Background
The proportions of older patients hospitalised in intensive care units (ICU) are increasing due to ageing populations [
1‐
3]. Older patients present specific conditions, such as frailty, polypharmacy and multimorbidity, making it more difficult for them to overcome acute medical stress and to survive and recover their previous functional autonomy [
4,
5]. For all these reasons, mortality is increased among older patients hospitalized in an ICU, comparatively with younger populations, with a 6-month mortality ranging from 21 to 58% [
1,
6‐
8]. Most studies evaluating short- and long-term mortality are heterogeneous in design and results, and factors associated with mortality vary widely across studies [
9]. Outcomes of older patients hospitalised in ICUs are influenced by premorbid conditions, notably frailty and comorbidities, as well as in-ICU events, such as duration of mechanical ventilation and decisions to withdraw life-sustaining therapies [
4,
5,
9‐
13]. In addition, among survivors, some may suffer from disabilities, cognitive impairment and decreased quality of life [
14,
15].
Currently, the benefits on mortality and qualitative outcomes (functional status and quality of life) of an ICU hospitalisation for a critically ill older patient are unclear. The question of which structure is best adapted to the needs of these patients has yet to be answered. We previously demonstrated in the randomised clinical trial ICE-CUB2 that promoting systematic ICU admission of critically ill older patients had no effect on mortality in comparison with standard practice, but analysis of the impact of hospitalisation in intermediate care units was not evaluated [
16]. Intermediate care units (IMCUs) are used to manage patients needing more care than a general ward, but requiring lighter monitoring than ICUs, which, theoretically, have more severely ill patients [
17]. Conflicting results are found in studies evaluating the effect of IMCU hospitalization on mortality [
17‐
21], and very few studies were performed specifically on older populations [
22,
23].
We, therefore, aim to evaluate whether the structure in which the older critically ill patient is hospitalised influences longer term outcomes.
Discussion
Although patient characteristics and severity differed according to ward destination, patients hospitalised in an AMW had a significantly worse 6-month survival than those admitted to an IMCU, after adjusting for baseline characteristics, severity and diagnosis, and propensity score matching. No significant differences in survival were found between patients admitted to an ICU and an IMCU. Hospital destination was not associated with the composite outcome of 6-month mortality or loss of functional ability. Mental and physical SF-12 was lowest in patients hospitalised in an AMW, although the difference was not clinically relevant.
Six-month mortality after IMCU admission was 31%, which is similar to the findings of other studies [
22,
23,
28,
29]. Survival was significantly worse when hospitalised in an AMW in comparison with IMCU admission, with no difference of survival between patients hospitalised in ICUs and IMCUs. The three groups differ in comorbidities, functional ability, diagnosis and severity. These differences of comorbidities and severity could be explained by the fact that patients eligible for IMCUs have a unique life-threatening organ insufficiency [
17], whereas patients with multiple organ failure were admitted in an ICU, or an AMW when ICU admission was refused because of life-sustaining treatment limitation. However, when adjusting on age, severity, comorbidities and diagnosis, we found that admission in an IMCU was still associated with a better survival than in an AMW, with no significant difference with ICU admission. These results were confirmed after propensity score matching, where admission to an IMCU was associated with better survival in comparison with AMW and ICU admission. In the context of aging of population, these results suggest that there could be a benefit of hospitalising older patients in IMCUs, notably in geriatric IMCUs [
30].
The SARS-CoV-2 pandemic has shown the importance of expanding critical care capacity, notably by creating IMCUs [
31‐
34]. With the aging of population, it is expected that by 2070, 20% of the French population will be aged 75 years and over [
30]. It is thus a possibility that tension in critical care capacity could occur again, even without a pandemic. An interactive session of the 2019 European Society of Intensive Care Medicine congress suggested that progress on pre-ICU triage and creation of dedicated intermediate care units for critically ill very old patients could be solutions to explore [
35]. These IMCUs can offer monitoring for less severe critically ill patients, or facilitate ICU discharge as a “step-down unit”, thereby freeing up ICU beds [
36]. Our study found better survival when patients with critical conditions were hospitalised in an IMCU than in an AMW, with no difference when admitted in an ICU. Thus, our results could suggest that there is no loss of opportunity for the older patient between being admitted in an ICU or an IMCU.
