Chapter 19. Long-Term Outcomes of Critical Illness

• Survivors of critical illness experience decreased health-related quality of life due to physical limitations, depression and anxiety, and cognitive impairments.
• There may be irreversible long-term neuromuscular dysfunction (e.g., muscle weakness, critical illness polyneuropathy, and myopathy).
• Other organ dysfunction (e.g., pulmonary) is present following critical illness but does not appear to have the same impact on patients' self-reported quality-of-life outcomes as other morbidities.
• Hypoxia and delirium are risk factors for poor long-term outcome resulting from cognitive impairments.
• Approximately one-third to one-half of survivors of critical illness will develop long-term cognitive impairments.
• Recent reports suggest that exercise capacity and cognitive function plateau at a lower than normal level at 1 year with limited improvement 2 years following ICU discharge.
• Long-term physical and neuropsychological dysfunction may be remediable through the implementation of a multidisciplinary and family-centered rehabilitation program. This is currently being evaluated.

In the United States, 55,000 patients are hospitalized in the ICU on any given day,1 and approximately one-half million Americans undergo protracted (>96 hours) mechanical ventilation in an ICU each year. Historically, outcome studies in adult critically ill patients have focused on mortality. Recently, survival in some of our highest-acuity patients (e.g., acute respiratory distress syndrome, sepsis) has improved significantly2 through novel ventilation strategies,3 early interventions for sepsis,4,5 daily administration of renal replacement therapy,6 tight glycemic control,7 and other emerging therapeutic modalities. These dramatic improvements in ICU survival have reinvigorated interest in understanding the nature, determinants, and modifiers of long-term morbidity in ICU survivors.

Patients who survive critical illness are at risk for permanent physical, functional, emotional, and neurocognitive deficits, of which some or all may contribute to decreased health-related quality of life (HRQL). The reasons for this late morbidity after ICU care are multifactorial and include but are not limited to the following: (1) nature of and treatment for the inciting critical illness; (2) multiple-organ-dysfunction syndrome and hypoxemia; (3) physiologic and emotional stress in the ICU related to the illness itself, sleep fragmentation, psychoactive medications, and impaired drug metabolism owing to simultaneous administration of multiple medications; and (4) prolonged immobility and long ICU stay.

Patients with the acute respiratory distress syndrome (ARDS) represent some of the most complex, highest-acuity, and long-stay ICU patients (see Chap. 38). ARDS affects an estimated 150,000 people per year in the United States and is manifested by acute lung injury and severe hypoxemic respiratory failure.8 ARDS is associated with a variety of insults, including, pneumonia, sepsis, trauma, massive transfusion, and other medical/surgical conditions.9 It is a systemic illness involving inflammatory and coagulopathic disturbances that may induce dysfunction of multiple organ systems, including skeletal muscle and the peripheral and central nervous systems.10,11 Because of the significant potential for morbidity, ARDS patients have been the focus in long-term outcome studies in survivors of critical illness. We are in the early stages of understanding the long-term impact of ARDS on physical, emotional, and cognitive functioning and how each contributes to the patients' HRQL. Most studies in ARDS survivors have focused on 6- to 12-month outcomes, and there is limited information on morbidity beyond this time point. Comprehensive 5- and 10-year follow-up data are not available for ARDS patients, and it is unclear whether all survivors of critical illness—even with a severe episode—will suffer from the same morbidity as observed in ARDS survivors. Despite these limitations, the ARDS survivor data are some of the most complete long-term outcome data available and represent the current state of the art in the critical care outcomes literature. As such, they will form the primary basis for this review.

Historically, the assessment of outcome in critical care has focused largely on mortality and to a lesser extent on short-term physiologic and radiologic measures of impairment. As the focus has shifted to include the evaluation of longer-term outcomes, more investigators have moved toward patient-centered measures of functional status and HRQL. One advantage of these measures is that many are self-administered, and valuable information pertaining to physical, emotional, and cognitive well-being can be obtained without an in-person visit. This strength is also a limitation because these data will not facilitate understanding of the many and varied determinants of reported impairments in HRQL. This limitation is further compounded by heterogeneity across studies related to study sample, case mix, follow-up time for the population of interest, and difficulties comparing studies owing to the different HRQL measures administered. One potential solution to this dilemma is to focus on relatively homogeneous populations of patients, characterize their HRQL, and then proceed with in-person natural history cohort studies to identify and describe the specific determinants of reported morbidity. One example of data on a relatively homogeneous group is the ARDS long-term outcomes literature. Many recent publications describe decreased quality of life in ARDS survivors. Most studies suggest that physical and/or cognitive impairments are the main determinants of long-term disability. Through natural history cohort data in ARDS survivors, we have gained important insights into the heterogeneity of pulmonary, neuromuscular, and neuropsychological sequelae that contribute—to a greater or lesser extent in individual patients—to decreased HRQL. Ultimately, a detailed understanding of the spectrum of contributors to morbidity and their determinants will allow development, testing, and implementation of interventions that will lead to improved functional status and HRQL (Fig. 19-1).

