctly. Age group did not enhance the ability to predict mortality. To see if differences in treatment of older and younger patients might confound the relation of age to mortality, we added pulmonary arterial catheterization to the logistic regression model. Pulmonary arterial catheterization independently predicted mortality (adjusted relative risk, 1.47;959b CI: 1.05,2.06)and increased correct classification from 699'0 to 72%. However, older patients still did not have higher adjusted relative risk of dying, and this model's predictive ability was also the same with and without age. To explore whether age group and mortality might be associated for subgroups within the population, age and mortality were cross-tabulated while controlling for gender, private/service status, and severity of illness (dividing APACHE IIM score into two groups: 115 and >15). Older patients had higher mortality rates in only two out of the eight private/service status-genderseverity of illness subsets of our sample. Among male private patients with APACHE IIM scores 515 (n = 46), 30% of younger versus 65% of older patients died in the hospital (Chi-square = 5.6, P = .018). Among female private patients with APACHE IIM scores >15 (n = 44), 41% of younger vs. 73% of older patients died (Chisquare = 4.5,P = .033). For all other subsets, both age groups had very similar mortality rates. As mentioned earlier, more older patients had an admission diagnosis of chronic obstructive lung disease, but these patients did not die at a higher rate than their younger counterparts. Admission diagnoses for our two age groups were otherwise similar. Because our sample size did not allow inclusion of individual admitting or secondary diagnoses as predictor variables in the logistic regression analysis, we examined how age was related to mortality within specific diagnostic groups. Crosstabulation of mortality and age group for specific diagnoses revealed that admission diagnoses modified the relation of age group to mortality independent of severity of illness. For example, despite lower mean APACHE IIM scores, older patients with pneumonia had twice the mortality of younger patients (884b vs. 444b,P 1 .01). Among those with sepsis, congestive heart failure, and chronic obstructive lung disease, a larger proportion of younger patients died (differences not significant) despite having similar or lower mean APACHE IIM scores. Among those with secondary diagnoses of malignancy, mean APACHE IIM scores were similar for older and younger groups and a slightly larger proportion of older patients died (differences not sigruficant). Thus, the relationship of age to mortality differed for patients with different diagnoses.
MEDICAL INTENSIVE CARE IN ELDERLY PEOPLE
625
DISCUSSI 0N In our study of a general medical ICU, a modified APACHE I1 score calculated without points for age (APACHE IIM) was an excellent predictor of mortality in a retrospectiveanalysis. When severity of illness, presence of a high-risk admission diagnosis, and underlying malignancy were considered, older patients did not have higher odds of dying than younger patients, and age group did not add to predicting mortality. Furthermore, older patients had a similar or even better prognosis than younger patients in some diagnostic subgroups, suggesting that advanced age does not uniformly confer a worse prognosis. A number of studies have examined intensive care outcomes for elderly patients.l-12 All but two3J0found that elderly patients had higher ICU or hospital mortality. Four of these studies attempted to control for the effect of severity of illness on the relationship between age and mortality. Fulkerson et a1 reported that age did not substantially contribute to ICU mortality when severity of illness, as estimated by the need for mechanical ventilation or cardiopulmonary resuscitation, was taken into account.6 Nicholas and associates found that although ICU mortality increased progressively with age group, the mortality rate in patients over 75 was equal to that observed in patients aged 55 to 59." To control for severity of illness, the authors stratified by the Acute Physiology Score (APS) component of the APACHE I1 into low and high severity groups (515 and >15). In both groups, older patients had a significantly higher death rate. However, this study did not include the chronic health evaluation (CHE) component of APACHE 11. Because elderly patients generally have more chronic health problems than younger patients, excluding CHE points reclassifies more older patients as "low severity." This reclassificationwould result in a "Will Rogers phenomenon," inflating the mortality rate for older patients in both low and high severity groups.19 McClish and associates employed both the acute and chronic health components of APACHE IL9They found that ICU mortality increased from 18%for patients aged 55 to 64 to 25% for patients 75 years of age or older, but that this difference disappeared when logistic regression was used to control for demographic and diagnostic variables. These results appear consistent with our findings. The original validation of Knaus et a1 of APACHE I1 examined whether age independently predicted survival.13They demonstrated an increasing risk of mortality with increasing age, adjusting for severity of illness, surgical status, and major diagnoses. Although the authors did not directly compare the two age groups we examined, recalculation from their regression coeffi-
1AGS-1UNE 1990-VOL. 38, NO. 6
626 W U ET AL
