Abstract
Background: Chronic critical illness (CCI) is a new and increasing entity that accounts for substantial cost despite its low incidence. We hypothesized that patients with end-stage liver failure undergoing liver transplant would be at high risk for developing CCI. With limited liver donors it is essential to understand pre- and peritransplant predictors of CCI.
Methods: To accomplish this we performed a retrospective cohort study at a large academic transplant center of all adult liver transplant patients from 2011 to 2017. We defined CCI as the need for mechanical ventilation for seven days or more post-transplant. Recipients who had re-transplantation during their index admission, acute rejection, or who died during transplant surgery were excluded. Logistic regression was performed using the Akaike information criterion (AIC) and the likelihood ratio test.
Results: We identified 382 transplant recipients. Forty-five (11.8%) developed CCI. Univariable analysis identified 16 pre-transplant factors associated with post-transplant CCI. Subsequent multivariable logistic regression identified eight independent factors associated with CCI in liver transplant recipients including previous liver transplant, acute renal failure, frailty, lower albumin level, higher international normalized ratio, need for mechanical ventilation, and higher systolic pulmonary artery pressure. Pre-transplant factors associated with protection against CCI included higher Model for End-Stage Liver Disease (MELD) score.
Conclusion: The incidence of CCI post-liver transplant is similar to the general population admitted to the intensive care unit. Pre-transplant factors associated with CCI can help identify at-risk patients, and furthermore, promote further research and interventions with the goal to decrease the incidence of CCI in the liver transplant recipients.
Keywords: critical illness, frailty, liver transplantation, mechanical ventilation, prognosis, respiratory insufficiency
Chronic critical illness (CCI) is a newly described disease entity that occurs in patients with a variety of diseases and has a mortality rate that approaches 50%. The term “CCI” was first coined three decades ago in a poignantly subtitled article, “The chronically critically ill: To save or let die?” [1].
In the United States, 5% to 10% of critically ill patients develop CCI. The cost of care is nearly $35 billion annually and 50% will die within three months of discharge from the respiratory care unit [2–4]. Furthermore, only 4% to 8% will be at home and fully functional at one year [5,6]. The persistently poor outcomes highlight the need not only to predict which patients are most likely to develop CCI, but also to appropriately inform, care for, and respond to their life directives.
In this study, we focused on patients with end-stage liver failure undergoing liver transplantation—a population with a unique set of challenges, especially in terms of the allocation of donor livers. Currently, to determine priority, the World Health Organization (WHO) uses the Model for End-Stage Liver Disease (MELD) [7,8] score to identify which liver transplant candidates are closer to death. Consequently, its use has also increased post-transplant morbidity and cost [9]. The burden of CCI post liver-transplant is not known but CCI may impact post-transplant cost and outcomes. We sought to determine the incidence of CCI in liver transplant recipients and to identify pre- and peritransplant predictors for developing CCI post-liver transplant.
Patients and Methods
Study design
We reviewed medical records of all liver transplant recipients 18 years or older at the University of Minnesota from January 2011 through May 2017. Excluded from our study were recipients who underwent multiple transplants during the same admission, those experienced acute rejection, or those who died during their transplant surgery (Fig. 1). Our study was approved by the institutional review board of the University of Minnesota (STUDY00000819).
FIG. 1.
Inclusion and exclusion criteria.
Definitions and outcomes
We defined CCI by the need for mechanical ventilation for seven days or more post-transplant in accordance with a recent randomized controlled trial that evaluated different forms of support and their effects on loved ones of the chronically critically ill [10]. Re-intubations were defined by a 24-hour period free of mechanical ventilation with subsequent need for mechanical ventilation again for more than 24 hours. Patients extubated for less than 24 hours would still be considered a day of requiring mechanical ventilation.
Data collection
We collected the following clinical and demographic data from recipients' electronic medical record: age, gender, past medical history, cause of liver failure (hepatocellular carcinoma, viral hepatitis, alcoholic hepatitis, non-alcoholic steatohepatitis, autoimmune hepatitis, primary biliary cirrhosis, α1-antitrypsin deficiency, Wilson disease, and hemochromatosis), pre-transplant dialysis or mechanical ventilation, mean systolic pulmonary arterial pressure (sPAP) measured prior to transplant by right heart catheterization, laboratory values (bilirubin, international normalized ratio [INR], hemoglobin, creatinine, platelet count, sodium, and albumin), vital signs, pre- and peritransplant complications, MELD score on admission and day of transplant, Braden Scale score on admission (normal ≥18, mild/moderate 13–17, high/severe ≤12), and Prognosis for Prolonged Ventilation (ProVent) score on the day of transplant. Pre- and peritransplant complications included the following: sepsis, acute renal failure, gastrointestinal bleeding, pneumonia, respiratory failure, and higher estimated blood loss during surgery (defined as >2.5 L).
Statistical analysis
For recipients with versus without CCI, we computed summary statistics, including means and standard deviations (for continuous parameters) and counts and percentages (for categorical parameters). We identified clinically pertinent cutpoints in continuous parameters to be used as categorical parameters for our primary analysis. To compare continuous parameters, we used the Student t-test, without assuming equal variances. We used the Fisher exact test to compare categorical parameters.
To evaluate any association between CCI and each parameter, we used a simple logistic regression model for computing the log-odds ratio (OR) including 95% confidence intervals (CI). We subsequently used a multivariable logistic regression model to estimate predictors of CCI by applying forward selection utilizing the Akaike information criterion (AIC) and the likelihood ratio test. We then used re-sampling validation to compute bias-corrected fit.
