STRENGTHS AND LIMITATIONS OF THIS STUDY
-
The outcome area under curve for hypoxaemia selected in the study characterises hypoxic duration and severity.
-
Wireless wearable devices, which provided continuous 48-hour monitoring for patients with patient-controlled intravenous analgesia, can collect more accurate hypoxic data.
-
The single-centre setting may limit the general significance of the results.
Introduction
Patient-controlled intravenous analgesia (PCIA) has exhibited great effectiveness, convenience and patient satisfaction for acute postoperative pain management. Thus, it remains a commonly used system, especially for moderate to severe pain.1–3 However, PCIA is accompanied by side effects such as nausea, vomiting and postoperative opioid-induced respiratory depression (POIRD),4 which are of great concern. Therefore, basal infusion PCIA of opioids is not routinely recommended for opioid-naive patients according to clinical practice guidelines.1 Previous studies have shown that basal infusion, compared with no-basal infusion PCIA, did not improve the analgesic effect for patients but rather increased the risk of POIRD in adults.5 6 However, many of the studies supporting this observation5 7–10 employed a relatively high rate of basal infusion of morphine (~1–2 mg/hour) for gynaecological surgeries. It remains unclear whether low-basal infusion PCIA of opioids may benefit the patients without increasing such risk.
Currently, hydromorphone is widely used in postoperative PCIA. It is a semisynthetic opioid exhibiting approximately 5–10 times greater analgesic potency and fewer side effects than morphine.11 12 As early as 1989, Sinatra et al observed that patient-controlled administration of an extra low-basal infusion of hydromorphone significantly reduced the pain scores reported by patients postcaesarean delivery.13 Moreover, a recent randomised controlled trial (RCT)14 found that PCIA with a low-basal hydromorphone infusion (~0.12 mg/hour) provided better pain relief for patients undergoing video-assisted thoracoscopic surgery than no-basal infusion PCIA. Similar results were reported for patients undergoing major abdominal surgery.15 16 We thus hypothesised that patients undergone ‘more painful’ surgery might benefit from low-basal infusion (0–0.1 mg/hour of hydromorphone) PCIA. However, few RCTs have explored the effects of low-basal infusion PCIA on the incidence of POIRD. Answering this question will give evidence on the safety of using low-basal infusion PCIA in such group of patients. Since monitoring of SpO2 is commonly used to assess the severity of POIRD in clinical and research settings,17–19 the area under curve (AUC) for SpO2 value (lower than predefined threshold) is used as a scientific assessment for both the severity and duration of hypoxaemia, because both longer duration and lower SpO2 would result in a larger area under the threshold of predetermined SpO2.19 20 Based on the above, this RCT is aimed to investigate whether a low-basal infusion of hydromorphone PCIA would be non-inferior to no-basal infusion concerning AUC for hypoxaemia within the first 48 hours after gastrointestinal tumour surgery.
Material and methods
Study design
We conducted a randomised parallel-group non-inferiority trial. It was successfully registered with the Chinese Clinical Trial Registry (ChiCTR2100054317; URL: https://www.chictr.org.cn/showproj.html?proj=141259) on 13 December 2021, prior to the initial enrolment of patients. The complete trial protocol21 has been published.
Participants
A total of 160 patients scheduled to undergo major gastrointestinal tumour surgery at Xijing Hospital (Xi’an, China) were enrolled in this trial that was conducted between December 2021 and August 2022. Inclusion criteria were: >18 years old, American Society of Anesthesiologists grade I–III, body mass index (BMI) between 18.5 kg/m2 and 30 kg/m2, underwent elective gastrointestinal tumour resection and received postoperative PCIA. Exclusion criteria were as follows: patients with chronic severe respiratory disease (SpO2<90% in room air), other conditions for careful opioid use (including abuse of sedatives and analgesics), opioid allergy, hepatic (liver enzyme concentrations twice the normal values) or kidney disease (serum creatinine concentration >140 (males) or 130 (females) µmol/L and oliguria/anuria) or renal replacement therapy, pregnancy or breastfeeding. Patients with an operation time >5 hours or who were transferred postoperatively to the intensive care unit were also excluded. Patients who participated in a clinical trial within 3 months prior to the current trial were excluded as well. The eligible subjects were explained the purpose and risks associated with the trial and provided with education about PCIA and wireless wearable devices.