significance of a relative risk of 1.05 would still be doubtful. In addition, because the average risk of older versus younger patients differed by demographic and diagnostic subgroups, the conclusion that age confers uniform risk would still be suspect. Two incidental findings merit discussion. First, in both age groups, patients of private physicians had worse outcomes than service patients despite similar severity of illness and treatment intensity. However, multivariate analysis demonstrated that the increased incidence of underlying malignancy among private patients (37 private vs. 4 service)accounted for their excess mortality. Possibly, private attending physicians prevailed upon house staff to admit higher risk patients to the ICU. Second, pulmonary arterial catheterization was an independent predictor of hospital mortality in our study. There are several ways to interpret this finding. Because hemodynamic derangement led to the placement of pulmonary artery catheters, this intervention may indicate an increased disease severity that is not reflected by APACHE IIM. Furthermore, catheter placement is often an intervening outcome; if it occurred after the first 24 hours of ICU care (when APACHE scores were calculated), it may simply reflect a deteriorating course. In a retrospective population-based study of acute myocardial infarction, Gore et a1 demonstrated that patients receiving a pulmonary artery catheter for congestive heart failure or hypotension had substantially increased mortality and length of hospital stay compared with those who were not catheterized.20Like our study, their investigation was retrospective, and they observed that they could not determine if patients who received a pulmonary artery catheter were sicker than those who did not. Nevertheless, our results support the recent call for a randomized trial of pulmonary arterial catheterization2l Variables we did not measure may also affect out-
cients gives a relative risk of dying of 1.4 for patients 75 years of age or older compared with patients aged 55 to 65. This is greater than the adjusted relative risk of 1.05 that we found in our study. The study of Knaus et a1 did not examine whether specific diagnoses modified the relation of age to mortality. The varying relation of age to mortality that we found within different diagnostic subgroups argues against the routine incorporation of age into APACHE scores. A recent review of severity of illness measures concludes that “although intrinsic biological severity may one day be measurable, currently it is an abstraction.” l7 Owing to the heterogeneity of patients within older groups, chronologic age is probably not as important in determining survival as biologic age, which we are unable to measure. Our study had some methodologic constraints. We had no other source of demographic or clinical information about the patients whose charts were unavailable, who may have differed from those we studied. In addition, we do not know how our patients compared with those denied or not considered for admission to the ICU. If elderly patients were more carefully selected based on ”hardiness” or clinical factors not reflected in our diagnostic or severity-of-illness measures, this might bias our results. To answer this question would require detailed accounting of all ICU consults, both formal and informal, about patients considered for admission; such data could not be collected in a retrospective chart review but would be important to consider in future prospective studies. It is also possible that we failed to find a significant relationship of age to mortality in multivariate analysis because our sample size was too small. The lower confidence limit for the adjusted relative risk (Table 5) is close to 1. Had we collected more observations we might have demonstrated that the slightly higher adjusted relative risk of death of the older group was statistically different from 1. However, the clinical
TABLE 5. LOGISTIC REGRESSION ANALYSIS USING ADMISSION CHARACTERISTICS AS PREDICTORS OF MORTALITY __
.. .
~
- .-
~
-~ - -.
~
_
~~
_
-~
-
-
~
Risk Factor
P Value
Adiusted Relative Risk (95%CI)
APACHE IIM Age group Gender Status Primary diagnosis Malignancy
,000 .187 ,232 ,278 ,082 .016
Low/high* = 1.40 (1.10, 1.67) Old/young = 1.05 (0.97, 1.12) Male/female = 1.04 (0.98, 1.12) Private/service = 1.05 (0.96, 1.12) High/low risk = 1.07 (0.99, 1.14) Yes/no = 1.11 (1.03, 1.15)
.... ....
.
- -
~~~~~~~~
~~
~
~~
~~
~
~~~~~~~~
~~
~
-~ ~
~~
~
~
~
. . .
~
~
~~~
~~~~~~
~
APACHE IIM scores 1 SD above and below the mean. Logistic regression equation: In (aliveldead) = 1.4643 - 0.2380 (age group) - 0.1783 (APACHE IlU, 0.2105 (gender) - 0.2243 (status) - 0.7514 (malignancy) - 0.3244 (primary diagnosis). Coding ofvariables: Agegroup: 55 to 65 = 0, 2 75 = 1; male = 0, female = I ; service status = 0, private = I ; no malignancy = 0, malignancy = I; low-risk diagnosis = I , high-risk diagnosis = 2. All relative risks calculated using most typical valuesfor other riskfactors, ie, mean APACHE IIM,older, female, private, high-risk primary diagnosis, and no underlying malignancy.
+
JAGS-]LINE 1990-VOL 38, NO. 6
come, such as the experience and expertise of ICU personnel and the attitudes of staff toward patients. It would be valuable to examine the relationship of doctors‘ and nurses’ expectations about patient prognoses to outcome, controlling for severity of illness. In conclusion, our study validates APACHE I1 without the inclusion of age as an excellent predictor of mortality, even when applied retrospectively from chart review. Advanced age did not predict mortality once severity of illness, presence of a high-risk admitting diagnosis, and presence of underlying malignancy were taken into consideration.For specific diagnoses, elderly patients had similar prognoses to younger patients; for others, their outcome was worse despite a lower APACHE IIM score. Our findings illustrate that the relationship of age to mortality is complex, even for patients with similar severity of illness. Our findings suggest that elderly patients should not be denied consideration for intensive care on the grounds that they are less likely to benefit than their younger counterparts. Future studies should explore the effects of age separately for patients with different diagnoses as well as control for the severity of illness. The incorporation of age into the calculation of APACHE I1 score should be reconsidered because the assumption of an equal effect across groups of patients may be invalid.
