Introduction

Measuring outcome after critical illness is important for many reasons, and historically, mortality has been used particularly as a measure of the effectiveness of intensive care treatment [1]. However, the development of new medical procedures for critical care patients has led to increased survival despite more complex illnesses and extensive injuries. Therefore, the need for outcome measures other than survival has been claimed to be increasingly important [2]. Furthermore, the long-term patients’ perspective has in parallel been considered increasingly relevant in evaluating long-term post-ICU treatment outcomes. As a consequence, health-related quality of life (HRQoL) has gained its place as one of the most important outcome variables for assessing the effectiveness of intensive care, together with mortality and survival [3].

It is well known that HRQoL is reduced after intensive care compared with that in the general population [4]. However, a recurring issue is that ICU patients are compared with a healthy population, most often not properly adjusted for effects of comorbidities. In other HRQoL assessments comorbidities are a well-known, and probably the most important, factor for the perceived quality of life described by the individual [5,6,7]. In particular, up to 73% of patients admitted to the ICU have developed one or more chronic comorbidity in the years before admission to the ICU [8, 9]. In addition, a considerable proportion, approximately 15%, are diagnosed with a new “chronic” or long-term health condition at the event which has led to the critical care period [8]. Adding this together, a very large proportion, approximately 80–90%, of ICU patients have comorbidities [8, 9]. One confounding consideration for this issue is that, not seldomly, adjustment for age is claimed to be a substitute for comorbidities, because there is a well-known collinearity between the two [4]. Another comorbidity adjustment that needs to be mentioned is the Charlson comorbidity index. However, the Charlson comorbidity index is not an HRQoL adjustment score but a mortality prediction score with a low HRQoL sensitivity for most diseases other than the most lethal [10].

With a high proportion of comorbidities in the ICU cohorts, it may be argued that the decrease in the level of HRQoL seen post-ICU may already be present in the period before intensive care. It is then concomitantly uncertain when, and if, the pre-ICU HRQoL level is reached. Most of the published data so far suggest that HRQoL levels seem to reach a plateau at approximately one year after the ICU period [4]. This temporal relationship has just recently been extensively reviewed [11]. In most studies, the level of HRQoL in ICU patients is lower than in the control groups, which might be related to control groups that have not been fully adjusted for comorbidities.

Comorbidities, by definition, hold paramount significance in any context when assessing HRQoL. Consequently, in evaluating HRQoL among these individuals, it is imperative to account for the impact of pre-existing comorbidities. The objective of this targeted review is therefore to meticulously evaluate the extent to which current published research on HRQoL post-critical care has been adequately adjusted for the elevated comorbidity levels in former ICU patients.

Methods

This is a purpose specific review targeting the extent to which current published research on HRQoL post-critical care has adequately been adjusted for the elevated comorbidity levels in former ICU patients [12]. The choice of the specific review type, “purpose specific review” was made to target specifically the issue of comorbidity effects on post ICU HRQoL.

Data sources and search strategy

This purpose-specific systematic review was written in accordance with the Equator network PRISMA-statement and conducted according to the PICO (Population, Intervention, Controls Outcome) strategy to ensure a systematic search of the available literature. The search method used to identify all relevant articles was developed and discussed by two authors (LO and FS), and the final string was approved by all authors. An electronic search strategy was developed in collaboration with a librarian with extensive experience in systematic reviews. The process was done in three steps. Firstly, the literature searches were conducted in PubMed, Cinahl, Scopus, and the Cochrane library independently by two reviewers (LO and SW), who eliminated clearly irrelevant articles based on title, abstract, and full-text levels (Fig. 1). During this screening process a third author (FS) was involved and any disagreements were settled by a consensus process. The following terms were used, with all terms mapped to the appropriate MeSH term/equivalent function using the medical subject headings: for Population (P); (“intensive care unit” OR “critical care” OR “critical illness”) AND (“intensive care unit” OR “ICU”): for Intervention (I); (“follow-up studies” OR “aftercare”); Controls (C) is not applicable for this review: for Outcome (O); (“health-related quality of life” OR “HRQoL” OR “QoL”) AND (“physical ability”) OR (“cognitive dysfunction”) OR “ADL” OR mortality. Full-text articles published in peer-reviewed journals were considered for inclusion. Reference lists of retrieved articles and relevant publications of expert authors were screened to identify additional papers that met the inclusion criteria.

