Intensive Care Medicine

, Volume 31, Issue 5, pp 611–620

Quality of life in adult survivors of critical illness: A systematic review of the literature

Authors

  • David W. Dowdy
    • Department of EpidemiologyJohns Hopkins Bloomberg School of Public Health
    • School of MedicineJohns Hopkins University
  • Mark P. Eid
    • School of MedicineJohns Hopkins University
  • Artyom Sedrakyan
    • Department of Health Policy and ManagementJohns Hopkins Bloomberg School of Public Health
  • Pedro A. Mendez-Tellez
    • Department of Anesthesiology & Critical Care MedicineJohns Hopkins University
  • Peter J. Pronovost
    • Department of Health Policy and ManagementJohns Hopkins Bloomberg School of Public Health
    • Department of Anesthesiology & Critical Care MedicineJohns Hopkins University
    • Department of SurgeryJohns Hopkins University
  • Margaret S. Herridge
    • Interdepartmental Division of Critical Care MedicineUniversity of Toronto
    • Division of Pulmonary & Critical Care MedicineJohns Hopkins University
Systematic Review

DOI: 10.1007/s00134-005-2592-6

Cite this article as:
Dowdy, D.W., Eid, M.P., Sedrakyan, A. et al. Intensive Care Med (2005) 31: 611. doi:10.1007/s00134-005-2592-6

Abstract

Objective: To determine how the quality of life (QOL) of intensive care unit (ICU) survivors compares with the general population, changes over time, and is predicted by baseline characteristics. Design: Systematic literature review including MEDLINE, EMBASE, CINAHL and Cochrane Library. Eligible studies measured QOL ≥30 days after ICU discharge using the Medical Outcomes Study 36-item Short Form (SF-36), EuroQol-5D, Sickness Impact Profile, or Nottingham Health Profile in representative populations of adult ICU survivors. Disease-specific studies were excluded. Measurements and results: Of 8,894 citations identified, 21 independent studies with 7,320 patients were reviewed. Three of three studies found that ICU survivors had significantly lower QOL prior to admission than did a matched general population. During post-discharge follow-up, ICU survivors had significantly lower QOL scores than the general population in each SF-36 domain (except bodily pain) in at least four of seven studies. Over 1–12 months of follow-up, at least two of four studies found clinically meaningful improvement in each SF-36 domain except mental health and general health perceptions. A majority of studies found that age and severity of illness predicted physical functioning. Conclusions: Compared with the general population, ICU survivors report lower QOL prior to ICU admission. After hospital discharge, QOL in ICU survivors improves but remains lower than general population levels. Age and severity of illness are predictors of physical functioning. This systematic review provides a general understanding of QOL following critical illness and can serve as a standard of comparison for QOL studies in specific ICU subpopulations

Keywords

Critical careCritical illnessHealth status indicatorsIntensive care unitsOutcome assessment (health care)Quality of life

Introduction

As more patients survive critical illness, assessing quality of life (QOL) among intensive care unit (ICU) survivors has become a research priority [1, 2]. Despite a rapidly growing body of literature, previous reviews of QOL in ICU survivors have been descriptive [37] or methodological [8], and have not broadly synthesized research findings. Specifically, there has been no systematic review of how QOL in ICU survivors (1) compares with the general population, (2) changes over time, and (3) is predicted by baseline characteristics.

Ultimately, it is important to understand QOL in specific ICU subpopulations (e.g., acute respiratory distress syndrome (ARDS) and severe sepsis) in order to assess the impact of specific interventions on these patients [9]. To place this knowledge in context, however, an understanding of QOL among the general population of critically ill patients is needed as a standard against which to compare outcomes in ICU subpopulations. Consequently, in this systematic review, we studied QOL in patient populations representative of all adult ICU survivors.