Randomisation and blinding
The enrolled patients were randomly allocated (ratio 1:1) to receive PCIA with a low-basal or no-basal infusion. A randomisation sequence was generated using R software and randomly sized blocks of 2 and 4. All the patients and investigators were blinded to the group allocations. An anaesthetic nurse who did not participate in any other trial process assigned the PCIA pumps according to the randomisation number received in a sealed envelope. There was no difference in appearance between the PCIA pumps used in the two groups.
Anaesthesia
Both induction and maintenance of general anaesthesia were standardised. Every patient underwent a standard ECG, Narcotrend and invasive blood pressure monitoring. Midazolam (0.02 mg/kg), propofol (1–2 mg/kg), etomidate (0.15–0.3 mg/kg), sufentanil (0.3–0.5 µg/kg) and rocuronium (0.6–0.9 mg/kg) were used for induction of general anaesthesia. The anaesthesiologist adjusted concentrations of the anaesthetics to maintain the Narcotrend index within the range of 40–60. Perioperative multimodal analgesia regimen, postoperative nausea, vomiting prevention measures and reversal of muscle blocker actions on completion of the operation were performed according to routine practice.
Postoperative analgesia and monitoring
An allocated anaesthetic nurse prepared the PCIA pumps after confirming that the patients met the recruitment requirements at the conclusion of the procedure. The PCIA medication included 10 mg hydromorphone dissolved in 100 mL of saline, which made 0.1 mg/mL hydromorphone. Either a no-basal infusion PCIA with a demand dose of 2.0 mL (0.2 mg of hydromorphone) or a basal infusion PCIA of 1 mL/hour (0.1 mg/hour of hydromorphone) with a demand dose of 1.0 mL (0.1 mg of hydromorphone) was administered to the patients. The lockout interval for both groups was set as 10 min. When the patients entered the recovery room, they received PCIA treatment. Subsequently, wearable wireless devices provided continuous monitoring (ePM/ep pod; for more information, see published protocol)21 in the surgical ward for 48 hours, or until discharge from the hospital. This wireless system continuously monitored SpO2 and ambulation data, which could be downloaded to a laptop computer. Ward nurses performed standard care management with the patients receiving normal oxygen supplements with 2 L/min on arrival at the surgical wards. The attending surgeon administered oxygen therapy (including when to adjust the oxygen flow or stop treatment), and all of the ward medical staff were blinded to the allocation. According to the Enhanced Recovery After Surgery (ERAS) protocol, all subjects were encouraged to mobilise, take oral fluids and solids early and were prescribed intravenous flurbiprofen (50 mg) every 8 hours for 3 days postoperatively. Additional administration of hydromorphone (intravenous 0.2 mg) was available as a rescue analgesia regimen if required (ie, for a pain score >4). All postoperative analgesia was recorded.
Postoperative follow-up
A follow-up investigator visited the patients three times a day (including weekends and evenings), collected trial-related data, and ensured compliance with wireless monitoring. The following parameters were recorded during follow-up: numeric rating scale (NRS) scoring (ranging from 0 (no pain) to 10 (severe pain)) at 4 hours, 24 hours and 48 hours postoperatively; Overall Benefit of Analgesia Score (OBAS)22 (which measures analgesic effectiveness) at 24 hours and 48 hours postoperatively; opioid side effects and patient satisfaction. Additional parameters were recorded and included: withdrawal of PCIA (due to severe adverse reactions); duration of oxygen therapy during PCIA; postoperative time of first flatus and first liquid food intake without vomiting; patients’ satisfaction scores for wireless monitoring and other relevant records of trial-related postoperative medication. Another researcher used patient interviews and electronic medical records to determine patient demographics, preoperative laboratory test results, procedure-related data and postoperative length of stay.
