Medication Errors Involving Patient-controlled Analgesia

Rodney W. Hicks; Vanja Sikirica; Winnie Nelson; Jeff R. Schein; Diane D. Cousins


Am J Health Syst Pharm. 2008;65(5):429-440. 

In This Article


Between July 1, 2000, and June 30, 2005, Medmarx processed 919,241 medication error records from 801 facilities ( Table 1 ). Of these records, 9,571 (1%) were associated with PCA; these records came from 491 facilities (61.3%). Sixty-seven records in which "PCA" indicated something other than patient-controlled analgesia were excluded from the study. Even though there were annual increases in the number of records processed and the number of facilities reporting, the proportion of facilities reporting PCA errors remained flat at about 50%. The facilities that submitted records were mostly nongovernment, nonprofit hospitals and general community hospitals ( Table 2 ).

Error Severity. There were 624 records of PCA errors associated with harm (NCCMERP categories E through I), corresponding to 6.5% of the patients ( Table 3 ). Of the 624 errors, 19 (3%) were sentinel events (categories G through I), including 2 errors (reviewed in the qualitative analysis section) that could have caused or contributed to a patient's death. Harmful errors were most frequent between July 1, 2000, and June 30, 2001, the year when the fewest records were received.

Patient Demographics. More than 4200 PCA error records identified the sex of the patient; 59.6% of the records involved females and 40.4% involved males. Errors were reported for all age decades ( Table 4 ). More than half of the 5404 records identifying patient age concerned patients between 40 and 69 years. Percentages of errors peaked at an older age group for harmful errors than for nonharmful errors.

Types of Errors. In a review of records by year that identified at least one type of error, 88-100% of records included at least one selection from the multiselection error list. The leading type of error among all records was improper dosage or quantity (n = 3377 [38%]), indicating that a wrong amount of drug was involved ( Table 5 ). For the five-year period studied, two other leading error types had nearly equal frequencies, omission errors (17.4%) and unauthorized or wrong drug errors (17.3%); there was some minor variation in the rank order of these errors over the period. Prescribing errors were fourth most frequent (10.2%). Together, these four types of errors represented 7369 (82.9%) of all errors.

A cross-tabulation analysis of the types of errors explored their association with the medication-use process ( Table 6 ). A total of 682 errors (61.9%) were associated with prescribing, indicating that the medication order was the source of the error. Approximately one third (n = 516) of the omission errors originated during the transcribing phase. About a quarter (n = 255) of the dispensing errors involved an improper dosage or quantity. A majority of all errors (4807 [54.9%]) were associated with administration. Almost half (n = 2374) of these errors involved an improper dosage or quantity. Nearly one third of the errors in the monitoring phase were errors of omission (30.8%) or improper dosage or quantity (30.3%).

Causes of Errors. In a review of the percentage of records by year that identified at least one cause of errors, 85.7-100% of the records included at least one selection from the multiselection list. Each year a majority of PCA error records reported human factors as the leading causes of errors (n = 6669 [69.8%]). The second most frequent cause was faulty equipment (n = 1860 [19.5%]), followed by communication issues (n = 1563 [16.4%]); systems (n = 1216 [12.7%]); storage, labeling, and packaging (n = 1106 [11.6%]); documentation (n = 1022 [10.7%]); and name confusion (n = 112 [1.2%]).

Contributing Factors. A contributing factor influences the occurrence of an error but does not directly cause it. About one PCA error record in three (37.6%) identified at least one contributing factor ( Table 7 ). Two of the leading contributing factors were distraction (n = 713 [37.8%]) and workload increase (n = 371 [19.7%]). A distraction is any interruption in the medication-use process.

Six contributing factors were related to characteristics of the staff: inexperienced staff, shift changes, agency or temporary staff, insufficient staff, cross-covering staff, and floating staff. Collectively, staff-related issues accounted for 1100 (44.8%) of the errors for which contributing factors were identified. Inexperience was the most important staff-related contributing factor.

Even though patient transfer (patient movement from one clinical area to another) was not available to error reporters as a selection option in 2000 or 2001, by 2004 it was the fourth most frequent contributing factor, accounting for 42 (8.8%) of the errors in this category.

Actions Taken. Slightly less than half (45.9%) of the PCA error records identified at least one action taken as a direct result of the error. Reporters most often specified that the action consisted of informing the staff who made the error ( Table 8 ). Additional education and training followed an error 25.8% of the time. Actions taken were consistent in each of the years analyzed. "Informed patient's physician" was frequently specified in the last two study years.

Level of Care. An increased level of patient care is applicable only to errors that reach the patient (errors in categories C through I). Depending on the year, additional care was necessary as a direct result of the PCA error for 49.5-72.4% of the errors. Among the 458 harmful errors, the most common responses were changing the drug therapy regimen, including changing the analgesic drug or administering oxygen (n = 286 [62.5%]), and observation (n = 228 [49.8%]). In 26.9% of the cases (n = 123), antidotes or antagonists were needed. Other outcomes included increased length of stay (n = 54 [10.0%]), performance of a diagnostic test (n = 27 [4.8%]), a life-sustaining intervention [n = 18 (3.3%)], and interruption of care (n = 7 [1.5%]). Most frequently, no additional care was rendered in cases of nonharmful errors.

Drugs Involved. Between 75.3% and 95.6% of the error records identified at least one drug by generic name; morphine was named most frequently (n = 4495 [49.2%]) ( Table 9 ). Hydromorphone and meperidine were also frequently associated with the errors. The records also pointed to use of the opioid-receptor antagonist naloxone following the error, which is internally consistent with the patient outcome findings.

