Device-Related Error in Patient-Controlled Analgesia

Analysis of 82,698 Patients in a Tertiary Hospital

Hyo-Jung Son, MD; Sung-Hoon Kim, MD, PhD; Jeong-Ok Ryu, RN, MS; Mi-Ra Kang, RN, MS; Myeong-Hee Kim, RN, MS; Jeong-Hun Suh, MD, PhD; Jai-Hyun Hwang, MD, PhD


Anesth Analg. 2019;129(3):720-725. 

In This Article


This retrospective study was approved by the Asan Medical Center Institutional Review Board. The requirement for written informed consent was waived by the institutional review board. Data were obtained from the electronic medical record database of the acute pain service (APS). The APS team comprises 3 nurses and staff anesthesiologists. Twice a day, APS nurses perform rounds of all postoperative patients using PCA devices and visit the patients during the daytime on weekdays. During routine rounds, the APS nurses interview each patient and review their medical records. Using a numeric rating scale, the APS nurses also check and record the scores for pain with and without movement and also note the amount of PCA medication delivered, symptoms and signs of opioid overdose, and any additional pain control medications administered. When an error or adverse event suddenly occurs during the use of PCA device, the APS team members receive direct calls from the clinician or ward nurses. The APS nurses notify the staff anesthesiologist about significant safety errors or problems that they cannot manage. All events were systemically recorded during nurse rounding sessions.

Two anesthesiologist reviewers (H.J.S., S.-H.K.) obtained and independently reviewed all events of PCA device-related errors from January 1, 2011 to December 31, 2014. Intravenous PCA, epidural PCA, and nerve block PCA were applied to patients. The number of prescriptions for PCA devices during the study period was considered as the total number of cases. In our institution, we have established a routine PCA safety protocol to reduce and rule out human error during PCA administration. First, at least 2 health care providers, typically a doctor and a nurse, mix and dispense the medication together. Second, after dispensing, the providers check the PCA device together and record information in a time-out sheet; this includes PCA setting, drug dosage, and priming status. When the total volume of the drug cannot be identified, the whole PCA device, including the box case and accessory, should be weighed before the next administration to a patient. Third, when a patient is admitted to, and discharged from, the postanesthetic recovery unit (PACU), a PACU nurse routinely checks the PCA program and remaining drug volume. Fourth, no person, except the anesthesiologist and APS team nurse, may manipulate or change the PCA.

Human error was defined as prescription error (use of wrong analgesic, incorrect dosing, duplicate medication orders) and operating error (accidental pump misprogramming, false triggering by proxy). After ruling out human error, remaining cases were regarded as device-associated error, and each case was confirmed to be mostly associated with the PCA device component. Selected cases were classified into 4 categories based on the error mechanism: overflow (from any cause wherein >5% of the intended amount of drug was given), underflow (from any cause wherein <5% of the intended amount of drug was given), display error (error displayed on the window, or activated alarm signal, without actual flow rate change; eg, error message or alarm without any cause, frozen program key button, abrupt shut down with full charging), and broken pump device (physical breakdown or crack of PCA infuser or accessories). When any discrepancy regarding case categorization was noticed between the reviewers, it was resolved through discussions with a third reviewer (J.-H.H.). In our hospital, 4 types of PCA devices are applied to patients: elastomeric balloon infuser (Accufuser Plus; Woo Young Medical, Seoul, Korea); carbon dioxide gas-driven infuser (ANAPA; EHWA Biomedics, Seoul, Korea); semielectronic disposable pump (AutoMed3400; Ace Medical, Seoul, Korea); and electronic programmable pump (Accumate 1100; Woo Young Medical, Seoul, Korea). All 4 PCA device types have acquired Conformité Européene certificate, International Organization for Standardization certificate, and Good Manufacturing Practice certificate. Staff anesthesiologists decided which type of PCA device to apply depending on the type of operation, patient demographics, and expected postoperative pain severity.

The clinical outcomes were determined based on clinical records up to postoperative day 3. When the PCA error was detected later than postoperative day 3, additional data of 3 days before and 1 day after error detection were collected. The cases were classified into 4 categories: minimal physiological change, minor side effect of opioid, major physiological change, and inadequate analgesia. The definitions of each category are as follows. Minimal physiological change was defined as "no side effect or detectable patient symptom caused by the error in PCA error." Minor side effects of opioids included nausea, vomiting, pruritus, and other opioid side effects not involving any major systemic organs. Major physiological changes included side effects involving cardiovascular, respiratory, neurological, or other major system functions. Inadequate analgesia was defined as an numeric rating scale pain score of >5 or the administration of additional pain control. Cases could belong to ≥2 categories. We compared the clinical progress using vital sign records and pain scores. Although confirming the correlation between the error in PCA and the clinical outcome of the patient in each case can be challenging in a retrospective setting, we tried to include all adverse clinical events.

The rate of PCA device error of each type was calculated using confirmed cases of device error of each device type as the numerator, and the number of prescriptions of PCA devices as the denominator. Poisson regression was used to compare incidences of device error among various PCA device types during the entire interval. Based on the exact Poisson method, 95% confidence intervals (CIs) were calculated. A P value of <.05 was considered to indicate statistical significance.

The primary aim of this study is to estimate the incidences, so that we could justify the sample size in terms of the accuracy of estimating the incidence using CI width. According to the previous report,[8] error rate of electronic PCA was 0.25%. Based on our clinical experience, error rate of nonelectronic type PCA was expected to be approximately two thirds as large as the electronic type. Both types of devices attempted to achieve a CI width of 0.1%. Therefore, assuming that error rate with the electronic pumps is 0.25% and the error rate with nonelectronic pumps is expected relative 35% lower (0.16%), the required sample size is 40,414 and 26,692, respectively. Thus, we can see that the actual sample size in the study (N = 49,364, 33,334) is all appropriate.