Outcomes
The primary outcome was integrated AUC for hypoxaemia per hour (defined as SpO2<95%) during a 48 hour-postoperative period (or until patient discharge from the hospital). The same time frame was also applied to all secondary outcomes. The secondary outcomes were defined as integrated AUC values for hypoxaemia per hour (SpO2<90% and <85%). Belcher et al first used integrated AUC for hypoxaemia per hour as an important outcome to compare different effects of long versus short-acting opioids patient-controlled analgesia (PCA) on postoperative hypoxaemia and gave the reference to our study to set the non-inferiority margin.20 Continuous SpO2 data with a 1 min interval were cleaned using the Gaussian kernel in MATLAB R2016b (MathWorks, USA), with filtering of the raw SpO2-time curve. As shown in online supplemental figure S1 in the supplementary, an integrated AUC for hypoxaemia was calculated based on the smoothed SpO2-time curve. Integrated AUC values, which characterised the duration and severity of hypoxaemia, were subsequently calculated based on smoothed SpO2-time curves (more details were shown in the published protocol).21 Other secondary outcomes included dosage of hydromorphone administered and average ambulation time per hour within 48 hours postoperatively. In addition, NRS scores (evaluated during rest and movement) were recorded at 4 hours, 24 hours and 48 hours, while OBASs were recorded at 24 hours and 48 hours.
Supplemental material
Statistical analysis
According to a previous study,23 the primary outcome, AUC for hypoxaemia per hour (SpO2<95%), exhibited a log-normal distribution. In addition, non-inferiority tests for the ratio of two means were used to test the non-inferior efficacy of log-normal distribution variables.24 PASS 2015 (NCSS, USA) was used to calculate the sample size. A minimum of 160 patients (n=80/group) were necessary to reach 80% power to demonstrate non-inferiority (non-inferiority margin: ROM of 1.25; coefficient of variation: 0.5) of low-basal infusion PCIA compared with no-basal infusion PCIA concerning hypoxaemia. In addition, a one-sided significance level of 0.025 and an anticipated dropout rate of 10% were considered.
Continuous variables were analysed using the Kolmogorov-Smirnov test to test the normality first. Then, skewed distribution data were shown as median with IQR values, whereas normally distributed data were presented as mean±SD. Frequencies, percentages and ratios were used to represent count variables. For baseline variables, the continuous variables of normal distribution were compared using the Student’s t-test. The continuous variables of skewed distribution were compared using the Mann-Whitney U test, and χ2 test or Fisher exact probability method was used for the classified variables. Primary analysis was conducted on an intention-to-treat (ITT) and per-protocol (PP) set basis. The Mann-Whitney U test was employed to compare the primary outcome between the two groups, namely AUC for hypoxaemia per hour (SpO2<95%). If the difference was significant, the ROM between the groups was calculated with a 95% CI. CI limits with a predefined ROM of 1.25 were compared to decide whether or not to reject the null hypothesis. Subgroup analyses were conducted to consider baseline variables such as SpO2, age, BMI, surgery site and laparoscopic surgery (posthoc variable) using generalised linear models to test interactions between the treatment effect and these five factors. The Mann-Whitney U test was applied to assess any variations in AUC values for other hypoxaemia thresholds. Hydromorphone consumption and ambulation time per hour were subsequently compared using the Mann-Whitney U test and Student’s t-test, respectively. Due to the narrow ranges of the results, a generalised linear mixed model (GLMM) with multinomial distribution was used for within-group comparison of repeated measures variables, including OBAS and NRS scores. Patient identification was a random effect in the GLMM, while fixed factors included groups, time and groups-by-time.
