Factors that influence intensive care admission decisions for older people: A systematic review

A university liaison librarian, who specialises in the health field, was consulted throughout the development of the search strategy; their expertise guided the selection of the databases and development of search terms including use of Medical Subject Headings (MeSH) and synonyms most appropriate for the review. Database specific search terms, MeSH headings, and synonyms used to describe ‘intensive/critical care’, ‘oldest old’, ‘decision-making’, and ‘admission’ were combined using Boolean operators (AND and OR) to search for literature (Table 1). Three databases were used to search for relevant literature: Medline Complete via EBSCO Host (Supplement 1), Cumulative Index for Nursing and Allied Health Literature (CINAHL) Complete, and Embase via the Embase platform. The database searches were limited to English language, and no limiters were applied to the year of publication.

Inclusion criteria were as follows: (i) adult patients aged 85 years and above with a critical illness that may warrant admission to a facility that provides specialised intensive care (ICU, critical care, coronary care, high dependency); articles that referred to other age groups were also eligible if findings were separated into subgroups where those aged 85 years and above were reported separately; (ii) ICU physicians as participants who make decisions surrounding ICU admission or nonadmission; (iii) details admission criteria and/or factors that contribute to the decision to admit/refuse admission; (iv) peer-reviewed articles reporting on primary research to ensure only the best available evidence was included. No limiters were applied to research designs because the review question could be addressed by designs involving quantitative, qualitative, or mixed methods. No exclusion criteria were applied.

All records retrieved from the database searches were downloaded into EndNote X9 and deduplicated. These records were uploaded into Covidence, a web-based software platform that streamlines the process of systematic review, allowing authors to independently screen citations, undertake text review and together, resolve reviewer conflicts. Two authors independently screened the titles and abstracts against the inclusion criteria, followed by full-text review. Conflicts were resolved by discussion between two reviewers, using the inclusion criteria to guide decision-making. If conflicts could not be resolved, the third researcher was consulted to make a decision.

The quality of each article was assessed using an 11-item evaluative framework suitable for critiquing qualitative and quantitative research. Quality assessment was performed independently by two reviewers, with discrepancies reviewed and resolved by a third reviewer. Articles were not excluded on the basis of the quality assessment; rather the assessment results were used to describe the quality of the research evidence in this area.

The data extracted from the included articles were informed in part by the Centre for Reviews and Dissemination guidelines for undertaking reviews in health care. Data were extracted into a Microsoft Excel spreadsheet by the lead researcher, and the extracted data were checked for accuracy by a second researcher. Extracted data included year published, authors, country in which the study was conducted, study and participant characteristics, admission criteria, and factors contributing to decisions to admit, or refuse admission of, patients aged 85 years and above (Supplement 2).

The six included studies were published between 1991 and 2019; five were quantitative studies,^{,  ,  ,  ,}  and one was a qualitative study. Of the quantitative studies, three were observational,^{,  ,}  one was an online survey, and one was a multicentre clinical trial. Two studies were conducted in France by the same lead researcher,^, one in Switzerland, one in the United Kingdom (UK), one in the USA, and one in Australia and New Zealand (NZ). A convenience sampling approach was used in five studies.^{,,  ,  ,}  The remaining study used systematic sampling, involving consecutive data collection every 12th day for 1 calendar year, for each patient who met the inclusion criteria. Three studies focused on physicians as participants (N = 367),^,, and the other three on patients (N = 8461).^,, Of the studies of patients, two sought to determine whether admission to the ICU differed between older and younger participants; one study focussed on participants aged 40 years and above, and the other study included all age groups. The third study aimed to describe triage to the ICU for patients aged 80 years and above, whilst also investigating one-year survival and quality of life after ICU discharge. Of the studies that focused on physicians, one explored the triage process and barriers and facilitators to ICU admission, another assessed the variability in ICU physicians’ admission decisions, while the third study compared ICU triage practices between Australian and NZ physicians.

Final quality assessment scores ranged from 5 to 9, with a mean score of 8, out of a possible score of 11 (Table 2). Overall, identification and description of the methodology was considered poor with none of the publications scoring in this domain, owing to either the methodology being described in a separate publication or a lack of detail in this section. All publications received a score for author credibility, providing an abstract, providing a rationale for the study, and clearly stating the study aim.