Health-Related Quality of Life (HRQL)

HRQL can be defined as a set of causally linked dimensions of health, including biologic/physiologic, mental, physical, social function, cognitive, and health perception.12 Measures of HRQL assess how disease and its treatment are related to physical, social, emotional, or cognitive functioning. HRQL has emerged as an important measure of recovery from a variety of disease states, including critical illness, and has been used to evaluate patient-centered outcomes.

Most studies indicate that a significant proportion of ICU survivors experience some impairment in HRQL; however, this can be quite variable.13,14 Case mix may represent one important reason for these differences in reported HRQL. Critically ill patient populations are very diverse. The premorbid functional status of the patient and the etiology of the critical illness and its outcome represent important determinants of reported HRQL. Trauma patients with brain injury but normal cognitive function and intact social and work functioning reported a higher quality of life than trauma patients who were unemployed and had cognitive impairment.15 Critically ill multiple-trauma survivors experience decreased quality of life associated with cognitive impairments and decreased income.16 Elderly patients (>70 years) hospitalized in the ICU more than 30 days reported decreased physical functioning and poorer health and memory, but most were still functionally independent.17 Survivors of sepsis18 and prolonged mechanical ventilatory support (mean of 45 days) had compromised physical function, and the degree of dysfunction was related to premorbid functional status and the underlying disease.19

Although there is clearly some heterogeneity across different populations of ARDS patients, there appears to be less variability in reported HRQL compared with general populations of critically ill patients. In 1994, McHugh and her colleagues,20 using a prospective cohort study, serially evaluated pulmonary function and quality of life to assess the relationship between pulmonary dysfunction and functional disability. These authors found that the Sickness Impact Profile (generic quality-of-life measure of the subject's self-perceived physical and psychological condition) scores were very low at extubation, rose substantially in the first 3 months, and then exhibited only slight improvement to 1 year. When quality of life was assessed using a lung-related Sickness Impact Profile score, only a modest proportion of the patients' overall dysfunction was attributed to residual pulmonary problems. Weinert and coworkers21 identified functional impairment in a cohort of acute lung injury survivors. They administered the Medical Outcomes Study 36-item short-form health survey (SF-36), which yields scores in eight domains including physical and social functioning, role limitations because of emotional or physical problems, mental health, vitality, bodily pain, and general health perceptions.22 While all domains of the SF-36 were reduced, the largest decrements were in physical ability to maintain their roles (role-physical) and physical functioning. While some decreased quality of life was attributed to pulmonary dysfunction, many more patients attributed this to global and generalized disability. Schelling and colleagues23 made similar observations about impaired physical functioning and inferred that disability was due to pulmonary dysfunction; however, they did not assess this in their study. Davidson and colleagues24 designed a study to determine if there were differences in health-related quality of life in ARDS survivors and comparably ill controls. They used the SF-36 and a pulmonary disease–specific measure (St. George's Respiratory Questionnaire [SGRQ]) to determine the degree to which perceived physical disability in ARDS survivors was related to pulmonary dysfunction. Similar to previous reports, all domains of the SF-36 were reduced, and the largest decrement was in the role-physical domain. ARDS survivors had significantly worse scores on the SGRQ compared with critically ill controls. There appeared to be an ARDS-specific degree of physical disability, but it was not clear whether this was solely related to pulmonary dysfunction or there were other important extrapulmonary contributors.

Angus and colleagues25 used the quality-of-well-being (QWB) score in a prospective cohort of ARDS survivors to measure quality-adjusted survival in the first year after hospital discharge. The mean QWB scores for the ARDS cohort at 6 and 12 months were significantly lower than the scores of a control population of patients with cystic fibrosis. When QWB was disaggregated into its component subscores, post hoc analyses showed that the symptom-component scores of the QWB accounted for 70 percent of the decrement in perfect health at 6 and 12 months. Although respiratory symptoms were reported in almost half the patients, the most common complaints were musculoskeletal and constitutional.