SAPS-3 was significantly associated with poorer outcomes independently of hospital destination. No score predictive of mortality among critically ill patients was created specifically for older patients, mean age at inclusion in the SAPS-3 study being only 63 years [
26]. SAPS-3 is a score predicting hospital mortality, but contains no variables specific to older patients, such as functional status. We found that baseline ADL was associated with poorer outcomes only in patients hospitalised in ICUs and IMCUs, but not in AMWs. This could suggest that one should admit in an ICU or IMCU mainly patients with preserved functional ability. Other factors specific to older patients are known to influence ICU outcomes, such as the Clinical Frailty Scale, its components being items reflecting functional ability [
10]. Age and pre-existing comorbidities weigh heavily in mortality prediction models [
37], thus leading to lower discrimination of severity of these models in older patients [
38]. We observed a difference between observed mortality and probability of mortality using SAPS-3 which suggests that this score is not suited for older patients.
We found that when adding loss of functional status with mortality as an outcome, admission in an IMCU was no longer associated with this outcome. After propensity score matching, admission to an AMW or an ICU were associated with a higher risk of death or loss of ADL, in comparison with IMCU admission. Although it was not the study’s primary outcome, this finding could encourage to perform specific studies. In all three destinations, age, baseline ADL and severity were similarly associated with this outcome, which could suggest that other unmeasured factors could be involved, such as post-critical care management. Loss in functional ability is a reversible factor that can be prevented during and after hospital stay [
39,
40]. A prospective observational study documented that older patients had worse health-related quality of life over time, while younger patients showed spontaneous improvement [
14]. This finding suggests that studies evaluating the benefit of early rehabilitation specifically tailored for older population could be of interest. Post-ICU management could, therefore, possibly influence patient outcomes in the older critically ill population.
Large studies evaluating quality of life after a critical condition are scarce. In a study of 23 ICU survivors aged ≥ 80 years, quality of life (using WHOQOL-BREF and WHOQOL-OLD) was similar or better than that of a general population matched on age and gender [
15]. On the other hand, among 54 patients aged ≥ 75 years prospectively followed-up after ICU admission, quality of life evaluated using SF-36 was equal or better than that of the general population, except for physical functioning, vitality and mental health SF-36 components [
41]. To summarise, large studies evaluating long-term autonomy and quality of life after ICU admission are scarce, and use various scales, thus limiting reproducibility and comparability. Our study is the first to have prospectively used, on a large cohort, both validated and reproducible scales. We found that physical and mental SF-12 was lowest among patients hospitalised in an AMW, although a variation of one point on a 56-point scale is not clinically relevant. Physical and mental SF-12 in our cohort was similar to those of other cohorts of frail, older patients living in the community or in care homes [
42‐
45]. Therefore, after a critical condition, quality of life seems comparable among the 3 hospital destinations, and similar to that of the general older population [
42‐
45].
The present study has several strengths. This study included patients in real-life settings, with high age (median 85 years), multiple comorbidities and high severity at admission. The cohort originates from a prospective multicentre study, with many patients and a low loss of follow-up, thereby limiting bias [
25]. Our study used a pragmatic clinical outcome which is 6-month survival and loss of 1-point ADL. Robust, validated and reproducible scales of functional ability and quality of life were used. Our results were confirmed using three different statistical models.
