BACKGROUND:
Ketorolac-refractory pain behavior following bilateral myringotomy and pressure equalization tube placement (BMT) is associated with the absence of middle ear fluid. Intraoperative fentanyl/ketorolac affords more reliable pain control than ketorolac alone. We hypothesized that middle ear condition would correlate with postoperative pain despite such combination therapy. We further sought to demonstrate seasonal variation in ear condition and its influence on pain.
METHODS:
We conducted a single-institution retrospective cohort study of healthy children (9 months–7 years), who underwent BMT by a single surgeon from 2015 to 2020. Anesthetic care included sevoflurane/nitrous oxide/oxygen/air by mask and intramuscular fentanyl/ketorolac. Left/right middle ear fluid status was recorded at the time of BMT, and ear condition (primary exposure) was dichotomized as bilateral infected (mucoid or purulent) or normal/unilateral infected. The primary outcome was maximum postanesthesia care unit Face, Legs, Activity, Cry, and Consolability (FLACC) score: 4–10 (moderate-to-severe pain) versus 0–3 (no-to-low pain). Rescue oxycodone, acetaminophen administration, and emergence agitation were secondary outcomes. Statistical analysis incorporated generalized linear mixed-effect models (GLMMs) with random intercepts to account for clustering by anesthesia provider. A year-over-year monthly time-series analysis was conducted using an autoregressive integrated moving average (ARIMA) regression model.
RESULTS:
Excluding recurrent cases, 1149 unique evaluable subjects remained. Bilateral infection prevalence was 39.8% (457/1149; 95% confidence interval [CI], 37.0–42.6). Probability of moderate-to-severe pain behavior was 23.5% (270/1149; 95% CI, 21.1–26.0) overall. Compared to patients with bilateral infected middle ears, those with normal/unilateral infected ears were more likely to have a FLACC score ≥4 (26.7% [185/692] versus 18.6% [85/457]; odds ratio [95% CI], 1.7 [1.2–2.3]; P = .002). Variability in pain outcome explained by the multivariable GLMM was 4.7%. Fentanyl dose response was evidenced by oxycodone administration differences (P ≤ 0.002). Moderate-to-severe pain and emergence agitation were more likely with reduced fentanyl dosing. Bilateral infection prevalence exhibited seasonality, peaking in March and nadiring in July. However, pain outcomes did not vary by season.
CONCLUSIONS:
Normal/unilateral infected ears at time of pediatric BMT are associated with higher incidence of moderate-to-severe postoperative pain following intraoperative fentanyl/ketorolac administration, but the predictive value of ear condition on pain is limited. Infections were less common in the summer.
KEY POINTS.
Question: Does middle ear condition affect the prevalence of moderate-to-severe pain in the postanesthesia care unit following bilateral myringotomy tube placement when intraoperative fentanyl and ketorolac have been administered in combination?
Findings: Non/unilaterally infected middle ears are associated with increased risk of moderate-to-severe postoperative pain, and increasing fentanyl dose correlates with reduced need for oxycodone rescue.
Meaning: Fentanyl is an important component of an effective preemptive analgesic regimen for bilateral myringotomy tube placement and may be of particular value when at least 1 ear shows no signs of active infection.
Bilateral myringotomy and pressure equalization tube placement (BMT) is the most common pediatric day surgical procedure in the United States.1 Indications include recurrent or refractory acute otitis media, chronic otitis media with middle ear effusion, and conductive hearing loss due to middle ear fluid.2 These conditions are most prevalent in young children and are often associated with viral/bacterial infections, school attendance, and day-care use, each of which feature seasonal variation.3–11
Middle ear condition at the time of BMT may be associated with postoperative pain behavior. Pediatric patients for whom ketorolac is the sole intraoperative analgesic are more likely to experience moderate-to-severe acute postoperative pain (Face, Legs, Activity, Cry, and Consolability [FLACC] scores between 4 and 10/10) if they have normal ears without effusion.12 Whether such an association stands in the setting of the more efficacious fentanyl/ketorolac analgesic regimen is unknown.13
Understanding pain patterns associated with middle ear fluid status may help clinicians and staff better prevent or treat pain following BMT. We sought to investigate such associations under our current practice standard using intraoperative fentanyl/ketorolac. To eliminate pain outcome variation (~9%) attributable to individual surgeon and better standardize ear condition assessment and surgical technique, we chose to study the caseload of a single, high-volume surgeon.12 We hypothesized that middle ear condition (primary exposure) would be associated with moderate-to-severe pain (FLACC scores ≥4, the primary outcome) in the postanesthesia care unit (PACU). We further hypothesized that middle ear condition would be associated with pain-related secondary outcomes of oxycodone rescue, acetaminophen administration, and emergence agitation. Finally, we wished to demonstrate seasonal variation in middle ear condition at the time of BMT and determine whether it was consequential for the above outcomes.
METHODS
This retrospective observational study was not considered human subjects research and was exempted from parental consent and continued institutional review board monitoring by the Children’s Hospital of Philadelphia (CHOP) institutional review board. It adheres to applicable STrengthening the Reporting of Observational studies in Epidemiology guidelines.14
Patient Selection
The CHOP electronic medical records system (Epic Systems Corporation‚ Verona‚ WI), live in the perioperative environment on January 25, 2015, was accessed with the following search terms: procedure category “myringotomy,” procedure performed by a single surgeon (Dr Sobol), and procedure date between February 1, 2015 and December 31, 2020. The dataset was then filtered for: surgery duration <15 minutes, completed PACU FLACC scores and medication administration records, patient age between 9 months and 7 years, and American Society of Anesthesiologists (ASA) physical status I or II. Patients were excluded if they had procedures in addition to BMT, involvement of surgical/anesthesia house staff or certified registered nurse anesthetists, or additional/erroneous medication dosing. For patients with multiple BMTs over the study epoch, only the first was included. Patients with documented prior BMT (before study start) or mention of existing tympanostomy tubes in the operative note were recorded as having a history of BMT.
Coronavirus disease 2019 (COVID-19) era patients were defined as those with BMTs between May 1, 2020 and December 31, 2020. Beginning in March 2020, pediatric otolaryngology divisions throughout the United States were forced to postpone all “nonemergent” surgical procedures, including BMT. Based on state guidelines, limited elective surgery was allowed to resume in May with the most urgent cases scheduled first. For BMT surgery, priority was given to patients at higher risk of communication difficulties related to hearing loss and those who continued to have episodes of otitis media despite social distancing.
