According to the World Health Organization, older people are a rapidly growing proportion of the world's entire population. In fact, in 2015, this population rose by 55 millions and the proportion of the older people reached 8.5% of the total population.1 This trend unavoidably leads to an increasing demand for health care resources, including those provided by intensive care units (ICU). Actually the median age of the entire ICU population is already above 65 years, as shown in large, recent epidemiological studies.2–4 Nonetheless, ICU beds are limited and will probably decrease even more.5 These limitations pose great challenges to the ICU triage decision-making process. Indeed, older patients are usually subjected to strict triage before admittance to the ICU and this selection is partly due to the fact that biological age is often not precisely estimated by chronological age. Therefore, intensivists are shifting their attention from the chronological age itself and the traditional co-morbidity measures to the more comprehensive concept of frailty, as a predictor of poor prognosis. Frailty is a relatively new issue for the intensive care field and is defined as a status of increased susceptibility to stressors associated with a decline in reserve and function of a wide range of physiological systems.6,7 The Clinical Frailty Scale (CFS), in particular, is a simple and visual scale that allows categorizing individuals in nine classes of increasing frailty. Such tool was first used on a large scale within intensive care in a Canadian study6 and has been associated with 6-month mortality.8,9 Moreover, studies evaluating CFS in respect to health-related outcomes suggested that the impact of frailty was more important than chronological age itself.10 This, again, emphasizes that age should not be used as an exclusive criteria to decide ICU admission.11,12

Although frailty has been largely investigated in geriatric epidemiological studies, the recent introduction of this concept in ICU settings has led to have only scarce available literature on this issue. Supporting the hypothesis that frailty may be a relevant predictor of poor prognosis of older patients admitted to an ICU, this study aimed to evaluate the association between frailty at ICU admission with 30-days, ICU and one-year mortality.

The baseline characteristics of patients included in the study are reported in Table 1. As shown, the median age of the sample was 84 (82–87) years and 153 (50.7%) patients were men. The most frequent diagnosis at ICU admission was respiratory failure, followed by cardiovascular insufficiency. In the sample as a whole, the most common chronic conditions were COPD, diabetes and cardiovascular diseases, and around a quarter of patients reported a physician-based diagnosis of cognitive impairment. Compared with patients alive at the 30-days follow-up, those who deceased were more likely to have had lower GCS at ICU admission (median GCS 13 vs. 15, p<0.001) and to suffer from ischemic heart diseases. More than half of patients were classified as frail at ICU admission (55.3%), and the prevalence was higher among those who died than those who were alive after a 30-days period (62.3% vs. 48.3%, p=0.01). As regards the medical interventions performed during the ICU stay (Table 2), most of patients required intubation, mechanical ventilation and administration of vasoactive drugs, especially those who had worse prognosis at the follow-up. The median length of ICU stay was 5 (3–10) days in the overall population and was significantly longer in 30-days survivors compared with those who deceased [6 (4–14) vs. 4 (2–9) days, p<0.001].

Differences in clinical outcomes between the 167 frail and 135 non-frail patients are reported in Table 3. As shown, frail individuals were more likely to be men, to have lower GCS, and a higher number of chronic diseases, especially congestive heart failure, peripheral vascular disease, hemiplegia and cognitive impairment. No significant differences between groups were found concerning the median length of hospital and ICU stay and the required medical interventions, while one-year mortality resulted to be higher in the frail (84.4%) than in the non-frail patients (65.2%, p<0.001).

At univariate logistic regression analysis (Table 4), the factors that demonstrated to be significantly associated with 30-days mortality were GCS at ICU admission (OR=0.83, 95%CI: 0.76–0.91, per each 1-unit increase in GCS), and the need for non-invasive and invasive mechanical respiratory support (OR=3.51, 95%CI: 0.10–12.40; OR=3.79, 95%CI: 1.15–12.50, respectively). At the multivariable logistic regression models, only GCS at ICU admission and connective tissue disease but not frailty, were significantly associated with 30-day mortality (Table 4). Similar results were observed when considering ICU mortality (Supplementary Table S1). On the contrary, as reported in Supplementary Table S2, frailty resulted to be significantly associated with one-year mortality both at univariable (Model 1, OR=2.70, 95%CI: 1.22–5.99) and multivariable regression models (Model 2, OR=2.69, 95%CI: 1.48–4.89; and Model 3, OR=2.15, 95%CI: 1.15–4.03).