REFERENCES 1. Campion EW, Mulley AG, Goldstein RL, et a1 Medical intensive care of the elderly: a study of current use, costs, and outcomes. JAMA 246:2052,1981 2. Thibault GE, Mulley AG, Bamett 0,et al: Medical intensive care: indications, interventions and outcomes. N Engl J Med 302:938,
1980 3. FedulloAJ, Swinbume AJ: Relationship of patient age to cost and survival in a medical ICU. Crit Care Med 11:155,1983 4. Goldstein RL, Campion EW, Mulley AG, et a1 Nursing home patients admitted to a medical intensive care unit. Med Care
222354, 1984
MEDICAL INTENSIVE CARE IN ELDERLY PEOPLE
627
5. McLean RF, McIntosh JD, Kung GY, et al: Outcome of respiratory intensive care for the elderly. Crit Care Med 13:625, 1985 6. Fulkenon WJ, DeLong E, Young SL Elderly admissions to a medical ICU: 8 year experience (abstract). Am Rev Resp Dis 133:A351,1986 7. Smithies M, Bihari D, Catling J, et a 1 Intensive therapy of the elderly (abstract). Crit Care Med 14:397,1986 8. Franklin C, Nightingale S, Mamdanni B, et al: Increases in mortality in the oldest old critically ill patients (abstract). Crit Care Med 14:398,1986 9. McClish DK, Powell SH, Montenegro H, et al: The impact of age on utilization of intensive care resources. J Am Geriatr Soc
35983, 1987 10. Papadakis MA, Browner W S Prognosis of noncardiac medical 11.
12. 13. 14. 15. 16. 17. 18.
patients receiving mechanical ventilation in a veterans hospital. Am J Med 83:687,1987 Nicholas F, LeGall JR, Alperovitch A, et al: Iduence of patients’ age on survival, level of therapy and length of stay in intensive care units. Intensive Care Med 13:9, 1987 Fedullo AJ, Swinbume AJ: Intensive care for elderly patients with myocardial infarction: age-related comparison of clinical features and outcome. J Intensive Care Med 3:265, 1988 Knaus WA, Draper EA, Wagner DP, et al: APACHE I 1 a severity of disease classification system. Crit Care Med 13:818, 1985 Knaus WA, Draper EA, Wagner DP, et al: An evaluation of outcome from intensive care in major medical centers. Ann Intem Med 104:410,1986 Chang WS, Jacobs S, Lee B Use of APACHE I1 severity of disease classification to identify intensive care unit patients who would not benefit from total parenteral nutrition. Lancet 1:1483, 1986 Jordan DA, Miller CF, Kubos KL, et a 1 Evaluation of sepsis in a critically ill surgical population. Crit Care Med 15897,1987 Stein REK, Gortmaker SL, Perrin EC, et al: Severity of illness: concepts and measurements. Lancet 2:1506, 1987 Johnson MH, Gordon PW, Fitzgerald FT Stratificationof prognosis in granulqtopenic patients with hematologic malignancies using the APACHE I1 severity of illness score. Crit Care Med
14:693,1986 19. Feinstein AR, Sosin DM, Wells CK The Will Rogers phenomenon. Stage migration and new diagnostic techniques as a source of misleading statistics for survival in cancer. N Engl J Med
312:1604,1985 20. Gore JM, Goldberg RJ, Spodick DH, et al: A community-wide assessment of the use of pulmonary artery catheters in patients with acute myocardial infarction. Chest 92:721,1987 21. Spodick DH: Analysis of flow-directed pulmonary artery catheterization. JAMA261:1946,1989
CLINICAL INVESTIGATION
Are Elderly People Less Responsive to Intensive Care? Albert W.Wu,MD,* Haya R. Rubin, MD, PhD,* and MarkJ. Rosen, M D t
Older patients may be excluded from intensive care units because of the perception that they will benefit less than younger patients. To determine if advanced age is associated with increased mortality independent of severity of illness, we compared older and middle-aged patients admitted to a medical intensive care unit. W e reviewed the charts of 130 patients age 75 years or older and 135 patients age 55 to 65 admitted over a 30-month period. W e controlled for severity of illness using the Acute Physiology Assessment and Chronic Health Evaluation (APACHE 11) system without including points for age (APACHE 11M).The groups were similar with regard to gender, whether or not they had a private attending physician, mean APACHE 11M score, and diagnoses, except that older patients had more chronic obstructive pulmonary disease. Hospital stay was slightly longer in the older group (37 us. 39 days, rank sum, P < .02).Hospital mortality was significantly greater in the older group (39% us. 51 %, Chi-square P < .05) with a crude relative risk of
1.32 (95% confidence interval [Cl]: 1.01, 1.73). However, the relation of age group to mortality differed for patients with different diagnoses. When we used logistic regression to adjust for APACHE UM,whether the patient had a private attending physician, primary admitting diagnosis, or presence of cancer, older patients did not have a significantly greater risk of dying (adjusted relative risk, 1.05; 95% Cl: 0.97, 1.12). When pulmonary artery catheterization was added to the model, it independently predicted mortality (adjusted relative risk, 1.47; 95% Cl: 1.05, 2.06). APACHE 11M (calculated without the inclusion of age) was an excellent predictor of mortality. Older age did not predict mortality once severity of illness, admitting diagnosis, and the presence of underlying malignancy were taken into consideration. Further studies should explore the effects of age separately for patients with different diagnoses as well as control for severity of illness. J Am
I