Fig. 1
figure 1

PRISMA study flow diagram

Full size image

Applied inclusion criteria were; articles evaluating HRQoL after intensive care, published in the period of January 1, 2010, to December 31, 2021, adult patients (≥ 18 years), and treated at a general intensive care unit. The search period from 2010 was chosen due to catch the newest research in the field. Articles describing paediatric follow up and follow up after specialised intensive care treatment such as burn unit, cardiothoracic- or neuro-intensive care were excluded. Articles were also excluded if they described a subgroup of the general ICU population regarding diagnoses, treatment, sex, or specific age groups. In addition, articles were excluded if they described just one dimension for HRQoL, such as physical function only. Language restriction was applied in the end stage of search to reduce language selection bias. The languages selected was English or Swedish (when appropriate). The PubMed search strategy is available in Supplemental file 1.

Study selection

In the second stage, all full articles were evaluated using a predefined data extraction form developed for this study in a Microsoft Access-database, with the following predefined criteria; (1) assessment of HRQoL, (2) exclusion criteria including, (3) demographic characteristics and comparisons of responders vs. non-responders, and (4) HRQoL comparisons adjusted for age and sex.

In the third stage, all the studies identified in the previous stage were selected for review if they met the final predefined inclusion criteria; HRQoL comparisons adjusted for age or comorbidity. The final selection of articles was made by all authors, based on the reporting of all necessary data and in accordance with the predefined inclusion and exclusion criteria. Discrepancies were resolved by consensus.

Results

Search results and characteristics of the included studies

A total of 1588 references were identified by the database searches, and in addition one reference was identified through reference lists. After duplicated articles (same findings by different search engines) were removed (n = 1041) 548 abstracts were screened. Of these, 496 articles were excluded after reading the title and the abstract because they did not meet the predefined study criteria. The remaining 52 articles were read in full and of these 41 articles were excluded because they did not meet the final inclusion criteria; HRQoL comparisons adjusted for age or comorbidity. Finally, a total of 11 articles were included in the review (Fig. 1).

Most studies were conducted in Europe [13,14,15,16,17,18,19], two in Australia [20, 21], and one each in Canada [22] and USA [23]. All studies were observational prospective cohorts, except one that was a case–control study [21], one a cross-sectional comparison study [15], and one that was a retrospective cohort study (Table 1).

Table 1 Characteristics of included studies
Full size table

Instrument and methods for measure of HRQoL

HRQoL was assessed using two general tools (SF-36 or SF-12 and EQ-5D) and one palliative care-specific tool (EORTC QLQ-C15-PAL). The most frequently used instrument was the Medical Outcomes Study 36-Item Short-Form health Survey (SF-36) [13, 15, 17, 19,20,21, 23] or its short-version 12-Item Short-Form Health Survey (SF-12) [22] (62%), followed by the EuroQol-5D (EQ-5D) [14, 16, 18, 22] (31%), whereas the European Organization for Research and Treatment of Cancer (EORTC) QLQ-C15-PAL were used in one study [21] (7%). Two studies used a combination of HRQoL instruments: EORTC QLQ-C15-PAL with the SF-36 [21], and the EQ-5D with the SF-12 [22] (Table 1).

Method of assessment for HRQoL varied between the 11 included studies. Almost half of them (46%) used mailed questionnaires [15, 16, 18, 19, 23], whereas two used telephone interview (18%) [17, 22], two used face-to-face interview (18%) [13, 14], and two used a combination of face-to-face interview and/or telephone interview (18%) [20, 21] (Table 1).

An overview of HRQoL assessment after ICU discharge is shown in Table 2. Inclusion periods were for most of the studies between one and four years [13, 15, 18, 20,21,22,23] and ≤ 1 year in two of the studies [16, 19], whereas the inclusion period was not presented in two of the studies [14, 17].

Table 2 Assessment of health-related quality of life after ICU
Full size table

A total of 18,566 critically ill patients were included in the 11 studies, and the response rate ranged between 16 and 94%.

