Last week in our article on some of the hazards of alarms (February 23, 2010 Patient Safety Tip of the Week “Alarm Issues in the News Again”), we mentioned using alarms to pick up clinical deterioration earlier. This week we’ll discuss a couple articles where monitoring devices are used to identify that early clinical deterioration but highlight the problem of alarm sensitivity: balancing early detection vs. triggering too many false alarms that might lead to development of alarm fatigue.
The first article is one we already discussed in our What’s New in the Patient Safety World for February 2010 “Rapid Response Teams Still Not Cutting It”. The article by Taenzer et al 2010 discussed a new system of monitoring post-op orthopedic patients with continuous pulse oximetry at Dartmouth. Data from continuous pulse oximetry was analyzed by computer and tied to a system of notifying the patient’s nurse by pager. Key concepts were setting thresholds that met a balance between high sensitivity and numbers of false alarms. They also incorporated a delay into the notification system to further minimize the number of false alarms that nurses would have to respond to. The ultimate triggers used were an oxygen saturation of less than 80% and a heart rate below 50 or greater than 140. Their preliminary results show a reduction in rescue events and fewer transfers to the ICU. There were also fewer deaths, though the numbers were too small to be considered significant. This is an exciting concept and suggests that noninvasive monitoring tied to computer algorithms might someday operate in the background to help identify patients in need of early intervention.
Their discussion on balancing high sensitivity against false positives is excellent. They did observations for a month and noted that using an O2 saturation of 93% would have led to too many alarms. They found that 12% of data points in their postoperative patients occurred below that O2 saturation level. They note that such sensitivity might be appropriate in a setting with 1:1 care, but in a more general care setting it would distract nurses from other activities and likely lead to development of alarm fatigue where alarms might be responded to in a delayed fashion or not responded to at all. They also note that you must take into account individual patient characteristics as well. While the parameters they selected above may work well for the general population, those patients with pre-existing physiological abnormalities may need different settings. They therefore allowed nursing staff to make adjustments of parameters +/- 10% but any greater adjustments required a physician order.
The second article (Deitch et al 2010) looked at use of capnography during procedural sedation in the emergency department. Monitoring end-tidal CO2 during procedures in which IV sedation or IV opioids are used is theoretically very attractive since it is a better indicator of respiratory depression in such cases and should identify patients at risk before they develop oxygen desaturation. Their randomized controlled trial in adults undergoing propofol sedation showed significantly fewer patients with hypoxia in the group monitored by capnography. The 17% absolute risk reduction translates to a number needed to treat (NNT) of 6 cases to prevent one episode of hypoxia. Capnography recognized 100% of the patients who developed hypoxia and did so a median time of 60 seconds before hypoxia developed. That delay meant the physicians usually had adequate time to intervene.
It is extremely difficult to identify respiratory depression just by watching a patient. We have had medical students, residents, and attendings all observe patients on whom we had arterial blood gases and they did little better than chance at picking out which patients had respiratory depression. So the ability to identify patients with respiratory depression before they become hypoxic is a major potential advantage.
The editorial (Green & Pershad 2010) accompanying the Deitch study describes the pros and cons of adopting capnography routinely for such procedural sedation. The editorial notes that capnography clearly identifies respiratory failure before it is apparent by either clinical examination or pulse oximetry. And it acknowledges the significant contribution of the Deitch study, though it points out that some methodological issues (eg. using O2 at a flow rate of only 3L/minute and using an O2 saturation of 93% as the definition of hypoxia) may have overestimated the frequency of hypoxic events. And many episodes of hypoxia are self-limited and resolve without any intervention. In fact, 41% of the episodes in the Deitch study resolved without any intervention.
But Green & Pershad also note the substantial number of “nuisance” alarms. Combining the number of capnographic alarms that were not followed by hypoxia with the number of patients excluded because of data loss during capnography, they estimate the rate of “nuisance” alarms could be as high as 39%. They note that this may be even higher in procedures done in pediatrics because children tend to be less tolerant of the cannula and verbalization or crying will interfere with recordings. They point out it is unclear how physicians will respond to these nuisance alarms. So, while they like the concept they suggest that rather than becoming a routine standard, capnography might be best used in targeted situations such as sedation for radiology studies.
We have some of our own concerns. Will responding to the capnography alarms distract physicians from the procedure they are performing? First of all, we consider it very dangerous when the physician performing the procedure is also the person doing the monitoring. We find it hard to believe that you can keep full attention to both simultaneously. But does it happen (i.e. that one physician will be doing both)? Of course it does. And even if there is a separate person doing the monitoring, the alarms may still distract the physician from the procedure at hand.
But might there be other unintended consequences? One of the phenomena we like to point out is that described by Charles Perrow in his classic book “Normal Accidents” (Perrow 1999) where he talks about how new technologies often simply “push the envelope”. He cites as an example how the introduction of maritime radar simply encouraged boats to travel faster and did little to reduce the occurrence of maritime accidents. So will practitioners now give higher doses of sedatives or opioids, knowing that capnography will warn them of problems before hypoxia occurs? Will the safety margin engendered by capnography thus actually lead to more frequent episodes of serious respiratory depression?
The real lesson here is that when you plan to implement a new monitoring system with alarms you need to how often those alarms will trigger, what sort of responses you are likely to see, and how workflow will likely be impacted. Then you need to anticipate what workarounds may develop and what other unintended consequences are likely to occur. You can never anticipate all of the latter. So careful surveillance and auditing during the period after implementation of a new alarm system is critical.
Taenzer AH, Pyke JB, McGrath SP, Blike GT. Impact of Pulse Oximetry Surveillance on Rescue Events and Intensive Care Unit Transfers: A Before-and-After Concurrence Study. Anesthesiology 2010; 112(2): 282-287
Deitch K, Miner J, Chudnofsky CR, Dominici P, Latta P. Does End Tidal CO2 Monitoring During Emergency Department Procedural Sedation and Analgesia With Propofol Decrease the Incidence of Hypoxic Events? A Randomized, Controlled Trial. Annals of Emergency Medicine 2010; 55(3): 258-264
Green SM, Pershad J. Should Capnographic Monitoring Be Standard Practice During Emergency Department Procedural Sedation and Analgesia? Pro and Con. Annals of Emergency Medicine 2010; 55(3): 265-267
Perrow C. Normal Accidents: Living with high-risk technologies. Princeton, New Jersey: Princeton University Press, 1999
Patient Safety Tips of the Week pertaining to alarm-related issues:
March 5, 2007 “Disabled Alarms”
March 26, 2007 “”
April 2, 2007 “”
June 19, 2007 “Unintended Consequences of Technological Solutons”
April 1, 2008 “Pennsylvania PSA’s FMEA on Telemetry Alarm Interventions”
February 23, 2010 “Alarm Issues in the News Again”
March 2, 2010 “Alarm Sensitivity: Early Detection vs. Alarm Fatigue”
November 2010 What’s New in the Patient Safety World “Alarms in the Operating Room”