Introduction

The traditional goal of intensive care has been to decrease short-term mortality. This is an appropriate and important end-point when caring for critically ill patients. However, less attention has been paid to the issue of what it means to survive intensive care for patients and relatives. Key questions include whether intensive care survivors have good long-term outcomes, whether these outcomes are the best possible results from our care, and whether our care would change if we knew more about outcomes beyond ICU discharge. The 2002 Roundtable was designed to highlight these issues, summarizing the available evidence on natural history and risk factors, and discussing future directions for research and improved care.

Methods

The Roundtable chairs outlined the key questions and developed a syllabus and list of speakers in the late spring of 2001 (see "Appendix"). The syllabus was divided into four sessions: the natural history of critical illness; the predictors and modifiers of long-term outcomes; future research issues, and; approaches to improve long-term outcomes. Speakers were invited in the summer and fall of 2001. Each speaker was given instructions and details regarding their proposed topic and asked to prepare a manuscript and 20-min oral summary. Speakers involved in reviewing the existing literature (the first two sections) were asked to conduct a systematic review as part of their tasks. Manuscripts were circulated to all participants prior to the Roundtable. The Roundtable was held in Brussels 16–18 March 2002. Each participant presented his or her topic followed by a 30-min discussion. At the end of each of the four sessions, further discussion focused on the key themes emerging from that session. At the end of the Roundtable a further 2-h discussion was held to discuss the common themes of the entire Roundtable and summation of key points. These key points were presented at the Opening Session of the 22nd International Symposium on Intensive Care and Emergency Medicine. Following the conference the manuscripts were revised in accordance with thoughts and comments raised during the Roundtable and resubmitted to the two chairs. The entire collection of manuscripts is published under separate cover by Springer-Verlag in the Update in Intensive Care and Emergency Medicine series [1]. This document summarizes the key points emerging from the Roundtable. It was drafted and edited by the Chairs in the spring and summer of 2002 and circulated to all Roundtable participants for feedback and critique. It is organized following the four sessions of the Roundtable.

What do we know about patient-centered outcomes after critical illness?

There are a wide variety of important, serious sequelae in survivors of intensive care. These include late mortality [2, 3], on-going morbidity [3, 4, 5], neurocognitive defects [6], impaired mental health [3, 5, 6], poor functional capabilities [3, 4, 5], decreased quality of life [7], decreased return to work and usual activities [6], burden and stress on families and informal caregivers [8], and economic costs to the patient, the family and society [9].

Several studies have suggested that patients who survive intensive care, such as those surviving severe sepsis, are at a higher risk of death than control cohorts for many years [2, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27]. Furthermore, ICU survivors often report debilitating problems with poor functional status [28, 29, 30, 31, 32]. For example, Montuclard et al. [28] recently reported in a small cohort of elderly ICU survivors that only 41% were alive at 1 year, and 23% had difficulty bathing, 15% had difficulty toileting, 26% had difficulty transferring from chairs or beds, and 19% were incontinent. Even at hospital discharge, many patients have considerable problems. Ely et al. [33] recently demonstrated that almost half of all ICU patients leave the hospital with abnormal mental status. Hopkins et al. [34] showed that neurocognitive problems can persist long after discharge. At 2 years, one-fifth to one-third of acute respiratory distress syndrome (ARDS) survivors demonstrated problems with memory, mental processing, executive decision-making, and attention deficit [35, 36]. ARDS survivors are also likely to have a wide array of symptoms ranging from respiratory problems, such as hoarseness and dyspnea, to general constitutive complaints, such as weakness and fatigue [13]. Herridge [37] has also reported considerable weight loss and debilitation of ARDS survivors, including flexion contractures and heterotopic ossification, presumably due to the prolonged immobilization of protracted ICU stays. These problems impair exercise ability and interfere with the ability to return to work. Indeed, Herridge [37] reported that only 38% of survivors had returned to work 1 year after discharge. It is not surprising, therefore, that numerous studies have demonstrated that ICU survivors and survivors of ARDS or sepsis endure a poorer quality of life than controls [13, 14, 15, 30, 31, 32, 34, 38]. Studies have suggested ICU survivors are more likely to suffer posttraumatic stress disorder and depression with poor quality of life related to both physical and mental domains [32].