Variable Definitions and Derivations
The highest postoperative FLACC score for each subject was retained for analysis, with moderate-to-severe pain defined as a FLACC score between 4 and 10 and no-to-low pain defined as a FLACC score between 0 and 3. Fentanyl and ketorolac entered the dataset as total micrograms or milligrams, respectively, and were then converted to weight-based doses. Emergence agitation was defined as a 3-4 on the 4-point Watcha scale.15,16
Middle ear fluid status was abstracted from the operative notes and categorized, by ear, as normal (no fluid), serous, mucoid, or purulent. A 4 × 4 contingency table was generated showing right and left middle ear fluid characteristics and the frequencies (percentages) of patients with moderate-to-severe pain (Supplemental Digital Content 1, Table 1, http://links.lww.com/AA/E52). The majority of subjects (n = 824/1149, 71.7%) had concordant middle ear findings between right and left ears, with 183/824 (22.2%) found to have a FLACC 4–10. Of these, 98/381 (25.7%) with normal, 14/45 (31.1%) with serous, 51/267 (19.1%) with mucoid, and 20/131 (15.3%) with purulent ears exhibited moderate-to-severe pain in the PACU. The remaining study participants had discordant ears, with a normal or serous finding in one behaving as a dominant trait regarding pain outcome. Observing fundamental pathophysiology of the middle ear, we grouped both normal and serous fluid as “noninfected” and consolidated mucoid and purulent as “infected,” which when combined with the pain dominance pattern of normal/serous middle ear condition justified reduction of middle ear condition into a dichotomous variable as “bilateral infected” or “non/unilateral infected.” These ear condition categorizations are shown in Table 1. Within-group homogeneity with respect to the distribution of subjects with moderate-to-severe pain among the 3 bilateral infected subgroups and across the 7 non/unilateral infected subgroups was assessed separately using Fisher exact tests (both P > .35).
Table 1.
Moderate-to-Severe Postoperative Pain by Bilateral Middle Ear Status
| Categories | Composites | n/N (%) | P |
|---|---|---|---|
| Non/unilateral infected | Normal/normal | 98/381 (25.7) | .928 |
| Normal/serous | 18/66 (27.3) | ||
| Serous/serous | 14/45 (31.1) | ||
| Normal/mucoid | 37/134 (27.6) | ||
| Serous/mucoid | 10/31 (32.3) | ||
| Normal/purulent | 7/27 (25.9) | ||
| Serous/purulent | 1/8 (12.5) | ||
| Bilateral infected | Mucoid/mucoid | 51/267 (19.1) | .362 |
| Mucoid/purulent | 14/59 (23.7) | ||
| Purulent/purulent | 20/131 (15.3) |
Composites represent combinations of right and left middle ear fluid.
Anesthetic Management and Monitoring
During the entire study period, and as described in previous reports,12,13 attending anesthesiologists followed CHOP standard anesthetic protocol for patients undergoing BMT including inhalation induction and maintenance with sevoflurane in air/oxygen or nitrous oxide/oxygen delivered via facemask. No intravenous catheters were placed unless indicated for emergent situations, fluid replacement, or ondansetron use. All such cases receiving additional medications or fluids were excluded at the time of data abstraction. Preemptive fentanyl and ketorolac were administered in the deltoid muscle after induction and just before speculum insertion into the external auditory canal. Fentanyl dosage was targeted at 1.5–2 µg/kg and ketorolac at 1 mg/kg.
Following surgery, FLACC scores were recorded by nursing staff on PACU arrival and at 15-minute intervals thereafter until patient discharge criteria were met. Emergence agitation scores were similarly captured, although not routinely on PACU arrival. Oral acetaminophen (10–15 mg/kg) was administered by nursing staff prophylactically or for mild pain; oxycodone (0.05–0.15 mg/kg), however, was reserved for FLACC scores 4–10 or trajectories toward them. Staff worked to quickly address pain by administering acetaminophen or oxycodone to patients demonstrating pain-specific behaviors including ear touching or tugging.
Statistical Analysis
Patients’ characteristic data were grouped by postoperative pain scores (ie, no-to-low pain, FLACC 0–3 and moderate-to-severe pain, FLACC 4–10). Mean (standard deviation [SD]) was used for continuous variables, while frequency and percent were used for categorical variables. Middle ear fluid status was collapsed into a dichotomous variable (non/unilateral infected versus bilateral infected) as described above. Primary and secondary outcomes were summarized and compared across the dichotomous middle ear fluid status.
To quantify the adjusted estimate of the association between middle ear fluid status and incident moderate-to-severe postoperative ear pain, a generalized linear mixed-effect model (GLMM) with anesthesiologist-specific random intercept was developed. The model regressed the logit of the probability of having moderate-to-severe ear pain on middle ear fluid status, age, season, COVID-19 era, selected patient clinical variables, and an anesthesiologist-specific random intercept to adjust for homogeneity in patient outcomes by anesthesiologist. Patient clinical measures that were unbalanced between ear conditions (P < .2) in the preliminary bivariable analysis served as the criteria for inclusion as covariates in the final multivariable GLMM. Age category, season, and COVID-19 era variables were included a priori in the model. Linearity assumption was verified with plots of log odds of moderate-to-severe pain over 10 equal-length ranges of each numeric variable. Multicollinearity issues were assessed using Spearman correlation coefficient matrix (all absolute values <0.8). GLMM analysis with the same specified random and fixed effects as above was developed for secondary outcomes, including oxycodone administration (yes/no), acetaminophen administration (yes/no), and emergence agitation score (1–2/3–4). Bonferroni adjustment was used to account for increased type I error rate resulting from multiple outcomes. Thus, with 4 outcomes, P value less than .05/4 = 0.0125 was considered statistically significant. With a 2-sided test at an alpha level of 0.0125 and reference rate being the outcome rate among patients with bilateral infected middle ear, our current sample has 80% power to detect a difference of 8.41% in the proportions of moderate-to-severe pain (with reference rate of 85/457 [18.6%]), 5.28% in those of oxycodone administration (with reference rate of 23/457 [5.0%]), 7.50% for acetaminophen administration (with reference rate of 60/457 [13.1%]), and 7.80% (with reference rate of 37/360 [10.3%]) for emergence agitation.