tients has been examined previouslyI1-l2few studies have compared the outcome of intensive care in elderly patientswith that of younger patients with similar severity of illness. The purpose of this study is to determine if advanced age is associated with increased mortality in a general medical intensive care unit, after accounting for severity of illness. The APACHE I1 (Acute Physiology Assessment and Chronic Health Evaluation)system provides a quantitative estimate of severity of illness and accurately indicates prognosis in the intensive care unit (ICU)setting when used in prospective and retrospective eva1uations.l3-l* The APACHE I1 score includes points for age group, acute physiologic derangements, and chronic health status. The original validation studies refer to age as "a well-documented risk factor for death from acute illness, that is independent of the severity of disease," n and older age groups were found to have a higher mortality rate independent of the acute physiology and chronic health scores. However, subsequent studies of the effect of age and severity of illness on mortality have yielded conflicting r e s ~ l t s . ~ . 9 . ~ ~ We conducted a retrospective study to determine the
s a patient's age an appropriate criterion to use in making decisions about intensivecare? In most hospitals, the limited number of intensive care beds requires physicians to decide who will be admitted to them. Patients considered to have an extremely poor prognosis may not be offered admission because of limited beds and the low likelihood of benefit. Many feel that elderly patients, because of chronic illnesses and impaired physiologic reserve, have a worse prognosis then younger individuals, and there is evidence to suggest that physicians may be less aggressive in treating elderly patients.' Although intensive care in elderly pa-
From the *Department of Medicine, University of California, San Francisco, San Francisco, California and tDeparhnent of Medicine, The Mount Sinai Medical Center, New York, New York. Supported in part by a grant from the Catherine and Henry Gaisman Foundation and the Veterans Administration/Robert Wood JohnsonClinical ScholarsProgram. Presented at the society of Critical Care Medicine Annual Educational and Scientiiic Symposium, Anaheim, California, May 26-29,1987. Address correspondenceand reprintrequests to Albert W.Wu,MD, Robert Wood JohnsonClinical Scholars Program, 350 Pamassus Avenue, Room 407, San Francisco, CA 94117. 0 1990 by the American Geriatrics Society
Geriatr SOC38:621-627,1990
0002-8614/90/$3.50
622 W U ET AL
IAGS-JUNE 1990-VOL 38, NO. 6
relation between age and mortality among individuals in a general medical ICU with comparable severities of illness, as quantified by APACHE I1 scores calculated without including age.
MATERIALS AND METHODS Hospital and Intensive Care Unit The Mount Sinai Medical Center is a 1,212-bed tertiary care hospital in New York City. The Medical Special Care Unit consists of eight beds used for general medical intensive care. Coronary care and surgical patients are usually treated in other units. Decisions regarding admission to the unit are made by a resident in internal medicine, with consultation from a fellow or from the director of the unit in cases of uncertainty or objections by other medical staff. Decisions to discharge patients are usually made jointly by the attending physician and house staff rotating through the unit. Patients frequently are discharged because a bed is needed for another patient who may be more critically ill or more “salvageable.” Study Design Patients admitted to the ICU between January 1, 1982 and June 30, 1984 were eligible for study. All available charts of patients 75 years or older were analyzed and compared with those of patients between the ages of 55 and 65. These age groups were selected to compare elderly patients with a group with comparable diagnoses whose members would not have been refused ICU admission on the basis of age alone. Measures Independent variables included those hypothesized to affect mortality: age, gender, whether or not the patient had a private attending physician (private/service), primary ICU admitting diagnosis, severity of illness, and major diagnostic and therapeutic interventions. The primary admitting diagnosis was determined from the resident’s admission note. We classified the most common admitting diagnoses as having high or low risk; cardiopulmonary arrest, septic shock, and pneumonia were considered ”high risk’ whereas gastrointestinal hemorrhage, chronic obstructive pulmonary disease, congestive heart failure, and neuromuscular disease were classified as “low risk.” Less frequent diagnoses included asthma, adult respiratory distress syndrome, pulmonary embolism, restrictive lung dis-
ease, upper airway obstruction, hemoptysis, pneumothorax, hypertensive crisis, arrhythmia, hypovolemia, coronary artery disease, seizure, intracranial hemorrhage, stroke, acute renal failure, anemia, diabetic ketoacidosis, drug overdose, and metabolic imbalance. Major interventions were defined as intubation and mechanical ventilation, pulmonary or peripheral arterial catheters, peritoneal or hemodialysis, cardiac pacemakers, or tube thoracostomy. The APACHE I1 system point score is based on 11 routine physiologic measurements, the Glasgow coma score, age, and previous health status.’ To measure severity of illness, a score was calculated for each patient without including age (APACHE IIM), based on information recorded in the chart during the first 24 hours of ICU hospitalization. Physiologic measurements were abstracted from detailed nursing flow sheets. Glasgow coma score was determined using medical and neurologic data in the ICU chart. Outcomes recorded were length of ICU and hospital stay, readmission to the ICU, and ICU and hospital survival.