To compare the results of our simple and multivariable logistic regression models and to estimate the level of importance of each parameter under consideration, we computed a random forest. To estimate each tree, we used five randomly selected parameters; in all, we estimated 15,000 trees. Then, to identify the top five parameters of importance, we used the mean minimal depth and Gini index to estimate the out-of-bag error rate. Our analyses were exploratory; we made no corrections for multiple comparisons.
Results
Of the 382 liver transplant recipients included in our study, 45 (11.8%) developed CCI posttransplant. The mean ages were similar for those with and without CCI (53 and 54 years, respectively; p > 0.05). There was no statistically significant difference in patient characteristics (Table 1). Patients were predominantly white and male. Eighteen patients (5%) required re-intubation during their admission. The mean MELD score for those developing CCI was 23 and 21 for those that did not.
Table 1.
Patient Baseline Characteristics and Pre-Transplant Parameters in Patients with and without Development of Chronic Critical Illness after Liver Transplant
| Variables | Categories | None (n = 338) |
CCI (n = 45) |
p |
|---|---|---|---|---|
| Demographics | ||||
| Race | White | 256 (76.0%) | 32 (71.1%) | 0.71 |
| Black | 30 (8.9%) | 5 (11.1%) | ||
| Other | 51 (15.1%) | 8 (17.8%) | ||
| Age (y) | <45 | 57 (16.9%) | 8 (17.8%) | 0.93 |
| 45–60 | 152 (45.1%) | 21 (46.7%) | ||
| >60 | 128 (38.0%) | 16 (35.6%) | ||
| Gender | Female | 101 (30.0%) | 15 (33.3%) | 0.73 |
| Male | 236 (70.0%) | 30 (66.7%) | ||
| Alcohol | No | 251 (74.5%) | 32 (71.1%) | 0.59 |
| alcohol use | 86 (25.5%) | 13 (28.9%) | ||
| Tobacco | No | 124 (36.8%) | 13 (28.9%) | 0.15 |
| tobacco use | 93 (27.6%) | 19 (42.2%) | ||
| Former tobacco use | 120 (35.6%) | 13 (28.9%) | ||
| Illicit drug use | No | 248 (73.6%) | 32 (71.1%) | 0.72 |
| illicit drug use | 89 (26.4%) | 13 (28.9%) | ||
| Admission variables | ||||
| MELD | <15 | 96 (28.5%) | 11 (24.4%) | 0.57 |
| 15–30 | 119 (35.3%) | 14 (31.1%) | ||
| >30 | 122 (36.2%) | 20 (44.4%) | ||
| Braden Score | Good | 237 (70.3%) | 19 (42.2%) | <0.001 |
| Malnourished | 77 (22.8%) | 19 (42.2%) | ||
| Severely Malnourished | 23 (6.8%) | 7 (15.6%) | ||
| BMI | Obese | 143 (42.4%) | 26 (57.8%) | 0.27 |
| Overweight | 92 (27.3%) | 9 (20.0%) | ||
| Normal | 79 (23.4%) | 9 (20.0%) | ||
| Underweight | 23 (6.8%) | 1 (2.2%) | ||
| sPAP (mm Hg) | < 25 | 66 (19.6%) | 9 (20.0%) | 0.007 |
| 25–45 | 233 (69.1%) | 23 (51.1%) | ||
| > 45 | 38 (11.3%) | 13 (28.9%) | ||
| Pre-transplant location | Outpatient | 133 (39.5%) | 13 (28.9%) | 0.02 |
| Floor | 168 (49.9%) | 20 (44.4%) | ||
| ICU | 36 (10.7%) | 12 (26.7%) | ||
| Number of transplants | 1st | 287 (85.2%) | 30 (66.7%) | <0.001 |
| 2nd | 44 (13.1%) | 7 (15.6%) | ||
| 3rd | 4 (1.2%) | 4 (8.9%) | ||
| 4th | 2 (0.6%) | 4 (8.9%) | ||
| Hospital day of transplant (d) | <4 | 252 (74.8%) | 29 (64.4%) | 0.02 |
| 4–14 | 67 (19.9%) | 8 (17.8%) | ||
| >14 | 18 (5.3%) | 8 (17.8%) | ||
| Dialysis pre-transplant | No | 317 (94.1%) | 32 (71.1%) | <0.001 |
| Dialysis | 20 (5.9%) | 13 (28.9%) | ||
| Mechanical ventilation pre-transplant | No | 313 (92.9%) | 32 (71.1%) | <0.001 |
| MV | 24 (7.1%) | 13 (28.9%) | ||
| Hospital LOS, mean (SD) | <0.001 | |||
| Total LOS (d) | 16.7 (13.3) | 52.7 (49.6) | ||
| Post-transplant LOS (d) | 13.7 (9.8) | 46.1 (49.3) | ||
| Cirrhosis cause | 0.61 | |||
| Alcoholic hepatitis | 147 (43.6%) | 17 (37.8%) | ||