Several limitations should be taken into account. This study was a post-hoc analysis of a prospective study that was not designed for this analysis. Specific studies should be designed to confirm our findings. We did not assess frailty nor decisions of withdrawal of life sustaining therapy, which are two important prognostic factors in the older population [
10,
13]. We did not have information on the type of IMCU the patient was hospitalised (general or specialised high dependency units), thus adding heterogeneity in the IMCU group. It cannot be excluded that patients admitted in an ICU or an AMW have more cumulative number of organ failures at admission, than those admitted in an IMCU. Although we have adjusted on SAPS-3 and admission diagnosis, and matched patients according to baseline characteristics using a propensity score, no adjustment is perfect and one cannot exclude a potential indication bias. We also did not assess information on patient movement between each ward, and patients initially hospitalised in an AMW or IMCU could have then been transferred in an ICU, which could be a source of bias. It is possible that data on post-critical care management could influence patient outcome as well. Indeed, our aim was to find predictive factors of poor outcomes at admission, to help the clinician when making triage decisions on which ward to admit a critically ill older patient. A visualization of the post-ICU care trajectory of older patients is a key variable to take into account, as a specific post-ICU geriatric approach, including prevention of geriatric-specific complications and closer follow-up, could ameliorate both long-term survival and functional status.
Acknowledgements
Group Information: ICE-CUB 2 Study Network: Bertrand Guidet, Dominique Pateron, MD, Erwan Debuc, Youri Yordanov, (top 10 recruiter), Ariane Boumendil, Caroline Thomas, Didier Dreyfuss, Jean-Damien Ricard, Patrick Brun, Christophe Leroy, Yves Cohen, Frédéric Adnet, Maguy Woimant, MD (top 10 recruiter), Jean-Paul Mira, Benoît Doumenc, MD (top 10 recruiter), Khalil Tku, Thomas Similowski, Bruno Riou, Pierre Hausfater, Samuel Delerme, MD (top 10 recruiter), Jean-Pierre Quenot, Ph. D. (top 10 recruiter), Didier Honnart, Jean-François Timsit, Pierrick Guérin, Françoise Carpentier, Maxime Maignan, MD (top 10 recruiter), Foued Makhlou, Jean-François Poussel, Yohann Picard, François Braun, Pauline Trognon, MD (top 10 recruiter), François Fourrier, Patrick Goldstein, Marie Girot, Pierre Gosselin, Francesco Santoli, Pierre Charestan, MD (top 10 recruiter), Claire Poly, Hervé Mentec, Catherine Le Gall, Karima Sahraoui, Christophe Baillard, Nicolas Javaud, MD (top 10 recruiter), Benoît Misset, Maité Garrouste-Orgeas Olivier Ganansia, MD (top 10 recruiter), François-Xavier Rooryck, Jean Luc Aim, Abudlrazak El Rifai, Jean Reignier, Laurent Martin-Lefevre, Philippe Fradin, Claire Mauriat, Emelyne Cwicklinski, Michel Slama, Hervé Dupont, Christine Ammirati, Justine Gallou, Muriel Fartoukh, Michel Djibre, Patrik Ray, Ph.D. (top 10 recruiter), Edwin Rouff, Bertrand Souweine, Ali Ait Hssain, Jeannot Schmidt, Daniel Pic, Farès Moustafa, Alain Mercat, Nicolas Lerolle, Pierre-Marie Roy, Frédéric Baud, Patrick Plaisance, Sophie Montagnon, Bertrand Galichon, MD (top 10 recruiter), Michel Wolff, Bruno Mourvillier, Enrique Casalino, Christophe Choquet, Julien Bernard, Gaëlle Juillien, Jean-Yves Fagon, Emmanuel Guerot, Philippe Juvin, Anabela Patzak, MD (top 10 recruiter), Bruno Verdière, Vincent Ioos, Marie-Clément Kouka, Audrey Berthoumieu, Christian Richard, Raphael Maurice, Sophie Sarnel, Stéphane Diez, Antoine Vieillard Baron, Sébastien Beaune, Julie Grenet, Unité de Recherche Clinique Paris Île-de-France Ouest: Sylvie Azerad, PharmD; Department of Anesthesiolgy and Critical Care-Laval University: Guillaume Leblanc, Sorbonne Universités-Pierre et Marie Curie: Tabassome Simon
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.