In a recent discussion after a naloxone reversal of excess opioid therapy, the question was raised as to why a patient had not been on patient-controlled analgesia (PCA). PCA has been a major development in pain management and, for the most part, has resulted in improved pain control and patient satisfaction, shorter lengths of stay, and better utilization of resources. Though having the theoretical “built-in” advantage that should prevent inadvertent narcotic overdose (i.e. as the patient becomes drowsy, he/she cannot press the button to infuse more narcotic), use of PCA pumps have been associated with multiple problems of their own. PCA is another example where new technologies effectively eliminate some problems but introduce new problems of their own.
In our September 9, 2008 Patient Safety Tip of the Week “Less is More….and Do You Really Need that Decimal?” we discussed many of the safety considerations related to use of PCA pumps. We referred to ISMP’s outstanding monograph about patient safety issues involved in PCA, published in 2006.
So just how frequent are errors and safety issues associated with use of PCA? An analysis of one year’s worth of reports to the MEDMARX® database (Hicks et al 2008) found 1% of all reports related to PCA. However, 6.5% of these were associated with harmful outcomes, compared to 1.5% for all other errors in the MEDMARX® database. That makes sense because all the medications being delivered via PCA are high-risk medications. They also noted that a greater percentage of harm occurred in older patients.
One of the problems in determining the frequency of adverse events related to PCA is that the MEDMARX® database is a voluntary reporting database and likely considerably underestimates the true frequency. A new paper (Meissner et al 2009) has actually quantified both the frequency and costs of errors attributable to PCA. They used data from both the MEDMARX® database and the MAUDE database (an FDA database that collects mandatory reports on device-related errors), then used a correction factor to adjust for the likely underreporting in both databases. They estimated the rates of PCA-related errors and device-related errors, respectively, at 407 per 10,000 people and 17 per 10,000 people. The average cost per error was $733 for errors in the MEDMARX® database and $552 for those in the MAUDE database. Given that about 30% of surgical patients use PCA (the most common situation for PCA use), they estimate that errors related to PCA may cost the entire United States healthcare system about $400 million annually.
The Hicks paper categorized the types of errors seen. Improper dosage or quantity was the most frequent type of error, but omission errors, unauthorized or wrong drug errors, and prescribing errors were also frequent. Errors in the monitoring phase were also frequent. Errors related to human factors were the leading causes for errors, accounting for almost 70% of causes listed. But also seen were faulty equipment, communication issues, problems with storage/handling/packaging, documentation, and name confusion. Errors were seen with the gamut of opioid drugs.
Other contributing factors were distraction, workload, inexperienced or temporary staff, shift changes, and cross-coverage. Many errors were also noted to occur at the time of patient transfer from one clinical area to another.
They provide actual clinical examples of the errors in the various phases of the medication cycle. Of particular note, problems associated with dosage conversions (when converting from one narcotic to another), LASA (look-alike sound-alike drugs) confusion (eg. confusion morphine-hydromorphone and meperidine-morphine), and retrieving the wrong medication from automated dispensing machines were highlighted. The latter issue will hopefully be minimized as more facilities move to barcoding/bedside medication verification systems but still suggests that healthcare organizations need to pay careful attention to overrides and their automated dispensing systems. Restricting the number of medications used for PCA may also help minimize some of these errors. Tall-man lettering may help avoid some of the LASA drug-pair confusion.
Dose miscalculations and device programming errors are frequent. In our September 9, 2008 Patient Safety Tip of the Week “Less is More….and Do You Really Need that Decimal?” we discussed some specific examples of such errors. We noted the ISMP Safe Medication Alert “Misprogramming PCA concentration leads to dosing errors” that highlighted a paradoxical problem with PCA pump programming. If one programs in too high of a concentration, the patient tends to get underdosed (so may suffer continued pain). If one programs in too low a concentration, the patient actually gets overdosed! This seems counterintuitive. But think about it – the patient asks for a certain dose of the narcotic and the pump delivers the volume it is programmed for to meet that request. If the concentration was erroneously too low, the pump has now given a higher volume and, hence, a higher actual narcotic dose. And often a warning on the pump that the concentration is too low may be overridden because the nurse or physician feels less concerned about “too low” than “too high”. Also, in our March 12, 2007 Patient Safety Tip of the Week “10x Overdoses” we pointed out another potential problem with misprogramming PCA (or other infusion) pumps. The data entry person may double press a key (or the key may become stuck) resulting in, for example, “88” instead of “8”. Also, during data entry it is possible to think one hit a decimal point but it fails to print out. These types of data entry error have recently been noted in programmable intravenous infusion pumps and there have been several occurrences of 10x overdoses with those pumps.
For a variety of reasons, using independent double checks is very important when dealing with high-risk drugs and PCA. But keep in mind that the error rate for the person doing the second check may be as high as 10% (based on data on double checks in almost any industry). Such double checks should take place not only at the time the preparation is made and when it is first set up on the PCA pump but also any time there is a change in dosage or rate, when a patient is moved from one site to another, at change of shift, etc.
