Abstract
Rationale: The relationship between organ failure during critical illness and long-term survival is uncertain, especially among intensive care unit (ICU) survivors.
Objectives: To describe the relationship between individual organ failures, total organ failure burden, and mortality during the 5 years after an episode of critical illness.
Methods: We studied a cohort of sequential admissions to 10 Scottish ICUs (n = 872). Logistic regression was used to explore independent associations between organ failures and mortality over a 5-year time horizon, adjusting for potential confounders.
Measurements and Main Results: Daily Sequential Organ Failure Assessment scores described organ dysfunction during ICU stay. The sum of the worst scores at any time point during the ICU stay for each organ system except neurological dysfunction was used to calculate total organ failure burden. Mortality was obtained from the national death register. Five-year mortality was 58.2%; 34.4% of deaths occurred within 28 days. In adjusted analyses, cardiovascular (odds ratio [OR], 2.5; 95% confidence interval [CI], 1.8–3.7), liver (OR, 2.3; 95% CI, 1.1–5.0), and respiratory failure (OR, 2.1; 95% CI, 1.3–3.5) were independently associated with 5-year mortality. Organ failure burden was strongly associated with mortality; 81% of patients in the highest tertile died during follow-up (OR, 6.3 relative to lowest tertile; P < 0.001). Patients surviving more than 12 months post-ICU were still more likely to subsequently die if they experienced greater organ failure burden in the ICU (OR, 2.4; P = 0.02, highest vs. lowest tertile).
Conclusions: Cardiovascular, respiratory, and liver failures during critical illness strongly predict subsequent 5-year survival. Acute organ failure burden is associated with long-term mortality even among patients who survive up to 1 year after ICU admission.
There is increasing concern amongst critical care clinicians regarding the longer term effects of critical illness on patient outcomes. The impact of organ failure during an episode of critical illness on longer term outcomes is not known.
Our study provides evidence that a higher burden of organ failure during an episode of critical illness is strongly associated with greater 5-year mortality. This association remains strong in patients who are alive 1 year after their episode of critical illness, suggesting an important residual effect from the acute illness on long-term health.
General intensive care units (ICUs) care for patients with a wide range of disease processes, but a common goal is the prevention of further organ dysfunction and the management of established organ failures. It is well established that the numbers and severity of organ dysfunction correlate with short-term mortality (1, 2). Risk scores based on the first 8–24 hours in ICU, such as the Acute Physiology and Chronic Health Evaluation (APACHE) (1) and Simplified Acute Physiology Score (SAPS) (2) systems, are widely used for predicting hospital mortality and to standardize mortality outcomes (3). Similarly, organ failure scores such as the Sequential Organ Failure Assessment (SOFA) score have been used to describe organ dysfunction during ICU admission (4), and both worst values and changes over time correlate closely with short-term outcomes such as ICU and hospital mortality (5). Cardiovascular failure, in particular, is strongly predictive of higher short-term mortality (6). Trials of novel therapies frequently adjust for the degree of organ dysfunction at baseline or use progression of organ failure as a clinical outcome, highlighting the importance of this factor in critical care research.
It is increasingly appreciated that longer term mortality should be measured after critical illness, particularly when assessing new interventions (7, 8). Justifications for this include the recognition that many ICU survivors have significant physical impairment, impaired quality of life, and excess mortality during the post-ICU period compared with the general population (9–13). There is still a paucity of data describing long-term survival in general critical care populations, especially in relation to the association between the severity of critical illness and long-term mortality. Specifically, it is unknown whether the degree of acute organ dysfunction is related only to short-term survival, or whether this continues to be associated with an increased risk of death during the years after ICU admission. Detailed cohort studies, such as those describing survivors with acute respiratory distress syndrome, indicate that many patients experience long-term physical disability, which could impact on survival (14).
We hypothesized that more severe acute organ dysfunction would be associated with both short- and long-term mortality in unselected general intensive care patients. Specifically, we hypothesized that ICU organ failures were associated with an additional risk of death among patients who survived the ICU and early post-ICU period, perhaps through residual organ dysfunction and/or an effect on general health. The aims of this study were as follows: first, to describe the relationship between organ failure and long-term mortality during the 5 years after an episode of critical illness; second, to identify which organ failures, if any, were most strongly associated with higher long-term mortality; and third, to investigate the relationship between the total burden of organ failure and long-term mortality.
Detailed methods are provided in the online supplement.
We undertook a secondary analysis of a data set collected prospectively for a multicenter cohort study, the Audit of Transfusion in Intensive Care in Scotland (ATICS) study. Details of study entry criteria are reported elsewhere (15). In brief, all ICU admissions were recruited from 10 mixed medical/surgical ICUs during a 100-day period in 2001, accounting for 44% of all ICU admissions in Scotland during the study period. Their admission characteristics were similar to those of patients admitted to nonparticipating ICUs. We excluded patients with a neurological diagnosis, using admission diagnosis fields, together with patients less than 16 years of age.