These poor long-term outcomes likely place significant toll on spouses and families. Although there are no studies of the long-term impact of ICU care on informal caregivers [8], evidence from stroke survivors suggest that stroke caregivers (spouses and families) are three times more likely to develop depression [39]. This appears to be particularly problematic among women [40]. Based on studies in the care of other chronically ill patients, the ramifications for informal caregivers of ICU survivors might well include impaired physical and mental health, social isolation, impaired sexual relations, and forced loss of earnings [8]. In addition, families may experience serious psychological sequelae from a bad experience with the patient stay in the ICU.

What are the causes and modifiers of the poor patient-centered outcomes after intensive care?

Despite the importance of this question, the answers are generally complicated and poorly understood. Problems include delineating post-ICU sequelae that are due to the ICU course vs. part of the underlying illness, separating the myriad potential variables that might influence outcome and interact with each other, and a lack of comprehensive data. To consider this question, therefore, it is helpful to first create a conceptual model of critical illness. We propose that an episode of critical illness is not just the period of time a patient spends in an ICU but is the period of time that begins with the onset of the acute deterioration and ends when a patient's risk of late sequelae, such as on-going mortality, has returned to the baseline risk of a similar patient who had not incurred the acute critical illness [41]. This is represented graphically in Fig. 1.

Fig. 1.
figure 1

The episode of critical illness. The figure shows that an episode of critical illness is not just the period of time a patient spends in an ICU but is the period of time that begins with the onset of the acute deterioration and ends when a patient's risk of late sequelae, such as on-going mortality, has returned to the baseline risk of a similar patient who had not incurred the acute critical illness. (Adapted with permission from [41])

Full size image

Imposed upon this model, one can begin to delineate and classify some of the potential variables and their interactions that might influence subsequent patient-centered outcomes. We propose such a classification with illustrative examples in Table 1. Many of these pre- and intra-ICU factors have been studied previously. However, in many instances, the studies have only explored the relationship with short-term outcome using association analysis. For example, there are many observational studies suggesting that full-time intensivists improve hospital mortality but no prospective studies [42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67]. The same holds true for the use of many interventions in the ICU and obviously for variables suspected of causing harm. The relationship between nosocomial pneumonia and outcome will only be studied by association analysis because one would not prospectively assign a patient to develop a pneumonia.

Table 1. Variables potentially influencing long-term outcomes after critical illness
Full size table

The best approach to understand causality in clinical medicine is to conduct a prospective randomized trial. However, even variables that can be studied using prospective randomized trials, such as ICU therapies [68] and protocols [69, 70, 71], have typically not evaluated the effect on long-term outcomes. Research in this area is also complicated by the inherent difficulties of isolating pre-ICU factors in patients who are often only identified once they have been admitted to the ICU. Similarly, obtaining long-term follow-up can be difficult and expensive.

Given the difficulties of measuring long-term outcomes, and the existing and comprehensive focus on short-term outcomes, it would be appealing to believe that these short-term outcomes would be valuable proxies for subsequent outcomes. This, however, is a very dangerous assumption. As shown in Table 2, there are many examples from many diseases where an intervention appeared to engender a good short-term outcome yet was subsequently shown to have no effects or even harmful effects on long-term outcomes. Indeed, Clermont et al. recently showed that, in patients surviving a hospitalization for community-acquired pneumonia, the degree of acute organ dysfunction (as a measure of the 'burden of critical illness') was not an independent predictor of subsequent patient-centered outcomes [72].

Table 2. Paradoxical short- and long-term effects after certain interventions (CHF congestive heart failure, AMI acute myocardial infarction) (Adapted with permission from [91]
Full size table

In other words, although we are very concerned about the poor long-term outcomes of ICU survivors, it is not clear to what extent these outcomes are due to the ICU care, the ICU disease, or underlying characteristics of the types of patients who develop critical illness. These are crucial distinctions that must be addressed in future research if we are to optimize long-term outcomes after critical illness.

What are the implications for future research?

We organized our recommendations for future research into four areas: observational studies, conduct of future ICU interventional trials, innovations to improve long-term outcomes, and methodological research.

