Three profs from UPMC have published a paper (full text) that surveyed the impact of an eICU installation in two ICUs between 1999 and 2001.

Hospital mortality for ICU patients was
lower during the period of remote ICU care (9.4% vs. 12.9%; relative
risk, 0.73; 95% confidence interval [CI], 0.55–0.95), and ICU length of
stay was shorter (3.63 days [95% CI, 3.21–4.04] vs. 4.35 days [95% CI,
3.93–4.78]). Lower variable costs per case and higher hospital revenues
(from increased case volumes) generated financial benefits in excess of
program costs.

With documented results like these, it's no wonder hospitals are
adopting eICU. The study investigators note that factors other than the
introduction of the off-site intensivists may have contributed to the
observed results.

These results are provocative,
particularly since improvements were seen in both the medical and
surgical ICUs. However, several major limitations warrant mention. The
authors admit that the "actual basis for the observed changes is not
known." They note that the introduction of computer systems and
decision support tools and the increased institutional focus on ICU
care that accompanied the implementation of the eICU program may have
affected the results. Although the authors state that no other major
changes in care paradigms or protocols occurred during the study
period, some unmeasured or unappreciated changes may have occurred. The
use of historical controls, despite the similarity of admission
criteria and APACHE III scores between the baseline and intervention
periods, raises questions regarding potential differences in case-mix.
Although the patient population included medical and surgical ICU
patients, the results were based on eICU physicians staffing a total of
18 beds in a single institution that had preexisting daytime on-site
intensivist coverage. The clinical impact on hospitals completely
lacking intensivist coverage may not be the same. Furthermore, the
financial benefits seen herein may not be realized in smaller hospitals
with fewer ICU beds due to economies of scale.

The foregoing commentary should not take anything away from the
potential benefits of eICU. This kind of secondary effect is seen with
the introduction of many types of information systems (I'm thinking of
patient flow software applications, especially). When broad workflows
that entail multiple actors are addressed with an IT solution, the
inevitable focus frequently brings a benefit that is beyond the
capabilities of the system being implemented. A broad systematic review
of operations, and the resulting optimizations that result from changes
made during system implementation, can have a synergistic impact creating positive changes and outcomes.