This study was conducted to study the impact of frailty on survival after intensive care in patients of different ages. We show that while frailty is common in critically ill patients of all ages, a large proportion of adults aged 64 years and under were considered frail. The negative impact of frailty in patients is more important in the non-elderly, more specifically in the middle-aged, where it was independently associated with a twofold increased hazard of 180-day mortality. Our study thereby provides new insights into the prevalence of frailty and its impact on the risk of death in younger critically ill patients.
While the concept of frailty as a medical syndrome was developed in geriatric medicine [1, 16], there has been a rapidly increasing interest in other specialties . Most studies on frailty in the intensive care field have focused on older patients [3, 5, 7, 18] or on specific patient groups such as those with kidney injury, cardiac surgery or, more recently, COVID-19 [19,20,21]. Very few studies have focused on younger patients [7, 22] where it is plausible that frailty may even play a prognostic role independent of conventional severity of illness scores. In a recent, large study the implementation of routine frailty screening in intensive care patients of all ages in Australia and New Zealand was reported. This study also demonstrated the negative prognostic impact of frailty in younger patients . Our study adds to the limited body of knowledge regarding the impact of frailty in intensive care patients of all ages, further supporting a broad approach to the use of frailty assessment in intensive care.
The proportion of frail patients was larger in all age groups in our study than in previous similar studies. In one large international study on intensive care patients aged 80 years or older , 40% of the patients were categorized as frail, markedly less than the 60% in our study. One study on frailty in middle-aged intensive care patients demonstrated frailty in 28% of the patients , and in the large binational study previously mentioned, 23% of patients aged 50 or older were considered frail . Among younger patients the prevalence of frailty differed even more between studies, with 6% of patients younger than 50 considered frail in Australia and New Zealand, but 19% in our study . Notably over half of patients aged 64 or under were considered frail in our cohort. Despite the larger proportion of frail patients in our study, mortality at 180 days or more from ICU admission did not differ markedly from previous comparable studies . These differences and similarities may have been influenced by factors such as differences in healthcare organization in general, ICU thresholds and occupancy as well as the availability of ICU beds .
Unsurprisingly, the proportion of patients who had died within 180 days of being admitted to the ICU was lower in younger patients. Equally unsurprising, a larger proportion of frail patients were dead within 180 days in all age groups, all in line with previous studies [5, 7, 24]. An interesting finding, however, was that frailty remained an independent risk factor for mortality after adjustment for illness severity, comorbidities, and decisions to withhold or withdraw care, only in the group best described as middle-aged. In previous studies, the probability of survival for younger frail patients has appeared higher than for older patients , whereas our results coincide with the large study where frailty was an independent predictor of negative outcome, also in younger patients, after adjustment for severity of illness . Indeed, since frailty is strongly correlated with age and often seen as an age-associated decline in physiological reserves and function [25, 26], it has been suggested to add frailty to the clinical assessment in elderly patients . However, as shown here, frailty is relevant among younger critically ill patients [11, 22, 23]. It has indeed been suggested that the manifestation of frailty earlier in life conveys information about an accumulation of deficits that biological age and comorbidities alone does not . Our study thereby adds to previous indications that frailty can be seen as a reflection of biological age, rather than chronological, and provides further support to the use of frailty assessment in intensive care patients of all ages.
Comorbidity and frailty are linked, but not completely synonymous [27, 28]. Although the prevalence of both comorbidity and frailty is higher in the elderly, there are indications that there is a greater overlap of the two in younger patients [7, 9, 27]. The comorbidities considered in this study, however, were limited to those included in SAPS3, all of which mirror relatively severe disease . In fact, frailty seems to be more related to activities of daily life, which we have not studied. Nonetheless, our study indicates an independent value of frailty assessment in younger patients. Our results also support previous findings that frailty adds to conventional risk assessment scores [12, 23, 30, 31]. Thus, in outcome prediction, frailty mirrors a dimension not accounted for by comorbidities or by risk scores, especially in younger patients. This suggests that frailty assessment should be applied across the whole adult age span and, particularly in middle-aged ICU patients. Further research is warranted to determine how frailty may complement the clinician’s careful assessment of an individual’s functional and physiological reserve, multimorbidity and response to treatment, irrespective of age.
Limitations and strengths
Our study has some limitations that must be considered. First, all patients were recruited following ICU admission to one single ICU. Differences in hospital characteristics, admission criteria and case-mix may therefore limit the generalizability of the study. Second, the identification of premorbid frailty in patients presenting in an intensive care setting may be hampered by factors related to the acute illness and events leading up to it, which may be difficult to separate from true frailty. We cannot exclude that the assessment may have been affected by events leading up to the admission to, and the patient’s status in, the intensive care unit. Third, sociodemographic factors and more detailed data on premorbid comorbidities were not available, all of which could have been used for statistical adjustment purposes. Importantly, we were not able to study interactions between individual comorbidities, concomitant diagnoses, and frailty. Last, other outcome measures than death, such as the risk of complications after ICU stay, return to home or previous level of independency in frail and non-frail patients, were not available. The lack of patient-related outcome measures such as health related quality of life is a notable limitation. Strength of the study lies in the inclusion of all patients presenting in a general ICU, without limitation to patients of a certain age or diagnosis, and in the overall number of patients included.