Introduction

The elderly population is growing in developed countries. For example, the number of persons ≥80 years old in Paris, France, increased threefold between 1962 and 1990 [1]. In addition, elderly patients are known to use more hospital, particularly intensive care unit (ICU), resources [2]. The decision to admit elderly patients, especially the oldest-old (≥85 years), to the ICU is probably influenced by atypical presentations of their clinical conditions [3], variable evaluations of the potential benefit of admission in the ICU [4] and the consideration of such patients as potential “bed-blockers.” However, little is known about the short- and long-term outcomes of the elderly, particularly of the oldest-old, patients admitted to the ICU [4].

Age is generally considered a major determinant of the outcome of ICU patients [5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19]. Most previous studies, however, did not distinguish between the old (75–79 years), very old (80–84 years) and oldest-old (≥85 years) patients. The results of one study indicated that the ICU mortality rate of 65–75-year-old patients did not differ significantly from that of older (>75 years) patients [10]. The authors of other studies focusing on the outcome of the oldest-old population concluded that age alone was not associated with ICU mortality [20, 21, 22, 23]. Finally, in the APACHE III study, age was not a major determinant of hospital mortality [24].

Studies evaluating the relationship between age and post-ICU discharge mortality provided conflicting results [4]: six studies did not find a significant association between age and post-ICU mortality [6, 7, 15, 19, 21, 22], whereas seven others identified age as a determinant of post-ICU mortality [5, 11, 12, 20, 25, 26, 27]. In most of those studies, ICU and post-ICU mortality rates were pooled for multivariate analysis of outcome [5, 6, 7, 11, 12, 21, 26, 27]. As a consequence, factors associated with deaths of the elderly in the ICU on the one hand, and post-ICU on the other hand, have not yet been clearly distinguished in the literature.

Therefore, we initiated this study to analyze the characteristics of the old, very old and oldest-old patients admitted to an ICU, and to assess the relative effects of age on ICU and long-term outcome.

Materials and methods

Patients and age groups

All critically ill patients ≥75 years old who were admitted to the ten-bed medical ICU at Broussais University Hospital, Paris, between 1 January 1991 and 31 October 1996 were included in the study. There was no written ICU admission policy regarding patient’s age. The study population was divided into three groups by age: 75–79, 80–84 and ≥85 years old. For patients admitted more than once during the study period, only the last stay was included in the analysis in view of not underestimating ICU and post-ICU mortalities.

Data collection

The following data were prospectively collected: age; sex; underlying medical condition(s) at admission, according to the criteria of McCabe and Jackson [28], considered to be rapidly fatal, ultimately fatal or not fatal; previous health status [24] before hospitalization, classified as no limitation of activity, moderate limitation, severe limitation or bedridden or institutionalized; the primary diagnosis; the presence or absence at admission of cardiac, respiratory, renal, hepatic, neurologic and/or hematologic dysfunction(s) as defined by the organ dysfunction(s) and/or infection (ODIN) score [29]. The severity of illness was estimated using the acute physiology assessment and chronic health evaluation (APACHE II) score [30]. After 75 years of age, this score is not modified by age. Therapeutic activity was estimated using the OMEGA score [31, 32], which is composed of therapeutic items, each accorded 1 to 10 points, and divided into three categories as follows: category 1 includes 26 items entered only at the time of the first application, like central venous catheter (3 points) or massive blood transfusion (10 points); category 2 includes seven items entered at each application, like intrahospital transport (3 points) or hemodialysis (10 points) and category 3 includes five items entered every day of application like continuous ICU surveillance (4 points) or mechanical ventilation (10 points). The total score, covering the entire length of stay, is calculated by adding all the points. The OMEGA score per day was calculated by dividing the total OMEGA score by the length of ICU stay. Outcome variables were: ICU mortality, ICU length of stay, duration of mechanical ventilation and occurrence of nosocomial infections, using predefined diagnostic criteria [32].

Follow-up telephone interviews were conducted with ICU survivors or their relatives to ascertain the vital status of studied patients on 1 January 1997. When phone contact was impossible, a letter was sent to the primary care physician and/or to the Registry Office of the patient’s birth district where the birth and death of every French citizen is recorded.

Statistical analyses

Results are expressed as means ± standard error (m±SE), medians [range] or numbers of patients. Comparisons between groups were made with one-way analysis of variance (ANOVA) for continuous variables and the chi-square test for categorical variables. The long-term mortality of ICU survivors was analyzed using Kaplan-Meier survival curves and the log-rank test.