In our study, considering older patients admitted to an ICU, we found that frailty at ICU admission was not associated with 30-days mortality. On the contrary, frailty was significantly associated with one-year mortality, and such relationship seemed to mostly depend on the presence of pre-existing chronic diseases and on patients’ clinical complexity. Of note, one-year mortality was extremely high in this population and reached 84.4% among those classified as frail. In this regard, although no longitudinal data on health-related quality of life were collected, it should be pointed out that a not irrelevant number of patients who survived at the 30-day follow-up underwent tracheostomy, and were thus likely to experience a severe reduction of self-sufficiency and quality of life over their last years.

In this study, we decided to focus on a population of inpatients aged 80 years or older. These patients constitute one of the categories with the fastest growing representation in ICU. However, indications for ICU admission, triage criteria and the level of care for the oldest old are still matter of intense debate, making this issue of great relevance in the daily clinical practice. As well known, triage process should differ when dealing with younger vs. older patients.10 However, current evidence suggests that age per se is not necessarily associated with negative health-related outcomes, especially in critically ill patients.10 Therefore, there is increasing awareness among intensivists that the assessment of older patients in ICU should be performed with a comprehensive evaluation, including the concept of frailty.

In this regard, recent studies in older frail patients found an increased risk of mortality and a reduced quality of life.11,14,15 Conversely, other works reported contrasting results and strengthened the need for further investigations on this issue.16,17 Our work corroborates a recent retrospective study that considered a large sample of ICU patients aged 75 years or older.16 In that study, ICU survivors had significantly lower hospital frailty risk score than non-survivors and, as expected, frailty was significantly associated with negative health-related outcomes. However, hospital frailty risk score did not independently predict ICU mortality risk after adjustment for APACHE-II scores or SAPS-II scores. Other authors showed, instead, an independent association between frailty and adverse outcomes in ICU or Emergency Department settings.11,14,15 In a large prospective multicentric study, frailty, not-elective admission to the ICU and high SOFA score emerged as the three most important factors associated with short-term mortality in older patients.11

Overall, in line with previous findings, our study suggest that triage criteria for ICU admission should not depend only on chronological age, but that a comprehensive evaluation of older patients inclusive of frailty assessment, may play a fundamental role in predicting survival in such special population. Considering 30-day mortality as an outcome that could more likely capture the clinical vulnerability of older adults, our work supports the potential benefit of ICU treatments on short-term survival as long as clinical conditions at admission are not too much compromised, especially neurological functions. Even in that case, however, one-year mortality remains extremely high, and whether an extension of lifespan could preserve individuals’ self-sufficiency and quality of life is still a matter of debate. For this population of patients, indeed, so far possible alternatives are not admitting them to the ICU at all, privileging a hospitalization in a regular/acute care ward; or admitting them to the ICU and conducting all efforts to ensure a rapid ICU discharge, as suggested by previous studies.17–18

Our study has some limitations that need to be mentioned. First of all, biases linked to the monocentric and retrospective nature of our study are well-known and may impact on the generalizability of our results. A further related issue concerns the inclusion in the study only of patients ≥80 years, which could have influenced both the prevalence of frailty in our sample and the strength of its association with mortality. Second, the matching between cases and controls was not performed.19 However, the adjustment of our analyses for different sets of potential confounders should have reduced this possible bias and support our results. Third, severity of illness at ICU admission, evaluated with validated tools such as the Simplified Acute Physiology Score (SAPS) II score or the Sequential Organ Failure Assessment (SOFA) score, were not available for most patients, especially for the most critical ones, who died a few hours after ICU admission. Finally, we had no information about triage decision process before ICU admission. On the other hand, the high number of included patients in a relatively short period ensures uniformity in admission policy and clinical practice. Moreover, having collected data on a number of clinical parameters and procedures may represent a further strength of our work.

In conclusion, the admission of geriatric patients to ICUs is increasing over time and their clinical evaluation for triage need to be comprehensive considering parameters more likely associated with clinical outcomes, such as survival and quality of life. Frailty assessment may play an important role in this regard, but it should not be considered a priori as an exclusion criterion for admission to ICU. Further studies investigating this issue in specific subgroups of frail older patients in the ICU setting are warranted.

Laura Pasin: study concept and design, acquisition of data, analysis and interpretation of data, preparation of manuscript.

Gianlorenzo Golino: study concept and design, acquisition of data, preparation of manuscript.

Sabrina Boraso: study concept and design, analysis and interpretation of data, preparation of manuscript.

Safaee Fakhr: study concept and design, acquisition of data, analysis and interpretation of data, preparation of manuscript.

Ivo Tiberio: study concept and design, analysis and interpretation of data, preparation of manuscript.

Caterina Trevisan: study concept and design, analysis and interpretation of data, preparation of manuscript.

None.

The authors declare that they have no competing interests.