| Hepatocellular carcinoma | 19 (5.6%) | 2 (4.4%) | ||
| Non-alcoholis steatohepatitis | 28 (8.3%) | 3 (6.7%) | ||
| Viral hepatitis | 115 (34.1%) | 16 (35.6%) | ||
| Other | 28 (8.3%) | 7 (15.6%) | ||
| Past medical history | ||||
| HE | None | 174 (51.8%) | 15 (33.3%) | 0.03 |
| HE | 162 (48.2%) | 30 (66.7%) | ||
| HRS | none | 331 (99.1%) | 43 (95.6%) | 0.11 |
| HRS | 3 (0.9%) | 2 (4.4%) | ||
| PVT | none | 322 (96.4%) | 43 (95.6%) | 0.68 |
| PVT | 12 (3.6%) | 2 (4.4%) | ||
| SBP | None | 308 (92.2%) | 44 (97.8%) | 0.23 |
| SBP | 26 (7.8%) | 1 (2.2%) | ||
| Varices | None | 251 (75.1%) | 34 (75.6%) | 1 |
| Varices | 83 (24.9%) | 11 (24.4%) | ||
| AFib | None | 314 (94.0%) | 40 (88.9%) | 0.20 |
| AFib | 20 (6.0%) | 5 (11.1%) | ||
| COPD | None | 318 (95.2%) | 43 (95.6%) | 1 |
| COPD | 16 (4.8%) | 2 (4.4%) | ||
| CKD | None | 309 (92.5%) | 36 (80.0%) | 0.01 |
| CKD | 25 (7.5%) | 9 (20.0%) | ||
| DM | None | 235 (70.4%) | 31 (68.9%) | 0.86 |
| DM | 99 (29.6%) | 14 (31.1%) | ||
| Heart failure | None | 323 (96.7%) | 44 (97.8%) | 1 |
| Heart failure | 11 (3.3%) | 1 (2.2%) | ||
| Complications (prior to transplant) | ||||
| Acute hemorrhage | None | 331 (98.2%) | 43 (95.6%) | 0.24 |
| Acute hemorrhage | 6 (1.8%) | 2 (4.4%) | ||
| Acute renal failure | None | 233 (69.1%) | 15 (33.3%) | <0.001 |
| AKI | 104 (30.9%) | 30 (66.7%) | ||
| GI bleed | None | 314 (93.2%) | 37 (82.2%) | 0.02 |
| GI bleed | 23 (6.8%) | 8 (17.8%) | ||
| Pneumonia | None | 334 (99.1%) | 42 (93.3%) | 0.02 |
| Pneumonia | 3 (0.9%) | 3 (6.7%) | ||
| Pulmonary failure | None | 284 (84.3%) | 27 (60.0%) | <0.001 |
| Pulmonary failure | 53 (15.7%) | 18 (40.0%) | ||
| Sepsis | None | 317 (94.1%) | 34 (75.6%) | <0.001 |
| Sepsis | 20 (5.9%) | 11 (24.4%) | ||
| VTE | None | 323 (95.8%) | 40 (88.9%) | 0.06 |
| VTE | 14 (4.2%) | 5 (11.1%) | ||
| Transplant day laboratory tests | ||||
| MELD | Mean (SD) | 21.0 (9.8) | 22.7 (10.2) | 0.29 |
| <15 | 96 (28.5%) | 13 (28.9%) | 0.70 | |
| 15–30 | 154 (45.7%) | 18 (40.0%) | ||
| >30 | 87 (25.8%) | 14 (31.1%) | ||
| Albumin | >3 | 40 (11.9%) | 1 (2.2%) | 0.03 |
| Low (2–3) | 192 (57.0%) | 23 (51.1%) | ||
| Very low (<2) | 105 (31.2%) | 21 (46.7%) | ||
| Bilirubin total | <2 | 95 (28.2%) | 10 (22.2%) | 0.67 |
| 2–4 | 76 (22.6%) | 12 (26.7%) | ||
| >4 | 166 (49.3%) | 23 (51.1%) | ||
| Creatinine | Normal | 250 (74.2%) | 36 (80.0%) | 0.81 |
| >2 | 63 (18.7%) | 7 (15.6%) | ||
| >4 | 24 (7.1%) | 2 (4.4%) | ||
| INR | Normal | 208 (61.7%) | 23 (51.1%) | 0.20 |
| >2 | 129 (38.3%) | 22 (48.9%) | ||
| Sodium | Normal | 204 (60.5%) | 28 (62.2%) | 0.08 |
| <135 | 128 (38.0%) | 14 (31.1%) | ||
| >145 | 5 (1.5%) | 3 (6.7%) | ||
| Platelets | >75 | 55 (16.3%) | 4 (8.9%) | 0.27 |
| <75 | 282 (83.7%) | 41 (91.1%) | ||
| Transplant complications | ||||
| Estimated blood loss (mL) | <1000 | 51 (15.1%) | 1 (2.2%) | 0.02 |
| 1000–2500 | 64 (19.0%) | 6 (13.3%) | ||
| >2500 | 222 (65.9%) | 38 (84.4%) |
CCI = chronic critical illness; MELD = Model for End-Stage Liver Disease; BMI = body mass index; sPAP = systolic pulmonary artery pressure; SD = standard deviation; LOS = length of stay; GI = gastrointestinal; INR = international normalized ratio; HE = hepatic encephalopathy; HRS = hepatic renal syndrome; PVT = portal venous thrombosis; SBP = spontaneous bacterial peritonitis; AFib = atrial fibrillation; COPD = chronic obstructive pulmonary disease; CKD = chronic kidney disease; DM = diabetes mellitus; AKI = acute kidney injury; VTE = venous thromboembolism.