Factors that influenced the decision to admit the oldest old to ICU were examined in three of the six studies; two studies examined factors associated with admission,^, whereas one study examined ICU admission decisions by proxy, by assessing ICU physician admission decisions surrounding the use of life-sustaining treatments (LSTs) and factors associated with this decision. Four of the studies examined admission rates and admission scenarios.^{,  ,  ,}

Factors associated with the decision to refuse ICU admission were examined in all six included studies.^{,  ,  ,  ,  ,}  Two studies included data about admission refusal rates,^, and a third examined refusal rates by proxy, using rates of refusal for LSTs. The three studies that focused on physicians explored the triage process based on common situations arising in the healthcare setting and used either fictional or actual scenarios.^,, Of the studies that focused on patients, two examined differences in access to intensive care when comparing older and younger participants,^, whilst another described triage to the ICU for patients aged 80 years and above.

Factors associated with the decision to admit the oldest old to the ICU were examined in three of the included studies, two of which were published by the same lead author.^,, In the latter two studies, greater premorbid self-sufficiency was found to be associated with the decision to admit.^, These findings correlate with the SCCM guidelines for ICU admission, discharge, and triage, which specify the decision to admit a patient older than 80 years to the ICU should be based on, among other factors, the patient's prehospital functional status and not age. In the only qualitative study to meet the inclusion criteria, when the patient's goals of care aligned with the ICU physician's assessment of appropriate treatment, the likelihood of ICU admission was increased. Patient preferences were also found to strongly influence ICU physicians’ decisions surrounding the use of LSTs. These findings align with those of a recently published systematic review of decision-making in the ICU, which found understanding patient wishes and a process of shared decision-making involving the patient, when possible, were of paramount importance in reaching an ICU admission decision. A multisite study, which did not meet the inclusion criteria for this review, examined whether patients older than 80 years were asked their opinion regarding their care before admission to the ICU. Interestingly, the researchers found that only 13% (n = 270) of patients were asked their opinion during triage. A dementia diagnosis reduced the likelihood of patients being asked their opinion, whereas patients with good baseline functional status or those with a relative present were asked for their opinion more frequently. These findings highlight the importance of advance care planning, setting agreed goals of care, and open discussions regarding both the benefits and consequences of ICU admission with the critically unwell older patient, their families, and the ICU physician. Although care planning conversations with critically unwell patients and their families are a high priority, ultimately the final decision to admit a patient to the ICU lies with the ICU physician, which is the position taken by the World Federation of Societies of Intensive Care Medicine triage decision recommendations. However, these decisions may be easier if the ICU physician has some insight into the preferences of the patient who potentially requires ICU care.

Other factors associated with the decision to admit were age less than 85 years, absence of cancer, and absence of previous ICU admissions. The SCCM guidelines state that chronological age should not be used as a deciding factor when admitting a patient to the ICU. Interestingly, this recommendation was not reflected in the findings of a French simulation study included in this review, in which decisions for LSTs were used as a proxy for the decision to admit to ICU, and age less than 85 years was independently associated with the decision to implement LSTs. The differences, however, may be associated with differences in age range, with the French study focused on those aged 80 years and above and the SCCM guidelines applying to all age groups. The absence of previous ICU admissions was also identified as a factor associated with the likelihood of ICU admission. A diagnosis of metastatic cancer may also influence admission decisions, with a retrospective cohort study identifying that metastatic cancer in patients older than 80 years was associated with 1-year mortality.

Factors associated with the decision to refuse admission to the ICU were measured in all six of the included studies.^{,  ,  ,  ,  ,}  Limited ICU bed availability was associated with the decision to refuse admission in three of the six included studies.^,, Limited bed availability was also associated with the decision to refuse admission for all three forms of LSTs; NIV, IMV, and RRT following IMV. Yet, when a bed became available in the simulation, some ICU physicians who had first refused admission for NIV and IMV changed their decision to admit the patient to the ICU. Decisions made in simulated scenarios, however, may differ to those made in real life, affecting the transferability of the results to real practice. Another study by the same lead author produced similar findings, with no bed availability statistically associated with the likelihood of ICU admission refusal. Limited bed availability also led to delayed admission decision-making, in the hope that admission could be avoided. A systematic review of factors important to ICU clinicians’ triage decisions identified bed availability to be a factor that influenced the admission decision-making process. However, patients are at an increased risk of longer ICU admission and increased mortality when there is a delay in admission from a ward.^{,  ,}  Consideration for bed availability in the ICU admission decision-making process may reflect a systemic organisational issue that needs addressing to ensure all patients have equitable access to intensive care in their time of need. The SCCM guidelines suggest that emergency physicians should be prepared to deliver critical care in the emergency department to work around this barrier to ICU admission.