Methods

Search strategy

To identify studies eligible for the systematic review, we searched Medline (1966–2004), EMBASE (1974–2004), CINAHL (1982–2004), pre-CINAHL, and the Cochrane Library (2004, Issue 3) on August 13, 2004. The following search strategy was used, with all terms mapped to the appropriate MeSH/EMTREE subject headings and “exploded”: (“quality of life” OR “health status indicators”) AND (“intensive care units” OR “critical care” OR “critical illness” OR “sepsis” OR “adult respiratory distress syndrome”). Sepsis and ARDS were included in the strategy because one eligible study identified before conducting the search was not captured using the initial search terms. No limits regarding language or publication type were applied. We also hand-searched personal files and the reference lists of relevant review articles [38]. For all articles included in the systematic review, we reviewed the reference lists and conducted an Expanded Science Citation Index (ISI Web of Knowledge, Thomson Corporation) search of publications citing the articles.

Study selection

Two authors (D.W.D., M.P.E.) independently reviewed citations, abstracts, and full articles to select eligible studies for review. All citations selected by either author for abstract review were included, and subsequent disagreement regarding eligibility was resolved by consensus. Agreement between the two reviewers was calculated by percentage agreement and the kappa statistic. For all foreign language articles, English translations of abstracts were reviewed. Eligible articles written in German, French, and Spanish were reviewed, in full, by a single author (D.W.D.). One potentially eligible article written in Czech [10] was excluded without full-text review.

Studies were selected for review if they met two inclusion criteria: (1) study of a representative population of adult ICU survivors (defined as a population consisting entirely of medical and/or surgical ICU patients ≥14 years old and representing a majority of all patients with an ICU length of stay >24 h), and (2) measurement of QOL ≥30 days after ICU discharge, using one of four widely used instruments: Medical Outcomes Study 36-item Short Form General Health Survey (SF-36) [11], EuroQol-5D (EQ-5D) [12], Sickness Impact Profile (SIP) [13], or Nottingham Health Profile (NHP) [14]. In order to minimize heterogeneity, studies focusing on specific diseases (e.g., ARDS), treatments (e.g., parenteral nutrition), or conditions (e.g., length of stay >7 days) were excluded.

Description of measurement instruments

SF-36, EQ-5D, SIP, and NHP are measures of health-related quality of life commonly used in critical care research. Patrick’s Perceived Quality of Life (PQOL) questionnaire [15] was excluded because it measures qualitatively different QOL domains (e.g., family, community, leisure, help, income, and respect) than the four selected instruments, thereby precluding a meaningful synthesis of results. Other instruments, such as the Fernandez questionnaire [16], were excluded because they lacked specific domains or because they were used in ≥5 critical care studies, including studies of specific subpopulations [3].

SF-36 [11] uses 36 items to measure eight QOL domains: physical functioning, role limitations due to physical problems, bodily pain, general health perceptions, energy/vitality, social functioning, role limitations due to emotional problems, and mental health. SIP [13] uses 136 questions to evaluate twelve QOL domains: work, recreation, emotional behavior, alertness, home management, sleep, body care, eating, ambulation, mobility, communication, and social interaction. Both SF-36 and SIP allow calculation of summary scores from their multiple domains: physical and mental component scores for SF-36, and physical and psychosocial summary scores for SIP. SF-36 and SIP have been comprehensively validated in critically ill patients [17,18].

NHP [14] measures QOL with two parts. Part I evaluates subjective functional status with 38 yes/no statements in six domains: physical mobility, pain, sleep, energy, emotional reactions, and social isolation. Part II is no longer recommended for use by its developers, due to poor measurement properties, and is therefore excluded from this review. EQ-5D [12] is a brief, two-part instrument. The EQ-5D self-classifier is a five-item questionnaire that elicits one of three responses (“no”, “some”, or “extreme”) for problems with mobility, self-care, usual activities, pain/discomfort, and anxiety/depression; the EQ-VAS assesses self-rated health status using a single visual analogue scale (VAS) scored from 0 to 100. Although some studies [19,20] have examined limited aspects of NHP validity, NHP and EQ-5D have not been comprehensively validated for ICU patients.