Statistical Analysis For continuous variables, differences between group means were determined using the unpaired two-tailed Student’s t test. Where distributions were not normal, the Wilcoxon rank-sum test was used. The Chi-square test was employed to compare proportions for different groups. Logistic regression analysis was used to examine the independent relation to mortality of age, severity of illness, gender, private/ service status, and treatment. We calculated adjusted relative risk for each independent variable using the most typical values for each of the other variables. RESULTS During the 30-month study period there were 895 admissions to the ICU. One hundred seventeen of these were readmissions; only first ICU admissions were considered to ensure that all observations were independent. The mean age for the 778 patients was 57 years, witharangeof 15 to97; 181(23%)wereage55 to65and 164 (21%) were age 75 or older. Eighty of these charts were incomplete or unavailable for review (46 in the younger and 34 in the older group). The age and gender distribution for those patients whose charts were avail-
TABLE 1. STUDY GROUP CHARACTERISTICS (55 to 65 Years)
Group I1 (275 Years)
135 63 (47) 90 (67) 15.8 f 9.5
130 54 (42) 99 (76) 16.1 & 9.1
Group I Number Male (%) Number with private attending physician (96) Mean APACHE IIM score & SD (does not include age) There were no significant differences between groups I and II.
IAGS-JUNE 1990-VOL 38, NO. 6
MEDICAL INTENSIVE CARE IN ELDERLY PEOPLE
623
TABLE 2. PRIMARY ADMITTING DIAGNOSES Group I
(55 to 65 Years)
High-risk diagnoses Cardiopulmonaryarrest (%) Septic shock (%) Pneumonia (%) Low-risk diagnoses Gastrointestinalhemorrhage (%) Chronicobstructive lung disease (%)* Congestive heart failure (%) Neuromuscular disease (%) Other diagnoses (%)t Total
Group I1 (2:75 Years)
19(14) 15(11) 16(12)
21(16) 18(14) 16(12)
Chi-square, P < .005. t Chi-square, P < .02.
able did not differ significantly from those whose charts were missing.
Sample Characteristics Charts were available for 135 patients aged 55 to 65 years (mean, 60.2) and 130 aged 75 years or older (mean, 80.5). We found no differ-
ences between the groups in gender, private/service status, and mean APACHE IIM score (Table 1).
Admission Diagnoses Table 2 lists the primary diagnoses at the time of ICU admission in both age groups. The distribution of diseases was similar in the two study groups except that more elderly patients had chronic obstructive pulmonary disease whereas fewer had "other" diagnoses. Hospital Course and Major Interventions Two hundred eleven patients (8096) had invasive interventions, including mechanical ventilation, dialysis, chest
tube placement, or invasive hemodynamic monitoring (Table 3). There was no significant difference in the proportion of younger and older patients who had an invasive intervention or in the mean number of procedures performed per patient. However, pulmonary arterial catheterization was performed more frequently in the older patients (76% vs. 62Y0, P 5.02). There were no significant differencesbetween groups in length of ICU stay or ICU readmission. However, once discharged from the ICU, older patients stayed longer in the hospital (Table 3).
Relation of Age to Mortality For both age groups, the APACHE IIM score correlated well with ICU and hospital death rates (Figure 1). Mean APACHE IIM score was 12.3 f 7 for survivors compared with 20.5 f 10 for those who died (t test, P < .0001). Overall ICU mortality was 21Yo and total hospital
TABLE 3. PROCEDURES, INTENSIVE CARE UNIT (ICU), AND HOSPITAL COURSE Major intervention (%) No.of interventions/patient Pulmonary artery catheterization (%y Endatracheal intubation (%) Mean ICU days Mean hospital days after ICU discharget Total hospital dayst ICU readmission (%) Mortality (%)$ ICU mortality (%) Hospital mortality after ICU discharge (%)€j Chi-square, P 5 . 0 2
t Runk sum, P 5.02.
$ Chi-square, P 5.05. 5 Chi-square, P 5.001.
55 to 65 Years
2 7 5 Years
Total
105(78) 1.8 84(62) 35(26) 7.4 21.8 37.4 12 (9) 52(39) 34(25) 18(13)
106(82) 2.0 99(76) 42(32) 6.3 25.3 39.1 10 ( 8 ) 66(51) 22(17) 44(34)
211(80) 183(69) 77(29)
22 (8) 118(45) 56(21) 62(23)
624 W U ET AL
IAGS-IUNE 1990-VOL 38, NO. 6
100
Percent Died
6 2
80 60
FIGURE 1. Hospital mortality by APACHE IIM score (without age) comparing patients aged 55 to 65 years with those 75 years of age and older. Numbers above each bar indicate numbers of patients.
40 20 0
0-4
5-9
10-14
15-19
20-24
25-29
30-34
35*
Apache II Score Without Age 55-65
mortality was 45% (Table 3). For younger and older groups, the difference between rates of ICU mortality was not significant. However, a sigdicantly greater proportion of elderly patients died in the hospital after discharge from the ICU, so elderly patients had a significantly greater total hospital mortality rate (39%vs. 51%, P 5 .05). Table 4 lists crude relative risks of mortality depending on the patient's age group, gender, private/service status, diagnostic risk group (as classified in Table 2), presence or absence of underlying diagnosis of malig-
75 and older
nancy, and whether or not the patient received pulmonary arterial catheterization. All factors (except gender) were significantly associated with mortality. Logistic regression was performed using admission characteristics: APACHE IIM, age group, gender, private/service status, presence of cancer, and presence of a high-risk admission diagnosis as independent variables (Table 4). Older patients did not have a sigdicantly greater risk of dying controlling for other variables. Patients with higher APACHE IIM scores were more likely to die as were those with underlying cancer.