For the duration of the ICU stay, each of the five organ systems (cardiovascular, respiratory, renal, coagulation, and liver) was prospectively evaluated on a daily basis and assigned a score ranging from 0 (no dysfunction) to 4 (severe dysfunction) according to the original SOFA definitions (see Table E1 in the online supplement) (4). We decided a priori not to collect a neurological organ dysfunction score, because of the difficulty in assessing sedated patients (16); this approach has been used by others and we considered it justified by our exclusion of patients with primary neurological diagnoses (17). Individual organ SOFA scores were grouped into two categories: 3–4 was defined as organ failure and 0–2 was considered absence of organ failure, a dichotomization used in other studies (18, 19). A measure of total organ failure burden was derived as the sum of the worst SOFA scores at any time point during the ICU stay for each organ system except neurological dysfunction (maximal cumulative SOFA score). Total organ failure burden was grouped using tertiles of the maximal cumulative SOFA score: mild (0–5), moderate (6–10), and severe (≥11). Details relating to other measures of SOFA score and time of maximal score are provided in the online supplement.
Death status was retrieved by linking patient data from the original study data set to the Scottish Death Registry database. This method of data linkage has been shown to have high accuracy (20). Short- and long-term mortality were defined as death at 28 days and 5 years from ICU admission, respectively.
We adjusted for potential confounders collected in the original data set, namely age, sex, admission source, and a count of preexisting comorbidities. Data were available for APACHE II and SAPS II scores, but we decided it would not be appropriate to adjust for these in addition to SOFA (see the online supplement for explanation). However, we compared the association between long-term mortality and tertiles of maximal cumulative SOFA score, SAPS II score, and APACHE II score. We explored possible effect modification for specific patient subgroups by producing stratum-specific odds ratios (ORs) for categories of age, comorbidity, and surgical status. Significance was assessed by including the interaction term in regression models.
The associations between SOFA score and mortality were presented using unadjusted ORs. The group with the lowest SOFA score was the reference category (SOFA score 0–2 for individual organ SOFA scores, 0–5 for maximal cumulative SOFA score). Adjusted ORs were calculated by binary logistic regression. To explore the residual effects of ICU organ failures on survivors remaining in the cohort over 5 years of follow-up, we examined the relationship between maximal cumulative SOFA score and long-term mortality for the patient cohorts who survived to 28 days, 90 days, 180 days, and 1 year after ICU admission. This removed the effect of deaths before this date on the relationship between ICU organ failure burden and mortality and focused on progressively longer term survivor cohorts.
Of 1,023 admissions during the study period, 151 (15%) were excluded (neurological diagnosis, n = 141; <16 yr old, n = 10) (Figure 1). The worst SOFA score during ICU admission could be calculated for more than 97% of patients for each individual organ system other than liver, for which 11.8% of patients had no bilirubin value available and no score recorded (Table E2). Maximal cumulative SOFA score could be calculated in the complete case analysis for 754 patients (86.5%). Five-year mortality from data linkage was available for 861 patients (98.7%).
Figure 1. Flow diagram of derivation of study cohort for complete case analysis and sensitivity analysis with multiple imputation (MI) for missing values. ATICS = Audit of Transfusion in Intensive Care in Scotland; ICU = intensive care unit.
[More] [Minimize]Demographic and descriptive data for the cohort are shown in Table 1. Patients admitted with a respiratory diagnosis comprised the largest group (n = 244, 28.0%). Respiratory failure was the most common organ failure, occurring in 82.6% of patients at some point during their ICU stay, followed by cardiovascular failure (45.5%). Only 8.7% of patients attained a SOFA score consistent with liver failure. Overall mortality at 28 days, 1 year, and 5 years post-ICU admission was 34.4, 46.5, and 58.2%, respectively (Figure 2). The time from admission at which the maximal individual SOFA score occurred differed by organ system, with cardiovascular scores peaking earlier during the ICU stay and liver and coagulation peaking later (Table E3). Baseline characteristics stratified by the maximal cumulative SOFA score tertiles differed in terms of age, recorded comorbidities, admission source, and ICU length of stay (Table E4).
Figure 2. Mortality rate 28 days, 1 year, and 5 years after intensive care unit (ICU) admission for each organ system and for tertiles of cumulative Sequential Organ Failure Assessment (SOFA) score. The denominator is presented on the y axis for each graph and is different because of variation in missing data for each organ system.
