Secondary Logo

Journal Logo

Featured Articles: Original Clinical Research Report

Exposure to Surgery and Anesthesia in Early Childhood and Subsequent Use of Attention Deficit Hyperactivity Disorder Medications

Ing, Caleb MD, MS*,†; Ma, Xiaoyue MS; Sun, Ming PhD*,§; Lu, Yewei MS; Wall, Melanie M. PhD§,‖; Olfson, Mark MD, MPH†,‖; Li, Guohua MD, DrPH*,†

Author Information
doi: 10.1213/ANE.0000000000004619

Abstract

See Articles, p 720 and 734

KEY POINTS

  • Question: Do children exposed to surgery and anesthesia at an early age have subsequent increased need for attention deficit hyperactivity disorder (ADHD) medications?
  • Findings: In this observational study of children enrolled in Medicaid from Texas and New York, those with a single exposure to surgery and anesthesia were 37% more likely than unexposed children to use ADHD medications.
  • Meaning: The finding of increased ADHD medication use after early exposure to anesthesia is consistent with findings from other published studies and suggests that deficits in behavior and executive function should be further explored in anesthetic-exposed children.

Millions of children in the United States are exposed to general anesthetic agents each year, with millions more exposed around the world.1,2 Recent clinical studies have found that anesthetic exposure in early childhood has minimal to no effect on a variety of neurocognitive functions including general intelligence and academic achievement.3–7 However, of the 3 recent studies that assessed prospectively collected outcome measures, all 3 found worse scores in secondary outcomes including behavioral and executive function.5–8 Assessment of behavior is challenging because evaluations typically require direct reporting by caregivers, so most studies have small sample sizes. Large health care administrative databases, however, can also be harnessed to identify children with behavioral problems. Attention deficit hyperactivity disorder (ADHD) is a clinical diagnosis that can be evaluated in large administrative data sets, is associated with deficits in behavior and executive function,9 and has also been found to be associated with anesthetic exposure.10–12

While clinical diagnoses are available in administrative databases, a criticism is that these outcomes may be subject to inaccuracies or misclassification.13 Pharmacy claims including claims for filled ADHD medication prescriptions, however, are highly accurate in reflecting medication use, particularly for chronic medications.14,15 Also, while ADHD diagnosis and medication use are related, these outcomes are not equivalent. Significant discrepancies between ADHD diagnoses and medication use have been reported16,17 and may be due to a desire to avoid stigma associated with a psychiatric diagnosis or to parents resistant to ADHD medication use despite a diagnosis of ADHD.18,19 Therefore, compared to diagnoses, use of stimulants and other ADHD medications in children more accurately identifies children with severe psychiatric disease20 and specifically those with symptoms concerning enough that parents and physicians recognize a need for pharmacological treatment. In this study, we evaluate the association between anesthetic exposure and behavioral disorders by assessing whether anesthetic-exposed children have an increased use of stimulants and other medications commonly prescribed for treating ADHD.

METHODS

This study was approved by the Columbia University Institutional Review Board, which waived the requirement for written informed consent. The Strengthening The Reporting of OBservational studies in Epidemiology (STROBE) statement was followed.

Generating the Birth Cohort Data Set

The data set was generated using Medicaid Analytic eXtract (MAX) files from the Centers for Medicare and Medicaid Services (CMS) and included children enrolled in Texas and New York Medicaid from 1999 to 2010. This cohort has been previously used to evaluate the association between anesthetic exposure and mental disorder diagnoses,10 but, in the present study, pharmacy claims containing National Drug Codes (NDC) were linked to each child.

Identifying the Anesthetic and Surgical Exposure

Exposure was defined as the presence of an International Classification of Diseases, Ninth Revision(ICD-9) or Current Procedural Terminology Fourth Edition (CPT-4) coded claim for 1 of 4 commonly performed procedures in young children: pyloromyotomy, inguinal hernia repair, circumcisions outside the perinatal period (where 91% have been reported to require general anesthesia),21 and tonsillectomy and/or adenoidectomy (T&A). Exposure to anesthesia and surgery was evaluated in 5 age-at-exposure categories: infancy, defined as the time between discharge from the birth admission and ≤1 year old, >1 year and ≤2 years, and 1-year age intervals thereafter until 5 years old. Children with >1 instance of the 4 procedures of interest, or those with any other surgical procedure in the operating room before age 5 years old, as defined by the Agency for Healthcare Research and Quality (AHRQ) Healthcare Utilization Project (HCUP) Surgery Flags, were excluded from analysis.22