Data extraction, synthesis, and study quality

For each eligible study, two authors (D.W.D., M.P.E.) independently abstracted measures of study quality, baseline variables (age, severity of illness, gender, ICU length of stay, and admission diagnosis), QOL instrument, method of administration, and QOL outcomes in an unmasked manner [21] with differences resolved by consensus.

Due to quantitative and qualitative heterogeneity in the reporting of QOL results, a quantitative synthesis was not possible. We therefore summarized study results as either statistically significant ( p <0.05) or nonsignificant. In addition, differences in SF-36 scores of more than 5 points were reported as clinically meaningful [11]; comparable measures of clinical meaning were not available for the other instruments.

We assessed study quality using four criteria adapted from the United States Preventive Services Task Force [22]: (1) no systematic exclusion of >10% of the adult ICU population; (2) description of patients lost to follow-up and comparison with those remaining in the study; (3) measurement and reporting of specific QOL domains at baseline and follow-up; and (4) adjustment for confounders—including age—by stratification, statistical adjustment, or comparison with a matched population. Each criterion was assessed independently without reporting an overall score. Quality criteria were not used in decisions regarding inclusion or exclusion of eligible studies.

Results

Search results and study characteristics

We identified 8,894 citations, of which 352 abstracts and 111 full-text publications were reviewed (Fig. 1). A total of 26 reports (25 articles and one dissertation) describing 21 independent patient populations were eligible for the review (Table 1). Reviewer agreement on selection of eligible citations was 97.4% (kappa=0.71), and on selection of publications for final review was 99.0% (kappa=0.97).
Fig. 1

Flow diagram of literature search results

Table 1

Characteristics of studies measuring quality of life in adult ICU survivors ( ICU intensive care unit, LOS intensive care unit length of stay, SF-36 Medical Outcomes Study 36-item Short Form General Health Survey, EQ-5D EuroQol-5D, SIP Sickness Impact Profile, NHP Nottingham Health Profile, CABG coronary artery bypass graft)

First author (year)

Na

Key exclusion criteria

Patient age

(years) b

ICU LOS

(days) b

Follow-up

(months) c

SF-36

Pettila (2000) [23] and Kaarlola (2003) [24]

298

None

53 (16)

5 (6)

12, 72

Kleinpell (2003) [25]

199

<45 years old; LOS <24 h

>45 yearsd

3 (3)

1, 3, 6, 12

Wehler (2003) [26]

171

LOS ≤24 h

57 (17)

11 (19)

6

Graf (2003) [27]

164

Delirious/comatose; LOS <24 h

64 (13)

--

1, 9

Eddleston (2000) [28]

136

None

49 (12)

4 (1–13)

3

Kvale (2003) [29,30]

126

LOS ≤24 h

52 (16)

3 (mean)

6, 24

Vedio (2000) [31]

115

Overnight stay

65 (56–70)

2 (1–4)

6

Ridley (1997) [32]

95

“Precipitous” discharge

62 (mean)

1 (1–1)

6

Flaaten (2001) [33] and Kvale (2002) [34]

51

Cardiac surgery

33 (22)

5 (7)

12 years

Chaboyer (2002) [35]

16

No family proxy

61 (18)

3 (4)

6, 12

EQ-5D

Badia (2001) [36]

334

Proxy not available; LOS <12 h

57 (44–65)

5 (3–16)

12

Granja (2002) [37] and Granja (2004) [38]

275

LOS <24 h [37]

57 (median)

2 (median)

6

Garcia Lizana (2003) [39]

96

Uncomplicated elective surgery

60 (42–75)

3 (3–6)

18

Sznajder (2001) [40]

64

None

55 (20)

8 (11)

6

SIP

Tian (1995) [17] and Miranda (1994) [41]

3,655

LOS <24 h

60 (15)

3 (4)

6

Short (1999) [42]

853

None

45 (32–62)

2 (1–4)

12

Kleinpell (1991) [43]