Of the 45 recipients with CCI, 26 (57.8%) had a mild/moderate or high/severe Braden Scale score (mean, 16.4 ± 3.9) vs. 100 (29.6%) without CCI (mean, 18.8 ± 3.3), p < 0.001. Mean systolic pulmonary artery pressure (sPAP) in patients who developed CCI was 45.8 ± 42 mm Hg vs. 36.8 ± 18.7 for those who did not (p = 0.007). For patients with CCI versus without: a total of 32 (71.1%) versus 204 (60.6%) had been hospitalized before their transplant admission, (p = 0.02), 15 (33.4%) versus 50 (14.9%) had undergone at least 1 previous liver transplant (p < 0.001), 13 (28.9%) versus 20 (5.9%) had needed pre-transplant dialysis (p < 0.001), 13 (28.9%) versus 24 (7.1%) had needed pre-transplant mechanical ventilation (p < 0.001), and 16 (35.6%) versus 85 (25.1%) had undergone their transplant 96 hours after their admission (p = 0.02). Discharge data were only available on a subset of patients (n = 54). For patients with CCI versus without: 0 versus 4 (9.1%) were discharged home (p = 1.0) and 7 (70%) versus 4 (9.1%) were discharged to a long-term acute care hospital (p < 0.001).
Per our univariable analysis, the incidence of the following pre- and peritransplant complications was different for recipients with versus without CCI (the rates in parentheses below are for those with CCI vs. without): sepsis (24.4% vs. 5.9%, p < 0.001), acute renal failure (66.7% vs. 30.9%, p < 0.001), gastrointestinal bleeding (17.8% vs. 6.8%, p = 0.01), pneumonia (6.7% vs. 0.9%, p = 0.02), and higher estimated blood loss during surgery (84.4% vs. 65.9% p < 0.001). In addition, those with (versus without) CCI had lower albumin (mean, 2.2 vs. 2.5 g/dL on admission, p = 0.01; 2.1 vs. 2.3 g/dL on the day of transplant, p = 0.02), higher sodium (mean, 137 vs. 135 mmol/L on the day of transplant, p < 0.001), and lower platelet count (mean, 38.7 vs. 51.5 × 109/L on the day of transplant, p < 0.001). Laboratory parameters were also categorized to more clinically relevant groups and only albumin remained statistically significant on admission (p = 0.03) (Table 1.)
We identified eight pretransplant parameters using forward selection via AIC and likelihood ratio (LR) tests (Table 2). These eight parameters were included in the final multivariable analysis listed in order of selection: acute renal failure (OR 11.7, 95% CI 4.2–36.6, p < 0.001) at least one previous liver transplant (OR 16.4, 95% CI 4.6–67.8, p < 0.001), the need for mechanical ventilation (OR 2.7, 95% CI 1–7.6, p = 0.05), lower Braden Scale score on admission (OR 3.9, 95% CI 1.1–13.7, p = 0.03), higher sPAP (OR 1.5, 95% CI 0.5–5.2, p = 0.4), lower MELD score on day of transplant (OR 0.02, 95% CI 0.002–0.1, p < 0.001), lower albumin level on admission (OR 7.3, 95% CI 2.1–32.6, p = 0.004), and higher day of transplant INR (OR 3.1, 95% CI 1.2–8.4, p = 0.02). Our final multivariable model had a resampling calibrated bias-corrected C statistic or area under the curve (AUC) of 0.87 (Table 3).
Table 2.
Parameters Associated with Chronic Critical Illness
| Item added | AIC | LR χ2 | DF | p |
|---|---|---|---|---|
| Acute renal failure pre transplant | 259.8 | 21.2 | 1 | <0.001 |
| Number of transplants | 247.5 | 16.3 | 2 | <0.001 |
| Pretransplant ventilator needs | 241.7 | 7.8 | 1 | 0.005 |
| Albumin (category) | 238.4 | 7.2 | 2 | 0.03 |
| MELD (transplant) | 228.8 | 13.7 | 2 | 0.001 |
| Braden Score (category) | 225.1 | 7.7 | 2 | 0.02 |
| sPAP (category) | 220.8 | 8.3 | 2 | 0.02 |
| INR | 217.4 | 5.3 | 1 | 0.02 |
| Past Medical History HEa | 215.5 | 3.7 | 1 | 0.06 |
The multivariable model below was built using forward selection via the AIC and LR tests. The name in the left column is each variable added. Acute renal failure was added to an intercept only model for the first row, then number of transplants was added to the model with Acute renal failure and the AIC, LR statistic, and p value from the LR test are shown, etc.
The last row with Past medical history did not have a statistically significant LR test so it was not included in the final model and forward selection was stopped.
AIC = Akaike information criterion; LR = likelihood ratio; DF = degrees of freedom; MELD = ; Model for End-Stage Liver Disease; sPAP = systolic pulmonary artery pressure; INR = international normalized ratio; HE = hepatic encephalopathy.
Table 3.