The Hicks paper also gave an example of “PCA by proxy”. “PCA By Proxy” is perhaps the best known safety issue with PCA. This, of course, means the pressing of the infusion administration button by someone other than the patient. This is most often a friend or family member but could be a member of the healthcare team. In most cases, the person pressing the PCA button thinks they are helping the patient avoid pain. They may not recognize the problem of overdosage from the narcotics. One of the “built-in” safety features of PCA is that when a patient gets sedated from too much analgesic, they can no longer press the PCA button to get more analgesic. That safety measure is bypassed in “PCA by Proxy”. In fact, the occurrence of incidents involving “PCA by Proxy” was significant enough for Joint Commission to issue a Sentinel Event Alert in 2004.
The ISMP monograph about patient safety issues involved in PCA published in 2006 also discusses other issues regarding misprogramming PCA pumps and several other issues, including selecting appropriate patients for PCA, monitoring patients, setting up quality indicators, and performing FMEA.
Recently, ISMP had a Medication Safety Alert: Beware of basal opioid infusions with PCA therapy. This included both a case presentation and excellent discussion of many of the problems that can be associated with PCA. In particular, the patient had both obesity and sleep apnea, 2 conditions which predispose to significant risk during PCA, and monitoring for respiratory depression was inadequate. Use of a basal infusion in this opioid-naïve patient also predisposed to the ultimate respiratory depression that occurred. The ISMP article, and several before it, point out that basal infusions are seldom necessary in PCA and may be dangerous. That article also has a nice table of the risk factors for respiratory depression in PCA patients including:
· use of basal infusion
· advanced age
· upper abdominal surgery
· obstructive sleep apnea
· concurrent use of CNS depressants
· impaired renal/pulmonary/hepatic/cardiac function
· pump programming errors
· PCA by proxy
· lack of opioid tolerance
Actually, all the risk factors for hypercapnic respiratory failure we noted in our June 10, 2008 Patient Safety Tip of the Week “Monitoring the Postoperative COPD Patient” and our January 27, 2009 Patient Safety Tip of the Week “Oxygen Therapy: Everything You Wanted to Know and More!” should also be considered relative contraindications for PCA. These include COPD, certain neuromuscular disorders, chest wall deformities, massive obesity, and obstructive sleep apnea. These are not absolute contraindications for PCA, but they do require that such patients would have more intensive monitoring. The same applies to patients requiring high doses of opiates regardless of whether they are receiving them via PCA or otherwise. After identifying high-risk patients, one must use sedation scales properly, and consider using capnography in addition to pulse oximetry in such high-risk patients.
Pasero and McCaffery (2002) discuss risk factors for opioid-induced respiratory depression and keys to monitoring such patients. Infants less than 6 months old, opioid-naïve elderly patients, and patients with coexisting conditions such as COPD, sleep apnea, or major organ failure are at increased risk of respiratory depression. In addition, drugs such as intramuscular opioids, muscle relaxants and anxiolytics, benzodiazepines, sedating antihistamines, and some antiemetics may increase the risk for opioid-induced respiratory depression. In their discussion of monitoring, they provide the following very practical sedation scale:
S = Sleep, easy to arouse (acceptable; no action necessary)
1 = Awake and alert (acceptable; no action necessary)
2 = Slightly drowsy, easily arousable (acceptable; no action necessary)
3 = Frequently drowsy, arousable, drifts off to sleep during conversation (unacceptable; decrease opioid dose by 25-50%, add an opioid-sparing analgesic, and monitor the patient’s level of sedation and respiratory status closely)
4 = Somnolent, minimal or no response to physical stimulation (unacceptable; stop opioid, consider administering naloxone)
PCA pumps or other judicious use of opiates may be very important for patient care. However, one must recognize the potential dangers and ensure proper patient and drug selection, appropriate training of staff, patients and families, proper safeguards and good monitoring techniques.
Cohen MR, Weber RJ, Moss J (Institute for Safe Medication Practices). Patient-Controlled Analgesia: Making it Safer for Patients. A continuing education program for pharmacists and nurses. ISMP. April 2006 http://www.ismp.org/profdevelopment/PCAMonograph.pdf
Hicks RW, Sikirica V, Nelson W, Schein JR, Cousins DD. Medication errors involving patient-controlled analgesia. American Journal of Health-System Pharmacy 2008; 65: 429-440
Meissner B, Nelson W, Hicks R, Sikirica V, Gagne J, Schein J. The rate and costs attributable to intravenous patient-controlled analgesia errors. Hosp Pharm. 2009;44:312–324
ISMP. Misprogramming PCA concentration leads to dosing errors. Medication Safety Alert Newsletter (Acute Care Edition). August 28, 2008 http://www.ismp.org/Newsletters/acutecare/articles/20080828.asp
Joint Commission. Sentinel Event Alert. Patient controlled analgesia by proxy. Issue 33. December 20, 2004
Institute for Safe Medication Practices (ISMP). Beware of basal opioid infusions with PCA therapy. Medication Safety Alert. Acute Care Edition. March 12, 2009.
Pasero C, McCaffery M. Monitoring Sedation: It's the key to preventing opioid-induced respiratory depression. American Journal of Nursing. 2002; 102(2):67-69