TABLE 4. RELATION OF PATIENT DEMOGRAPHIC AND CLINICAL VARIABLES TO MORTALITY Independent Variab1e Age group 55 to 65 years 2 75 years Gender Male Female Status Service Private Primary diagnosis Low risk High risk Malignancy Absent Present Pulmonary artery catheter No Yes Intuba tion
No Yes
Died
(%)
Crude Relative Risk (95%CI)
52/135 66/130
(39) (51)
Old/young = 1.32 (1.01, 1.73)
59/117 59/148
(50) (40)
Male/female = 1.265 (0.97, 1.66)
26/76 92/189
(34) (49)
Private/service (1.03, 1.96)
26/82 69/105
(32) (66)
High/low risk = 2.07 (1.52, 2.82)
89/224 29/41
(40) (71)
Present/absent = 1.78 (1.31, 2.42)
18/82 100/183
(22) (55)
Yes/no = 2.49 (1.73, 3.57)
57/188 61/77
(30) (79)
Yes/no = 2.61 (2.02, 3.39)
= 1.42
JAGS-IUNE 1990-VOL.38, NO. 6
APACHE IIM score alone classified 69% of cases correctly; APACHE IIM for patients with cancer and a high-risk admission diagnosis classified 74% of cases correctly. Age group did not enhance the ability to predict mortality. To see if differences in treatment of older and younger patients might confound the relation of age to mortality, we added pulmonary arterial catheterization to the logistic regression model. Pulmonary arterial catheterization independently predicted mortality (adjusted relative risk, 1.47;959b CI: 1.05,2.06)and increased correct classification from 699'0 to 72%. However, older patients still did not have higher adjusted relative risk of dying, and this model's predictive ability was also the same with and without age. To explore whether age group and mortality might be associated for subgroups within the population, age and mortality were cross-tabulated while controlling for gender, private/service status, and severity of illness (dividing APACHE IIM score into two groups: 115 and >15). Older patients had higher mortality rates in only two out of the eight private/service status-genderseverity of illness subsets of our sample. Among male private patients with APACHE IIM scores 515 (n = 46), 30% of younger versus 65% of older patients died in the hospital (Chi-square = 5.6, P = .018). Among female private patients with APACHE IIM scores >15 (n = 44), 41% of younger vs. 73% of older patients died (Chisquare = 4.5,P = .033). For all other subsets, both age groups had very similar mortality rates. As mentioned earlier, more older patients had an admission diagnosis of chronic obstructive lung disease, but these patients did not die at a higher rate than their younger counterparts. Admission diagnoses for our two age groups were otherwise similar. Because our sample size did not allow inclusion of individual admitting or secondary diagnoses as predictor variables in the logistic regression analysis, we examined how age was related to mortality within specific diagnostic groups. Crosstabulation of mortality and age group for specific diagnoses revealed that admission diagnoses modified the relation of age group to mortality independent of severity of illness. For example, despite lower mean APACHE IIM scores, older patients with pneumonia had twice the mortality of younger patients (884b vs. 444b,P 1 .01). Among those with sepsis, congestive heart failure, and chronic obstructive lung disease, a larger proportion of younger patients died (differences not significant) despite having similar or lower mean APACHE IIM scores. Among those with secondary diagnoses of malignancy, mean APACHE IIM scores were similar for older and younger groups and a slightly larger proportion of older patients died (differences not sigruficant). Thus, the relationship of age to mortality differed for patients with different diagnoses.