[More] [Minimize]| n | Value | |
| Age (yr), mean (SD) | 872 | 58.8 (17.4) |
| Female, n (%) | 872 | 380 (43.6) |
| Admission SAPS II score, median (IQR) | 854 | 35 (25, 48) |
| Number of comorbidities, n (%) | 872 | |
| 0 | 636 (72.9) | |
| 1 | 211 (24.2) | |
| 2+ | 25 (2.9) | |
| Diagnosis on ICU admission, n (%) | 872 | |
| CVS | ||
| Cardiac | 97 (11.1) | |
| Peripheral vascular | 42 (4.8) | |
| Other CVS | 9 (1.0) | |
| Respiratory | ||
| Pneumonia | 140 (16.1) | |
| ARDS | 15 (1.7) | |
| Other respiratory | 89 (10.2) | |
| GI/liver | ||
| GI perforation | 52 (6.0) | |
| GI bleeding | 40 (4.6) | |
| Pancreatitis | 28 (3.2) | |
| Liver failure or transplant | 25 (2.9) | |
| Other GI | 88 (10.1) | |
| Other | ||
| Septic shock | 61 (7.0) | |
| Trauma, non-CNS | 43 (4.9) | |
| Overdose | 39 (4.5) | |
| Renal | 19 (2.2) | |
| Miscellaneous | 85 (9.8) | |
| Admission source, n (%) | 872 | |
| Theater | 321 (36.8) | |
| Ward or noncritical care area | 178 (20.4) | |
| High-dependency unit | 150 (17.2) | |
| Emergency department | 121 (13.9) | |
| Other hospital | 102 (11.7) | |
| Organ failure incidence during ICU stay, n (%) | ||
| Coagulation | 845 | 132 (15.6) |
| Cardiovascular | 869 | 395 (45.5) |
| Liver | 769 | 67 (8.7) |
| Renal | 858 | 201 (23.4) |
| Respiratory | 845 | 698 (82.6) |
| Organ failure burden during ICU stay, n (%) | 754 | |
| Total SOFA score, 0 to 5 | 218 (28.9) | |
| Total SOFA score, 6 to 10 | 279 (37.0) | |
| Total SOFA score, 11 to 20 | 257 (34.1) | |
| ICU length of stay | 872 | |
| Mean (SD) | 5.8 (9.1) | |
| Median (IQR) | 2.3 (0.9, 7.3) | |
| Mortality, n (%) | ||
| ICU discharge | 872 | 227 (26.0) |
| 28 d post-ICU admission | 861 | 296 (34.4) |
| 5 yr post-ICU admission | 861 | 501 (58.2) |
Patients with liver failure had the highest short-term mortality (70.1%) and those with respiratory failure had the lowest mortality (38.7%). In unadjusted analyses each individual organ failure was strongly associated with short-term mortality in univariable analyses (P < 0.001 for all; Table 2). Cardiovascular SOFA score had the strongest association (OR, 6.2) and coagulation score the weakest (OR, 2.8). In the multivariable analysis adjusting for age, sex, admission source, and comorbidity and mutually adjusting for each organ failure, cardiovascular and liver organ failure retained a strong association, but the association with respiratory and renal organ failure was statistically weaker and there was no association with coagulation score. Imputation of missing values made no difference to the predictors of short-term mortality (Table E5).
| Mortality | ||||||||||
| No Failure | Failure | Univariable Analysis | Multivariable Analysis | |||||||
| Organ System SOFA Score | n | % | n | % | OR | 95% CI | P Value | OR | 95% CI | P Value |
| Short-term mortality* | ||||||||||
| Coagulation | 214 | 30.4 | 71 | 54.6 | 2.8 | 1.9–4.0 | <0.001 | 1.3 | 0.8–2.1 | 0.32 |
| CVS | 78 | 16.7 | 216 | 55.2 | 6.2 | 4.5–8.4 | <0.001 | 3.3 | 2.2–4.8 | <0.001 |
| Liver | 212 | 30.6 | 47 | 70.1 | 5.3 | 3.1–9.2 | <0.001 | 3.2 | 1.6–6.2 | 0.001 |
| Renal | 175 | 27.0 | 114 | 57.3 | 3.6 | 2.6–5.0 | <0.001 | 1.6 | 1.1–2.5 | 0.02 |
| Respiratory | 21 | 14.6 | 267 | 38.7 | 3.7 | 2.3–6.0 | <0.001 | 2.1 | 1.1–3.9 | 0.03 |
| Long-term mortality* | ||||||||||
| Coagulation | 391 | 55.5 | 97 | 74.6 | 2.4 | 1.5–3.6 | <0.001 | 1.4 | 0.8–2.4 | 0.23 |
| CVS | 199 | 42.6 | 300 | 76.7 | 4.4 | 3.3–6.0 | <0.001 | 2.5 | 1.8–3.7 | <0.001 |
| Liver | 388 | 56.1 | 55 | 82.1 | 3.6 | 1.9–6.8 | <0.001 | 2.3 | 1.1–5.0 | 0.04 |
| Renal | 343 | 52.9 | 150 | 75.4 | 2.7 | 1.9–3.9 | <0.001 | 1.3 | 0.8–2.0 | 0.32 |
| Respiratory | 55 | 38.2 | 435 | 63.0 | 2.8 | 1.9–4.0 | <0.001 | 2.1 | 1.3–3.5 | 0.004 |
For the tertiles of maximal cumulative SOFA, there was a strong association between increasing burden of organ failure and increasing short-term mortality (Table 3), with particularly high risk for the 11–20 category compared with the lowest tertile after adjustment for potential confounders (OR, 8.8; 95% CI, 5.3 to 14.7; P < 0.001).