All data were analysed using IBM SPSS V.25 (IBM, USA), and all the figures were made by R software, GraphPad Prism 7.0 (GraphPad Software, La Jolla, California) and Matlab R2016b software (MathWorks, USA). The two-sided significance level was set at p<0.05.
Patient and public involvement
No patients were involved in setting the research question nor in the design, conduct or interpretation of the study.
Results
Between December 2021 and August 2022, 195 patients were screened for enrolment in this trial. A total of 160 patients qualified and were randomised to receive either low-infusion PCIA or no-infusion PCIA (n=80/group) (figure 1). For PP analysis of the primary outcome, one patient in the no-basal infusion group was excluded due to a second operation for postoperative haemorrhage. Patient characteristics and perioperative details were similar between the low-basal and no-basal infusion PCIA groups (table 1).
CONSORT flow chart. CONSORT, Consolidated Standards of Reporting Trials; ICU, intensive care unit; PCIA, patient-controlled intravenous analgesia.
Patients’ characteristic
Table 2 presents the main findings. The results of the ITT analysis demonstrate that AUC for hypoxaemia per hour (SpO2<95%) in the low-basal infusion group was significantly higher than that in the no-basal infusion group, with a median difference of 0.097 (95% CI 0.001 to 0.245; median (IQR): 0.42 (0.09–0.82) %-hour versus 0.22 (0.03–0.57) %-hour, respectively; p=0.035) (table 2, online supplemental figure S2). The ROM between the two groups was 2.146 (95% CI 2.138 to 2.155) (table 2 and figure 2), with the lowest limit being worse than the ROM 1.25. Thus, low-basal infusion PCIA was inferior to no-basal infusion PCIA with regard to postoperative hypoxaemia (table 2 and figure 2). The AUC value for hypoxaemia was also higher in low-basal group than in no-basal group in the PP analysis (0.42 (0.09–0.82) %-hour versus 0.22 (0.03–0.57) %-hour, respectively; p=0.043) (table 2), consistent with the ITT analysis results. The PP analysis results (ROM: 2.089; 95% CI 2.080 to 2.097) (table 2 and figure 2) further indicate that non-inferiority of low-basal infusion PCIA was not determined. A subgroup analysis for primary outcome did not find a significant interaction between the treatment effect of low-basal infusion versus no-basal infusion and four variables (age, BMI, baseline SpO2 and surgery site) or a post-hoc variable (laparoscopic surgery), with p>0.10 for all of the interactions (online supplemental figure S3). As for secondary outcomes related to hypoxaemia (AUC per hour at SpO2 of 90% and of 85%, respectively), there was no significant difference between the two groups (table 2).
Ratio of means of primary outcome for ITT and PP populations in the two treatment groups. The small blue squares and the black short lines connected to them represent the estimated value of ROM and 95% CI, respectively. The dotted line indicates the preset non-inferiority threshold of 1.25. When the ROM and 95% CI overlap with the blue shaded area, low-basal infusion is non-inferior to no-basal infusion with regard to hypoxaemia. ITT, intention-to-treat; PP, per-protocol; ROM, ratio of means.
Overview of statistics regarding primary outcome according to study group
The distribution of SpO2 for both patient groups is shown in online supplemental figure S4 in the supplementary as a function of postoperative time, from which it is obvious that more patients with hypoxaemia are in the low-basal group. For example, when the hypoxaemia threshold was set at SpO2<95%, approximately 62.5% of the patients in low-basal group and 48.8% of the patients in no-basal group experienced hypoxaemia for at least 10 min/hour (online supplemental figure S4A). Furthermore, 13.8% of the patients who received no-basal infusion and 21.5% of the patients who received low-basal infusion experienced hypoxaemia lasting at least 20 min. Similarly, when the same duration of hypoxaemia and a threshold SpO2 between <90% and 85% was considered, there were more patients in the low-basal group than in the no-basal group (online supplemental figure S4B,C).
