Evaluating Telemedicine in the ICU

Telemedicine, defined as the use of electronic information and communication technologies to provide health care when the caregiver and patient are geographically distanced,¹ is an increasing trend in inpatient medicine. One of the most common applications is in the intensive care unit (ICU), where ICU telemedicine (tele-ICU) can be used to increase access to intensivist physicians. In this care model, intensivist physicians and nurses, located in a central monitoring facility, are able to visually monitor patients across multiple ICUs in many hospitals using bedside cameras while also electronically tracking relevant clinical information, such as vital signs or laboratory data. The remote care team is then able to communicate orally with bedside caregivers to provide real-time, around-the-clock patient care. Research has shown that dedicated on-site intensivist staffing is associated with reductions in hospital and ICU length of stay (LOS) and mortality.² Thus, the appeal of tele-ICU is intuitive: it extends intensivist coverage to hospitals that do not have it and may increase the timeliness of intensivist input in other settings.

Despite the attractiveness of tele-ICU and its increasing adoption, few trials have evaluated its effect on outcomes. Therefore, it is not yet clear in what situations, if any, the potential benefits might be realized. Early studies, each performed at a single center and comparing care before and after implementation of a tele-ICU system, documented decreases in LOS, mortality, and cost, first in an academic-affiliated community hospital and subsequently in a tertiary-care hospital.^{3 - 4} Presumably because few hospitals would submit to being randomized to implement or not implement tele-ICU, all subsequent tele-ICU studies also involve before-and-after study designs. Furthermore, they are limited by lack of specificity about what the intervention includes, by the inclusion of multiple changes to ICU care in the tele-ICU as implemented, and by the absence of a detailed description of preintervention ICU care.

In this issue of JAMA, Thomas et al⁵ report the first federally funded multicenter evaluation of tele-ICU, expanding application of tele-ICU to 6 ICUs of varying case mix, size, and staffing. Using a before-and-after design, they measured the effect of tele-ICU on LOS, mortality, and complications and reported no overall benefit with the tele-ICU program. Although the investigators carefully describe their ICUs as “open” (admitting privileges open to physicians other than intensivists) or “closed” (admitting privileges restricted to intensivists), the “open” units were variable in their degree of exposure to intensivist care. Furthermore, their particular intervention, a proprietary tele-ICU system, included not only remote monitoring by intensivists aided by electronic patient data, but also nursing, technician, and alert-system services. The authors do not report whether similar services existed before tele-ICU implementation or whether such services varied among hospitals. Their findings may also be explained by the limited degree of treatment authority delegated to the tele-ICU by the majority of attending physicians. One interesting unanticipated observation was an apparent benefit among sicker patients, which highlights the need for careful evaluation of what each tele-ICU means, in terms of specific changes in ICU care relative to other alternative approaches and of which patients might benefit. Thus, even though the study by Thomas et al is a step forward with its range of hospitals and description of preintervention care, it nonetheless remains difficult to interpret.

Some of the uncertainty about tele-ICU reflects the lack of consistent reference in the literature to a unifying conceptual framework of what ICU care is and how tele-ICU could improve it. Without such reference, it is difficult to delineate which components of care a tele-ICU intervention (or any other attempt to redesign ICU care) addresses. In terms of a conceptual framework of ICU care, structural issues such as staffing levels and availability of computerized order entry likely contribute to outcomes.⁶ Within a given structure, team interactions—not just among physicians, but also among physicians and nurses, respiratory therapists, and pharmacists—influence the eventual processes of care undertaken. These processes of care—such as using low tidal volumes in acute respiratory distress syndrome (ARDS)—determine clinical outcomes,⁶ while communication by the ICU team with the patient and family shape patient and family experiences of care.

Reference to this framework in research publications would help in interpretation of results. For instance, multicenter studies could analyze their study hospitals separately based on preintervention staffing levels. This would facilitate comparison of different staffing levels with the team structure in their intervention. Similarly, when tele-ICU includes care protocols,⁷ investigators should report the extent of preintervention protocol use and whether tele-ICU changed protocol adherence. If improved protocol adherence led to better ICU outcomes, the contribution of protocol adoption could be isolated from that of remote monitoring.

Given the heterogeneity of tele-ICU systems and the hospitals adopting them, it is unlikely that any single study can definitely address the benefits of telemedicine for the critically ill. Rather, literature syntheses will be the most important approach to improving the understanding of the effects of tele-ICU support. It would be helpful for researchers to synthesize the literature by proceeding through sequential and parallel hypothesis testing.⁸ Sequential hypothesis testing in tele-ICU could start with studies in which remote monitoring, the primary component of tele-ICU, is most likely to be beneficial, such as in rural settings⁷ or institutions lacking not only intensivists but also other in-house physicians. A negative result in these hospitals would be an important signal that tele-ICU is not an effective means of increasing intensivist coverage, whereas a positive result would justify adding tele-ICU intensivists to similar settings. Given positive results in such institutions, the researcher synthesizing the literature could then move sequentially to settings with increasing preintervention coverage (perhaps next to community hospitals with hospitalists, then to teaching hospitals). Positive results even in these institutions would suggest that benefits of remote monitoring are robust, while negative results would not negate the findings in rural situations.

A similar approach could be taken with the patient populations studied. In theory, remote monitoring would add the most to care for the sickest patients who need timely interventions. Showing benefit for these patients should be followed by evaluation of less critically ill individuals to improve understanding of the group of patients for whom remote monitoring is beneficial. However, the existing literature does not yet delineate enough of the key variables in the conceptual framework to allow such sequential evaluation of hypotheses.

The concept of parallel hypothesis evaluation⁸ would call for isolation of remote monitoring from other elements of ICU care that might confound studies if they were included in tele-ICU but that should be evaluated in parallel. Protocol-based care (eg, ventilator weaning protocols) and computerized order entry have been included in tele-ICU and could, based on the conceptual framework, improve outcomes. However, both also could be implemented locally and therefore should be evaluated in parallel with remote monitoring but with the recognition that they are distinct interventions.

Tele-ICU is a potentially valuable change in ICU care, but its complexity means that “tele-ICU improves care” is not a testable hypothesis. Therefore, performing and synthesizing tele-ICU research will be challenging. If future studies include more description of which components of ICU care were present before tele-ICU and which were added, it would be easier to interpret the results. Synthesis of this literature with an eye to sequential hypothesis testing of remote monitoring and parallel hypothesis testing for other important elements of ICU care would help to determine when and how to use tele-ICU.