| Mortality | Univariable Analysis | Multivariable Analysis | |||||||
| Cumulative SOFA Score Category | n | Total | % | OR | 95% CI | P Value | OR | 95% CI | P Value |
| Short-term mortality* | |||||||||
| 0 to 5 | 26 | 214 | 12.1 | 1.0 | 1.0 | ||||
| 6 to 10 | 78 | 277 | 28.2 | 2.8 | 1.7–4.6 | <0.001 | 2.4 | 1.4–4.0 | 0.001 |
| 11 to 20 | 152 | 254 | 59.8 | 10.8 | 6.7–17.4 | <0.001 | 9.1 | 5.5–15.0 | <0.001 |
| Long-term mortality* | |||||||||
| 0 to 5 | 75 | 214 | 35.0 | 1.0 | 1.0 | ||||
| 6 to 10 | 159 | 277 | 57.4 | 2.5 | 1.7–3.6 | <0.001 | 1.9 | 1.3–2.9 | 0.001 |
| 11 to 20 | 205 | 254 | 80.7 | 7.8 | 5.1–11.8 | <0.001 | 6.3 | 4.0–10.0 | <0.001 |
All individual organ failures were associated with 5-year mortality (P < 0.001 for all). In the multivariable analysis mutually adjusting for each organ failure and confounders, only cardiovascular, liver, and respiratory failure were independently associated with greater 5-year mortality (Table 2). Multiple imputation of missing values under missing at random assumptions (imputation 1) made no difference to the significance of predictors (Table E5). However, imputing missing liver SOFA scores as zero (imputation 2) attenuated the association of liver SOFA score and mortality (OR, 2.1; P = 0.06 compared with OR, 2.3; P = 0.04 in complete case analysis).
The survival curves over 5 years of follow-up for each tertile are shown in Figure 3. Tertiles of maximal cumulative SOFA score were a strong predictor of long-term mortality in both univariable and multivariable analyses adjusted for confounders (P < 0.005 for all; Table 3). Patients with a maximal cumulative SOFA score of 11–20 had an 80.7% five-year mortality rate. Imputation of missing values in sensitivity analyses increased the magnitude of the association (Table E6).
Figure 3. Survival plot for 5-year follow-up of the study cohort stratified by tertiles of total Sequential Organ Failure Assessment (SOFA) score. Shaded areas indicate 95% confidence interval. ICU = intensive care unit.
[More] [Minimize]For the SOFA tertile 0–5, 489 patients survived to 28 days, of whom 37.4% died during the remaining follow-up period (4 yr, 11 mo). Compared with tertile 0–5, this late mortality was significantly higher for the SOFA tertile 6–10 (81 of 199; 40.7% of 28-d survivors) and tertile 11–20 (53 of 102; 52.0% of 28-d survivors). In a multivariable model adjusting for the available confounders, the SOFA 11–20 tertile was associated with greater long-term mortality when compared with the SOFA 0–5 tertile (Table 4). This association remained when nonsurvivors were excluded from the analysis at 90, 180, and 365 days. These data suggested that greater organ failure during ICU admission, based on the SOFA score system, was predictive of an ongoing excess risk of death for patients surviving to at least 1 year after ICU discharge. Imputation of missing values showed similar results (Table E7).