Odds Ratios and 95% Confidence Intervals for Multivariable Model for Development of Chronic Critical Illness after Liver Transplant
| Odds ratio | Lower CI | Upper CI | |
|---|---|---|---|
| Intercept | 0.02 | 0.004 | 0.09 |
| ↓ Braden Score, mild | 3.71 | 1.52 | 9.22 |
| ↓ Braden Score, severe | 3.98 | 1.12 | 13.67 |
| ↑ MELD (15–30) | 0.13 | 0.04 | 0.43 |
| ↑ MELD (>30) | 0.02 | 0.002 | 0.11 |
| ↓ Albumin, Low (2–3) | 4.2 | 1.24 | 18.04 |
| ↓ Albumin, very low (<2) | 7.32 | 2.05 | 32.65 |
| Mechanical ventilation | 2.76 | 0.99 | 7.61 |
| Transplant number, 2nd | 1.46 | 0.48 | 4.04 |
| Transplant number, 3rd+ | 16.38 | 4.27 | 67.75 |
| Acute renal failure | 11.66 | 4.21 | 36.56 |
| ↑ sPAP, mild/moderate | 0.41 | 0.15 | 1.17 |
| ↑ sPAP, severe | 1.57 | 0.49 | 5.16 |
| ↑ INR (>2) | 3.1 | 1.189 | 8.39 |
The final multivariable model is shown above. The C-statistic or area under the curve (AUC) was 0.872. Braden Score, albumin, sPAP, were all on admission, MELD and INR were calculated based on laboratory tests the day of transplant.
sPAP = systolic pulmonary artery pressure; MELD = ; Model for End-Stage Liver Disease; INR = international normalized ratio.
Discussion
Our purpose was to identify parameters that predict patients most likely to develop CCI. In this population of liver transplant patients, our multivariable model identified acute renal failure before transplant, prior liver transplant, need for mechanical ventilation before transplant, a lower Braden Scale score on admission, higher sPAP, a lower MELD score, a lower albumin level on admission, and a higher INR on the day of transplant as predictors of developing CCI.
The definition of CCI remains vague and undefined but is characterized by the body's maladaptive, long-term response to illness. Mechanical ventilation is often used as a surrogate marker of CCI but the required duration ranges from 96 hours to 21 days [1,10–13]. Other definitions have also been described using administrative data (Disease Related Group 483/541/542 [tracheostomy] or International Classification of Diseases Version 9.0—Clinical Modification procedure code 96.72 [mechanical ventilation >4 days] along with intensive care unit [ICU] length of stay) or organ dysfunction with mechanical ventilation needs [14–16]. In our study, we defined CCI as the need for mechanical ventilation for seven or more days post-transplant, consistent with a recent randomized controlled trial [10].
Chronic critical illness is associated with high morbidity and mortality. It is also expensive. In the United States, the median hospital charge for such patients is more than $120,000, with an annual cost of $22 billion in 2004 and nearly $35 billion currently [3,17]. Those numbers are staggering, especially because 50% of patients with CCI die within three months after leaving the respiratory care unit and 70% within one year [4,18]. The incidence and cost justify immediate attention and research. Thus far research has primarily focused on defining which patients have and are at risk for CCI.
Understanding the predictors of CCI can improve patient care and enhance communication with patients and their loved ones. Previously identified predictors of CCI include: higher body mass index, prolonged mechanical ventilation, sepsis, an abnormal Glasgow Coma Scale score, and inadequate nutrition [19]. To our knowledge, only 14 prior studies have examined risk factors for prolonged mechanical ventilation; furthermore, ours is the first to examine predictors of CCI in liver transplant recipients specifically [20].
The overlap between liver failure and CCI is large. Liver transplant recipients are an ideal population for evaluating factors that contribute to CCI. Their pretransplant condition can highlight physiologic derangements that should be improved, if not be ameliorated, with the intervention of a new liver. By assessing the non-transplant CCI population, we can start to uncover underlying issues that lead to the inability to recover from liver transplants.
Worldwide, chronic liver disease is the twelfth most common cause of death; it is the fifth most common cause in certain age groups [21,22]. The treatment for end-stage liver disease is transplantation. In an ever-evolving effort to allocate the limited supply of donor livers fairly to the sickest transplant candidates, scoring systems have been devised, such as the Child-Turcotte-Pugh classification [23], the Chronic Liver Failure-Sequential Organ Failure Assessment (CLIF-SOFA) tool [24], and the MELD score [8]. Unfortunately, those scoring systems have proven to be poor at predicting post-transplant outcomes [25,26]. Only a few studies have evaluated prediction models for post-transplant mortality; fewer yet have predicted post-transplant morbidity. Most of this limited literature is based on variations of the already defined severity scores [25,27]. To date, the literature evaluating pre-transplant predictors of CCI remains scant. In light of the increase in post-transplant morbidity after the introduction of the MELD score, studying such predictors is more necessary than ever [28].
In patients with cirrhosis, malnutrition has been associated with mortality [29]. The Braden Scale is an indicator of nutritional status originally created to help predict the risk of developing pressure ulcers; more recently it has been extended to help predict overall frailty and nutritional status [30,31]. The Braden Scale score has been used as an independent predictor of, in older post-operative patients, such outcomes as the complication rate, hospital length of stay (LOS), and type of discharge facility; in heart failure patients, such outcomes as mortality, hospital LOS, and type of discharge facility [32,33]. We used the Braden Scale score to focus on the general frailty of our cohort and found an association with the development of CCI. Other markers of nutritional status include albumin levels. Recent studies have found that serum albumin is a better predictor of persistent malnutrition posttransplant than global assessments or body mass index [34]. Our results mimic those findings: the albumin level had a higher OR in our prediction model and thus was superior to the Braden Scale score. This result is not surprising, given that CCI has recently been proposed to be a syndrome, with systemic sequelae that stem from the continuous catabolic state [35].
Pulmonary hypertension and an elevated transpulmonary gradient are associated with increased mortality in liver transplant recipients [36]. In contrast, recent studies of pre-transplant echocardiography did not demonstrate any association between pulmonary hypertension and mortality [37,38]. Pulmonary hypertension was associated with substantial morbidities, such as an increased risk of hospitalization for myocardial infarction or heart failure, increased ICU LOS, increased hospital LOS, pulmonary complications, and increased duration of mechanical ventilation. Although sPAP has not been identified as an independent factor for CCI, we noted a substantial point estimate and therefore included sPAP in our final model. Further investigation involving a larger number of patients is needed.