To explore whether the COVID-19 pandemic onset altered the monthly and yearly trends of ear condition and postoperative pain distribution, we aggregated both variables into monthly proportions and fitted separate autoregressive integrated moving average (ARIMA (p,d,q)(P,D,Q)[m]) regression models to the data. Pre-COVID-19 data collected from February 2015 to March 2020 were used to develop the ARIMA model and then used to forecast outcomes from May to December 2020 (the counterfactual: following COVID-19 outbreak).17,18 The observed proportions from May to December 2020 were then compared with the forecasted point estimates and their associated 95% CIs for incidental changes in outcomes due to COVID-19 pandemic. The choice of parameters in the ARIMA (p,d,q)(P,D,Q)[m] model followed Hyndman and Khandakar’s step-wise approach to select the model with the lowest Akaike's information criterion and was conducted using the automatic model selection tool from the “forecast” R package.19 We allowed the order of AR and MA, p and q, respectively, to take values from 0 to 5, and the order of seasonal AR and MA, P & Q, to take values from 0 to 2 for model selection. We also added total number of patients per month N (ie, <10 vs ≥10) as an additional regressor into the ARIMA model to control for variation of monthly patient volume. All analyses were conducted using R software version 3.5.1 (R Core Team).
RESULTS
Baseline cohort characteristics stratified by middle ear condition are presented in Table 2. In total, 1149 patients met inclusion criteria. Median age was 1 year (interquartile range [IQR], 1–2) and 702 (61.1%) were male. The majority of subjects were White (n = 856; 75.5%). Eighty (7.0%) patients were reported as having Hispanic ethnicity. Most patients had a history of mild systemic disease (ASA physical status II) (n = 733; 63.8%). A few patients had prior BMT (n = 49; 4.3%), and only 1 patient exhibited preoperative ear pain. Preoperative midazolam was administered to 916 (79.7%) patients. All patients received intraoperative fentanyl and ketorolac with mean doses of 1.7 (SD, 0.39) μg/kg and 0.88 (SD, 0.17) mg/kg, respectively. Younger children (<2 years) were more likely to have bilateral ear infections at the time of BMT, n = 354/824 (43.0%), than their 2–7-year-old counterparts, n = 103/325 (31.7%); P < .001. Patients with bilaterally infected middle ears were less likely to have had a prior BMT (12 [2.6%] vs 37 [5.3%]; P = .037) or to have received preoperative midazolam (337 [73.7%] vs 579 [83.7%]; P < .001). In addition, such patients received marginally higher intraoperative fentanyl dosages (mean, 1.736 µg/kg [0.399] vs 1.673 [0.383]; P = .007) and had longer procedure times (P < .001). Bilaterally infected middle ears were more common than non/unilateral infected ears in the winter and less common in the summer (P < .001).
Table 2.
Demographic Data and Baseline Variables by Middle Ear Condition
| Middle ear status | |||||
|---|---|---|---|---|---|
| Variable | Level | Overall count(N = 1149) | Non/unilateral infected (n = 692) | Bilateral infected (n = 457) | P |
| Age | 9–12 mo | 159 (13.8) | 86 (12.4) | 73 (16.0) | .001 |
| 13 mo–2 y | 665 (57.9) | 384 (55.5) | 281 (61.5) | ||
| 2–7 y | 325 (28.3) | 222 (32.1) | 103 (22.5) | ||
| Sex | Female | 447 (38.9) | 275 (39.7) | 172 (37.6) | .513 |
| Male | 702 (61.1) | 417 (60.3) | 285 (62.4) | ||
| Race, missing = 15/1149 (1.3%) | White | 856 (75.5) | 526 (77.0) | 330 (73.2) | .303 |
| Black | 110 (9.7) | 64 (9.4) | 46 (10.2) | ||
| Other | 168 (14.8) | 93 (13.6) | 75 (16.6) | ||
| Ethnicity, missing = 12/1149 (1%) | Hispanic/Latino | 80 (7.0) | 51 (7.4) | 29 (6.4) | .608 |
| Not Hispanic/Latino | 1057 (93.0) | 636 (92.6) | 421 (93.6) | ||
| ASA status | I | 416 (36.2) | 261 (37.7) | 155 (33.9) | .212 |
| II | 733 (63.8) | 431 (62.3) | 302 (66.1) | ||
| History of BMT | No | 1100 (95.7) | 655 (94.7) | 445 (97.4) | .037 |
| Yes | 49 (4.3) | 37 (5.3) | 12 (2.6) | ||
| Preoperative midazolam | No | 233 (20.3) | 113 (16.3) | 120 (26.3) | <.001 |
| Yes | 916 (79.7) | 579 (83.7) | 337 (73.7) | ||
| Fentanyl dose (μg/kg)‚ mean (SD); missing = 3/1149 (0.3%) | 1.698 (0.391) | 1.673 (0.383) | 1.736 (0.399) | .007 | |
| Fentanyl dose (μg/kg), classified, missing = 3/1149 (0.3%) | (0.6,1.42) | 284 (24.8) | 182 (26.4) | 102 (22.3) | .004 |
| (1.42,1.76) | 286 (25.0) | 182 (26.4) | 104 (22.8) | ||
| (1.76,2) | 283 (24.7) | 175 (25.4) | 108 (23.6) | ||
| (2,2.8) | 293 (25.6) | 150 (21.8) | 143 (31.3) | ||
| Ketorolac dose (mg/kg)‚ mean (SD); missing = 3/1149 (0.3%) | 0.883 (0.165) | 0.882 (0.164) | 0.883 (0.168) | .926 | |
| Ketorolac dose (mg/kg), classified, missing = 3/1149 (0.3%) | (0.05,0.83) | 278 (24.3) | 166 (24.1) | 112 (24.5) | .977 |
| (0.83,0.94) | 301 (26.3) | 184 (26.7) | 117 (25.6) | ||
| (0.94,0.99) | 281 (24.5) | 169 (24.5) | 112 (24.5) | ||
| (0.99,1.5) | 286 (25.0) | 170 (24.7) | 116 (25.4) | ||
| Surgery duration (min)‚ mean (SD) | 5.581 (1.656) | 5.198 (1.443) | 6.162 (1.787) | <.001 | |
| Surgery duration (min), classified | (1,4) | 298 (25.9) | 230 (33.2) | 68 (14.9) | |
| (4,5) | 346 (30.1) | 231 (33.4) | 115 (25.2) | ||
| (5,6) | 245 (21.3) | 133 (19.2) | 112 (24.5) | ||
| (6,14) | 260 (22.6) | 98 (14.2) | 162 (35.4) | ||
| COVID-19 | Before COVID-19 | 1103 (96.0) | 648 (93.6) | 455 (99.6) | |
| COVID-19 | 46 (4.0) | 44 (6.4) | 2 (0.4) | ||
| Season | Q1 | 343 (29.9) | 172 (24.9) | 171 (37.4) | |
| Q2 | 354 (30.8) | 215 (31.1) | 139 (30.4) | ||
| Q3 | 216 (18.8) | 168 (24.3) | 48 (10.5) | ||
| Q4 | 236 (20.5) | 137 (19.8) | 99 (21.7) | ||
Data are number of patients and column percent unless otherwise specified.
