From the June 26, 2003 issue
Information is key to preventing medication errors.
As demonstrated almost a decade ago, lack of information about
patients or their prescribed medications contributes to over
half of all serious, preventable adverse drug events (Leape
LL, Bates DW, Cullen DJ et al. Systems analysis of adverse
drug events. JAMA 1995; 274: 35-43).
Today, automated clinical decision-support systems are
available to help bring the most crucial information to
the attention of clinicians so they can provide the best
care. For example, a physician who prescribes an antibiotic
using an automated system with decision support may be alerted
to the need for a dose adjustment based on the patient's
most recent laboratory values. A pharmacist who enters an
order for a diuretic into a pharmacy computer system that
is integrated with the laboratory system may be alerted
to a patient's low potassium level. Clearly, these systems
help ensure that clinicians have the right information to
make the right decisions. However, most systems cannot ensure
that clinicians receive crucial information at the right
One of the limitations of most clinical decision-support
systems is that the provision of crucial information occurs
only when the clinician is actively engaged in their use.
In that regard, their effectiveness is limited to an episodic
encounter. For example, if a physician prescribes an antibiotic
early in the day before a patient's lab results are available,
he may not be alerted to an elevated creatinine level and
the need for a dose adjustment until the next day, when
he again logs into that patient's profile. What's needed
is technology that brings this type of crucial information
to the clinician as soon as a potential problem is detected.
That's exactly what a new line of clinical data-monitoring
technology can do. For example, automated systems offered
by Misys (Insight), VigiLanz (Dynamic Pharmacomonitoring),
and Cerner (Discern Expert for Cerner Millennium
and Classic systems), can "listen" to a
wide variety of information sources across the organization,
"watch" for specific problems that can be predefined
by the organization, and "notify" clinicians of
situations that may represent a risk to their patients as
soon as this information becomes available.
The technology utilizes computer industry data-compatibility
standards (HL7 compliant) to constantly communicate information
from various hospital clinical departments' computer systems
(e.g., radiology, laboratory, pharmacy, nutrition), tying
it to the hospital's patient information database and proprietary
software in a central system server. Each piece of data
is then checked to see if a relationship exists among pre-established
clinical rules to trigger an alert. If an imminent or existing
adverse clinical or drug event is detected, the system sends
a critical message via PDA, pager, cell phone, e-mail, or
fax, to alert the appropriate clinician(s) to the problem.
These alert messages can be sent to a department or designated
individuals (e.g., physicians, clinical pharmacists, nurses)
as determined by the professional staff in each hospital.
Notification can be provided in real time for situations
deemed critical, or the system may allow for a specified
time interval before notification while scanning to see
if appropriate action has been taken.
The powerful impact of these clinical data-monitoring systems
is clear. Hundreds of rules can be designed to alert clinicians
immediately to safety issues such as elevated liver enzymes
in a patient receiving a drug that is metabolized in the
liver, an abnormal potassium level in a patient taking digoxin,
a decreased platelet count in a patient receiving heparin,
or signs that an allergic response or drug interaction may
be occurring. The system could even alert staff about a
patient with pneumonia who didn't receive the first dose
of antibiotic within 8 hours of admission, or prompt a change
in antibiotic for a patient on vancomycin with a culture
and sensitivity report that doesn't support its continued
use. The failure to act on these types of data in a timely
fashion can have serious consequences for both patients
and staff. A case in point from a hospital that uses this
technology was the rapid notification of a positive acid-fast
bacilli result, which led to the immediate isolation and
treatment of the patient, and significantly reduced staff
exposure to tuberculosis.
Arguably, most hospitals have a call system in place for
these types of "panic" values or other acute situations.
However, non-automated call systems are quite fallible and,
too often, the appropriate clinician does not receive the
information in a timely fashion. Also, meaningful changes
in physiologic state due to disease or drug therapy may
not be noticed quickly if staff relies on "panic"
lab values. More often, health systems rely on nurses, pharmacists,
and physicians to detect problems, each of whom must navigate
through massive charts, large sets of daily lab results,
and multiple computer systems to collect this information.
Added to these challenges is the reality of staffing shortages
and budget restraints that can stretch the organization
to levels that threaten success with this labor-intensive
There's no question that automated decision-support systems
can be successful in reducing the risk of medication errors.
The key is making sure these systems provide timely advice
that is tailored to the needs of each individual patient,
and that the advice is provided unobtrusively to the clinician
who is best able to fix the problem quickly. To this end,
clinical data-monitoring technology represents a powerful
tool to effectively manage the dynamically changing relationship
between the patient's underlying physiologic state and corresponding
medication orders. And its use is not limited to health
systems where computerized prescribing is available and
used. Working seamlessly to provide an early warning system
in an environment that is chronically under stress, this
technology may well result in vastly improved patient outcomes,
and at a cost most hospitals would find affordable.(Installation
costs approximate that of a bedside bar-code scanning system.)
It's also likely that these systems will pay for themselves
through savings realized by preventing adverse events and
improving drug utilization and staff efficiency.
Surprisingly, most clinical staff we spoke with recently
were not aware of this technology. To learn more, visit
the vendor websites (e.g.,