In our study, we found acute renal failure pre-transplant and the need for dialysis pre-transplant were both strong predictors of CCI post-transplant. Kidney function, both before and after a liver transplant has a major effect on patient outcomes (including mortality, liver graft failure, post-transplant renal failure, and resource utilization) [39–41]. Most previous studies of liver transplant recipients have looked at long-term mortality (up to three to five years) in association with renal failure; the high mortality rate of recipients with CCI would have been part of the signal for a higher mortality rate posttransplant. [42,43].
Re-transplants have been associated with improved outcomes in the earlier literature but were largely influenced by acute indications, such as acute rejection or liver graft failure [44,45]. In our study, after excluding acute rejection, we found retransplants to be an independent predictor of CCI. This is more in line with recent studies that have found retransplants to be associated with increased hospital LOS, higher resource utilization, and decreased overall survival rates at one and five years [46–48]. We recognize the inevitably emergent nature of retransplants for acute rejection, rendering them impractical in any prediction model. Therefore, we excluded all recipients who needed retransplants during their admission. The ethical controversy about retransplants will continue, but it should be noted that our findings suggest an association between the number of previous transplants and CCI.
Interestingly, we found a higher MELD score was associated with protection against CCI. Originally, the MELD score was developed to predict outcomes after transjugular intrahepatic portosystemic shunt (TIPS) placement [8,49]. Later, its ability to predict mortality led to its use to prioritize organ allocation, which led to an increase in post-transplant morbidity (300%) and overall cost (55%) [9,28]. Because the goal of scoring systems is to determine which patients are sickest, they inherently increase the risk of post-operative morbidity. Studies evaluating transplant outcomes found higher MELD scores correlated with increased ICU LOS, increased hospital LOS, and increased duration of mechanical ventilation [37]. We hypothesize that patients with high MELD scores do not develop CCI because they possess the reserve to survive the severe physiologic derangements of liver failure and transplant, whereaw those susceptible to CCI do not survive as the MELD increases. In short, they die of liver failure before they receive a transplant and are allowed to develop CCI.
Limitations
Our study was exploratory and retrospective in design. Our single-center and relatively small sample decreased the power of our statistical analysis and may not be broadly generalizable. The data, obtained from the hospitals' electronic health record, rely on administrative data, diagnosis codes, and data manually entered into the chart by healthcare providers. Our study excluded 12 patients who died within 24 hours, had acute rejection, or received multiple transplants in order to adjust for potential perioperative complications. There may be competing risk in these patients because they may have been sicker and/or more likely to develop CCI; however, the concern for confounding outweighed the potential bias. Our data will need to be validated with a much larger database of similar liver transplant recipients.
Defining CCI is still in flux. Since 2004, when a consensus definition of CCI [11] as mechanical ventilation for 21 days or more was chosen, it has become clear this is an under-representation of the population with CCI. Furthermore, some CCI experts have noted that there are patients with complex respiratory complications who require prolonged mechanical ventilation but do not truly have CCI [15]. Our definition was chosen based on more recent control trials and reviews, however, as research and awareness increases, the definition of CCI will be further refined.
We chose liver transplant recipients because their complex physiologic derangements should improve post-transplant. The deficiencies not resolved by their transplant highlighted pre-transplant factors that represented their underlying inability to recover. In view of the limitations of our study, future work will need to involve other types of transplant recipients and larger ICU databases. Predictors of CCI will likely vary in different populations with different physiologic derangements.
Furthermore, predictors may highlight areas for modification prior to transplant. Renal failure, respiratory failure, and nutrition are important in any patient who is critically ill. In pre-transplant patients there is a fine line between sick enough and too sick. These parameters are crucial as they contribute to the decreased physiologic reserve. It remains difficult to determine transient physiologic derrangements from liver failure from those that are causing prolonged or even unrecoverable damage. Goals will continue to include aggressively preventing renal and pulmonary failure, promoting optimal nutrition and functional status when able in our wait-listed patients.
The complexity of CCI is apparent when clinicians and family members discuss the prognosis of an individual patient who remains on mechanical ventilation for a week or more. With better information to guide such discussions, everyone's expectations would be more accurate, and truly informed decisions would be more probable.
Conclusion
Pre-transplant predictors of CCI in our liver transplant recipients included a previous liver transplant, acute renal failure, frailty, a lower albumin level, a higher INR, the need for mechanical ventilation, and higher sPAP. Pre-transplant factors associated with protection against CCI included a higher MELD score. In an endless pursuit to decrease the mortality of our patients, we must remember that a subset of the patients saved may have a longer life but with limited quality of life. These patients are known as the chronically critically ill. Our findings should catalyze research to further define CCI, thereby helping identify at-risk patients and, ultimately, enabling us to fine-tune appropriate interventions and goals of care.
Acknowledgments
All authors were instrumental in study creation, data analysis, and manuscript writing and editing.
Funding Information
Funded by the University of Minnesota's Critical Care Research and Programmatic Development Program. N.E.I. is supported by NIH NHLBI T32HL07741.
Research reported in this publication was supported by the National Center for Advancing Translational Sciences of the National Institutes of Health, Award Number UL1TR000114. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Author Disclosure Statement
No competing financial interests exist for any author.