164

<45 years old

>45 yearse

4 (6)

7

Sage (1986) [44]

140

CABG patients

56 (mean)

4 (mean)

15

Frick (2002) [45]

85

None

65 (median)

2 (median)

6

NHP

Hurel (1997) [19]

223

None

52 (18)

9 (10)

6

Bell (1994) [46]

60

Transferred elsewhere

54 (mean)

3 (mean)

3

a Sample size at first follow-up visit after hospital discharge

b Data are reported as median (interquartile range) or mean (standard deviation) if not specified.

c Length of time from ICU or hospital discharge until quality of life measurement

d 53% of patients were 45–64 years old, and 47% were 65–86 years old.

e 22% were 45–64 years old, 46% were 65–79, and 32% were ≥80 years old.

Of the 21 independent studies reviewed, five were multicenter studies [17, 19, 25, 40,45]; 16 were conducted in Europe [17, 19, 23,2634,3640, 45,46], three in the United States [25, 43,44], one in Australia [35], and one in Hong Kong [42]. Ten studies used SF-36, four used EQ-5D, five used SIP, and two used NHP (Table 1). Thirteen of 14 (93%) studies using SF-36 or EQ-5D were published in 2000 or later, vs one of seven (14%) using SIP or NHP. One study [33] had a mean patient age of 33, while the mean or median age in all other studies ranged from 45 to 65. Three studies (14%) had a mean ICU length of stay >5 days [19, 26,40]. Nineteen studies (90%) had a follow-up period ranging from 3–24 months, while the remaining two studies had extended follow-up periods of 6 years [24] and 12 years [33]. The median follow-up time was 7 months. QOL was assessed at follow-up by mailed survey in 16 studies, by telephone in three [26, 36,39], and by personal interview in two [28,37].

Assessment of study quality

A majority of the studies met each of the four quality criteria (Table 2). However, pre-admission QOL domains were measured in only five studies (24%). Four studies excluded >10% of the eligible adult ICU population; these exclusions were based on age <45 years [25,43] and cardiac surgery status [33,44]. Two studies [42,44] reported only a summary QOL measure, instead of specific domains. Agreement on assessment of study quality was 93.0% (kappa=0.85).
Table 2

Assessment of study methods and reporting for quality of life in adult ICU survivors ( ICU intensive care unit, QOL quality of life, SF-36 Medical Outcomes Study 36-item Short Form General Health Survey, EQ-5D EuroQol-5D, SIP Sickness Impact Profile, NHP Nottingham Health Profile)

Source

No major exclusion criteriaa

Losses to follow-up described

Specific QOL domains measured

Adjusted for age and gender

Pre-ICU

Post-ICU

SF-36

Pettila [23]

--

Kleinpell [25]

--

Wehler [26]

--

Graf [27]

--

Eddleston [28]

--

Kvale [29]

--

--

Vedio [31]

--

Ridley [32]

--

Flaaten [33]

--

--

Chaboyer [35]

--

--

--

EQ-5D

Badia [36]

--

--

Granja [37]

--

Garcia Lizana [39]

--

--

Sznajder [40]

--

--

SIP

Tian [17]

--

--

Short [42]

--

--

Kleinpell [43]

--

--

Sage [44]

--

--

--

--

Frick [45]

--

--

--

NHP

Hurel [19]

--

--

Bell [46]

--

--

Total (of 21 studies)

17

14

5

19

12

aDefined as systematic exclusion of >10% of the adult ICU population with a length of stay >24 h

Comparing quality of life in ICU survivors versus the general population

Seven studies [23, 24, 26, 27,3133] compared QOL in ICU survivors versus a matched general population (Table 3). All studies used the SF-36 survey. Three of these studies retrospectively examined QOL prior to ICU admission, with each study reporting statistically significant ( p <0.05) and clinically meaningful (>5 points) decrements in all eight QOL domains.
Table 3

Quality of life measurements in adult ICU survivors versus age- and gender-matched general population ( ICU intensive care unit, QOL quality of life)