Abbreviations: ASA, America Society of Anesthesiologists; BMT, bilateral myringotomy tube; COVID-19, coronavirus disease 2019; Q, quarter; SD, standard deviation..
Table 3 presents primary outcome distribution, stratified by patient and procedural variables, together with bivariate GLMM analysis results. Non/unilateral infected ear was associated with increased proportion of moderate-to-severe pain as compared to bilateral infected ear (185/692 [26.7%] vs 85/457 [18.6]; odds ratio [OR], 1.6; 95% CI, 1.2–2.1; P = .002). Moderate-to-severe pain is associated with lower fentanyl dose (μg/kg) as compared to mild pain (mean [SD], 1.63 [0.40] vs 1.72 [0.39]; OR, 0.5 [0.4–0.8]; P = .002). Seventeen percent (49/293) patients with highest quartile of fentanyl dosing (2–2.8 µg/kg) had moderate-to-severe pain, while 28.2% (80/284) patients with lowest quartile (0.6–1.42 µg/kg) had moderate-to-severe pain.
Table 3.
Associations Between Patient and Procedural Characteristics and Moderate-to-Severe Postoperative Pain
| Bivariate GLMM analysis | ||||||
|---|---|---|---|---|---|---|
| Variables | Level | Overall count | FLACC 0–3 | FLACC 4–10 | OR (95% CI) | P |
| Middle ear status | Bilateral infected | 457 (39.8) | 372 (81.4) | 85 (18.6) | Reference | |
| Non/unilateral infected | 692 (60.2) | 507 (73.3) | 185 (26.7) | 1.6 (1.2–2.1) | .002 | |
| Age | 9–12 mo | 159 (13.8) | 125 (78.6) | 34 (21.4) | Reference | |
| 13 mo–2 y | 665 (57.9) | 504 (75.8) | 161 (24.2) | 1.2 (0.8–1.8) | .48 | |
| 2–7 y | 325 (28.3) | 250 (76.9) | 75 (23.1) | 1.1 (0.7–1.7) | .74 | |
| Sex | Female | 447 (38.9) | 346 (77.4) | 101 (22.6) | Reference | |
| Male | 702 (61.1) | 533 (75.9) | 169 (24.1) | 1.1 (0.8–1.5) | .48 | |
| Race, missing = 15/1149 (1.3%) | White | 856 (75.5) | 643 (75.1) | 213 (24.9) | Reference | |
| Black | 110 (9.7) | 91 (82.7) | 19 (17.3) | 0.6 (0.4–1) | .07 | |
| Other | 168 (14.8) | 137 (81.5) | 31 (18.5) | 0.7 (0.4–1) | .07 | |
| Ethnicity, missing = 12/1149 (1%) | Not Hispanic/Latino | 1057 (93) | 809 (76.5) | 248 (23.5) | Reference | |
| Hispanic/Latino | 80 (7) | 64 (80) | 16 (20) | 0.8 (0.5–1.5) | .52 | |
| ASA physical status |
I | 416 (36.2) | 318 (76.4) | 98 (23.6) | Reference | |
| II | 733 (63.8) | 561 (76.5) | 172 (23.5) | 1.1 (0.8–1.5) | .61 | |
| History of BMT | No | 1100 (95.7) | 842 (76.5) | 258 (23.5) | Reference | |
| Yes | 49 (4.3) | 37 (75.5) | 12 (24.5) | 1.1 (0.5–2.1) | .86 | |
| Preoperative midazolam | No | 233 (20.3) | 182 (78.1) | 51 (21.9) | Reference | |
| Yes | 916 (79.7) | 697 (76.1) | 219 (23.9) | 1.1 (0.7–1.5) | .73 | |
| Fentanyl dose (μg/kg), missing = 3/1149 (0.3%) | 1.698 (0.391) | 1.721 (0.386) | 1.625 (0.398) | 0.5 (0.4–0.8) | .002 | |
| Fentanyl dose (μg/kg), classified, missing = 3/1149 (0.3%) | (0.6,1.42) | 284 (24.8) | 204 (71.8) | 80 (28.2) | Reference | |
| (1.42,1.76) | 286 (25) | 212 (74.1) | 74 (25.9) | 0.9 (0.6–1.3) | .54 | |
| (1.76,2) | 283 (24.7) | 216 (76.3) | 67 (23.7) | 0.8 (0.5–1.2) | .22 | |
| (2,2.8) | 293 (25.6) | 244 (83.3) | 49 (16.7) | 0.5 (0.3–0.8) | .003 | |
| Ketorolac dose (mg/kg), missing = 3/1149 (0.3%) | 0.883 (0.165) | 0.889 (0.163) | 0.863 (0.172) | 0.5 (0.2–1.1) | .08 | |
| Ketorolac dose (mg/kg), classified, missing = 3/1149 (0.3%) | (0.05,0.83) | 278 (24.3) | 200 (71.9) | 78 (28.1) | Reference | |
| (0.83,0.94) | 301 (26.3) | 230 (76.4) | 71 (23.6) | 0.8 (0.6–1.2) | .34 | |
| (0.94,0.99) | 281 (24.5) | 219 (77.9) | 62 (22.1) | 0.8 (0.5–1.2) | .21 | |
| (0.99,1.5) | 286 (25) | 227 (79.4) | 59 (20.6) | 0.7 (0.5–1.1) | .09 | |
| Surgery duration (min) | 5.581 (1.656) | 5.568 (1.67) | 5.626 (1.612) | 1 (0.9–1.1) | .55 | |
| Surgery duration (min), classified | (1,4) | 298 (25.9) | 231 (77.5) | 67 (22.5) | Reference | |
| (4,5) | 346 (30.1) | 263 (76) | 83 (24) | 1.1 (0.7–1.6) | .67 | |
| (5,6) | 245 (21.3) | 192 (78.4) | 53 (21.6) | 1 (0.6–1.4) | .83 | |
| (6,14) | 260 (22.6) | 193 (74.2) | 67 (25.8) | 1.2 (0.8–1.8) | .31 | |
| COVID-19 | Before COVID-19 | 1103 (96) | 843 (76.4) | 260 (23.6) | Reference | |
| COVID-19 | 46 (4) | 36 (78.3) | 10 (21.7) | 0.8 (0.4–1.7) | .63 | |
| Season | Q1 | 343 (29.9) | 267 (77.8) | 76 (22.2) | Reference | |
| Q2 | 354 (30.8) | 270 (76.3) | 84 (23.7) | 1.1 (0.7–1.5) | .71 | |
| Q3 | 216 (18.8) | 155 (71.8) | 61 (28.2) | 1.4 (0.9–2.1) | .1 | |
| Q4 | 236 (20.5) | 187 (79.2) | 49 (20.8) | 0.9 (0.6–1.4) | .66 | |
Data are number of patients and column percent unless otherwise specified.