Future observational studies of long-term outcomes

Quality of life, function, morbidity, and survival in different ICU populations

Although we have considerable information on the natural history of critical illness, there is still a need for better observational study to catalog the potential sequelae more carefully. For example, although there are a number of studies of quality of life following ARDS, the general rigor is poor and inconsistent [73, 74]. And, there are still very few studies cataloging functional status and on-going morbidity. We would therefore urge investigators to better study the quality of life, functional status, morbidity, and survival in different ICU populations.

Neurocognitive abnormalities in ICU survivors

The recent reports of neurocognitive abnormalities in ICU survivors are particularly concerning [6, 33, 34, 75, 76, 77]. We use a wide array of powerful neurosedative and neurotropic agents in the ICU and there is a distinct possibility that our use of such agents and management of delirium in the ICU could be influencing subsequent neurocognitive function. We therefore strongly recommend greater study in this area.

Burden on family and other informal caregivers of ICU survivors

Many ICU survivors incur poor patient-centered outcomes akin to those suffered by patients with many chronic and debilitating conditions. Informal caregiver burden is considerable in the latter instance yet there is essentially no evaluation of the burden and stress on informal caregivers of ICU survivors [8].

Long-term economic costs

There are no studies of the long-term costs following ICU discharge. It is likely that these costs could be high, given the potential for numerous debilitating sequelae. If so, there may be a considerable financial impetus to find alternative treatment strategies that improve outcomes and decrease these costs. Alternatively, new strategies may improve outcomes but only at considerable increase in cost. If so, determining the value or cost-effectiveness of these strategies will be important when evaluating whether such interventions can be adopted [78].

Future interventional trials (e.g., antisepsis trials)

New interventions in intensive care are commonly studied in large, multicenter (and frequently multinational) randomized trials. To better understand the long-term ramifications of these interventions, yet not unduly burden the study with excessive additional data collection, we recommend the following.

Prolong follow-up for survival to at least 6 months

Clinical trials would ideally include a follow-up period that is comparable to the time interval of risk [2]. Based on survival curves from prior studies [10] and in keeping with the recommendations of the International Working Party on clinical trials in sepsis convened by the UK Medical Research Council [79], we recommend that all ICU interventional trials designed to test efficacy include survival follow-up to at least 6 months (and ideally longer, especially for those enrolled early in the trial).

Telephone-based assessment of quality of life

Survival alone is an inadequate measure of patient-centered outcome [7, 80, 81, 82]. We would therefore recommend also assessing quality of life. We recommend using a standard, well-validated instrument that is straightforward and appropriate to administer by telephone (given the complexities of long-term follow-up in a multicenter study) and, ideally, is applicable in different countries and languages. The SF-36 [83] is an example of a comprehensive instrument that meets these criteria while the EQ-5D by the EuroQOL group [84] is an example of a simple instrument that meets these criteria [7, 78, 80]. The EQ-5D provides a measure of both quality of life and utility, the latter being necessary for the calculation of quality-adjusted survival, a key measure of health effect for cost-effectiveness assessments [85]. Recently, Brazier et al. demonstrated a method to calculate utilities from the SF-36 as well [86]. Consistent adoption of common instruments across studies would also facilitate interstudy comparisons [74].

Telephone-based assessment of function, morbidity, resource use, return to work and usual activities, and other domains

Quality of life is not the only important long-term outcome. If researchers are conducting a telephone-based assessment of quality of life, we would recommend the interview be complimented with additional questions relating to the other domains listed above [9, 78]. Instruments are less well-standardized in this arena and may require tailoring to specific diseases or interventions. Nevertheless, a short set of additional questions could add valuable information at little additional cost to the study.

Innovations to improve long-term outcomes

When data characterizing the nature and extent of decreases in long-term patient-centered outcomes after intensive care are still somewhat lacking, it is difficult to articulate a list of specific interventions directed at improving care. Nevertheless, there are examples of potential interventions that might be developed with reasonable ease. One example is the ICU follow-up clinic. [87] Such clinics already exist sporadically in some countries [88]. The organization and purpose of these clinics is quite varied but generally conducted at least in part by intensive care staff. The premise for these clinics is that: (a) there are sequelae following intensive care that can be identified and potentially improved, and (b) specific post-ICU follow-up is better suited to detecting and treating these sequelae than general medical care. However, there is little formal evaluation at this point and a number of questions regarding the benefits and costs of such follow-up are unanswered.