Two separate multivariate analyses based on the Cox proportional hazards model were performed to identify factors independently associated with ICU and/or long-term mortality. For ICU mortality, multivariate analysis included age, sex, severity of illness (APACHE II score), previous health status classified as no limitation versus limitation of activity and underlying medical conditions (McCabe and Jackson score) as dependent variables. For long-term mortality, we also included variables related to ICU stay: ICU length of stay, therapeutic activity (OMEGA score per day), use of mechanical ventilation and nosocomial infection(s). Statistical analyses were performed with the JMP 3.2 Statistical Software System (SAS Institute Corp., Cary, N.C.). All P values were two-tailed, with significance set at 0.05.

Results

Four hundred and thirty-four of the 1,795 (24.2%) admissions during the study period concerned patients ≥75 years, representing 3,486 days of ICU stay (21.8% of the total number of ICU days). When readmissions were excluded, 410 patients constituted the study population (Fig. 1), with 182 75–79 years old, 137 80–84 years old and 91 ≥85 years old. Comparing the periods 1991–1994 and 1995–1996, the overall percentages of elderly patients were similar (24.0 and 24.6% of all admissions, respectively; P=0.762), with similar representations of the three age classes (44.9, 35.2 and 20.3% for 1991–1994 and 44.4, 32.0 and 23.6% for 1995–1996; P=0.785).

Fig. 1.
figure 1

Distribution of ICU admissions by age

Full size image

Clinical characteristics of the studied patients are given in Table 1. The mean APACHE II score for all studied patients was significantly higher than that calculated for patients <75 years of age admitted to the ICU during the same period (20.1±9.4 vs. 16.1±10.5, respectively; P<0.001); however, when points related to age were excluded, the part of APACHE II score related to physiologic perturbation and previous health status were comparable (14.1±9.4 vs. 13.3±10.2, respectively; P=0.147). The primary diagnosis was respiratory failure in 111 patients, including acute exacerbation of chronic obstructive pulmonary disease in 29, pulmonary embolism in 10, post cardiac surgery acute respiratory distress syndrome in 34 and pneumonia in 30, without differences between the three groups of age.

Table 1. Patients’ characteristics at ICU admission
Full size table

One hundred and nineteen patients who were ≥75 years died during their ICU stays (ICU mortality, 29.3%), but no significant between-group difference was found (Table 2). The mortality rate observed for the 1,361 admissions of patients <75 years old was 19.2% (P<0.001 vs. the study population). Among the 85 patients who stayed ≤2 days in the ICU, a significantly higher percentage were ≥85 years old; (P=0.012, Fig. 1). The overall mortality rate of elderly patients hospitalized ≤2 days was 45.9% (62.5, 37.5 and 34.4%, respectively; P=0.056) as compared to 24.6% of the 325 patients hospitalized >2 days (P<0.001) (25.3, 22.1 and 27.4% for the three age groups, respectively; P=0.711). No difference was observed between survivors and non-survivors for age, sex, previous health status or McCabe and Jackson score. In contrast, the mean APACHE II score was higher for non-survivors (Table 2). Multivariate analysis of factors associated with ICU mortality identified the APACHE II score as the only factor independently associated with ICU mortality [odds ratio (OR): 1.11, 95% confidence interval (CI): 1.09–1.14); P<0.001].

Table 2. Characteristics of ICU non-survivors and survivors
Full size table

Use of ICU resources by the three groups differed significantly (Table 3). ICU length of stay differed among survivors of the three groups (9.7±0.7, 7.0±0.7 and 4.6±0.9 days, respectively; P<0.001), but not non-survivors (10.4±1.8, 11.1±2.4 and 9.3±2.7 days, respectively, P=0.9). Nosocomial infections complicated the course of 22.6% of 75–79-year-old patients, 17.5% of 80–84-year-old patients and 9.9% of ≥85-year-old patients (P=0.037).

Table 3. Use of ICU resources
Full size table

Long-term follow-up was obtained for 283 of the 291 ICU survivors (97.3%) (Fig. 1). Their mean follow-up was 649 days (median: 438 days; range: 1–2,151 days). For all 291 ICU survivors, the survival time after discharge, evaluated using the Kaplan-Meier method, progressively decreased with age (P=0.007) (Fig. 2A). Examining only the 185 patients admitted in 1991 to 1994 who had a mean follow-up of 866 days (median: 861 days, range: 1–2,151 days), did not modify our estimates (Fig. 2B) (P=0.013, comparing the three groups). This latter cohort included 68 (36.8%) patients who were still alive on 1 January 1997 (mean follow-up: 1,366 days; median: 1,360 days; range: 722–2,136 days).