Abbreviations: ASA, American Society of Anesthesiologists; BMT, bilateral myringotomy tube; CI, confidence interval; COVID‚ coronavirus disease; FLACC, Face, Legs, Activity, Cry, and Consolability pain scale; GLMM, generalized linear mixed-effect model; OR, odds ratio; Q‚ quarter.
Table 4 shows the results from the multivariable GLMM analyses for all 4 outcomes. After adjusting for season, COVID-19 pandemic, age, history of BMT, preoperative midazolam, fentanyl dose, surgery duration, and within-anesthesiologist correlation, non/unilateral infected ears were related to increased odds of moderate-to-severe pain as compared to bilateral infected ears (OR, 1.7; 95% CI, 1.2–2.3; P = .002). Although trending similarly for rescue analgesics and emergence agitation, middle ear condition was not significantly associated with any of these secondary outcomes. The models also demonstrated a strong fentanyl dose-response relationship with postoperative oxycodone administration: ORs ranged from 0.2 to 0.4 with P ≤ .002 for all 3 upper fentanyl dosing quartiles compared to the lowest. Highest dose fentanyl also was associated with diminished incidence of moderate-to-severe pain: OR, 0.5 (0.3–0.9); P = .008, as well as emergence agitation: OR, 0.4 (0.2–0.8); P = .006. Neither season nor the COVID-19 pandemic was significantly associated with moderate-to-severe postopeartive pain or secondary PACU outcomes.
Table 4.
Multivariable Generalized Linear Mixed-Effect Modeling for Primary and Secondary Outcomes
| FLACC 4–10 | Oxycodone | Acetaminophen | Emergence agitation 3–4 | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Predictor | Level | Odds ratio (95% CIs) | P | Odds ratio (95% CIs) | P | Odds ratio (95% CIs) | P | Odds ratio (95% CIs) | P |
| Observations used | 1146 | 1146 | 1146 | 917 | |||||
| Middle ear status | Bilateral infected | Reference | Reference | Reference | Reference | ||||
| Non/unilateral infected | 1.7 (1.2–2.3) | .002 | 1.9 (1–3.4) | .035 | 1.3 (0.9–1.9) | .215 | 1.7 (1.1–2.6) | .028 | |
| Age | 9–12 mo | Reference | Reference | Reference | Reference | ||||
| 13 mo–2 y | 1.1 (0.7–1.8) | .559 | 0.8 (0.3–2) | .696 | 1.2 (0.7–2.1) | .593 | 1 (0.5–1.9) | .982 | |
| 2–7 y | 1 (0.6–1.7) | .95 | 1.2 (0.5–3) | .685 | 1.3 (0.7–2.5) | .389 | 0.5 (0.2–1) | .041 | |
| History of BMT | No | Reference | Reference | Reference | Reference | ||||
| Yes | 1 (0.5–2) | .923 | 1.9 (0.7–5) | .198 | 0.8 (0.3–2) | .653 | 1 (0.4–2.8) | .989 | |
| Preoperative midazolam | No | Reference | Reference | Reference | Reference | ||||
| Yes | 0.9 (0.6–1.4) | .769 | 1.3 (0.6–2.9) | .453 | 0.6 (0.4–1) | .049 | 2 (1–3.9) | .045 | |
| Fentanyl dose (μg/kg) | (0.6, 1.42) | Reference | Reference | Reference | Reference | ||||
| (1.42,1.76) | 0.9 (0.6–1.3) | .554 | 0.4 (0.2–0.7) | .002 | 0.9 (0.5–1.4) | .562 | 1 (0.6–1.6) | .917 | |
| (1.76,2) | 0.8 (0.5–1.2) | .302 | 0.2 (0.1–0.5) | <.001 | 0.6 (0.3–1) | .058 | 0.8 (0.5–1.3) | .362 | |
| (2,2.8) | 0.5 (0.3–0.9) | .008 | 0.2 (0.1–0.5) | .001 | 0.6 (0.3–1) | .07 | 0.4 (0.2–0.8) | .006 | |
| Surgery duration (min) | (1,4) | Reference | Reference | Reference | Reference | ||||
| (4,5) | 1.2 (0.8–1.7) | .394 | 1.3 (0.6–2.5) | .491 | 0.8 (0.5–1.3) | .486 | 1.2 (0.7–2.2) | .433 | |
| (5,6) | 1.1 (0.7–1.7) | .607 | 1.3 (0.6–2.8) | .488 | 1.1 (0.7–1.8) | .75 | 1.6 (0.9–2.9) | .109 | |
| (6,14) | 1.6 (1–2.4) | .039 | 2 (0.9–4.1) | .073 | 1.2 (0.7–2) | .491 | 1.9 (1.1–3.4) | .028 | |
| Season | Q1 | Reference | Reference | Reference | Reference | ||||
| Q2 | 1 (0.7–1.5) | .806 | 0.9 (0.4–1.7) | .677 | 0.9 (0.6–1.4) | .634 | 0.7 (0.4–1.2) | .154 | |
| Q3 | 1.4 (0.9–2) | .142 | 1.2 (0.6–2.7) | .573 | 1.2 (0.7–2) | .465 | 1.4 (0.8–2.4) | .219 | |
| Q4 | 1 (0.6–1.5) | .89 | 0.6 (0.3–1.5) | .308 | 0.9 (0.5–1.6) | .763 | 1 (0.5–1.7) | .898 | |
| COVID | Before COVID | Reference | Reference | Reference | Reference | ||||
| During COVID | 0.7 (0.3–1.4) | .313 | 0.3 (0–2.7) | .3 | 0.7 (0.3–1.7) | .389 | 1.6 (0.7–3.8) | .28 | |
Bolded P values indicate significance by .0125 threshold.