In a prospective cohort study of 78 survivors of ARDS, Orme and coworkers26 evaluated HRQL and pulmonary function outcomes in patients treated with higher tidal volume versus lower tidal volume ventilation strategies. Both groups (higher and lower tidal volumes) reported decreased HRQL in physical functioning, role-physical, bodily pain, general health, and vitality (energy) on the SF-36. The pulmonary function abnormalities correlated with decreased HRQL for domains reflecting physical function.

Not only is the observation of impaired physical functioning robust across studies and investigators, but it also appears to persist for long periods of time following ICU or hospital discharge. The paper by Davidson and coworkers24 discussed earlier reported outcomes at 23 months after discharge, and Herridge and colleagues27 also have reported persistent physical dysfunction at 2 years after ICU discharge.

Hopkins and colleagues28 were the first to rigorously evaluate cognitive dysfunction in ARDS survivors and to report the significant impact this had on reported HRQL outcomes. Fifty-five consecutive ARDS survivors completed detailed neuropsychological testing and questionnaires relating to health status and cognitive and psychological function at hospital discharge and 1 year after ARDS onset. These authors reported that decreased HRQL was related to cognitive dysfunction. Impaired cognitive function following ARDS also has been reported by others.29

Decreased HRQL has been associated with posttraumatic stress disorder (PTSD) and is manifest in the emotional domains of the SF-36 (e.g., role-emotional, mental health, and vitality). PTSD may represent yet another important contributor to subsequent disability and loss of employment.30

HRQL in ARDS survivors is affected by physical limitation, cognitive impairment, and emotional dysfunction. The HRQL data have had an enormous impact on the critical care community and have helped to focus attention on long-term morbidity after critical illness. However, these data provide limited insights into the specific determinants of morbidity. Natural history cohort data—evaluating both functional and cognitive long-term outcomes—have helped us to begin to understand the heterogeneous nature of reported morbidity and the complexity of interaction among physical, emotional, and cognitive domains in individual patients (see Fig. 19-1).

Physiologic and Functional Outcomes in ARDS Survivors

Many authors have focused on residual pulmonary function abnormalities as a probable explanation for long-term functional impairment in ARDS patients. While most ARDS follow-up studies report pulmonary dysfunction, the pulmonary function abnormalities tend to be modest and may not fully explain functional limitation. More recently, neuromuscular dysfunction sustained as a result of the inciting critical illness and its attendant ICU care has been associated with ongoing physical impairment in ARDS survivors. These data have been gleaned from in-person prospective natural history cohort data.

Pulmonary Function Abnormalities

The lung was the obvious focus of outcome studies early after the first description of ARDS in 1967. The studies of pulmonary function have suffered from several limitations, including a lack of consecutive patient recruitment and loss to follow-up, limited sample size, limited follow-up times, and studies of pulmonary function in isolation without any concurrent functional evaluation. Evaluation of pulmonary function on its own or coupled with HRQL measures has remained a dominant theme until recently.

Many ARDS survivors have persistent pulmonary function impairments that typically are mild to moderate, with restrictive changes and a reduction in diffusion capacity.31–33 Orme and colleagues26 reported that ARDS survivors had abnormal pulmonary function associated with decreased health-related quality of life 1 year following hospital discharge, and Schelling and colleagues34 recently reported no additional improvement in pulmonary function after the first year following ARDS. In a recent publication, Neff and colleagues35 reviewed 30 studies that evaluated pulmonary function in ARDS survivors. They reported significant variability in obstructive (0% to 33%) and restrictive (0% to 50%) defects, as well as compromised diffusion capacity (33% to 82%). This spectrum of pulmonary dysfunction may relate to population heterogeneity with respect to evolving definitions or severity of ARDS, severity of lung injury, ICU ventilatory strategy, prior history of lung disease or smoking, and the presence of other pulmonary processes that fulfill the ARDS definition but have a very different natural history (e.g., bronchiolitis obliterans organizing pneumonia).

Most outcome studies found that ARDS survivors frequently are unable to resume their prior lifestyle, but the degree of pulmonary dysfunction does not fully explain their functional limitation. This observation has led investigators to explore other possible contributors to physical disability.