Source

Na

Follow-up timeb

Physical QOL domainsc

Mental QOL domainsc

Physical function

Role physical

Bodily pain

General health

Vitality

Social function

Role emotional

Mental health

Studies of QOL prior to ICU admissiond

Wehler [26]

318

--

*

*

*

*

*

*

*

*

Graf [27]

153

--

*

*

*

*

*

*

*

*

Ridley [32] e

75

--

*

*

*

*

*

*

*

*

Studies of QOL after ICU stay

Wehler [26]

171

6 months

*

*

--

*

*

*

*

*

Ridley [32] e

75

6 months

*

*

*

*

*

*

*

*

Vedio [31] f

  Elective

66

6 months

--

*

--

--

--

--

--

  Emergency

49

6 months

*

*

--

*

*

--

*

Graf [27]

153

9 months

*

*

--

*

*

*

*

*

Pettila [23]

298

12 months

*

*

*

*

*

Kaarlola [24] g

169

6 years

--

--

--

--

--

Flaatten [33]

51

13–14 years

*

*

*

*

*

*

aSample size at the time of follow-up

bLength of time from ICU or hospital discharge until quality of life measurement

c ↓/↑ clinically meaningful (i.e., >5-point) decrement/improvement in quality of life;*clinically meaningful and statistically significant ( p <0.05) decrement in quality of life; -- non-clinically meaningful (i.e., ≤5-point) change in quality of life

dQOL prior to ICU admission was measured retrospectively from patient or proxy.

eIncludes only patients <65 years old

fSeparately analyzed patients with emergency and elective diagnoses on ICU admission

gThe study population in [24] is a subset of that in [23]. No measure of significance was reported in [24].

QOL was measured at 6 months to 14 years after ICU admission in eight patient populations (Table 3). For each domain, except bodily pain, a majority of populations reported significant and meaningful decrements in QOL versus the general population. For bodily pain, a clinically meaningful decrement was observed in four populations (50%), but this finding was statistically significant in only one study (14%). Compared with the general population, no study reported a statistically significant or clinically meaningful increase in any QOL domain among its complete population of ICU survivors.

Changes in quality of life over time

Five studies compared QOL at follow-up with pre-admission baseline QOL measured retrospectively. Four of these studies [25, 26, 32,36] used proxies to obtain baseline QOL for patients who were unable to respond, and one study [27] excluded such patients. Four studies used the SF-36 survey (Table 4). Social functioning and role performance due to both physical and emotional problems showed clinically meaningful decreases from baseline at 1-month follow-up, but clinically meaningful increases from baseline at 6 months and 12 months. Vitality also showed clinically meaningful improvement between 1 and 6 months, between 1 and 9 months, and between 1 and 12 months follow-up [25,27]. For both physical functioning and bodily pain, clinically meaningful improvement from baseline was observed by 1 month and seen throughout follow-up in five of seven measurements. Only one study [26] showed a clinically meaningful change in mental health, and no study showed a meaningful change in general health perceptions.
Table 4

Change in quality of life from baseline for adult ICU survivors a ( ICU intensive care unit, QOL quality of life)

Follow-up timeb

Domain

1 month

6 months

9 months

12 months

Kleinpell [25]

Graf [27]

Kleinpell [25]

Wehler [26]

Ridley [32]

Graf [27]

Kleinpell [25]

Physical domains

Physical functioning

*

*

*

--

--

*

*

Role – physical

*

*

--

*

*

Bodily pain

--

*

--

*

*

General health

--

--

--

--

--

--

--

Mental domains

Vitality

--

--

--

*

*

Social functioning

--

*

*

*

--

*

Role – emotional

--

*

*

--

--

Mental health

--

*

--

--

--

*

--

aBaseline measurements were obtained retrospectively by interview with patient [2527, 32] or proxy [25, 26,32]. Follow-up times are from ICU or hospital discharge.

b ↑/↓clinically meaningful (i.e., >5-point) improvement/decline in quality of life;*/↓*clinically meaningful and statistically significant ( p <0.05) improvement/decline in quality of life; -- non-clinically meaningful (i.e., ≤5-point) change in quality of life. All quality of life measurements at follow-up are compared against baseline.