Abbreviations: ASA, American Society of Anesthesiologists; BMT, bilateral myringotomy tube; CI, confidence interval; FLACC, Face, Legs, Activity, Cry, and Consolability pain scale; GLMM, generalized linear mixed-effect model; OR, odds ratio; Q‚ quarter.
Figure 1 demonstrates monthly proportions of stratified middle ear conditions and moderate-to-severe pain prevalence from 2015 to 2020. From 2015 to 2019, bilateral infections (and, inversely, normal and combined non/unilateral infections) exhibited seasonality through the year, peaking in March with a median of 58.8% (IQR, 50%–61.8%; n = 5) and nadiring in July, median 14.3% (IQR, 14.3%–28.6%; n = 5). Between January and March, the proportion of patients having bilateral ear infections, when comparing infections in 2020 to prior years (42.6% [20/47] vs 51.0% [151/296]; P = .36), was not significant. Although overall case load fell to 35.7% of normal volume between May and December of 2020, bilateral infected ears comprised only 4.3% (2/46) of patients in the study compared to 35.3% (228/645) in prior years (P < .001), and ARIMA modeling demonstrated significant disruption in the seasonal middle ear condition pattern during this time (Figure 2A). Incidence of moderate-to-severe pain did not show a defined seasonal pattern throughout the study period. Between May and December, when comparing 2020 and previous years, moderate-to-severe pain incidences were similar: 21.7% (10/46) vs 22.5% (145/645); P > .99. There was no disruption in the static ARIMA modeling during COVID-19 (Figure 2B). Multivariable GLMM modeling was consistent, showing no effect of season or the COVID-19 pandemic on moderate-to-severe pain incidence (Table 4).
Figure 1.
Monthly middle ear status proportions and moderate-severe pain incidence. The proportion of patients with noninfected (A), unilateral infected (B), bilateral infected (C), and combined non/unilateral infected middle ears (D) is reported by month and year. E, The proportion of patients with moderate-severe postoperative pain (FLACC 4–10). FLACC indicates Face, Legs, Activity, Cry, and Consolability pain scale.
Figure 2.
Incidence and prediction of bilaterally infected middle ears and moderate-to-severe pain. The proportion of patients found to have bilaterally infected middle ears (A) and the proportion with moderate-to-severe pain (B) (black) is reported from February 2015 to March 2020. An ARIMA regression model was fitted (blue) and used to predict expected (red) proportions of bilaterally infected middle ears (A) and proportion with moderate-to-severe pain (B) during the COVID-19 pandemic between May and December 2020. ARIMA indicates autoregressive integrated moving average; FLACC, Face, Legs, Activity, Cry, and Consolability pain scale.
DISCUSSION
Our data demonstrate a strong association between non/unilateral infected middle ear condition and increased incidence of moderate-to-severe postoperative ear pain in infants and children given intramuscular fentanyl/ketorolac during BMT. This middle ear condition nearly correlated with, but did not reach statistical significance for, both increased oxycodone administration and emergence agitation. Increasing fentanyl doses associated with decreasing likelihood of oxycodone administration in dose-response fashion and highest fentanyl dose quartile (2–2.8 µg/kg) was associated with reduced incidence of moderate-to-severe postoperative ear pain and emergence agitation compared to lowest dose quartile (0.6–1.42 µg/kg).
Middle ear condition was found to be a significant predictor of postoperative pain behavior after adjusting for age, history of BMT, midazolam administration, fentanyl dose, surgery duration, season, COVID-19 era, and clustering by anesthesiologist in our single-surgeon cohort. Combined adjustment for age and fentanyl dose was particularly important because of their clear interrelation with middle ear condition: younger children were more likely to have acute otitis media and were more likely to receive a higher fentanyl dose per kilogram absolute body weight. We suspect that the small decrement in per kilogram dosing for older and larger children may have resulted from estimated ideal body weight corrections. Intraoperative fentanyl/ketorolac was again shown to provide substantially better analgesia than ketorolac alone13: the overall incidence of moderate-to-severe pain in our current investigation was 23.5% (95% CI, 21.1–26.0) compared to 52.4% (50.2–54.6) in the earlier single-agent ketorolac study with its pain probability low point estimate at 32.3% (25.8–39.4) under the most favorable ear and surgical conditions.12 The magnitude of ear condition effect on pain appears to be diminished with fentanyl: the moderate-to-severe pain OR of 2.2 (1.6–2.9) from the previous study falls to 1.7 (1.2–2.3) here when comparing bilateral normal or non/unilateral infected versus bilateral abnormal or bilateral infected, respectively. However, this comparison is confounded by our updated ear categorization.
Ear condition was dichotomized as non/unilateral infected versus bilateral infected and, although empirically derived, reflects pathophysiology. Purulent middle ear fluid signifies acute inflammation with active infection.20 Mucoid fluid, with accumulation of thick effusions, follows from chronic infection and inflammation with poor drainage, often from eustachian tube dysfunction.21,22 Serous fluid typically occurs as acute otitis media clears and represents a different inflammatory state.21,23 We hypothesize that ongoing infections and active inflammation may downregulate pain pathways, causing desensitization. Additionally, tympanic membrane pressure relief from myringotomy and fluid evacuation may reduce pain if higher fluid volumes and pressures are associated with infectious effusions. Nonetheless, current data reinforce that “the worst side from a postoperative pain perspective may be the normal one,”12 acting as the dominant pain trigger.