Limitation in Physical Functioning

One of the limitations in the ARDS morbidity literature has been an absence of data that objectively quantify functional disability. These data would be most useful in the context of in-person assessment in addition to concurrent physiologic and HRQL outcome measures. Multiple outcome measures may result in a better understanding of the determinants of functional morbidity and how this might be ameliorated. The Toronto ARDS Outcomes group evaluated exercise capacity (distance walked in 6 minutes with continuous oximetry) and pulmonary function and conducted an interview, physical examination, and HRQL measure in 109 ARDS survivors at 3, 6, and 12 months after ICU discharge.36,37 Similar to other pulmonary function studies, the ARDS patients had mild restrictive disease and reduced diffusion capacity at 3 months following ICU discharge. By 6 and 12 months, they had normal to near-normal lung volumes and spirometric measures with a persistent mild reduction in carbon dioxide diffusion capacity—lung impairment similar to that noted by others. The ARDS survivors had profound muscle weakness and wasting and were only able to achieve 66% of their predicted exercise capacity 1 year after ICU discharge. This functional disability was reflected in the HRQL assessment, in which patients reported profound reduction in the physical functioning and role-physical domains of the SF-36. Impaired exercise capacity was related to burden of comorbid disease, exposure to systemic corticosteroid treatment during the ICU period and the rate of resolution of lung injury, and multiple-organ dysfunction during the ICU stay. The precise determinant(s) of the observed muscle wasting and weakness were not clear, but possibilities included critical illness polyneuropathy, ICU-acquired myopathy, and entrapment neuropathies.

Critical Illness Polyneuropathy

An acute polyneuropathy was described in the 1980s in association with multiple-organ dysfunction and sepsis. This was called critical illness polyneuropathy and has been characterized electrically and morphologically by a primary axonal degeneration of motor and sensory fibers.38,39 The prevalence of critical illness polyneuropathy is 70 percent and has been documented in populations with sepsis, multiple-organ dysfunction, and ARDS.40 The precise etiology is unknown, but it may represent ischemic nerve injury secondary to a disturbance in the microcirculation. There has been a recent report that critical illness polyneuropathy may persist for years following ICU discharge and contribute to long-term physical limitation.41

ICU-Acquired Myopathy

The incidence of an ICU-acquired myopathy and its impact on disability and prolonged rehabilitation in the post-ICU period are uncertain. A recent report described a 25% incidence of ICU-acquired paresis in patients remaining on the mechanical ventilator for 7 or more days.42 Myopathic changes have been documented both in the presence43 and the absence44 of corticosteroid and neuromuscular blockade use. Several patients from the Toronto ARDS Outcomes study underwent open muscle biopsy45 in an attempt to better understand the nature of the observed muscle wasting and weakness. The median time to biopsy was almost 1 year after ICU discharge, and all patients had histopathologic evidence of a chronic myopathic process. Muscle injury is likely multifactorial, and it may represent an important determinant of long-term functional impairment.

Entrapment Neuropathy

The Toronto ARDS Outcomes study observed a 6% prevalence of peroneal and ulnar nerve palsies.37 Although this represents only a small proportion of patients, these nerve palsies complicated rehabilitation therapy and precluded return to original work in some cases. Other studies also have found detrimental long-term consequences resulting from compression neuropathies.46

Heterotopic Ossification

Heterotopic ossification is the deposition of para-articular ectopic bone and has been associated previously with polytrauma, burns, pancreatitis, and ARDS.47,48 It has been linked with paralysis and prolonged immobilization. There was a 5% prevalence of heterotopic ossification in the Toronto ARDS Outcomes study, with all patients having large joint immobilization, leading to important functional limitation (Fig. 19-2). Natural history cohort studies facilitate unexpected observations—such as heterotopic ossification—and link them to functional disability. Heterotopic ossification is remediable with appropriate surgical intervention, and screening for this might be an important part of a multidisciplinary intervention to improve functional outcomes.

Emotional Outcomes

Emotional Function after ARDS

The relationship between critical illness and emotional (mood) disorders is being recognized increasingly. Mood disorders represent important contributors to long-term HRQL impairments in survivors of critical illness. However, it is unclear whether these disorders are a psychological reaction to extraordinary emotional and physiologic stress, sequelae of brain injury sustained due to a critical illness and its treatment, or both.