Follow-up period

Follow-up periods for the assessment of HRQoL varied between the included studies from short follow-up (3 months) to long-term follow-up (5 years). All studies had a strict time-point for follow up except for two that had wide ranges of 18–24 months [17] and 4–5 years [20] after ICU discharge, respectively. The strict time points varied between 6 months [15, 16], 12 months [14, 18], and 18 months [13, 19]. Two of the studies assessed HRQoL over time at several occasions: 6 and 12 months after hospital discharge [22], and 3, 6, 12, 18, 24, and 30 months after enrolment [23] (Table 2).

Study criteria (Table 3 )

Table 3 Study specifics
Full size table

All the included studies met the third, and final quality criteria stage for inclusion; the HRQoL comparisons adjusted for age or comorbidity. All but one study [20] did HRQoL comparisons adjusted for comorbidity, and more than half of them (54%) did HRQoL comparisons adjusted for both comorbidity, age, and sex [13, 15, 18, 19, 22, 23].

However, only two (18%) of them met all the predefined study criteria at the second stage; assessment of HRQoL, exclusion criteria, including demographic characteristics and comparisons of responders vs. non-responders, and HRQoL comparisons adjusted for age and sex [15, 19].

Six of the studies showed detailed HRQoL scores [15, 18,19,20, 22, 23], but for one of them [22] only the component scores for SF-12 were given. Exclusion criteria were given in all but two studies [21, 23], and a description of the non-responder group and comparison with patients who responded to the HRQoL survey were given in four of the studies [15, 16, 19, 23] (Table 3).

Health-related quality of life adjusted for comorbidity

A common finding amongst the studies that fulfilled all the inclusion and none of the exclusion criteria (n = 11), was that comorbidities were found to be the most important factor affecting HRQoL after critical care. Amongst the eleven studies, seven used comorbidity scoring systems (Charlson (n = 7) and APACHE (n = 4)). A significant observation was that three studies (n = 3) included a comorbidity-adjusted control group (Table 2) [15, 21, 23]. For these three studies, there were difficulties in documenting any significant effect of the critical care period itself on the registered HRQoL after the critical care period. A summary of the major findings for long-term HRQoL per article is shown in Table 4.

Table 4 Major findings and factors influencing long-term health-related quality of life
Full size table

Discussion

The ambition of this purpose-specific review has been to examine how comorbid conditions affect long-term (HRQoL) outcome after critical illness. From a definition-specific perspective, it is obvious that health status alters the HRQoL. However, in the present literature little emphasis is put on comorbidities [24], and consequently it may be difficult to assess the effect of ICU care specifically on HRQoL. The main finding of this review is that when comorbidity is added properly in the modelling of post-ICU HRQoL, it significantly alters the interpretation of HRQoL after intensive care. In the 11 studies included, 7 were adjusted for comorbidities by Charlson score and 4 were by use of APACHE. In all eleven studies looking specifically at comorbidities the comorbidity factor was the most important for the long term HRQoL outcome. In contrast to this effect and finding, the three studies that also included a comorbidity adjusted control group, found an even stronger effect of the comorbidities as the difference between the patient group and the control group completely disappeared and thus stronger and differently supported the comorbidity effect. From a methodological perspective it then seems important to include comorbidity adjusted control groups. A significant shortcoming of the research field shown in this review is the heterogeneity of the studies included. Despite these, an overarching finding is the significant effect of comorbidity acquired before the intensive care period on HRQoL outcome after critical care. As only a fraction of studies of HRQoL after critical care include a comorbidity estimate (See Fig. 1) the present study underlines the need to always include such a measure in future studies of HRQoL after intensive care. Furthermore, a more stringent approach to the protocols used may be asked for as the studies finally selected in this review had large variations in the HRQoL instrument used, the extent of the follow-up time, the comorbidity assessment technique, and the age and sex adjustments (Fig. 2).

Fig. 2
figure 2

Summary of Health-Related Quality of Life (HRQoL) Effects Post-ICU Care. This review, despite the limited number of underlying studies and their variations, suggests minimal impacts on Health-Related Quality of Life (HRQoL) following ICU care, beyond those typically attributable to comorbidities alone. However, this generalization may occasionally mask significant deterioration of an existing comorbidity or the onset of a newly diagnosed comorbidity. The common early, transient decline in HRQoL typically observed post-ICU care is included. The HRQoL levels depicted in the figure are based on RAND-36/SF-36 protocols. These levels are representative of patients across three categories: those previously healthy, those with 1–2 comorbidities, and those with more than 2 comorbidities. HRQoL: Health-Related Quality of Life; ICU: Intensive Care Unit