Middle ear status weakly correlated with oxycodone administration and emergence agitation. Although oxycodone rescue should correspond with increased pain, the nonsignificant association with ear condition could be due to infrequent administration (n = 84; 7.3%) and/or variability in use thresholds. Connections between middle ear condition and emergence agitation may be pain-mediated. Theories that increased emergence agitation results from sudden regained ability to hear and increased noise sensitivity are not borne out; however, those with non/unilateral infected ears, who presumably have minimal hearing loss preoperatively and no immediate improvement post-BMT, more frequently exhibit emergence agitation. Lack of association significance could be due to relatively small sample size (n = 125/919;13.6%) given observed effect size (15.7% [88/559] vs 10.3% [37/360]) or contributing factors beyond pain, such as young age, emotionality, impulsivity, and anxiety.24
Rescue analgesic administration patterns were consistent with a highly effective intraoperative regimen. For each quartile jump in fentanyl dose, between a minimum of 0.6 µg/kg and maximum of 2.8 µg/kg, the OR for oxycodone rescue halved. These data suggest that fentanyl dosage could be effectively titrated to meet BMT pain requirements while remaining within pediatric dosing standards of 1.5–2 µg/kg. However, preemptive PACU administration of some analgesics may have tempered maximum pain scores and primary outcome associations. Although nurses generally limit oxycodone rescue to those reaching FLACC 4–6 pain thresholds, preventative acetaminophen use is encouraged. The lack of association between acetaminophen administration and middle ear condition, and also fentanyl/ketorolac dosing, is consistent with the latter practice.
With known seasonal patterns of pediatric otitis media and its exposure variables, we investigated the effects of seasonality on intraoperative middle ear status and pain.3–11 Previous studies largely relied on otoscopy and tympanometry, which are less accurate than operative visualization and cannot differentiate effusion type.4,10,11,25 Intraoperative findings from the only study comparable to ours showed no significant seasonal differences in middle ear status, but were derived from only 1 year of data.26 Over the 5 years in our study before the COVID-19 pandemic, bilateral infection peaked in March and nadired in July, consistent with established winter/summer relationships between day-care, school, and respiratory virus infection and the incidence of otitis media.5–9 This pattern was disrupted by the COVID-19 pandemic, as demonstrated by the ARIMA model. However, moderate-to-severe postoperative pain did not demonstrate the same seasonality over any portion of the study period. While middle ear condition influences postoperative pain, the model only explains 4.7% of pain variance and may not contribute strongly enough to drive seasonal pain patterns. Other unmeasured variables, including seasonal allergies, cerumen content variation, or lifestyle factors, may significantly contribute to the remaining variance.
LIMITATIONS
As with all retrospective observational cohort studies, associations can be demonstrated but causation cannot be proven. While a single-surgeon cohort was chosen to focus on ear condition and eliminate surgeon-driven variability, such a design limits the generalizability of our findings across the wider spectrum of surgical technique. Our model was adjusted for anesthesiologist but not for the PACU nurse who assessed pain and administered analgesics. This came of the large number (>100) of nurses involved and the possibility of multiple nurses caring for a single patient. Pain scoring variation and preemptive PACU oxycodone administration could have reduced the recorded incidence of moderate-to-severe pain behavior and lessened the potential effect size of ear condition on that primary outcome. With respect to our seasonality analysis, COVID-era patient volume was significantly restricted through the close of data collection and reduced the power and generalizability of our findings. Additionally, with monthly patient volume being limited for a single surgeon (median [IQR], 17.5 [12–22]), the observed monthly trend of bilateral infection rate and moderate-to-severe pain rate could be nonrepresentative of the population and unstable. Although we added patient volume as a regressor in the ARIMA model, uncertainty and bias were difficult to adjust for, and thus, the forecast confidence intervals remained wide.
CONCLUSIONS
In infants and young children who receive intramuscular fentanyl/ketorolac during BMT, our data showed that noninfected middle ear status is associated with increased incidence of moderate-to-severe postoperative pain behavior. Intramuscular fentanyl dose may be adjusted upward toward 2 µg/kg, in accordance with dosing guidelines, to minimize risk of PACU oxycodone administration. However, with 95% of variance in PACU pain incidence due to factors beyond those exposures studied here, middle ear condition cannot be used as the sole determinant of fentanyl dosing in this population. While the incidence of bilateral infected ears displays seasonality, postoperative pain risk did not follow a seasonal pattern and precludes us from making recommendations for preemptive analgesic changes based on the time of year the BMT is performed.
ACKNOWLEDGMENTS
The authors acknowledge and thank Lisa Morse, PsyD, MSCIS (Department of Anesthesiology and Critical Care Medicine, The Children’s Hospital of Philadelphia, Philadelphia, PA) for her help in data extraction.
DISCLOSURES
Name: William G. Cohen, BA.
Contribution: This author helped with data analysis, manuscript preparation, and manuscript review/editing.
Name: Bingqing Zhang, MPH.
Contribution: This author helped with data analysis and manuscript review/editing.
Name: David Lee, MD.
Contribution: This author helped with project conceptualization and manuscript review/editing.
Name: Steve B. Ampah, PhD.
Contribution: This author helped with data analysis and manuscript review/editing.
Name: Steven E. Sobol, MD.
Contribution: This author helped with project conceptualization and manuscript review/editing.
Name: Scott D. Cook-Sather, MD.
Contribution: This author helped with project conceptualization, data analysis, manuscript preparation, and manuscript review/editing.
This manuscript was handled by: James A. DiNardo, MD, FAAP.
Supplementary Material
GLOSSARY
- ARIMA
- autoregressive integrated moving average
- ASA
- American Society of Anesthesiologists
- BMT
- bilateral myringotomy tube
- CHOP
- Children’s Hospital of Philadelphia
- CI
- confidence interval
- COVID-19
- coronavirus disease 2019
- FLACC
- Face, Legs, Activity, Cry, and Consolability
- GLMM
- generalized linear mixed-effect model
- IQR
- interquartile range
- OR
- odds ratio
- PACU
- postanesthesia care unit
- SD
- standard deviation
Published online 2 November 2022.