Individuals with critical illness have to cope with a disease or injury that is life threatening as well as very burdensome interventions. The combination of medications, physiologic changes, pain, altered sensory inputs, and an unfamiliar environment may contribute to emotional changes following critical illness.49–51 Recent evidence suggests that mood disorders that occur secondary to medical illness may constitute discrete entities in which symptoms are similar to primary mood disorders, but there is a male predominance and earlier onset.52

The reported prevalence and severity of mood disorders including symptoms of depression, anxiety, and PTSD in survivors of critical illness are quite variable among patients following ICU hospitalization.49,50,53–54 Rincon and colleagues55 noted symptoms of depression and anxiety in 14% and 24%, respectively, of patients following critical illness. Similar prevalence rates of anxiety and depression have been reported by Scragg56 and Orme and coworkers.26 In contrast, Weinert and colleagues21 found that 43% of patients with acute lung injury reported symptoms of depression, and Angus and coworkers25 reported a 50% prevalence of depression and anxiety at 1 year in ARDS patients. The Toronto ARDS Outcomes study found that 58% of ARDS survivors reported depressive symptoms almost 2 years after ICU discharge.57 By contrast, Hopkins and coworkers28 found that ARDS patients reported minimal symptoms of depression or anxiety that were within the normal range in their natural history ARDS cohort study. The observed depression and anxiety after ICU treatment are likely multifactorial, and further study will be needed to better understand patient predisposition, illness, and treatment-specific determinants of affective morbidity.

PTSD is the development of characteristic symptoms that occur following a traumatic event(s) where triggers include a serious personal threat experienced with helplessness and intense fear.58,59 The diagnostic criteria include a history of traumatic event(s) accompanied by symptoms from each of three symptom clusters: hyperarousal symptoms, intrusive recollections, and avoidant/numbing symptoms.60 Schelling and colleagues30 were the first to introduce the concept of PTSD resulting from critical illness and ICU treatment to the critical care community. These authors evaluated HRQL and PTSD in a cohort of 80 ARDS survivors 4 years following discharge from the ICU. Almost a third of the ARDS survivors reported impaired memory, bad dreams, anxiety, and sleeping difficulties after ICU discharge, with a prevalence rate of PTSD of 28%. PTSD was related to the number of adverse ICU-related memories recalled by patients. Other authors61,62 also have noted this relationship. Memory for nightmares or delusions while in the ICU, as well as a complete absence of any ICU memories, also has been perceived as a traumatic event.63 The prevalence of PTSD has been reported to be as high as 38%61 and is a persistent complaint for years after ICU discharge.29,61,63a We are just beginning to fully appreciate how long-standing and debilitating mood disorders are following critical illness and the important contribution they have to decreased HRQL.

Cognitive Outcomes

Cognitive Impairment in ARDS Survivors

Cognitive impairment represents the major threat to both recovery and quality of life following an acute illness.29 Quality of life is largely determined by the ability to return to baseline level of cognitive performance.24,28,56 Long-term cognitive dysfunction—even when modest—results in vastly increased medical and disability costs. For example, a 3-point decrease on the Mini Mental State Exam was associated with increased overall health care expenditures of $6000 per year.29 The per-patient societal cost burden for even mild cognitive impairments is over $15,000 per year, and it is considerably higher ($34,515) for individuals with moderate to severe cognitive dysfunction.64

Cognitive impairments have been observed in a variety of patient populations with hypoxia.65–73 Hypoxia-related cognitive impairments include memory deficits,74 executive dysfunction,74,75 visual-spatial deficits,77 and intellectual decline.73,77 Critical illness, including ARDS, is associated with significant cognitive dysfunction.28,29,78,79 Approximately 33% of ICU survivors develop cognitive impairments that are similar to the cognitive dysfunction observed in mild dementia.66 Cognitive impairments are a major determinant of the ability to return to work, work productivity, and life satisfaction following cardiac surgery,80,81 traumatic brain injury,82 and ARDS.29 Even mild cognitive dysfunction results in clinically significant difficulties in driving, money management, and activities of daily living.83–87 Data obtained from interviews with seriously ill patients indicate that 90 percent of these patients would rather die than survive with cognitive disability.88

Hopkins and colleagues28 published the seminal long-term cognitive outcome study in ARDS survivors in 1999. In this natural history cohort, they found that 100% of ARDS survivors had cognitive impairments, including memory, attention, concentration, and decreased intellectual function, at the time of hospital discharge. At 1-year follow-up, 30% of the survivors had decreased intellectual function, and 78% had impaired memory, attention, concentration, and/or mental processing speed. ARDS survivors had significantly lower IQ than their estimated premorbid IQ (p ≤0.05) and the measured IQ 1 year later.