Perhaps complimentary to ICU follow-up clinics, a second idea is that of ICU discharge screening [89]. It is certainly possible that there are key risk factors potentially identifiable at ICU discharge for important sequelae. Screening for such risk factors may allow more targeted post-ICU care. Again, however, although the idea appears worthy, there are no data currently.

Methodological research

There are a number of methodological issues relating to the adequate capture of post-ICU patient-centered outcomes. For example, what are the best quality of life instruments? Are existing instruments suitable for capturing important nuances of post-ICU sequelae or should disease-specific instruments be captured? Although the panel has made recommendations for the use of such instruments above, we recognize there are limitations and would encourage investigators to continue evaluation of new instruments. Specific challenges, for example, include developing instruments that can adequately capture neurocognitive defects post-ICU in environments such as large multicenter trials where exhaustive neuropsychological testing may be impractical. Similar issues relate to tracking long-term economic costs, especially in multinational studies. Of course, some of these issues are not specific to intensive care and it is key that we take advantage of methodological breakthroughs in related fields as they occur.

How could we change care now?

Although there is considerable need for on-going research, there was a general perception among the panelists that there is adequate empirical evidence to support changing practice today. Indeed, attention to many common elements of ICU practice may improve long-term outcomes. However, even within the experience of the panel, there was wide variation in current practice patterns with regard to simple, and presumably straightforward, practice recommendations. Table 3 presents some examples of common problems that develop in the ICU, the potential long-term sequelae, and some relatively simple changes in care, or the focus of care, that could minimize unwanted sequelae.

Table 3. Examples of common ICU problems, potential late sequelae, and simple interventions
Full size table

These examples highlight the potential simplicity of the steps that could be taken to optimize long-term outcomes. At the same time, they also highlight the need to have a global awareness of critical illness both within and outside the ICU "box." This is illustrated in Fig. 2, which demonstrates the traditional "within the box" model of intensive care delivery (Fig. 2A) and an alternative role for the critical care team that thinks "outside the box" (Fig. 2B).

Fig. 2.
figure 2

Traditional and alternative models of intensive care delivery to the critically ill. In the traditional model of critical illness, critical care is delivered only by ICU clinicians in the ICU. Patient management is delivered according to distinct clinical roles during distinct, usually short-term, episodes of illness. In the alternative model, critical illness is viewed on a continuum. Patient management related to critical illness is offered at several points in the illness continuum. The focus is on optimizing long-term outcomes. ER Emergency room; D/C discharge. (Adapted with permission from [87])

Full size image

Encouraging this paradigm shift is not necessarily easy. As with many other quality improvement initiatives, we would recommend that intensivists begin with small, simple measures that are likely to work locally [87, 90]. We also recognize that care before and after the ICU is already provided by other medical disciplines [87, 90]. We do not suggest that intensivists should assume this expanded responsibility alone. Rather, we would suggest that intensivists seek to create partnerships with other caregivers. For example, primary care physicians may be very willing to screen for the late sequelae of ARDS, and simply require education regarding what to look for. Similarly, rehabilitation medicine is a well-established field that plays a key role in aiding recovery after a wide variety of conditions such as stroke or traumatic brain injury. Creating partnerships with rehabilitation services to promote follow-up of debilitated, underweight ICU survivors could foster improved outcomes at minimal expense. Changing the focus of the ICU team to consider long-term patient-centered outcomes, such as whether a patient will successfully return to work, may have the added advantage of facilitating communication with family members.

Summary

This year's Roundtable represented an opportunity to take stock of an existing literature base, debate conceptual issues regarding the focus and purpose of critical care, and recommend future steps for research and clinical care. It was very apparent that critical illness is associated with a wide array of serious and concerning long-term sequelae that interfere with optimal patient-centered outcomes. Although traditional short-term outcomes, such as hospital mortality, remain extremely important, they are not likely to be adequate surrogates for subsequent patient-centered outcomes. As such, it is important to focus specifically on how critical illness and intensive care affects a patient's and relatives' long-term health and well-being. Clearly, there are a large number of potential pre-, intra-, and post-ICU factors that may improve or worsen these outcomes, and delineating these effects sets the stage for a rich research agenda. At the same time, there are opportunities today to improve care. Key to taking advantage of such opportunities is the need to embrace a global awareness of critical illness as an entity that begins and ends outside the ICU "'box."