Using EQ-5D to measure change in QOL from baseline to 12 months after ICU discharge, Badia et al. [36] found a significant increase in mobility, significant decreases in self-care and usual activities, and no significant change in pain/discomfort, anxiety/depression, or the visual analogue scale.

Predictors of quality of life

Seventeen studies investigated associations of patient age, severity of illness, admission type, gender, or ICU length of stay with QOL in ICU survivors. Regarding age, a majority of studies found significantly lower physical functioning (SF-36, Table 5), usual activities (EQ-5D, Table 6), and physical or total QOL (SIP, Table 6) in older versus younger ICU survivors. No study found a significant association between age and mental health (SF-36), anxiety/depression (EQ-5D) or psychosocial QOL (SIP).
Table 5

Predictors of quality of life using SF-36 ( SF-36 Medical Outcomes Study 36-item Short Form General Health Survey, ICU intensive care unit, QOL quality of life, SOFA Sequential Organ Failure Assessment, APACHE Acute Physiology and Chronic Health Evaluation, SAPS Simplified Acute Physiologic Score)

Source

Follow-Upa

Measure

Physical QOL domainsb

Mental QOL domainsb

Physical function

Role physical

Bodily pain

General health

Vitality

Social function

Role emotional

Mental health

Age

Eddleston [28]

3 months

>65 vs. ≤65 years of age

*

--

--

--

--

--

--

--

Kleinpell [25]

6 months

≥65 vs. <65 years of age

--

--

--

--

--

--

--

--

Wehler [26]

6 months

4 strata of age

*

*

*

*

*

*

--

--

Graf [27]

9 months

≥66 vs. <66 years of age

*

--

--

--

--

--

--

--

Pettila [23]

12 months

Agec

*

*

*

*

*

--

*

--

Severity of Illness

Kleinpell [25]

6 months

APACHE IIIc

*

--

--

*

--

--

--

--

Wehler [26]

6 months

SOFA ≥6 vs. <6

*

--

--

*

--

--

--

--

Vedio [31]

6 months

“Chronic problem” on APACHE II

*

--

--

*

*

--

--

--

Graf [27]

9 months

SAPS II and SOFAc, d

--

--

--

--

--

--

--

--

Pettila [23]

12 months

SOFA ≥6 vs. <6

*

*

*

*

*

*

*

*

Kvale [29]

24 months

SAPS IIc

--

--

--

--

--

--

--

--

aLength of time from ICU or hospital discharge until quality of life measurement

b *Statistically significant ( p <0.05) decrease in quality of life with increase in age or severity of illness; -- No statistically significant change in quality of life ( p >0.05)

cMeasured as a continuous variable

dBoth measures were independently studied, and neither showed a significant association.

Table 6

Predictors of quality of life using EuroQol-5D and SIP ( SIP Sickness Impact Profile, NHP Nottingham Health Profile, ICU intensive care unit, Activ. activities, Pain pain/discomfort, Anx. Depr. anxiety/depression, VAS visual analogue scale, Phys physical subscore, Psych psychosocial subscore, APACHE Acute Physiology and Chronic Health Evaluation, SAPS Simplified Acute Physiologic Score)

Source

Measure

Follow-Upa

EuroQol-5Db

SIPb

Mobility

Self Care

Usual Activ.

Pain

Anx. Depr.