Fig. 2.
figure 2

Survival after ICU discharge for (A) all patients and (B) those with long follow-up admitted in 1991–1994

Full size image

Multivariate analysis including age, sex, McCabe and Jackson score, previous health status, APACHE II score, use of mechanical ventilation, length of ICU stay, therapeutic activity (OMEGA score per day) and occurrence of nosocomial infection, identified age [OR: 2.17 (95% CI: 1.36–3.46), P=0.001 for ≥85-year-old patients, and 1.82 (95% CI: 1.21–2.74), P=0.004 for patients 80–84 years old, with 75–79-year-old patients serving as the reference group] and limitation of activity [OR: 1.74 (95% CI:1.09–2.76); P=0.020] as factors independently associated with long-term mortality. Inclusion of age as a continuous variable in the model did not change our results [OR: 1.06 (95% CI: 1.02–1.10), P=0.006].

The mortality rates 3 months after discharge from the ICU were 21.6% for 75–79 year olds, 26.7% for 80–84 year olds and 28.9% for patients ≥85 years old, as opposed to 0.9, 1.6 and 3.7% for the same age group in the general French population (data obtained from the Institut National de la Santé et de la Recherche Médicale (INSERM); at http//www.inserm.fr) (Fig. 3A). For patients surviving 1 year after ICU discharge, mortality observed 1 year later (i.e., 2 years after ICU discharge) were 14.9, 16.9 and 19.4 for the three studied groups, vs. 3.6, 6.3 and 14.8% for persons of the same age in the general French population (Fig. 3B).

Fig. 3. A
figure 3

Mortality rates of ICU survivors (open bars) compared to those of the general French population of the same age (closed bars) at 3 months. B Mortality rates of survivors 1 year after ICU discharge (open bars) compared to those of the general French population of the same age (closed bars) at 2 years

Full size image

Discussion

Our study of patients ≥75 years old admitted to an ICU showed that age is not the major determinant for ICU survival. During the first 3-month period following ICU discharge, the mortality rate of these “post-ICU” patients was much higher than that of individuals of the same age in the general French population. This difference decreased by about 50% or more for 1-year mortality of patients surviving 1 year after ICU discharge. Age and limitation of activity before admission were the two factors identified as being independently associated with long-term mortality.

The patient population ≥75 years old is increasing in all ICUs, with admission rates varying from 12% as reported in 1989 [27] to 24% in this study. Few demographic data on the oldest-old (≥85 years) patients are available: Knaus et al. [30], Chelluri et al. [22] and Ridley et al. [5] reported respective ICU-admission rates of 5.3, 1.1 and 0.6% of the oldest-old patients; in our population 5% of the admitted patients were ≥85 years.

The three populations defined by age in our study (75–79, 80–84 and ≥85 years old) were comparable in terms of severity of illness and previous health status. The oldest-old population was similar to that previously described with 20% of the patients having unlimited activity and <5% being bedridden or institutionalized [9, 13, 21, 23]. Disease severity, as assessed by the APACHE II score and the type(s) of organ dysfunction(s) were not significantly influenced by age. A mean APACHE II score between 16 and 20 points was usually reported for patients ≥75 years old [2, 6, 7, 8, 9, 10, 22, 23]. No admission restriction based on age exists in our unit; however, as suggested by the significantly shorter length of stay of the oldest-old population, particularly the survivors, a different initial evaluation of the suspected diagnosis and/or the prognosis probably influenced the decision to admit these patients to the ICU, as previously suggested by others [33].

Mean ICU length of stay, use of mechanical ventilation and level of therapeutic activity evaluated by the OMEGA score declined with age in our study. This result cannot readily be explained by more early deaths among the oldest-old patients, since length of ICU stay did not differ significantly between age groups; in addition, no relationship was found between age groups and mortality rates of patients with short stays or in those with stays >2 days.