Staff involved. Nurses were involved in nearly three fourths of the errors (n = 6008 [73.0%]) ( Table 10 ). Dispensing personnel were the group involved second most frequently (n = 1150 [14.0%]).

Locations. Most errors (n = 6263 [71.3%]) originated on nursing units, followed by pharmacy locations (n = 995 [11.3%]) and the perioperative area ( Table 11 ). Within the perioperative area, the postanesthesia care unit was the locus of the most errors (n = 510 [83%]).

The qualitative review of each PCA error's description showed that a majority of the errors (n = 4687 [54.8%]) occurred during the administration phase, followed by transcribing (n = 1631 [19.1%]), prescribing (n = 1012 [11.8%]), dispensing (n = 1001 [11.7%]), and monitoring (n = 215 [2.5%]).

Prescribing. Prescribing errors comprised issues such as incomplete orders, duplicate orders, contradictory orders, and failure to adjust the order on the basis of laboratory test values. In one case, the prescriber failed to order the four-hour maximum-dose limit. Because of the oversight, the patient developed respiratory impairment, necessitating transfer to the intensive care unit (ICU). In another case, a patient was receiving morphine sulfate via PCA and had inadequate pain control, despite receiving 54 mg. The prescriber changed the product to hydromorphone but failed to perform dosage conversions. As a result, the patient received 36 mg before being found unresponsive. The patient was treated with two doses of naloxone and was mechanically ventilated in the ICU.

A surgical patient with underlying comorbidities of hypertension, renal impairment, and obesity received intraoperative fentanyl and two doses of hydromorphone in the postanesthesia care unit. The prescriber did not adjust the PCA for the renal impairment or the comorbid conditions. On the patient care unit, the patient received 12 mg of morphine sulfate over an eight-hour period and was found unresponsive. The patient had to be transferred to the ICU for respiratory support. The patient's history of sleep apnea was thought to have contributed to the episode.

Another patient was prescribed morphine sulfate by PCA. The case report described that family members also participated via PCA by proxy (people other than the patient activating the PCA system). Respiratory suppression resulted, necessitating intubation and transfer to the ICU. It was determined that the patient lacked the judgment to use PCA appropriately and that the family's PCA by proxy contributed to the event.

Excessive meperidine was implicated in one of the fatal cases involving a postoperative patient with multiple comorbidities being cared for by a surgical resident. Contributing to the event were the resident's inexperience and the continuation of multiple central-nervous-system medications in the surgical ICU.

Dispensing. Dispensing errors were associated with misfills of automated dispensing devices (having either the wrong concentration of a commercially available product or the wrong product in the machine's pocket), not having the product available when needed, and compounding the wrong strength. Dispensing errors generally did not result in harm. However, one harmful compounding error resulted from the lack of standard concentrations. In this case, a patient had terminal cancer, the pain of which was managed with morphine sulfate 20 mg every 30 minutes. The patient fell and sustained a fracture, resulting in a hospital admission accompanied by inadequate pain control. It was necessary to increase the morphine sulfate dosage to 26 mg every 30 minutes. As a result of the dosage increase, the infusion provided by hospice was completed earlier than expected, and a pharmacist had to resupply the morphine. The pharmacist changed the solution to a less concentrated infusion, labeled the bag, and dispensed it to the floor. The nurse caring for the patient replaced the empty infusion container with the new one but did not note the change in concentration and continued the infusion with the existing PCA-pump settings. As a result of the less concentrated solution, pain relief was inadequate.

Administration. A number of administration errors involved an improper dosage or amount. The error descriptions frequently cited incorrect PCA-pump programming. For example, in one case the intended dosage of meperidine was 2 mg every 7 minutes; however, the nurse programmed the pump to deliver 20 mg every 17 minutes. In another case, the intent was to deliver 0.5 mg of morphine sulfate every hour, but the pump was programmed to deliver 2 mg/hr. In yet another case, the PCA order was for meperidine 5 mg every 10 minutes, but the pump was programmed to deliver 5 mL every 10 minutes.

Several errors described a failure to match drug concentration with pump programming. For example, a patient was given a PCA order for morphine sulfate 1 mg/mL. The pump was correctly programmed for this; however, the nurse was using morphine sulfate 10 mg/mL. In a different case with slight variation, the nursing staff overrode the automated dispensing device and withdrew morphine sulfate 10 mg/mL. The staff then programmed the pump as if the concentration were 1 mg/mL.

Several case reports described omission errors that originated in the administration phase and indicated that the original order was overlooked. In other cases, the drug was not given because the machine was not turned on or the tubing was not connected to the patient. One patient's PCA was interrupted for a blood transfusion and not resumed when the transfusion was complete.

A cluster of records identified errors due to an unauthorized or wrong drug originating during administration. Many of these errors were attributed to look-alike or sound-alike drugs (e.g., morphine-hydromorphone and meperidine-morphine). Many of these errors involved nurses retrieving the wrong product from an automated dispensing device.

Morphine PCA was implicated in a fatality that occurred when the patient's spouse participated in pain management by depressing the button on the PCA pump, even while the patient was sleeping. A total of 48 mg was administered during one eight-hour shift, which was within the lockout amount ordered by the physician. During this shift, when a nurse questioned the spouse about the patient's snoring, the spouse answered that it was normal for the patient. At the change of shift, a nursing assistant found the patient apneic and unresponsive.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.