| Mortality | Univariable Analysis | Multivariable Analysis | ||||||||
| Alive at: | Cumulative SOFA Score Category | n | Total | % | OR | 95% CI | P Value | OR | 95% CI | P Value |
| ICU admission (whole cohort) | 0–5 | 75 | 214 | 35.0 | 1.0 | 1.0 | ||||
| 6–10 | 159 | 277 | 57.4 | 2.5 | 1.7–3.6 | <0.001 | 1.9 | 1.3–2.9 | 0.001 | |
| 11–20 | 205 | 254 | 80.7 | 7.8 | 5.1–11.8 | <0.001 | 6.3 | 4.0–10.0 | <0.001 | |
| Day 28 | 0–5 | 49 | 188 | 26.1 | 1.0 | 1.0 | ||||
| 6–10 | 81 | 199 | 40.7 | 1.9 | 1.3–3.0 | 0.002 | 1.5 | 0.9–2.4 | 0.10 | |
| 11–20 | 53 | 102 | 52.0 | 3.1 | 1.8–5.1 | <0.001 | 2.6 | 1.5–4.6 | 0.001 | |
| Day 90 | 0–5 | 39 | 178 | 21.9 | 1.0 | 1.0 | ||||
| 6–10 | 64 | 182 | 35.2 | 1.9 | 1.2–3.1 | 0.006 | 1.5 | 0.9–2.5 | 0.11 | |
| 11–20 | 36 | 85 | 42.4 | 2.6 | 1.5–4.6 | 0.001 | 2.3 | 1.3–4.3 | 0.007 | |
| Day 180 | 0–5 | 34 | 173 | 19.7 | 1.0 | 1.0 | ||||
| 6–10 | 56 | 174 | 32.2 | 1.9 | 1.2–3.2 | 0.008 | 1.6 | 0.9–2.7 | 0.11 | |
| 11–20 | 27 | 76 | 35.5 | 2.3 | 1.2–4.1 | 0.008 | 2.0 | 1.0–3.9 | 0.04 | |
| Day 365 | 0–5 | 23 | 162 | 14.2 | 1.0 | 1.0 | ||||
| 6–10 | 45 | 163 | 27.6 | 2.3 | 1.3–4.0 | 0.003 | 1.9 | 1.0–3.5 | 0.04 | |
| 11–20 | 23 | 72 | 31.9 | 2.8 | 1.5–5.5 | 0.002 | 2.4 | 1.1–4.9 | 0.02 | |
We explored how this association varied for different subgroups of the study cohort by stratifying by age, comorbidity, and surgical status. Although potentially underpowered for these analyses, it was apparent that the association between maximal cumulative SOFA and long-term mortality was more pronounced in the younger age group, which reached statistical significance in the 90–day survivor cohort and later (Figure E1). Furthermore, those patients with at least one recorded comorbidity (compared with none), and nonsurgical patients (compared with surgical) appeared to have a stronger association between maximal cumulative SOFA score and long-term mortality, although the interaction terms did not reach statistical significance (Figure E1).
We compared the associations of tertiles of maximal cumulative SOFA score, SAPS II score, and APACHE II score with long-term mortality. This demonstrated that, in contrast to maximal cumulative SOFA score, the association of long-term mortality with either severity of illness score did not persist when restricting the cohort to Day 90 survivors or later (Figure 4). Similarly, the maximal cumulative SOFA score appeared to have a stronger association with long-term mortality in the later survivor cohorts than the maximal daily total SOFA score in ICU or the SOFA score on the last day in the ICU (Figure E2).
Figure 4. Comparison of the association with long-term mortality (5-yr mortality after intensive care unit [ICU] admission) of various measures of illness severity and organ failure: Simplified Acute Physiology Score (SAPS) II score, Acute Physiology and Chronic Health Evaluation (APACHE) II score, and cumulative Sequential Organ Failure Assessment (SOFA) score. Point estimates represent the odds ratio for death of the highest compared with the lowest tertile of each score. Error bars represent 95% confidence limits. Categories on the x axis represent the cohort at ICU admission and cohorts surviving to 28, 90, 180, and 365 days after ICU admission.
[More] [Minimize]Compared with the maximal daily total SOFA score, the time to reaching the maximal cumulative SOFA score was longer (mean, 2.7 vs. 3.9 d; Figure E3). When patients experienced the maximal cumulative SOFA score on the final day of ICU admission they were at higher risk of long-term mortality compared with patients for whom it occurred earlier during the ICU stay (OR range, 2.0 to 3.0; Table E8). The association between maximal cumulative SOFA score and long-term mortality remained after adjusting for the time at which it occurred during ICU admission (Figure E4).
We have described the long-term consequences of organ failure after ICU admission for a cohort of patients with a high prevalence of respiratory failure at admission. After adjustment for confounders and other organ failures, liver failure was associated with the highest odds of short- and long-term mortality despite being the least prevalent organ failure. Total burden of organ failure during the ICU stay also predicted short- and long-term mortality, even when we restricted the cohort to those alive 1 year after ICU discharge and adjusted for important comorbidities. The total burden of organ failure was a better predictor of long-term mortality than the APACHE II or SAPS II scores, especially for the longer term survivor cohorts, for whom the admission illness severity scores performed poorly. Our data support a long-term impact on survival from an episode of critical illness, which is greatest for patients who develop more severe organ dysfunction while in ICU.