MEDICAL INTENSIVE CARE IN ELDERLY PEOPLE
625
DISCUSSI 0N In our study of a general medical ICU, a modified APACHE I1 score calculated without points for age (APACHE IIM) was an excellent predictor of mortality in a retrospectiveanalysis. When severity of illness, presence of a high-risk admission diagnosis, and underlying malignancy were considered, older patients did not have higher odds of dying than younger patients, and age group did not add to predicting mortality. Furthermore, older patients had a similar or even better prognosis than younger patients in some diagnostic subgroups, suggesting that advanced age does not uniformly confer a worse prognosis. A number of studies have examined intensive care outcomes for elderly patients.l-12 All but two3J0found that elderly patients had higher ICU or hospital mortality. Four of these studies attempted to control for the effect of severity of illness on the relationship between age and mortality. Fulkerson et a1 reported that age did not substantially contribute to ICU mortality when severity of illness, as estimated by the need for mechanical ventilation or cardiopulmonary resuscitation, was taken into account.6 Nicholas and associates found that although ICU mortality increased progressively with age group, the mortality rate in patients over 75 was equal to that observed in patients aged 55 to 59." To control for severity of illness, the authors stratified by the Acute Physiology Score (APS) component of the APACHE I1 into low and high severity groups (515 and >15). In both groups, older patients had a significantly higher death rate. However, this study did not include the chronic health evaluation (CHE) component of APACHE 11. Because elderly patients generally have more chronic health problems than younger patients, excluding CHE points reclassifies more older patients as "low severity." This reclassificationwould result in a "Will Rogers phenomenon," inflating the mortality rate for older patients in both low and high severity groups.19 McClish and associates employed both the acute and chronic health components of APACHE IL9They found that ICU mortality increased from 18%for patients aged 55 to 64 to 25% for patients 75 years of age or older, but that this difference disappeared when logistic regression was used to control for demographic and diagnostic variables. These results appear consistent with our findings. The original validation of Knaus et a1 of APACHE I1 examined whether age independently predicted survival.13They demonstrated an increasing risk of mortality with increasing age, adjusting for severity of illness, surgical status, and major diagnoses. Although the authors did not directly compare the two age groups we examined, recalculation from their regression coeffi-
1AGS-1UNE 1990-VOL. 38, NO. 6
626 W U ET AL
significance of a relative risk of 1.05 would still be doubtful. In addition, because the average risk of older versus younger patients differed by demographic and diagnostic subgroups, the conclusion that age confers uniform risk would still be suspect. Two incidental findings merit discussion. First, in both age groups, patients of private physicians had worse outcomes than service patients despite similar severity of illness and treatment intensity. However, multivariate analysis demonstrated that the increased incidence of underlying malignancy among private patients (37 private vs. 4 service)accounted for their excess mortality. Possibly, private attending physicians prevailed upon house staff to admit higher risk patients to the ICU. Second, pulmonary arterial catheterization was an independent predictor of hospital mortality in our study. There are several ways to interpret this finding. Because hemodynamic derangement led to the placement of pulmonary artery catheters, this intervention may indicate an increased disease severity that is not reflected by APACHE IIM. Furthermore, catheter placement is often an intervening outcome; if it occurred after the first 24 hours of ICU care (when APACHE scores were calculated), it may simply reflect a deteriorating course. In a retrospective population-based study of acute myocardial infarction, Gore et a1 demonstrated that patients receiving a pulmonary artery catheter for congestive heart failure or hypotension had substantially increased mortality and length of hospital stay compared with those who were not catheterized.20Like our study, their investigation was retrospective, and they observed that they could not determine if patients who received a pulmonary artery catheter were sicker than those who did not. Nevertheless, our results support the recent call for a randomized trial of pulmonary arterial catheterization2l Variables we did not measure may also affect out-
cients gives a relative risk of dying of 1.4 for patients 75 years of age or older compared with patients aged 55 to 65. This is greater than the adjusted relative risk of 1.05 that we found in our study. The study of Knaus et a1 did not examine whether specific diagnoses modified the relation of age to mortality. The varying relation of age to mortality that we found within different diagnostic subgroups argues against the routine incorporation of age into APACHE scores. A recent review of severity of illness measures concludes that “although intrinsic biological severity may one day be measurable, currently it is an abstraction.” l7 Owing to the heterogeneity of patients within older groups, chronologic age is probably not as important in determining survival as biologic age, which we are unable to measure. Our study had some methodologic constraints. We had no other source of demographic or clinical information about the patients whose charts were unavailable, who may have differed from those we studied. In addition, we do not know how our patients compared with those denied or not considered for admission to the ICU. If elderly patients were more carefully selected based on ”hardiness” or clinical factors not reflected in our diagnostic or severity-of-illness measures, this might bias our results. To answer this question would require detailed accounting of all ICU consults, both formal and informal, about patients considered for admission; such data could not be collected in a retrospective chart review but would be important to consider in future prospective studies. It is also possible that we failed to find a significant relationship of age to mortality in multivariate analysis because our sample size was too small. The lower confidence limit for the adjusted relative risk (Table 5) is close to 1. Had we collected more observations we might have demonstrated that the slightly higher adjusted relative risk of death of the older group was statistically different from 1. However, the clinical
TABLE 5. LOGISTIC REGRESSION ANALYSIS USING ADMISSION CHARACTERISTICS AS PREDICTORS OF MORTALITY __
.. .
~
- .-
~
-~ - -.
~
_
~~
_
-~
-
-
~
Risk Factor
P Value
Adiusted Relative Risk (95%CI)
APACHE IIM Age group Gender Status Primary diagnosis Malignancy
,000 .187 ,232 ,278 ,082 .016
Low/high* = 1.40 (1.10, 1.67) Old/young = 1.05 (0.97, 1.12) Male/female = 1.04 (0.98, 1.12) Private/service = 1.05 (0.96, 1.12) High/low risk = 1.07 (0.99, 1.14) Yes/no = 1.11 (1.03, 1.15)
.... ....
.
- -
~~~~~~~~
~~
~
~~
~~
~
~~~~~~~~
~~
~
-~ ~
~~
~
~
~
. . .