The SCCM guidelines state that the decision to admit patients older than 80 years to the ICU should not be based on chronological age. Yet, in three studies, older age was associated with the decision to refuse admission to the ICU.^{,  ,}  In one of these studies, the likelihood of admission refusal following an AMI increased with older age, with patients older than 85 years having the highest CCU refusal rate compared with the younger age groups. The data for this study, however, were collected in 1979–1981, which precedes the widespread use of thrombolytic therapy for patients with AMIs, so results may differ if the study were repeated in the year 2021. Age greater than 85 years was associated with ICU admission refusal in two studies.^, In contrast, age was not a limiting factor for those who needed intensive care in a study that included patients irrespective of age and all intensive care settings including CCUs and high-dependency units.

In studies of physicians, perceptions about the influence of age on ICU admission decision-making varied, with authors of some studies reporting age had no impact,^{,  ,}  whereas others found it to be important.^{,,  ,}  Physiological age may be more important than chronological age in the decision-making process. The finding that older age is a factor associated with ICU refusal contradicts the recommendations made in the SCCM guidelines. Interestingly though, the SCCM guidelines also state that patients should receive ICU care if their prognosis for recovery and quality of life is acceptable but say that factors that can influence survival, including age, should also be taken into account.

Other factors associated with the decision to refuse admission were reported in three of the included studies.^,, These included a presenting feature of dyspnoea, with no chest pain, in relation to AMIs, patients being deemed either too well or too ill to benefit from ICU care, being a medical patient, examination by an ICU physician, and no assistance required for toileting. Being deemed too ill to benefit from ICU was also found to be associated with ICU admission refusal in a study not included in this review. Although this study did not separate data according to age groups, these patients were typically older and had more comorbidities and worse baseline functional and mental status than those who were admitted to the ICU. According to the findings of a qualitative study, as ICU physicians gained more experience, they did not admit patients to the ICU as readily, and difficult triage scenarios typically involved patients with advanced diseases or older patients with multiple comorbidities. Physician experience was also an influential factor in ICU admission decision-making, with experienced physicians more likely to acknowledge inappropriate ICU admissions and refuse admission. In an Australian and NZ study, views sought when deciding to refuse ICU admission for a patient were most often those of the referring specialist, followed by ICU colleagues, the patient, the patient's family, and the patient's GP.

Rates of admission or refusal of admission were observed in four of the six studies, with rates varying greatly amongst the studies.^{,,  ,}  One study included patients older than 40 years and found that patients older than 85 years had the lowest rates of CCU admission (66%, n = 27) amongst the age groups examined. In another study, patients older than 85 years also had lower rates of ICU admission, based on the use of LSTs. Admission rates proposed by Australian and NZ ICU trainees and physicians involved in an online survey were much higher (90–95%) based on a 95-year-old participant; however, this was based on a fictional scenario and not real practice. Refusal rates for ICU admission in a French simulation study were 73% (n = 132) for patients aged 80 years and above, which was the highest amongst the included studies; however, no comparative age group was examined. ICU refusal rates based on the refusal of LSTs were higher for those older than 85 years for all three forms of LSTs than for those younger than 85 years. The average refusal rate in another study was similar at 31%; however, this centred around all age groups and was based on the admissions the physicians had refused in the week prior to the survey.

The variation in study designs and sample size makes the comparison of ICU admission and refusal rates between the studies difficult and may explain variations in findings.^{,,  ,}  Substantial variability in ICU admission rates was observed in another multisite observational study which focused on patients older than 80 years, with admission rates ranging from 6 to 39%, amongst the 13 hospitals included in the analysis. This variance could not be explained by geographical differences as all hospitals were located within Paris, or by patient or hospital characteristics. A systematic review of decision-making in the ICU found physician factors impacted decision-making, including cultural, gender, age, and religious differences, suggesting variability in ICU admission and refusal rates may exist primarily owing to ICU physicians’ personal beliefs, experience, and other individual factors.

Of the six studies published between 1991 and 2019, three were published in the last decade,^,, which may be indicative of increasing research in this field, owing to a greater emphasis on the needs of an ageing population and the pressure on ICUs in recent years. The ageing population and the potential for increased demand for ICU admission and the wider healthcare system care is particularly relevant during the COVID-19 pandemic. COVID-19 has disproportionately affected older people, who already have an increased risk of hospitalisation, morbidity, and mortality in comparison with younger people. The American Geriatric Society published a COVID-19 position statement regarding resource allocation strategies and age-related considerations in COVID-19, to guide healthcare systems in the development of emergency rationing strategies. Their position statement states that age should not be used as a means for exclusion from interventions or in resource allocation and “life years saved” or “long-term predicted life expectancy” should not be used to guide resource allocation as these criteria disadvantage older people.