VAS

Phys

Psych

Total

Age

Granja [37]

Agec

6 months

*

*

*

*

--

*

--

--

--

Garcia [39]

Agec

18 months

--

--

*

--

--

--

--

--

--

Tian [17]

6 strata of age

6 months

--

--

--

--

--

--

*

--

*

Kleinpell [43]

3 strata of age

7 months

--

--

--

--

--

--

*

--

*

Short [42]

3 strata of age

12 months

--

--

--

--

--

--

--

--

*

Sage [44]

3 strata of age

15 months

--

--

--

--

--

--

--

--

*

Severity of Illness

Granja [37]

APACHE IIc

6 months

*

*

*

--

--

*

--

--

--

Garcia [39]

APACHE IIc

18 months

--

--

*

--

*

--

--

--

--

Tian [17]

APACHE IIc

6 months

--

--

--

--

--

--

--

--

--

Kleinpell [43]

APACHE IIc

7 months

--

--

--

--

--

--

--

--

--

Short [42]

APACHE II (4 strata)

12 months

--

--

--

--

--

--

--

--

*

Sage [44]

APACHE IIc

15 months

--

--

--

--

--

--

--

--

--

aLength of time from ICU or hospital discharge until quality of life measurement

b * Statistically significant ( p <0.05) decrease in quality of life with increase in age or severity of illness; -- no statistically significant change in quality of life ( p >0.05)

cMeasured as a continuous variable

Of six studies investigating severity of illness and QOL measured by SF-36, four (67%) found a significant association between severity of illness and lower physical functioning or general health perceptions (Table 5). Similarly, both EQ-5D studies measuring usual activities and the one study measuring EQ-VAS found significant associations in these domains with severity of illness (Table 6). There was no significant association of severity of illness with the total QOL score in three of four studies using SIP (Table 6) and one study using NHP [38].

Five studies [19, 29, 36, 42,46] examined QOL in ICU survivors with medical versus surgical diagnoses. Only one study [19] found a significant difference in more than one QOL domain. In patients surviving trauma, three studies (100%) [36, 37,39] demonstrated significantly worse pain/discomfort on EQ-5D, compared with other ICU survivors, at 6–18 months after discharge.

Six studies [19, 31, 36, 37, 42,44] evaluated survivors of elective versus emergency surgical procedures. At 6–18 months after ICU discharge, three studies [19, 42,44] found no significant association between surgical status and overall QOL, and two [31,37] found that emergency surgical patients had significantly worse quality of life in a minority of domains. Two studies [31,36] evaluated changes in QOL over time. Badia et al. [36] observed significant improvements from baseline to 12 months in four of five EQ-5D domains (mobility, usual activities, pain/discomfort, and anxiety/depression) among scheduled, but not unscheduled, surgical patients. Similarly, Vedio et al. [31] found that 76% of elective, vs. 31% of emergency, surgical patients reported an improved QOL from baseline after 6-months follow-up.

Of nine studies investigating associations between QOL and gender, only two studies [26,39] found a significant association of gender with any domain. Of eight studies investigating ICU length of stay, only one [23] found a significant association with any QOL domain.

Discussion

This systematic review of quality of life in 7,320 adult ICU survivors has three major findings. First, compared with the general population, ICU survivors have lower QOL for all domains (except bodily pain) at baseline and at 6 months to 14 years after discharge. Second, QOL in ICU survivors improves with over time after discharge, but this improvement is not uniform across domains. At 6–12 months, clinically meaningful improvements from pre-admission baseline occur in six SF-36 domains; however, general health perceptions and mental health do not consistently demonstrate any clinically meaningful difference from baseline. Third, regarding predictors of QOL, older age and increased severity of illness may be associated with poorer physical function and general health perceptions (severity of illness only), but are not consistently associated with decrements in other QOL domains. Lower QOL at baseline may explain this finding. Wehler et al. [26] found that older patients, compared with younger patients, did not have a significantly worse change in QOL after discharge. Patient gender, ICU length of stay, and medical versus surgical diagnosis do not appear to be important predictors of QOL. Survival after trauma is associated with worse pain/discomfort, and elective surgical patients are more likely than emergency patients to improve their QOL from pre-admission values. Similarly, one excluded study of patients with prolonged ICU stay [47] found QOL at 6 months post-discharge to be lowest in respiratory and trauma patients and highest in patients undergoing cardiac surgery.