Hopkins and colleagues28 hypothesized that hypoxia may be an important contributor to cognitive dysfunction in ARDS survivors, and they undertook a detailed assessment of oximetry during the period of critical illness. The ARDS survivors' oximetry was measured for a total of 31,665 hours, with a mean of 609 ± 423 hours. The patients' mean oxygen saturations and their duration were outlined as follows: <90% = 122 ± 144 hours, <85% = 13 ± 18 hours, and <80% = 1 ± 3 hours. On average, these patients had 25 episodes of oxygen desaturation of less than 90% and 1 episode of desaturation of less than 85% for a duration of more than 2 hours. In this cohort, the degree of hypoxia was significantly correlated with neurocognitive sequelae (r2 = 0.25–0.45, all p <0.01).28

Since the Hopkins and colleagues report of cognitive impairments in ARDS survivors, other groups have confirmed their findings.29,57,78 In a retrospective study of 33 ARDS survivors, Marquis and coworkers78 reported impaired attention, visual processing, psychomotor speed, and cognitive flexibility compared with critically ill control subjects. Rothenhäusler and colleagues29 retrospectively evaluated 46 ARDS survivors and found that 24% had cognitive impairments and 41% were disabled and could not return to work. Limitations of this study were low follow-up rate and the administration of one brief cognitive test that assessed only memory and attention. A study of self-reported memory problems in the Toronto ARDS Outcomes cohort found that 20% of ARDS survivors rated their memory as poor 18 months following their ICU discharge.57

Cognitive impairments occur in a significant number of ARDS survivors at 1 year, with little improvement at 2 years after hospital discharge.89 In their recent prospective 2-year follow-up study, Hopkins and colleagues28 assessed cognitive outcome in 71 consecutive ARDS survivors treated with higher and lower tidal volume strategies. The result was that 59% and 43% of patients had evidence of cognitive dysfunction (>1.5 SD below the mean) in at least two cognitive domains at 1- and 2-year follow-up, respectively. Cognitive performance at hospital discharge was significantly lower than at 1- and 2-year follow-up (p <0.001). There were no significant differences between 1- and 2-year cognitive outcomes except improvement in performance IQ. Cognitive impairments at 1 and 2 years correlated with duration of hypoxemia, but they were not associated with gender, type of ventilator treatment, mean blood pressure, or time on sedative medications.

The recent report entitled, “Surviving Intensive Care: A Report from the 2002 Brussels Roundtable,” indicated that future investigations should prioritize studies of cognitive impairments in survivors of critical illness.90 Cognitive dysfunction in ARDS survivors is prevalent, with impairments reported in up to 78% of patients in some studies.28 It represents significant morbidity and is a major obstacle to recovery in severely ill patients. The etiology of the cognitive impairments is likely multifactorial and the subject of ongoing debate, but the paper by Hopkins and colleagues28 already has demonstrated a relationship to hypoxia, and several recent studies have demonstrated an association between delirium and cognitive impairments in critically ill patients.91,92 For a detailed review of delirium in this book, see Chap. 62. Future investigations are required to further delineate etiology and to identify interventions that may improve or prevent cognitive impairment in ARDS survivors.

Brain Tissue Loss after ARDS

Hypoxia and/or ischemia damage the hippocampus in humans,74,93–95 rats,96 and monkeys.97 Research in humans using quantitative magnetic resonance imaging (MRI) analysis of neural structures shows significant reductions in hippocampus volume associated with hypoxia but not in the nearby areas of the parahippocampal gyrus or the temporal lobe.75,94,95,98 Using quantitative MRI, brain morphologic changes, including ventricular enlargement, cerebral atrophy, and hippocampal atrophy, have been found in patients with pulmonary disorders following anoxia,98 asthma,99 carbon monoxide poisoning,73,100 and obstructive sleep apnea.68,69,101 Carbon monoxide poisoning and its concomitant hypoxia result in acute demyelination, generalized cortical atrophy,73 and discrete lesions in the basal ganglia,102 thalamus,103 substantia nigra,104 and white matter.72 Hopkins and colleagues reported recently generalized atrophy68 and atrophy of both gray and white matter structures, including the corpus callosum,105 fornix,106 and hippocampus,74,94,95 following carbon monoxide poisoning.