References
- 1. Girard K, Raffin TA. The chronically critically ill: To save or let die? Respir Care 1985;30:339–347 [PubMed] [Google Scholar]
- 2. Carson SS, Bach PB. The epidemiology and costs of chronic critical illness. Crit Care Clin 2002;18:461–476 [DOI] [PubMed] [Google Scholar]
- 3. Kahn JM, Le T, Angus DC, et al. The epidemiology of chronic critical illness in the United States. Crit Care Med 2015;43:282–287 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Nelson JE, Meier DE, Litke A, et al. The symptom burden of chronic critical illness. Crit Care Med 2004;32:1527–1534 [DOI] [PubMed] [Google Scholar]
- 5. Nelson JE, Cox CE, Hope AA, Carson SS. Chronic critical illness. Am J Respir Crit Care Med 2010;182:446–454 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Carson SS, Bach PB, Brzozowski L, Leff A. Outcomes after long-term acute care. An analysis of 133 mechanically ventilated patients. Am J Respir Crit Care Med 1999;159(5 Pt 1):1568–1573 [DOI] [PubMed] [Google Scholar]
- 7. Wiesner R, Edwards E, Freeman R, et al. Model for end-stage liver disease (MELD) and allocation of donor livers. Gastroenterology 2003;124:91–96 [DOI] [PubMed] [Google Scholar]
- 8. Kamath PS, Wiesner RH, Malinchoc M, et al. A model to predict survival in patients with end-stage liver disease. Hepatology 2001;33:464–470 [DOI] [PubMed] [Google Scholar]
- 9. Weismuller TJ, Fikatas P, Schmidt J, et al. Multicentric evaluation of model for end-stage liver disease-based allocation and survival after liver transplantation in Germany—Limitations of the 'sickest first'-concept. Transpl Int 2011;24:91–99 [DOI] [PubMed] [Google Scholar]
- 10. Carson SS, Cox CE, Wallenstein S, et al. Effect of palliative care-led meetings for families of patients with chronic critical illness: A randomized clinical trial. JAMA 2016;316:51–62 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. MacIntyre NR, Epstein SK, Carson S, et al. Management of patients requiring prolonged mechanical ventilation: Report of a NAMDRC consensus conference. Chest 2005;128:3937–3954 [DOI] [PubMed] [Google Scholar]
- 12. Spicher JE, White DP. Outcome and function following prolonged mechanical ventilation. Arch Intern Med 1987;147:421–425 [PubMed] [Google Scholar]
- 13. Zilberberg MD, de Wit M, Shorr AF. Accuracy of previous estimates for adult prolonged acute mechanical ventilation volume in 2020: Update using 2000–2008 data. Crit Care Med 2012;40:18–20 [DOI] [PubMed] [Google Scholar]
- 14. Kahn JM, Benson NM, Appleby D, et al. Long-term acute care hospital utilization after critical illness. JAMA 2010;303:2253–2259 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Carson SS. Definitions and epidemiology of the chronically critically ill. Respir Care 2012;57:848–856 [DOI] [PubMed] [Google Scholar]
- 16. Efron PA, Mohr AM, Bihorac A, et al. Persistent inflammation, immunosuppression, and catabolism and the development of chronic critical illness after surgery. Surgery 2018;164:178–184 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Cox CE, Carson SS, Govert JA, et al. An economic evaluation of prolonged mechanical ventilation. Crit Care Med 2007;35:1918–1927 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Hollander JM, Mechanick JI. Nutrition support and the chronic critical illness syndrome. Nutr Clin Pract 2006;21:587–604 [DOI] [PubMed] [Google Scholar]
- 19. Loss SH, Marchese CB, Boniatti MM, et al. Prediction of chronic critical illness in a general intensive care unit. Rev Assoc Med Bras 2013;59:241–247 [DOI] [PubMed] [Google Scholar]
- 20. Shin HJ, Chang JS, Ahn S, et al. Clinical factors associated with weaning failure in patients requiring prolonged mechanical ventilation. J Thorac Dis 2017;9:143–150 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Scaglione S, Kliethermes S, Cao G, et al. The epidemiology of cirrhosis in the United States: A population-based study. J Clin Gastroenterol 2015;49:690–696 [DOI] [PubMed] [Google Scholar]
- 22. Olson JC. Acute-on-chronic and decompensated chronic liver failure: Definitions, epidemiology, and prognostication. Crit Care Clin 2016;32:301–309 [DOI] [PubMed] [Google Scholar]
- 23. Pugh RN, Murray-Lyon IM, Dawson JL, et al. Transection of the oesophagus for bleeding oesophageal varices. Br J Surg 1973;60:646–649 [DOI] [PubMed] [Google Scholar]
- 24. Moreau R, Jalan R, Gines P, et al. Acute-on-chronic liver failure is a distinct syndrome that develops in patients with acute decompensation of cirrhosis. Gastroenterology 2013;144:1426–1437 [DOI] [PubMed] [Google Scholar]
- 25. Cholongitas E, Marelli L, Shusang V, et al. A systematic review of the performance of the model for end-stage liver disease (MELD) in the setting of liver transplantation. Liver Transpl 2006;12:1049–1061 [DOI] [PubMed] [Google Scholar]