~
~
~~~
~~~~~~
~
APACHE IIM scores 1 SD above and below the mean. Logistic regression equation: In (aliveldead) = 1.4643 - 0.2380 (age group) - 0.1783 (APACHE IlU, 0.2105 (gender) - 0.2243 (status) - 0.7514 (malignancy) - 0.3244 (primary diagnosis). Coding ofvariables: Agegroup: 55 to 65 = 0, 2 75 = 1; male = 0, female = I ; service status = 0, private = I ; no malignancy = 0, malignancy = I; low-risk diagnosis = I , high-risk diagnosis = 2. All relative risks calculated using most typical valuesfor other riskfactors, ie, mean APACHE IIM,older, female, private, high-risk primary diagnosis, and no underlying malignancy.
+
JAGS-]LINE 1990-VOL 38, NO. 6
come, such as the experience and expertise of ICU personnel and the attitudes of staff toward patients. It would be valuable to examine the relationship of doctors‘ and nurses’ expectations about patient prognoses to outcome, controlling for severity of illness. In conclusion, our study validates APACHE I1 without the inclusion of age as an excellent predictor of mortality, even when applied retrospectively from chart review. Advanced age did not predict mortality once severity of illness, presence of a high-risk admitting diagnosis, and presence of underlying malignancy were taken into consideration.For specific diagnoses, elderly patients had similar prognoses to younger patients; for others, their outcome was worse despite a lower APACHE IIM score. Our findings illustrate that the relationship of age to mortality is complex, even for patients with similar severity of illness. Our findings suggest that elderly patients should not be denied consideration for intensive care on the grounds that they are less likely to benefit than their younger counterparts. Future studies should explore the effects of age separately for patients with different diagnoses as well as control for the severity of illness. The incorporation of age into the calculation of APACHE I1 score should be reconsidered because the assumption of an equal effect across groups of patients may be invalid.
REFERENCES 1. Campion EW, Mulley AG, Goldstein RL, et a1 Medical intensive care of the elderly: a study of current use, costs, and outcomes. JAMA 246:2052,1981 2. Thibault GE, Mulley AG, Bamett 0,et al: Medical intensive care: indications, interventions and outcomes. N Engl J Med 302:938,
1980 3. FedulloAJ, Swinbume AJ: Relationship of patient age to cost and survival in a medical ICU. Crit Care Med 11:155,1983 4. Goldstein RL, Campion EW, Mulley AG, et a1 Nursing home patients admitted to a medical intensive care unit. Med Care
222354, 1984
MEDICAL INTENSIVE CARE IN ELDERLY PEOPLE
627
5. McLean RF, McIntosh JD, Kung GY, et al: Outcome of respiratory intensive care for the elderly. Crit Care Med 13:625, 1985 6. Fulkenon WJ, DeLong E, Young SL Elderly admissions to a medical ICU: 8 year experience (abstract). Am Rev Resp Dis 133:A351,1986 7. Smithies M, Bihari D, Catling J, et a 1 Intensive therapy of the elderly (abstract). Crit Care Med 14:397,1986 8. Franklin C, Nightingale S, Mamdanni B, et al: Increases in mortality in the oldest old critically ill patients (abstract). Crit Care Med 14:398,1986 9. McClish DK, Powell SH, Montenegro H, et al: The impact of age on utilization of intensive care resources. J Am Geriatr Soc
35983, 1987 10. Papadakis MA, Browner W S Prognosis of noncardiac medical 11.
12. 13. 14. 15. 16. 17. 18.
patients receiving mechanical ventilation in a veterans hospital. Am J Med 83:687,1987 Nicholas F, LeGall JR, Alperovitch A, et al: Iduence of patients’ age on survival, level of therapy and length of stay in intensive care units. Intensive Care Med 13:9, 1987 Fedullo AJ, Swinbume AJ: Intensive care for elderly patients with myocardial infarction: age-related comparison of clinical features and outcome. J Intensive Care Med 3:265, 1988 Knaus WA, Draper EA, Wagner DP, et al: APACHE I 1 a severity of disease classification system. Crit Care Med 13:818, 1985 Knaus WA, Draper EA, Wagner DP, et al: An evaluation of outcome from intensive care in major medical centers. Ann Intem Med 104:410,1986 Chang WS, Jacobs S, Lee B Use of APACHE I1 severity of disease classification to identify intensive care unit patients who would not benefit from total parenteral nutrition. Lancet 1:1483, 1986 Jordan DA, Miller CF, Kubos KL, et a 1 Evaluation of sepsis in a critically ill surgical population. Crit Care Med 15897,1987 Stein REK, Gortmaker SL, Perrin EC, et al: Severity of illness: concepts and measurements. Lancet 2:1506, 1987 Johnson MH, Gordon PW, Fitzgerald FT Stratificationof prognosis in granulqtopenic patients with hematologic malignancies using the APACHE I1 severity of illness score. Crit Care Med
14:693,1986 19. Feinstein AR, Sosin DM, Wells CK The Will Rogers phenomenon. Stage migration and new diagnostic techniques as a source of misleading statistics for survival in cancer. N Engl J Med
312:1604,1985 20. Gore JM, Goldberg RJ, Spodick DH, et al: A community-wide assessment of the use of pulmonary artery catheters in patients with acute myocardial infarction. Chest 92:721,1987 21. Spodick DH: Analysis of flow-directed pulmonary artery catheterization. JAMA261:1946,1989