We studied the effect of organ failure on long-term survival for a large cohort of general ICU patients. We used prospectively collected data rather than undertaking a retrospective analysis of health databases; our data, therefore, are more likely to be accurate. Prospective data collection provided a measure of worst organ failure during ICU stay, rather than being limited to illness severity on ICU admission. This is important given previous studies showing that admission illness severity has poorer predictive value when compared with subsequent organ failure scores, especially for long-term ICU patients (6). We studied a high proportion of general ICU admissions from a representative sample of units in Scotland, suggesting our findings are generalizable to other nonneurological critically ill populations (15). However, the relatively low number of ICU beds in Scotland compared with other countries may mean our study population is more severely ill, reflected by more than 60% requiring mechanical ventilation, and may differ in case mix and outcomes compared with ICU populations in other countries (23). This contrasts with other studies examining long-term consequences of organ failure that were limited to small, single-center studies, or to particular groups of patients, such as those with renal failure (24), trauma (18), or hematological malignancy (25). Although our study is one of the largest to examine this question, it was not adequately powered to examine more homogeneous subgroups, such as patients admitted with trauma or sepsis. The high rates of complete 5-year follow-up (98%) mean the chance of bias due to missing data is small.
We were limited by missing data for some organ failure scores, particularly liver failure, for which 11.8% of the cohort had no recorded score. These were probably patients in whom abnormalities in bilirubin were not expected clinically. However, we undertook sensitivity analyses using a well-recognized method of imputing missing data, which confirmed the findings of the complete case analysis. We were limited to using mortality as the outcome measure because of the retrospective nature of the study. Investigating the relationship between total organ failure burden and other outcome measures, including quality of life and functional status, would be of interest.
Although we were able to adjust for the severe comorbidities collected as part of APACHE II and SAPS II risk models, these are selected to be part of these scores because of their strong association with hospital mortality. Furthermore, we did not have information on other potentially important less severe comorbidities. We cannot, therefore, exclude the possibility that the association we have found is due to residual confounding rather than being causal. A high cumulative SOFA score could have been a marker of milder or unmeasured preexisting comorbidity: a patient with chronic respiratory failure due to chronic obstructive pulmonary disease, chronic liver disease due to cirrhosis, or chronic renal impairment could each attain a SOFA score of at least 2 in each category when stable. Respiratory SOFA scores of 3 or 4 and cardiovascular SOFA scores above 1 require organ support and are therefore less likely to be markers of comorbidity. The only way to resolve whether the association between cumulative SOFA score and mortality is due to confounding would be to accurately measure preexisting comorbidity and adjust the analyses accordingly. One study that investigated the effect of SOFA score and preexisting comorbidity (using the Charlson Comorbidity Index) found that both factors independently predicted 6-month mortality for a cohort of long-term ventilated patients (26). Other studies investigating the relationship between organ dysfunction and functional status (rather than mortality) in non-ICU populations have demonstrated that organ dysfunction remains independently associated with functional status even after controlling for preexisting functional status (27, 28).
We found that the total burden of ICU organ failure was a significant predictor of mortality after adjustment for a number of confounders. There are several possible explanations for this, especially the persisting association after removing patients who died during the first 12 months. First, it is possible that the acute organ failure necessitating ICU admission leads to residual organ damage in those who survive, which decreases long-term survival. For example, patients who develop acute respiratory distress syndrome in ICU may suffer persisting lung fibrosis and respiratory dysfunction (14). Similarly, a high cardiovascular SOFA score may be a marker for occult myocardial damage. Elevated serum troponin has been detected in critically ill patients without myocardial infarction (29) and has been shown to be associated with inotropic support (30) and long-term mortality (30, 31). Second, a high cumulative SOFA score may be a marker for those who are likely to develop chronic ICU-acquired morbidities. These patients often have a protracted recovery, typified by muscle wasting and chronic inflammation (32). Poor functional status has been found to be associated with multiorgan failure measured by SOFA score in a cohort of survivors of trauma (18). Our finding that excess mortality appears to be more pronounced in younger age groups may indicate that the newly acquired morbidities have a greater impact on outcome in a group of patients who would otherwise be expected to have a lower prevalence of comorbidity. Third, excess morbidity may be associated with other long-term consequences of more severe critical illness (9) such as neuropsychological illness (higher prevalence of depression [33], anxiety [34], and posttraumatic stress disorder [35, 36]) or socioeconomic consequences of critical illness such as a reduced income (7, 37). These consequences merit further investigation to ascertain whether they act as possible mediators in the relationship between organ failure burden and long-term mortality.
Total organ failure burden measured on the basis of maximal cumulative SOFA score had a stronger association with long-term mortality in the survivor cohorts compared with other measures of illness severity (APACHE II and SAPS II) or organ failure (maximal daily total SOFA and total SOFA at ICU discharge). By reflecting cumulative organ dysfunction throughout the episode of critical illness, this measure appeared to better capture the impact of the critical illness on subsequent mortality, particularly in those surviving the immediate period post-ICU discharge. The relationship was especially pronounced when the worst score occurred late in the ICU stay, consistent with patients leaving ICU with significant ongoing organ dysfunction. This may contribute to long-term poor health and excess mortality.
Future studies investigating the relationship between organ failure and long-term outcome would benefit from gathering comprehensive information on comorbidities that are present before admission to the ICU as well as those that develop after critical illness. This would help identify the causal mechanisms of ongoing morbidity in survivors of critical illness and determine the degree of residual confounding due to preexisting illness. Furthermore, measuring a wider range of outcomes, including functional status and socioeconomic data, would improve the sensitivity of identifying the consequences of organ failure.