An analysis of the secondary outcomes showed that hydromorphone consumption in the low-basal infusion group was significantly higher (up to three times higher) than that in the no-basal infusion group (5.2 (4.8–5.4) mg versus 1.6 (0.8–2.8) mg, respectively; p<0.001) (table 2). Similarly, as a post-hoc, hydromorphone consumption per hour was significantly higher in the low-basal group (59.1 (42.5–86.5) versus 14.6 (9.6–27.4) µg/hour, respectively; p<0.0001). In contrast, no significant differences were observed for the AUC per hour under SpO2 of 90% and 85%, as well as for ambulation time (p=0.407, 0.440 and 0.416, respectively) (table 2). For the repeated measures variables, the results of the GLMM are listed in table 3. The NRS scores at rest or with movement did not significantly differ between the two PCIA intervention groups (p=0.203 and 0.341, respectively). OBASs were also similar between the two groups. All of the scores showed a downward trend over time, with the time effect being independent of the patient groups (p<0.001) and interactions with the PCIA regimens were non-existent (p=0.088, 0.327 and 0.720, respectively).
Results of the GLMM for repeated measurement variables
Online supplemental table S1 in the supplementary presents descriptive information regarding the PCIA administered (eg, demand frequency, rescue analgesia and discontinuation of PCIA treatment), recovery of gastrointestinal function, in-hospital complications, monitoring of satisfaction and length of hospital stay. A greater number of patients (9 (11.3%)) in the low-basal group discontinued PCIA due to side effects compared with the no-basal group (1 (1.2%)). However, while 6 (7.5%) patients receiving no-basal infusion PCIA needed analgesia rescue, only one patient from the low-basal group needed it. Online supplemental table S2 in the supplementary provides a summary of the SpO2 monitoring data quality. For the low-basal infusion and no-basal infusion groups, 91% and 88% of the patients had successful SpO2 monitoring (ie, excluding gaps) for at least 90% of the total monitoring time.
Discussion
The results did not show that low-basal infusion hydromorphone PCIA was non-inferior to no-basal infusion with regard to postoperative hypoxaemia within 48 hours of gastrointestinal tumour surgery. The AUC for hypoxaemia per hour (SpO2<95%) for the patients who received low-basal infusion PCIA for 48 hours postoperatively was greater than two-fold that of the no-basal group. In addition, the effect of low-basal or no-basal infusion of PCIA on postoperative hypoxaemia did not differ between the various subgroups examined.
Although many factors contribute to postoperative oxygen desaturation, it is a good indicator, and consequently, is widely used for evaluating the severity of POIRD.25 26 In addition, most of the previous studies have analysed the duration of predefined oxygen desaturation. However, AUC for hypoxaemia describes both the duration and severity of oxygen desaturation. Therefore, AUC for hypoxaemia was designated as the primary outcome for this trial. To calculate AUC for hypoxaemia, continuous SpO2 monitoring was performed to collect data throughout the entire experimental period, including during sleep and when patients got out of bed. This additional aspect more accurately represents the desaturation of patients compared with intermittent monitoring.27 The value of the SpO2 cut-off in the present trial was set at 95% based on a study conducted by Belcher et al regarding long-acting and short-acting opioid effects on postoperative hypoxaemia.23 Previously, very little or no research had been performed using a similar method to calculate AUC. It is noteworthy that other studies have also defined postoperative hypoxaemia as SpO2<95%.28–30 However, we also recognised that 95% is not a common cut-off value for defining desaturation. Therefore, AUC values of hypoxaemia under 90% and 85% were included as secondary outcomes. We did observe that when 90% was used as a cut-off, the two groups had no difference in desaturation. However, it should be recognised that the duration of oxygen supplement during the observation period was 24 hours (range: 18–40) for the low-basal infusion group and 36 hours (range: 20–40) for the no-basal infusion group. Thus, 95% was set as a reasonable and conservative cut-off value. Additionally, the trend of postoperative oxygen desaturation in this trial reflects that even in the case of postoperative oxygen therapy, oxygen desaturation usually occurs in patients with opioid analgesia and mainly on the first postoperative day. Both of these observations are consistent with those reported in previous studies.23 27 28 31 Thus, the present results demonstrate that continuous SpO2 monitoring is necessary and feasible for patients with postoperative opioid analgesia.