In light of their prior work linking hypoxia to morphologic changes in the brain and cognitive dysfunction, Hopkins and colleagues sought to evaluate whether there was a structural correlate for hypoxia and impaired cognition in ARDS survivors. They compared computed tomographic (CT) brain scans of 15 ARDS patients with those of normal age- and gender-matched controls (with normal scans) using quantitative image analysis.107 ARDS survivors exhibited brain atrophy, significantly enlarged ventricles, and an increased ventricle-to-brain ratio (another measure of generalized atrophy and an indirect index of white matter integrity) compared with the matched controls. The generalized brain atrophy, which was similar morphologically to that found in carbon monoxide poisoning, was associated with cognitive dysfunction.

As more patients are surviving critical illness and are being discharged from hospital more quickly, the burden of caregiving has shifted to family members. It is important to note that in most cases the caregivers are the same individuals who have just completed a protracted bedside vigil in the ICU. They may be physically and emotionally spent and experiencing financial stress due to the cost of medical care and time spent at the hospital and away from their workplace. They are likely to have very little reserve and will now be responsible for all aspects of the patient's recovery.

It is surprising that there is such limited literature on caregiver and family burden in survivors of a severe episode of critical illness. Potential contributors to this paucity of information include the historic focus on mortality as the main outcome of interest in critical illness and exclusive focus on patient-centered care. Neonatologists and pediatric intensivists understand that critical illness disrupts families—not just the lives of their patients—and adult critical care physicians might do well to embrace the concept of family-centered care.

The following discussion has been excerpted from an excellent recent review on this subject by Levy.108 In this review, Levy notes that Grad and Sainsbury first introduced the concept of family burden in 1966.109 These authors defined burden as the cost or negative consequences sustained by the family or caregiver and whether these were of a subjective or objective nature. Subjective burden refers to the caregivers' attitude toward or emotional reaction to the caregiving experience, and objective burden relates to the nature or degree of family disruption that can be observed directly or verified.

Chou and colleagues110 introduced a conceptual model of caregiver burden. The model includes four components: critical attributes and predisposing factors that influence the occurrence of burden, moderating/mediating factors, and consequences or outcomes. Critical attributes refer to the primary determinants of burden, including (1) the subjective burden as outlined earlier, (2) the multidimensional nature of demands that arise from caregiving, (3) the concept that illness is dynamic and therefore that the demands of caregiving also will be dynamic, and (4) the concept of overload, where the demands of caregiving exceed available resources. Predisposing factors relate to caregiver characteristics, including economic resources, race, and ethnicity. Mediating or moderating factors refer to elements that may alter the impact of burden, such as social or spiritual support. Consequences and outcome include the impact of the caregiving responsibilities on the caregiver and family.

Two recent papers highlight the burden of caregivers of ICU survivors. In a comprehensive review, Johnson and colleagues111 indicated that a very high degree of caregiver burden is related to patient neuropsychological dysfunction as compared with severe physical disability. Caregiver stress is also related to lack of education about the demands of the caregiver role and the responsibility of managing complex medical treatments and technology in the home.

Pouchard and coworkers112 conducted a prospective evaluation of anxiety and depression in 836 family members of critically ill patients. The prevalence of symptoms of anxiety and depression was almost 70% and 35%, respectively. It is still unclear whether or how long these symptoms persist and how they affect the care of the patient and the ability of the family to function normally.

Case History from the Toronto ARDS Outcomes Cohort

(See ref. 37.) The following case history will help to emphasize the functional and neuropsychological morbidity in survivors of critical illness that has been the focus of this chapter.