Patients who develop severe organ failure during their ICU stay have an extremely high 5-year mortality, which in the present cohort was 81%. There is an excess mortality attributable to organ failure during ICU admission, which persists during 5 years of follow-up and appears to increase the risk of death well beyond the first few months after ICU admission. Cardiovascular, liver, and respiratory organ failures have the strongest association with long-term mortality.
The authors are grateful to the Scottish Intensive Care Society Audit Group for undertaking data linkage to the Scottish death register. The authors thank the Scottish intensive care units and staff that participated in the ATICS study, and for contributing to the data set used in this analysis.
| 1. | Knaus WA, Draper EA, Wagner DP, Zimmerman JE. Apache II: a severity of disease classification system. Crit Care Med 1985;13:818–829. |
| 2. | Le Gall JR, Lemeshow S, Saulnier F. A new Simplified Acute Physiology Score (SAPS II) based on a European/North American multicenter study. JAMA 1993;270:2957–2963. |
| 3. | Keegan MT, Gajic O, Afessa B. Severity of illness scoring systems in the intensive care unit. Crit Care Med 2011;39:163–169. |
| 4. | Vincent JL, Moreno R, Takala J, Willatts S, De Mendonca A, Bruining H, Reinhart CK, Suter PM, Thijs LG; on behalf of the Working Group on Sepsis-related Problems of the European Society of Intensive Care Medicine. The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. Intensive Care Med 1996;22:707–710. |
| 5. | Ferreira FL, Bota DP, Bross A, Melot C, Vincent JL. Serial evaluation of the SOFA score to predict outcome in critically ill patients. JAMA 2001;286:1754–1758. |
| 6. | Nfor TK, Walsh TS, Prescott RJ. The impact of organ failures and their relationship with outcome in intensive care: analysis of a prospective multicentre database of adult admissions. Anaesthesia 2006;61:731–738. |
| 7. | Angus DC, Carlet J; 2002 Brussels Roundtable Participants. Surviving intensive care: a report from the 2002 Brussels Roundtable. Intensive Care Med 2003;29:368–377. |
| 8. | Rubenfeld GD. Improving clinical trials of long-term outcomes. Crit Care Med 2009;37:S112–S116. |
| 9. | Desai SV, Law TJ, Needham DM. Long-term complications of critical care. Crit Care Med 2011;39:371–379. |
| 10. | Cuthbertson B, Roughton S, Jenkinson D, MacLennan G, Vale L. Quality of life in the five years after intensive care: a cohort study. Crit Care 2010;14:R6. |
| 11. | Williams TA, Dobb GJ, Finn JC, Knuiman MW, Geelhoed E, Lee KY, Webb SAR. Determinants of long-term survival after intensive care. Crit Care Med 2008;36:1523–1530. |
| 12. | Wright JC, Plenderleith L, Ridley SA. Long-term survival following intensive care: subgroup analysis and comparison with the general population. Anaesthesia 2003;58:637–642. |
| 13. | Graf J, Wagner J, Graf C, Koch K-C, Janssens U. Five-year survival, quality of life, and individual costs of 303 consecutive medical intensive care patients—a cost–utility analysis. Crit Care Med 2005;33:547–555. |
| 14. | Herridge MS, Tansey CM, Matte A, Tomlinson G, Diaz-Granados N, Cooper A, Guest CB, Mazer CD, Mehta S, Stewart TE, et al..; Canadian Critical Care Trials Group. Functional disability 5 years after acute respiratory distress syndrome. N Engl J Med 2011;364:1293–1304. |
| 15. | Walsh TS, Garrioch M, Maciver C, Lee RJ, MacKirdy F, McClelland DB, Kinsella J, Wallis C; Audit of Transfusion in Intensive Care in Scotland Study Group. Red cell requirements for intensive care units adhering to evidence-based transfusion guidelines. Transfusion 2004;44:1405–1411. |
| 16. | Tallgren M, Backlund M, Hynninen M. Accuracy of Sequential Organ Failure Assessment (SOFA) scoring in clinical practice. Acta Anaesthesiol Scand 2009;53:39–45. |
| 17. | Arts DG, de Keizer NF, Vroom MB, de Jonge E. Reliability and accuracy of Sequential Organ Failure Assessment (SOFA) scoring. Crit Care Med 2005;33:1988–1993. |
| 18. | Ulvik A, Kvale R, Wentzel-Larsen T, Flaatten H. Multiple organ failure after trauma affects even long-term survival and functional status. Crit Care 2007;11:R95. |
| 19. | Aggarwal AN, Agarwal R, Gupta D, Jindal SK. Nonpulmonary organ dysfunction and its impact on outcome in patients with acute respiratory failure. Chest 2007;132:829–835. |