Full size image

Methodology

Comorbidity and HRQoL

Most of the comorbidity assessments in the present review used the APACHE score comorbidity alternatives. These have been developed not to adjust for comorbidity effect on HRQoL but to adjust for mortality [25]. A similar problem exists for the mostly commonly used comorbidity instrument, the Charlson index [10]. Also, this index is based not on the outcome measure HRQoL, but the effect of comorbidities on mortality. This shortcoming is well known, and an alternative adjustment tool is the Mukherjee index that has been developed to adjust HRQoL outcome for comorbidity [26]. However, this index has not yet, to our knowledge, been applied in the critical care setting.

In the studies selected for this review, three papers tried to construct a comorbidity adjusted control group [15, 21, 23]. Interestingly, in these papers no significant effects of the critical care period itself could be linked to HRQoL after the intensive care period. In the present review we chose not to include smaller subgroup studies. However, it needs to be mentioned that there are studies that have investigated a specific diagnosis treated in ICU [27,28,29]. In such a study of chronic obstructive pulmonary disease (COPD) [30], a control group with the same COPD score, but not treated in the ICU, was compared with COPD patients treated in the ICU. The interesting finding was that the authors could not find a specific HRQoL effect of the ICU treatment period itself. This supports the findings in the above-cited three studies which included comorbidity adjusted control groups.

Comorbidity, age, and sex

In many studies, adjustments for HRQoL in the intensive care cohorts are made based on age. This is of course important, as we know there are age-related effects on HRQoL [31]. Furthermore, age adjustment has been claimed to be important to compensate for comorbidities as comorbidities increase with age [31]. However, it has repeatedly been shown that the cohorts that end up in the ICU are different compared to the general population in the rate of comorbidities, irrespective of age [9, 32]. Therefore, it needs to be pointed out that age is not specific enough for such an adjustment. Furthermore, sex adjustments are often made for the same reasons. This can make sense as it is known that there are sex-related differences in the selection, treatment, and outcome process of the ICU cohorts [33, 34]. However, it may not fully compensate for more detailed information about comorbidities.

Limitations

The major limitation of this review resides in the scant retrieval of studies reported. Furthermore, in the studies selected there was a large heterogeneity in important study related parameters, i.e., HRQoL instruments, comorbidity, age, and sex adjustments as well as follow-up period. This significantly reduces the strength of the conclusions made. However, for the three studies that included comorbidity adjusted control groups the findings were coherent, supporting the overall conclusion of our review [15, 21, 23].

Future aspects

This review underlines the need for proper adjustments of the comorbidity effects, as lacking comorbidity adjustments may reduce the validity of the HRQoL studies made in ICU populations. The effect of comorbidity is large, and the ICU population in most high-income countries is more significantly burdened by comorbidities the more they age. To properly adjust for this is essential. HRQoL data in individual patients prior to intensive care is rare. Further, prospective pre-ICU HRQoL measured by the patient or proxies after the intensive care period is at risk of bias and should be used with caution [35, 36]. Therefore, trying to construct an appropriate comorbidity adjusted control group appears to be the method of choice. If comorbidity adjustments are planned, it is important to use an HRQoL instrument that is comorbidity validated [26] rather than a mortality related one. In addition, age and sex adjustments are important as both affect HRQoL. From the present study it seems reasonable to suggest SF-36 as the HRQoL instrument, as it better covers HRQoL issues than e.g., EQ-5D. The appropriate follow-up time is more difficult to suggest. One alternative is to extend it beyond 6–12 months, when many patients are still recovering from their critical illness, and to also extend the time to at least three years so that observations can be gathered in a stable period [9].

Conclusions

In summary, comorbidity is a main factor affecting the patients self-perceived Health-Related Quality of Life (HRQoL) in individuals who have been in the ICU. A major challenge in this area of study is the considerable diversity in the methodologies of the reviewed research, especially regarding the tools used for HRQoL assessment, the follow-up period, the approach to evaluating comorbidity, and considerations for age and gender. Despite these methodological differences, the predominant finding suggests that comorbidity alone is responsible for most of the observed HRQoL changes in the critical care survivor population. The present lack of thorough research in this area highlights the increasing necessity for more expansive and detailed studies, particularly those that incorporate comorbidity into the final HRQoL evaluation.