Funding: This study was supported by departmental resources..
The authors declare no conflicts of interest.
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website.
W. G. Cohen and B. Zhang are co-first authors.
REFERENCES
- 1.Cullen KA, Hall MJ, Golosinskiy A. Ambulatory surgery in the United States, 2006. Natl Health Stat Report. 2009;11:1–25. [PubMed] [Google Scholar]
- 2.Rosenfeld RM, Schwartz SR, Pynnonen MA, et al. Clinical practice guideline: tympanostomy tubes in children. Otolaryngol Head Neck Surg. 2013;149:S1–35. [DOI] [PubMed] [Google Scholar]
- 3.Bluestone CD, Swarts JD. Human evolutionary history: consequences for the pathogenesis of otitis media. Otolaryngol Head Neck Surg. 2010;143:739–744. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Stockmann C, Ampofo K, Hersh AL, et al. Seasonality of acute otitis media and the role of respiratory viral activity in children. Pediatr Infect Dis J. 2013;32:314–319. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Winther B, Alper CM, Mandel EM, Doyle WJ, Hendley JO. Temporal relationships between colds, upper respiratory viruses detected by polymerase chain reaction, and otitis media in young children followed through a typical cold season. Pediatrics. 2007;119:1069–1075. [DOI] [PubMed] [Google Scholar]
- 6.Heymann A, Chodick G, Reichman B, Kokia E, Laufer J. Influence of school closure on the incidence of viral respiratory diseases among children and on health care utilization. Pediatr Infect Dis J. 2004;23:675–677. [DOI] [PubMed] [Google Scholar]
- 7.Kørvel-Hanquist A, Koch A, Lous J, Olsen SF, Homøe P. Risk of childhood otitis media with focus on potentially modifiable factors: a Danish follow-up cohort study. Int J Pediatr Otorhinolaryngol. 2018;106:1–9. [DOI] [PubMed] [Google Scholar]
- 8.Ladomenou F, Kafatos A, Tselentis Y, Galanakis E. Predisposing factors for acute otitis media in infancy. J Infect. 2010;61:49–53. [DOI] [PubMed] [Google Scholar]
- 9.van Ingen G, le Clercq CMP, Jaddoe VWV, et al. Identifying distinct trajectories of acute otitis media in children: a prospective cohort study. Clin Otolaryngol. 2021;46:788–795. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Castagno LA, Lavinsky L. Otitis media in children: seasonal changes and socioeconomic level. Int J Pediatr Otorhinolaryngol. 2002;62:129–134. [DOI] [PubMed] [Google Scholar]
- 11.Tos M, Holm-Jensen S, Sørensen CH. Changes in prevalence of secretory otitis from summer to winter in four-year-old children. Am J Otol. 1981;2:324–327. [PubMed] [Google Scholar]
- 12.Cook-Sather SD, Castella G, Zhang B, Mensinger JL, Galvez J, Wetmore RF. Principal factors associated with ketorolac-refractory pain behavior after pediatric myringotomy and pressure equalization tube placement: a retrospective cohort study. Anesth Analg. 2020;130:730–739. [DOI] [PubMed] [Google Scholar]
- 13.Stricker PA, Muhly WT, Jantzen EC, et al. Intramuscular fentanyl and ketorolac associated with superior pain control after pediatric bilateral myringotomy and tube placement surgery: a retrospective cohort study. Anesth Analg. 2017;124:245–253. [DOI] [PubMed] [Google Scholar]
- 14.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med. 2007;147:573–577. [DOI] [PubMed] [Google Scholar]
- 15.Bajwa SA, Costi D, Cyna AM. A comparison of emergence delirium scales following general anesthesia in children. Paediatr Anaesth. 2010;20:704–711. [DOI] [PubMed] [Google Scholar]
- 16.Watcha MF, Ramirez-Ruiz M, White PF, Jones MB, Lagueruela RG, Terkonda RP. Perioperative effects of oral ketorolac and acetaminophen in children undergoing bilateral myringotomy. Can J Anaesth. 1992;39:649–654. [DOI] [PubMed] [Google Scholar]
- 17.Friedman J, Hansen H, Bluthenthal RN, Harawa N, Jordan A, Beletsky L. Growing racial/ethnic disparities in overdose mortality before and during the COVID-19 pandemic in California. Prev Med. 2021;153:106845. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Schaffer AL, Dobbins TA, Pearson SA. Interrupted time series analysis using autoregressive integrated moving average (ARIMA) models: a guide for evaluating large-scale health interventions. BMC Med Res Methodol. 2021;21:58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Hyndman RJ, Khandakar Y. Automatic time series forecasting: the forecast package for R. J Stat Softw. 2008;27:1–22. [Google Scholar]
- 20.Paparella MM. Middle ear effusions: definitions and terminology. Ann Otol Rhinol Laryngol. 1976;85:8–11. [DOI] [PubMed] [Google Scholar]
- 21.Val S, Poley M, Anna K, et al. Characterization of mucoid and serous middle ear effusions from patients with chronic otitis media: implication of different biological mechanisms? Pediatr Res. 2018;84:296–305. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Iwano T, Kinoshita T, Hamada E, Doi T, Ushiro K, Kumazawa T. Otitis media with effusion and eustachian tube dysfunction in adults and children. Acta Otolaryngol Suppl. 1993;500:66–69. [DOI] [PubMed] [Google Scholar]
- 23.Hall-Stoodley L, Hu FZ, Gieseke A, et al. Direct detection of bacterial biofilms on the middle-ear mucosa of children with chronic otitis media. JAMA. 2006;296:202–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Aouad MT, Nasr VG. Emergence agitation in children: an update. Curr Opin Anaesthesiol. 2005;18:614–619. [DOI] [PubMed] [Google Scholar]
- 25.Harmes KM, Blackwood RA, Burrows HL, Cooke JM, Harrison RV, Passamani PP. Otitis media: diagnosis and treatment. Am Fam Physician. 2013;88:435–440. [PubMed] [Google Scholar]
- 26.Knopke S, Böttcher A, Chadha P, Olze H, Bast F. Seasonal differences of tympanogram and middle ear findings in children. HNO. 2017;65(suppl 1):68–72. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.