We conducted a retrospective cohort study on infants diagnosed as cataract, FEVR and ROP (stage 3-5) who received lensectomy, transpupillary laser treatment, vitreoretinal surgery with/without anterior segment surgery, collected demographic, surgical, and anesthetic features, and documented cardiac and respiratory adverse events associated with surgery. We analyzed the association between the features and the adverse events to reveal potential risk factors for each adverse event.
The study was approved by the Peking University People’s Hospital Institutional Review Board, and parental consent was waived because of the retrospective design.
Study population
We retrospectively included all infants who underwent ophthalmological procedures for cataract, FEVR and ROP at Peking University People’s Hospital from November 1, 2016 to October 31, 2017. The majority of them were referred to our hospital where the National Pediatric Eye Center is located. Ophthalmological procedures included laser surgery, vitreoretinal and anterior segment surgery. Infants were excluded if they had a recent history (up to 2 weeks) of upper airway infection or their parents were unwilling to give consent.
We collected information regarding their demographic, anesthetic and surgical features. Gestation age at birth and PCA at the time of surgery would be recorded. Body weight at birth and body weight at the time of surgery would also be documented. Gestational age was determined by questioning the parents and confirmed with documentation of physical examination at birth in previous medical records. PCA was the sum of the gestational and postnatal ages. A preoperative history of apnea would be recorded.
Anesthetic and Surgical management
All patients were admitted to hospital on the day of surgery. All parents were instructed on applying topical anesthetic and mydriatic to their children preoperatively. In the preparation room, intravenous cannula was inserted and an intravenous infusion of 2-4 ml·kg-1·h-1 ringer lactate was initiated and maintained during the surgery. Upon arrival at the operating room, standard monitor was established including electrocardiogram (ECG), non-invasive blood pressure, pulse saturation (SpO2) and end tidal carbon dioxide after anesthesia induction.
General anesthesia was induced either by intravenous propofol or inhalational sevoflurane. Airway management device could be either endotracheal tube or a face mask. If an endotracheal tube was used, pressure controlled ventilation was adopted, inspiratory pressure set as 20 cmH2O, and respiratory rate adjusted according to a target end tidal carbon dioxide of 35-45 cmH2O. Retrobulbar block with 0.5% ropivacaine 0.1 mL/kg was administered to all patients by an experienced ophthalmologic surgeon except for those undergoing laser surgery. Appropriate depth of anesthesia was ensured by central position of pupils6. Use of preoperative atropine, opioids, muscle relaxant or steroids was left to the discretion of the anesthesiologist in charge. After surgery, all infants were monitored in the ward for at least 12 hours.
Infants suffering from cataract would undergo lensectomy; laser coagulopathy or retinal cryotherapy was given to infants suffering from ROP stage 3 or threshold disease; scleral buckling would be applied to ROP stage 4a patients; vitrectomy with/without lensectomy would be applied to patients suffering from ROP stage 4b/ 5 and FEVR.
Outcome measurement
All episodes of bradycardia (HR less than 100 bpm), laryngospasm, bronchospasm, airway obstruction, apnea, oxygen desaturation (less than 90%) were recorded as perioperative cardiac and respiratory adverse events in electronic anesthetic database. Bradycardia caused by oculocardiac reflex (OCR) was identified. Usually OCR was caused by pressing the globe or manipulation of extraocular muscles, which could often be corrected by cessation of surgical stimulus, or, if it did not work, atropine 0.01mg/kg should be administered. When surgical procedures were finished, all children who had recurrent desaturation (less than 90%) and could not be relieved by oxygen supply through a face mask would be admitted to intensive care unit (ICU) for further monitoring and treatment. Apnea was defined as a pause in breathing >10 seconds or a pause >5 seconds if associated with oxygen saturation <90% or bradycardia (HR less than 100 bpm). This definition was adapted from GAS study [12]. Pulse oximetry would be monitored for infants transferred to surgical ward, who would be discharged home 24 hours after surgery, and those transferred to ICU would stay there more than 24 hours.
Statistical analysis
Statistical analysis was done with SPSS (version 20.0). We did univariate analysis with the student’s t test or Mann-Whitney U test for continuous variables and the χ² test for categorical variables. For all analyses, we used two-sided tests, with P values less than 0.05 denoting statistical significance.
In order to avoid problems of multicollinearity, exploratory factor analysis was used to reduce dimension of the confounders and to find the underlying factors with clinical significance which could extract at least 70% of squared loadings from the initial components cumulatively. Scores of those underlying factors were saved as the new variables to be used in the multivariate logistic regression models with other uncorrelated covariates. The method of backward stepwise (Likelihood Ratio), which meant variables remained in the model if they improved the model fit with the likelihood ratio test, was used in the logistic regressions models to result in selected factors.
If any underlying factor from the factor analysis was selected, all of its main components, which extracted more than 50% of loadings in the rotated component matrix, were used as selected covariates. These selected covariates would construct a new logistic regression model with other selected initial confounders. Thus by using the method of backward stepwise (Likelihood Ratio), RRs of the observable factors could be calculated. Hosmer-Lemeshow statistic and Area Under the ROC Curve (AUC) were used to measure the calibration and discrimination of the logistic regression model, and the outliers, whose absolute value of standardized residuals were greater than 2.3, were excluded from the final analysis.
In order to facilitate clinical application of our research findings, we converted the continuous variables, which are predictors for various adverse events, to binary variables. Therefore we could figure out the cut-off point for different risk variables, through Decision Tree Analysis, i.e. Classification and Regression Trees.
Patient and public involvement
Patients were not involved in setting the research questions or planning the study. Investigators do not know the identity of study participants.
Data sharing
Mendeley DOI 10.17632/ryc2vtswcp.1