Our finding of high study quality contrasts with Heyland et al.’s previous methodologic assessment [8]. However, we restricted our analysis to four specific QOL instruments and used different quality assessment criteria. In their review, Heyland et al. [8] noted that few QOL instruments have been adequately validated in ICU populations; this remains true of EQ-5D and NHP, and represents an important focus for future methodologic research.

This systematic review has a number of methodologic limitations. First, in order to reduce heterogeneity between studies, we excluded QOL studies that examined only specific subsets of ICU survivors. QOL outcomes in these subsets may differ in important ways from general ICU populations [19]. Thus, this systematic review cannot provide a comprehensive picture of QOL in all patient groups. However, this review does provide a general understanding of QOL after critical illness and serves as a foundation for comparison with future reviews of QOL in specific ICU patient subsets.

Second, in this review, we evaluated only four QOL instruments commonly used in critical care research. The exclusion of Patrick’s PQOL [15] and other instruments reduces the comprehensiveness of this review. However, our selection of QOL instruments allowed for greater comparability and more meaningful synthesis of study findings than would have been possible under a more comprehensive review. Standardizing the instrument(s) used for QOL assessment in critical care research would enable more effective comparison between studies [1].

Third, scores for some QOL domains reflect a greater number of survey items, and are thus more precise, than other domains. For example, on SF-36, physical functioning reflects ten items, whereas social functioning reflects two items. A 5-point difference on the latter domain may therefore be less meaningful than a 5-point difference on the former. However, our findings generally addressed the most precise domains (i.e., physical functioning, mental health, and general health perceptions on SF-36). Thus, we believe that our synthesized findings do not simply reflect precision differences between QOL domains. We caution, however, that distribution-based measures, such as statistical significance and 5-point change, are problematic in translating QOL changes into clinically meaningful terms [48]. Validation of instruments with anchor-based methods in ICU populations is required to assist in making inference regarding clinically meaningful changes [48].

In addition to methodological concerns, our findings are limited by the infrequent collection of baseline data and inconsistent reporting methods among reviewed studies. Only five studies [2527, 32,36] examined QOL prior to ICU admission, and each demonstrated marked impairments in baseline QOL compared with the general population. As a result, it is difficult to determine whether QOL decrements at follow-up reflect the impact of critical illness or simply a lower baseline QOL. Baseline QOL measurement is needed to control for preexisting impairment, but is difficult to perform since it must be obtained retrospectively from the ICU survivor (subject to recall bias) or from his/her proxy (subject to measurement error). Prior studies [49,50] have found fair to good agreement between proxy and patient responses on the SF-36. Further research comparing methods of estimating baseline QOL is needed to assist with future study design.

Due to inconsistent reporting methods, we were unable to quantitatively synthesize results. QOL scores, even from the same instrument, were reported in a heterogeneous fashion (e.g., absolute scores, z -scores, semiquantitative figures). Standardization of reporting methods for QOL scores, specifically reporting the mean score and standard deviation for each domain, would allow improved comparison and synthesis of study findings.

In conclusion, our systematic review of 21 studies involving 7,320 patients demonstrates that ICU survivors report a lower baseline (pre-admission) quality of life that improves over time in most domains, but remains lower than general population levels throughout long-term follow-up. In particular, physical functioning shows rapid improvement and is associated with patient age and severity of illness, whereas mental health shows no improvement and is independent of baseline characteristics. This synthesis provides a general understanding of quality of life following critical illness and can serve as a standard of comparison for QOL studies in specific ICU subpopulations.

Acknowledgements

This research is supported by the National Institutes of Health (ALI SCCOR Grant # P050 HL 73994–01). Dr. Needham is supported by Clinician-Scientist Awards from the Canadian Institutes of Health Research and the University of Toronto, and a Detweiler Fellowship from the Royal College of Physicians and Surgeons of Canada

Copyright information

© Springer-Verlag 2005