POIRD has previously been associated with opioid consumption.32 Consequently, we postulate that the AUC for hypoxaemia observed in our low-basal group that is approximately two-fold higher than that in the no-basal group is due to greater hydromorphone consumption (5.2 mg versus 1.6 mg, respectively). Belcher et al reported a higher AUC of hypoxaemia (1.28 (0.50, 2.23) %-hour) than the value reported in the present study, which might similarly be attributed to greater opioid consumption (hydromorphone averaging 16 mg or morphine 43 mg).23 Unlike our low-basal infusion group which consumed much more hydromorphone, White et al33 observed decreased morphine consumption in their basal infusion group compared with their bolus-only group (morphine: 40 mg versus 50.9 mg, respectively). Even our low-basal infusion group consumed a lower equivalent of morphine than any of their groups,33 indicating that the patients in the present study needed fewer opioids as part of their multimodal analgesia protocol. It is also possible that patient heterogeneity, the diversity of surgery types and subjectivity of the pain assessments may have contributed to the differences in results. Another perspective is that patients undergoing a much more ‘painful’ surgery might benefit from low-basal infusion PCIA, even though we considered the surgery performed in our study to be ‘painful’.
There was no difference between our two treatment groups in terms of analgesic effects, comprehensive analgesia benefit or postoperative ambulation. Adverse effects of PCIA were also similar between the two groups. However, a greater number of patients in the low-basal group discontinued PCIA due to the side effects (nine patients versus one patient, respectively). Meanwhile, demand doses of PCA in the no-basal group were higher than in the low-basal group, and more patients needed analgesia rescue (six patients versus one patient, respectively). The above findings suggest that PCIA with no-basal infusion was a better choice for our gastrointestinal tumour patients under the premise of ERAS practice. On this basis, the infusion parameters can be adjusted according to individual patient feedback during postoperative PCA follow-up.
Our trial did have several limitations. First, only patients from one hospital were enrolled, which may limit the general significance of the results. Second, a threshold of 95% to define hypoxaemia is not as commonly adopted as the threshold of 90% which is generally employed in clinical practice. This was due to the limited data we could refer to when we designed our trial. In further studies, RCTs with a larger sample size will be required to investigate PCIA-related hypoxaemia, defined as 90%, which is of greater clinical relevance.
Conclusion
Among the patients receiving hydromorphone PCIA following gastrointestinal tumour excision, the low-basal infusion was inferior to no-basal infusion with regard to hypoxaemia within 48 hours after surgery at a hypoxaemia threshold of SpO2<95%.
Data availability statement
Data are available upon reasonable request. Upon a reasonable request, the individual deidentified participant data, including the study protocol and statistical analysis plan, are available after team discussion. The data will become available beginning 3 months and ending 5 years following article publication. Proposals should be directed to the corresponding author ([email protected]) to gain access to data.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by Ethics Committee of Xijing Hospital (NO. KY20212163-F-1) on 22 November 2021. Participants gave informed consent to participate in the study before taking part.
Acknowledgments
We wish to thank Chong Lei and Chen Li for participating in the protocol discussion and the statistical analysis review. This trial received support from all of the patients and staff in the Department of Anesthesiology and Perioperative Medicine and the Department of General Surgery of Fourth Military Medical University Xijing Hospital, Xi’an, China who participated in this study. The authors also thank Medjaden Inc. for the scientific editing of this manuscript.
This post was originally published on https://bmjopen.bmj.com