Restricting access to care of elderly patients was previously reported in the APACHE III study [24] and, more recently, by Castillo-Lorente et al. [35]. Length of stay depends on admission diagnosis, ICU-acquired complications and probably organizational factors (including admission and discharge protocols). Hamel et al. [27] reported that limitation of care was significantly and independently associated with age >70 years, but did not influence the mortality rate observed 6 months later.

However, studies usually analyzed mortality of patients ≥75 years old without differentiating among age subgroups. Reported mortality rates, varying from 10% [12] to 39% [11] probably reflect differences in underlying diseases and disease severity at admission. The largest and most recent studies reported hospital mortality rates of 25.9% [30] and 30.8% [34] for patients ≥75 years old are close to the 29.3% mortality rate we observed. In the subgroup of patients who develops acute lung injury and/or acute respiratory distress syndrome, reported mortality rate was 49.7%, and age of 70 years or more was identified as a strong predictor of in-hospital mortality [36].

For the elderly population, as for the overall ICU population, the most important risk factor for ICU mortality is the severity of illness. Describing the APACHE III score, Knaus et al. [24] showed that 73.1% of the mortality prediction power was due to physiologic conditions, with the “admitting” disease contributing to 13.6% and age to only 7.3%. Similarly, Chelluri et al. [22] found that the APACHE II score was the most significant predictor of ICU mortality of patients ≥85 years old. The multivariate analysis conducted by Rockwood et al. [7] indicated that age contributed to <5% of the variance of absolute risk of death in each diagnostic category. In our study, univariate and multivariate analyses did not identify a significant relationship between age and ICU mortality. Similarly, Kass et al. [21] did not establish a significant difference between ICU mortality rates of patients 85–89 years old and those ≥90 years old. Previous studies focusing on the outcome of the oldest-old patients [20, 21, 22, 23, 36] found no significant relationship between age and ICU mortality. These data confirm the predominant role of the severity of acute illness and physiologic disturbances.

The 1-year cumulative mortality rate observed for our population (51.7%) was comparable to those reported in other long-term outcome studies with mortality rates for elderly patients varying from 43.4% [12] to more than 60% [2, 5]. In the oldest-old population, the 1-year cumulative mortality rate (59.3%) was also similar to previously reported rates: 63.8% [21] and 58.8% [22]. Age was not directly associated with ICU mortality in our study. In contrast, we found significant differences between age groups for long-term survival of discharged ICU patients, and the multivariate analysis demonstrated that, for our entire elderly population, age and limitation of activity before admission in the ICU were independently associated with death after ICU discharge. In contrast with previous multivariate analyses of long-term outcome after ICU discharge [5, 6], severity of illness at ICU admission was not identified as a significant risk factor for dying after ICU discharge. These results suggest that diseases leading to ICU admission and their severity play predominant roles during the ICU stay and the early phase following ICU discharge, when age and functional status are major determinants for long-term outcome. Our analysis of subgroups of elderly patients have provided new insights into the long-term outcome. But, unfortunately, most previous studies [5, 6, 7, 11, 12, 21, 26, 27] pooled ICU and post-ICU deaths, rendering subsequent comparisons with our findings difficult. Otherwise, the importance of functional status of the elderly population for long-term survival has already been clearly established [6, 11, 20, 25].

Deaths occur mostly during the first 3 months after ICU discharge, with a markedly higher mortality rate for ICU survivors compared with persons of the same age in the general French population. Pertinently, this rate declined as early as 3 months after discharge (more than eight times higher) to 1-year after discharge (less than four times higher). This pattern of early post ICU mortality was previously reported for young and old patients [2, 11], particularly in patients ≥65 years old [5, 20] and ≥85 years old [21]. One possible explanation was the possibly reduced “physiologic reserves” following ICU discharge, with more frequent early-onset complications or deterioration of chronic underlying medical conditions. Such findings are potential arguments for the development of intermediate-care units for elderly patients transferred from the ICU with specialized geriatric nurses, associated with complete geriatric assessment, including respiratory and musculoskeletal rehabilitation. This approach to optimizing resource use for the oldest patients merits being explored prospectively.

Admission of aged patients to the ICU raises important medical, ethical, sociologic and economic questions. Our findings suggest that old age does not markedly influence ICU mortality, which is predominantly linked to disease severity. Age and baseline functional status were the two factors associated with long-term mortality. Future longitudinal studies of long-term outcome are needed to differentiate among ICU, short- and long-term post-ICU mortality of the elderly and to evaluate functional status, so as to establish meaningful guidelines for ICU admission of the elderly.