- 26. Brown RS Jr, Kumar KS, Russo MW, et al. Model for end-stage liver disease and Child-Turcotte-Pugh score as predictors of pretransplantation disease severity, posttransplantation outcome, and resource utilization in United Network for Organ Sharing status 2A patients. Liver Transpl 2002;8:278–284 [DOI] [PubMed] [Google Scholar]
- 27. Bruns H, Lozanovski VJ, Schultze D, et al. Prediction of postoperative mortality in liver transplantation in the era of MELD-based liver allocation: A multivariate analysis. PLoS One 2014;9:e98782. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Dutkowski P, Oberkofler CE, Bechir M, et al. The model for end-stage liver disease allocation system for liver transplantation saves lives, but increases morbidity and cost: A prospective outcome analysis. Liver Transpl 2011;17:674–684 [DOI] [PubMed] [Google Scholar]
- 29. Sam J, Nguyen GC. Protein-calorie malnutrition as a prognostic indicator of mortality among patients hospitalized with cirrhosis and portal hypertension. Liver Int 2009;29:1396–1402 [DOI] [PubMed] [Google Scholar]
- 30. Brown SJ. The Braden Scale. A review of the research evidence. Orthop Nurs 2004;23:30–38 [DOI] [PubMed] [Google Scholar]
- 31. Bergstrom N, Braden BJ, Laguzza A, Holman V. The Braden scale for predicting pressure sore risk. Nurs Res 1987;36:205–210 [PubMed] [Google Scholar]
- 32. Cohen RR, Lagoo-Deenadayalan SA, Heflin MT, et al. Exploring predictors of complication in older surgical patients: A deficit accumulation index and the Braden Scale. J Am Geriatr Soc 2012;60:1609–1615 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Bandle B, Ward K, Min SJ, et al. Can Braden score predict outcomes for hospitalized heart failure patients? J Am Geriatr Soc 2017;65:1328–1332 [DOI] [PubMed] [Google Scholar]
- 34. Wagner D, Adunka C, Kniepeiss D, et al. Serum albumin, subjective global assessment, body mass index and the bioimpedance analysis in the assessment of malnutrition in patients up to 15 years after liver transplantation. Clin Transplant 2011;25:E396–400 [DOI] [PubMed] [Google Scholar]
- 35. Marchioni A, Fantini R, Antenora F, et al. Chronic critical illness: The price of survival. Eur J Clin Invest 2015;45:1341–1349 [DOI] [PubMed] [Google Scholar]
- 36. Ramsay MA, Simpson BR, Nguyen AT, et al. Severe pulmonary hypertension in liver transplant candidates. Liver Transpl Surg 1997;3:494–500 [DOI] [PubMed] [Google Scholar]
- 37. Kia L, Shah SJ, Wang E, et al. Role of pretransplant echocardiographic evaluation in predicting outcomes following liver transplantation. Am J Transplant 2013;13:2395–2401 [DOI] [PubMed] [Google Scholar]
- 38. Bushyhead D, Kirkpatrick JN, Goldberg D. Pretransplant echocardiographic parameters as markers of posttransplant outcomes in liver transplant recipients. Liver Transpl 2016;22:316–323 [DOI] [PubMed] [Google Scholar]
- 39. Markmann JF, Markmann JW, Markmann DA, et al. Preoperative factors associated with outcome and their impact on resource use in 1148 consecutive primary liver transplants. Transplantation 2001;72:1113–122 [DOI] [PubMed] [Google Scholar]
- 40. Nair S, Verma S, Thuluvath PJ. Pretransplant renal function predicts survival in patients undergoing orthotopic liver transplantation. Hepatology 2002;35:1179–1185 [DOI] [PubMed] [Google Scholar]
- 41. Suleymanlar I, Yilmaz VT, Kocak H, et al. The effect of kidney diseases on survival in liver transplant patients. Int Urol Nephrol 2011;43:827–833 [DOI] [PubMed] [Google Scholar]
- 42. O'Leary JG, Levitsky J, Wong F, et al. Protecting the kidney in liver transplant candidates: Practice-based recommendations from the American Society of Transplantation Liver and Intestine Community of Practice. Am J Transplant 2016;16:2516–2531 [DOI] [PubMed] [Google Scholar]
- 43. Narayanan Menon KV, Nyberg SL, Harmsen WS, et al. MELD and other factors associated with survival after liver transplantation. Am J Transplant 2004;4:819–825 [DOI] [PubMed] [Google Scholar]
- 44. Saito S, Langnas AN, Stratta RJ, et al. Hepatic retransplantation: University of Nebraska Medical Center experience. Clin Transplant 1992;6:430–435 [PubMed] [Google Scholar]
- 45. Shaw BW, Gordon RD, Iwatsuki S, Starzl TE. Hepatic retransplantation. Transplant Proc 1985;17:264–271 [PMC free article] [PubMed] [Google Scholar]
- 46. D'Alessandro AM, Ploeg RJ, Knechtle SJ, et al. Retransplantation of the liver—A seven-year experience. Transplantation 1993;55:1083–1087 [DOI] [PubMed] [Google Scholar]
- 47. Powelson JA, Cosimi AB, Lewis WD, et al. Hepatic retransplantation in New England—A regional experience and survival model. Transplantation 1993;55:802–806 [DOI] [PubMed] [Google Scholar]
- 48. Azoulay D, Linhares MM, Huguet E, et al. Decision for retransplantation of the liver: An experience- and cost-based analysis. Ann Surg 2002;236:713–721 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. Malinchoc M, Kamath PS, Gordon FD, et al. A model to predict poor survival in patients undergoing transjugular intrahepatic portosystemic shunts. Hepatology 2000;31:864–871 [DOI] [PubMed] [Google Scholar]