| 20. | Kendrick S, Clarke J. The Scottish record linkage system. Health Bull (Edinb) 1993;51:72–79. |
| 21. | Sterne JAC, White IR, Carlin JB, Spratt M, Royston P, Kenward MG, Wood AM, Carpenter JR. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. BMJ 2009;338:b2393. |
| 22. | White IR, Royston P, Wood AM. Multiple imputation using chained equations: issues and guidance for practice. Stat Med 2011;30:377–399. |
| 23. | Adhikari NKJ, Fowler RA, Bhagwanjee S, Rubenfeld GD. Critical care and the global burden of critical illness in adults. Lancet 2010;376:1339–1346. |
| 24. | Ahlstrom A, Tallgren M, Peltonen S, Rasanen P, Pettila V. Survival and quality of life of patients requiring acute renal replacement therapy. Intensive Care Med 2005;31:1222–1228. |
| 25. | Cornet AD, Issa AI, van de Loosdrecht AA, Ossenkoppele GJ, Strack van Schijndel RJM, Groeneveld ABJ. Sequential organ failure predicts mortality of patients with a haematological malignancy needing intensive care. Eur J Haematol 2005;74:511–516. |
| 26. | Lee K, Hong SB, Lim CM, Koh Y. Sequential Organ Failure Assessment score and comorbidity: valuable prognostic indicators in chronically critically ill patients. Anaesth Intensive Care 2008;36:528–534. |
| 27. | Wu AW, Damiano AM, Lynn J, Alzola C, Teno J, Landefeld CS, Desbiens N, Tsevat J, Mayer-Oakes A, Harrell FE, et al.. Predicting future functional status for seriously ill hospitalized adults: the support prognostic model. Ann Intern Med 1995;122:342–350. |
| 28. | Clermont G, Angus DC, Linde-Zwirble WT, Griffin MF, Fine MJ, Pinsky MR. Does acute organ dysfunction predict patient-centered outcomes? Chest 2002;121:1963–1971. |
| 29. | Noveanu M, Mebazaa A, Mueller C. Cardiovascular biomarkers in the ICU. Curr Opin Crit Care 2009;15:377–383. |
| 30. | Babuin L, Vasile VC, Rio Perez JA, Alegria JR, Chai H-S, Afessa B, Jaffe AS. Elevated cardiac troponin is an independent risk factor for short- and long-term mortality in medical intensive care unit patients. Crit Care Med 2008;36:759–765. |
| 31. | Lim W, Qushmaq I, Cook DJ, Crowther MA, Heels-Ansdell D, Devereaux PJ, Troponin TTG. Elevated troponin and myocardial infarction in the intensive care unit: a prospective study. Crit Care 2005;9:R636–R644. |
| 32. | Nelson JE, Cox CE, Hope AA, Carson SS. Chronic critical illness. Am J Respir Crit Care Med 2010;182:446–454. |
| 33. | Davydow DS, Gifford JM, Desai SV, Bienvenu OJ, Needham DM. Depression in general intensive care unit survivors: a systematic review. Intensive Care Med 2009;35:796–809. |
| 34. | Davydow DS, Desai SV, Needham DM, Bienvenu OJ. Psychiatric morbidity in survivors of the acute respiratory distress syndrome: a systematic review. Psychosom Med 2008;70:512–519. |
| 35. | Hopkins RO, Weaver LK, Pope D, Orme JF, Bigler ED, Larson-Lohr V. Neuropsychological sequelae and impaired health status in survivors of severe acute respiratory distress syndrome. Am J Respir Crit Care Med 1999;160:50–56. |
| 36. | Wallen K, Chaboyer W, Thalib L, Creedy DK. Symptoms of acute posttraumatic stress disorder after intensive care. Am J Crit Care 2008;17:534–543, quiz 544. |
| 37. | Kalassian K, Angus D. Long-term economic consequences of surviving intensive care. In: Angus DC, Carlet J, editors. Update in intensive care and emergency medicine, no. 39: Surviving intensive care. Berlin: Springer; 2002. |
Author Contributions: Both authors contributed to the conception and design of the study; N.I.L. undertook the analysis; both authors contributed to interpretation of data; both authors contributed to drafting the article or revising it critically for important intellectual content; both authors gave final approval of the version to be published.
Participating units: Aberdeen Royal Infirmary; Borders General Hospital; Glasgow Royal Infirmary; Ninewells Hospital, Dundee; Royal Alexandra Hospital, Paisley; Royal Infirmary, Edinburgh; Southern General Hospital, Glasgow; St. John’s Hospital, Livingston; Western General Hospital, Edinburgh; Western Infirmary, Glasgow. See the online supplement for names of individuals from each participating unit.
This article has an online supplement, which is available from this issue's table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.201201-0059OC on July 26, 2012
Author disclosures