• A 47-year-old previously healthy female teacher presented to the ICU with hypoxemic respiratory failure secondary to pneumococcal pneumonia complicated by ARDS.
• The patient was intubated and mechanically ventilated for 5 weeks in the ICU.
• Her ICU course was complicated by the development of multiple-organ dysfunction requiring pressor support and renal replacement therapy.
• Three chest tubes were inserted during the ICU stay for pneumothoraces resulting from her complicated ventilatory management.
• The patient underwent tracheostomy at 3 weeks into the ICU stay and had a protracted wean from the ventilator related to weakness and difficulty clearing secretions.
• The patient had received continuous sedation and analgesia during the first 3 ICU weeks to facilitate mechanical ventilation. She had been immobile and noncommunicative.
• At the time of ICU discharge, the patient had severe weakness and generalized muscle wasting. She was unable to sit up. She could not walk, bear weight, or even transfer on her own.
• Prior to transfer, she was confused and was only able to obey simple commands intermittently.
• On transfer to the medical ward, she complained of insomnia and disturbing and recurrent dreams that she had been physically and sexually assaulted during her ICU stay.
• After 2 weeks on the medical ward, she was able to stand and take a few steps and was transferred to a rehabilitation facility for ongoing physical therapy.
• At 4 weeks, she was decannulated and discharged to her friend's home.
• She did not return to her own home because she couldn't be left by herself, and her husband and children were at work/school for the entire day.
• At 3 months following ICU discharge, she continued to have significant proximal muscle weakness with inability to fully abduct her arms. She could not get in or out of cars nor climb stairs because of proximal leg weakness (Fig. 19-3).
• At 12 months following ICU discharge, she had regained her lost weight, had mild proximal muscle weakness, but now could abduct her arms and reported being quite functional (see Fig. 19-3). However, because of her need to stand most of the day as a teacher, she was unable to return to work because of residual weakness and fatigue.
• A more intensive physical rehabilitation program was organized for her, and she continued to experience slow and progressive improvement in her weakness.
• At 18 months following ICU discharge, she returned to her original job on a part-time basis. At that time, she found her tracheostomy scar stigmatizing and underwent scar revision by a plastic surgeon.
• She began to complain of very severe depression and inability to cope with the stressors in her workplace. She underwent formal neuropsychological and psychiatric assessments.
• At 20 months after ICU discharge, she was diagnosed with mild cognitive dysfunction and PTSD. She received psychiatric treatment.
• At 24 months—with modest residual weakness and mild cognitive dysfunction—she returned to full-time work as a teacher and is performing well in the workplace.
• Throughout her recovery period, her family has endured significant financial stress related to her inability to work, and her children have had significant behavioral and academic problems.

The goal of this chapter was to review the current state of the art in the outcomes literature in survivors of critical illness. Another goal was to emphasize the importance of viewing critical illness as a continuum—from inciting illness through the ICU course and through recovery and rehabilitation. Understanding long-term outcome following critical illness is in its infancy. We are just beginning to understand the profound nature of functional and neuropsychological morbidity that may be sustained by the patients under our care every day in the ICU. In addition, we are only beginning to appreciate the caregiver and family burden experienced during and following the ICU stay.

A multidisciplinary model has been proposed113 that identifies predictors and modifiers of morbidity outcomes in survivors of critical illness (Fig. 19-4). This model is centered on a longitudinal post–critical illness clinic in which the physician and his or her team provide a support network. The goal is to provide ongoing continuity of care as patients recover from the morbidity of critical illness and to facilitate patient access to appropriate diagnostic and treatment referrals.

The different components of this model reflect the different types of morbidity that we have described in this chapter. Given the burden of neuropsychological morbidity, one component of this intervention is to offer neuropsychological and psychiatric assessment and supportive counseling to survivors of critical illness and their family members. The determinants of muscle wasting and weakness would be further delineated through formal neurologic and physiatry assessments, and treatment programs would be organized through occupational and physical therapy. Social work supports also would facilitate return to work and management of other family or child-related concerns.Variations of this model are currently under study, and hopefully, we will soon be in a position to ameliorate the long-term morbidity of critical illness.

Survivors of critical illness experience decreased health-related quality of life due to physical limitations, depression and anxiety, and cognitive impairments. Neuromuscular dysfunction may represent an important—and potentially irreversible—determinant of long-term functional disability in these patients. Other organ impairment such as pulmonary dysfunction also has been documented following critical illness but does not appear to have the same impact on patients' health-related quality of life. We now understand that approximately one-third to one-half of survivors of critical illness will develop long-term cognitive impairments and that both this and functional capacity appear to plateau at a lower than normal level at 1 to 2 years following ICU discharge. Long-term physical and neuropsychological dysfunction may be remediable through the implementation of a multidisciplinary and family-centered rehabilitation program, and this hypothesis is currently being tested.