This study was approved by the SingHealth Centralized Institutional Review Board, Singapore (SingHealth CIRB Ref: 2015/3062), and registered on Clinicaltrials.gov (NCT02804022). Written informed consent was obtained from every patient before any study procedure. The study period was between January 2017 and June 2017 and was conducted at KK Women’s and Children’s Hospital, Singapore
We recruited female patients aged 21 to 70 years old with American Society of Anesthesiologists (ASA) status I or II, undergoing elective surgery and intending to use postoperative PCA with morphine for postoperative analgesia. The exclusion criteria were patients with allergies to morphine, history of significant respiratory disease or obstructive sleep apnea, unwilling to wear oxygen saturation monitoring devices throughout the study duration, pregnancy and unable to comprehend the use of PCA. Recruitment was performed either in the pre-operative assessment clinic or on the same day of surgery if they had not attended pre-operative assessment clinic. An information brochure describing the use of VPIA analgesic infusion pump for post-operative analgesia, including potential side effects and complications was provided to patients.
Setting up of Infusion Pump
The algorithm and the VPIA analgesic infusion pump (“Intellifuse pump”; Model: Opiva) was designed by Innovfusion Pte Ltd, Singapore (Figure 1). Intravenous morphine used in the VPIA analgesic infusion pump was administered according to institutional guidelines: morphine diluted in normal saline to a concentration of 1 mg/ml, with bolus doses of 1 mg morphine delivered as per patient demand. In the VPIA analgesic infusion pump, vital signs monitoring (oxygen saturation, pulse rate) was programmed into the VPIA variable lockout algorithm, in which a temporary pause to the pump was triggered when vital signs safety threshold was breached; and subsequently the lockout interval was increased upon re-starting thereby improving the safety of intravenous morphine administration. That means when the vital signs were within normal range, the system was able to increase or decrease the lockout interval according to the analgesic needs of the patient. This lockout interval was a safety mechanism that limited the frequency of demands. By allowing an adequately long interval between each dose, patients were given sufficient time to achieve the opioid’s effects before the next dose. However, if the interval were prolonged, the effectiveness of patient controlled analgesia would be reduced.
The detailed VPIA variable lockout algorithm was illustrated in Figure 2. The bolus lockout interval was empirically set at 7 minutes, and was adjusted automatically according to the patient successful demands and the patient safety in the event of abnormal vital signs monitoring. The monitoring data was performed by taking average epochs of 15 seconds to summarize the vital signs. Missing vital signs were dropped from the analysis. However, if there was no available vital sign for the whole duration of each epoch, a safety pause would be triggered. The VPIA variable lockout algorithm reassessed for the recovery of the patient’s’ vital signs at the end of the safety lockout period. If the patient’s vital signs did not recover to safe levels, the pump automatically raised the on-board alarm. Conversely, if the patient’s vital signs recovered beyond the threshold limits by the end of the safety lockout period, the lockout interval would be prolonged. If at any time, there were critically abnormal vital sign parameters, the system would trigger the “emergency safety stop” function to cease the patient’s boluses. The system would be manually restarted by the clinician or the nurse after reviewing the patient.
All patients had established intravenous access before surgery. The patients were instructed on the use of the VPIA analgesic infusion pump prior to the study and educated to press the demand button whenever they needed pain relief. While in the recovery room after surgery, the VPIA analgesic infusion pump was secured with a 50 ml syringe filled with1 mg/ml morphine that was connected to the patient’s intravenous line for analgesia. The patient’s oxygen saturation and heart rate was continuously monitored by the VPIA analgesic system at least for 24 hours after surgery.
We collected and analyzed three sets of data: (1) patient demographic, surgical and anesthetic characteristics; (2) VPIA analgesic infusion pump data including opioid consumption, patient demands, successful demands, pattern of demands, oxygen saturation and heart rate; and (3) VPIA analgesic infusion pump user feedback survey that was conducted at the end of the study.
Following the initiation of VPIA analgesic infusion pump, attending nurses who were educated on the usage and side effects of morphine therapy would monitor the patient at regular intervals and document pain scores (0-10 numeric rating scale), blood pressure, heart rate, oxygen saturation and sedation score (0 for “awake, alert”, 1 for “occasionally drowsy, easy to rouse, responds to calling”, 2 for “occasionally drowsy, difficult to rouse, responds to shaking only”; 3 for “unresponsive and unarousable-- defined as no response to voice or physical stimulation”; D for “distressed -- defined as awake and in great pain”). Side effects such as nausea and vomiting were also recorded.
An independent observer would assess the patient during the period she was placed on the VPIA analgesic infusion pump. The patient’s overall satisfaction (numerical score between 0-100%) with the postoperative analgesia provided, the feedback on pain relief effectiveness and any side effect from the therapy were also gathered. Once the indications for PCA opioid for pain management were deemed unnecessary by the patient’s primary care team, the VPIA analgesic infusion pump was disconnected.
Sample size calculation and statistical analysis
The primary outcome measure of the study was the incidence of oxygen desaturation (defined as oxygen saturation < 95% in a patient for more than 60 seconds) in patients using the VPIA analgesic infusion pump. The secondary outcome measures were bradycardia, sedation, nausea/vomiting, pain scores, total consumption of morphine, patient’s satisfaction score. Patient’s oxygen saturation and heart rate were measured every minute for at least 24 hours. Assuming that a patient had < 3% oxygen desaturation, 0.005 width of interval and 95% confidence interval, we would require 18,000 oxygen saturation readings. Each patient would provide at least 1200 readings. Therefore, the study was adequately powered for 18 patients with 1200 readings (= 18 X 1200 ~ 21,600) for both primary and secondary aims.
Patient demographics, surgical and anesthetic characteristics were summarized as frequency with corresponding proportion, as mean ± standard deviation (SD) or median [range], whichever applicable. The incidence rate and 95% confidence intervals (CI) of binary outcomes (such as desaturation, bradycardia) were estimated based on the exact method by Wilson , which demonstrated to have good statistical properties even for small number of subjects and/or extreme probabilities [12, 13]. Significance level was set at 0.05 and all tests were two-tailed. SAS version 9.3 software (SAS Institute; Cary, North Carolina, USA) was used for the analysis.