Identifying NDC-Coded Prescription Fills

Medication prescription fills were identified by decoding NDCs using First Data Bank (San Francisco, CA) classifications. Medicaid pharmacy claims were used to determine the presence of ADHD medication use. In our time-to-event analysis, the event was defined as the time when there was adequate concern from parents and physicians to initiate ADHD medication treatment and the first prescription was filled. However, because there may be situations in which a prescription was filled but the medication was not used, children who did not fill at least two 30-day prescriptions, for a total of at least 60 days (Supplemental Digital Content, Table 1, http://links.lww.com/AA/C999, for list of ADHD medications), were not considered to be persistent users of ADHD medication. In addition, children with filled ADHD prescriptions before surgical exposure were excluded from analysis. To reduce ascertainment bias due to increased medical visits in children immediately after surgery, or capturing ADHD medication prescriptions for postoperative maladaptive behaviors,23 exposed children with filled ADHD medication prescriptions in the first 6 months after the surgical procedure were also excluded. Our outcome of subsequent use of ADHD medications was therefore defined as children with filled ADHD medication claims beginning at least 6 months following surgery until censoring. To test the robustness of our definition of persistent ADHD medication use, sensitivity analyses were performed redefining persistent use as children with at least three 30-day prescriptions and at least four 30-day prescriptions. Because our prior study excluded children with ICD-9 coded diagnoses for mental disorders before exposure while this study excludes children with ADHD medication prescriptions before exposure, differences in sample size between the 2 studies will be seen.10

Identifying Children Unexposed to Anesthesia and Surgery

Children were considered to be unexposed if they did not have any surgical procedures before 5 years of age.22 For each age-at-exposure category, the median age at anesthetic exposure was calculated. Unexposed children were only eligible to be chosen as matched controls for exposed children in each age-at-exposure category if they did not have a filled ADHD medication prescription within 6 months following their calculated median age at anesthetic exposure. This accounted for bias that would be introduced if only exposed children with early ADHD medication use were excluded.

Generating the Propensity Score–Matched Cohort

Propensity scores were calculated for all eligible exposed and unexposed children. The variables used in the score calculation included sociodemographic characteristics, prior comorbidities including acute and chronic medical conditions as defined by the Healthcare Cost and Utilization Project Chronic Condition Indicators,24 the number of inpatient stays and outpatient visits before surgery, the reason for Medicaid eligibility (disability or poverty), and year of birth (Supplemental Digital Content, Table 2, http://links.lww.com/AA/C999, for a list of all covariates). The year of birth was included because psychotropic prescriptions tended to increase during the study period.25 Each exposed child in each age-at-exposure category was matched without replacement to 5 unexposed children using nearest neighbor propensity score matching. Controls were matched without replacement within each age-at-exposure category. To ensure close matches, the estimated log odds scores predicting surgical exposure were required to be within 0.2 standard deviation units between the exposed and matched controls. This caliper distance has been effective in reducing bias due to measured confounding variables,26 but can vary from study to study and is only effective for measured confounders.

Statistical Analysis

To assess appropriate balance, standardized differences for all covariates were calculated with differences <0.1 considered as negligible between cases and controls.27 Cumulative frequencies were calculated according to the Kaplan–Meier method with time defined as the time in days, from the date of surgery (or for unexposed control patients, the median age at which surgery was performed in each age-at-exposure category) until the first ADHD medication prescription was filled. Log-rank tests were used to test the statistical significance of observed survivorship between exposed and unexposed children. Subjects were censored at the end of their Medicaid eligibility or the end of the year 2010, whichever came first. Cox proportional hazards models were used to evaluate the hazard ratio (HR) of ADHD medication use following exposure to anesthesia and surgery, with separate models used for each age-at-exposure category. HRs were calculated with combined data from Texas and New York in each age category with state of residence as a covariate. Children in Texas and New York were also evaluated separately to determine if the association between anesthetic exposure and ADHD medication use varied by state. An interaction analysis was also performed to evaluate the presence of a significant increased use of ADHD medication based on state of residence. To account for the matched structure of the sample, a robust variance estimator accounting for clustering within matched sets was used.28 One overall exposure model with all 4 surgical procedures was calculated, followed by procedure-specific models. To evaluate the robustness of the results from our primary outcome, we assessed the amount of unmeasured confounding that would be required to negate our results and change our conclusions.29

A number of sensitivity analyses were performed to validate our findings. Because ADHD symptoms may result from obstructive sleep apnea (OSA),30 T&A patients were excluded with the remaining patients evaluated. To determine whether exposed patients in this cohort had higher medication utilization in general due to unmeasured confounding or bias, the increased requirement for nonpsychotropic medications was evaluated as a negative control outcome.31 In this analysis, we evaluated claims for commonly used nonpsychotropic medications including amoxicillin, azithromycin, and diphenhydramine. To determine if the use of other types of psychotropic medications were elevated, medication claims for sedatives/anxiolytics, antidepressants, antipsychotics, and mood stabilizers were also evaluated (Supplemental Digital Content, Table 3, http://links.lww.com/AA/C999, list of psychotropic medications for treating conditions other than ADHD). The use of medications following exposure to surgery and anesthesia was defined as at least 1 prescription for nonpsychotropic medications and at least two 30-day prescriptions for other psychotropic medications. A post hoc interaction analysis was also performed to see if risk differed based on sex. In this analysis, circumcisions were excluded because the procedure is not performed in girls. All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc, Cary, NC).

RESULTS

Psychotropic Medication Prescriptions in the Texas and New York Medicaid Birth Cohort

Using Medicaid MAX data from 1999 to 2010, we built a birth cohort data set of 1,929,393 children born in Texas and New York. For children with any psychotropic prescription medication fill, 57.2% of children filled a prescription for ADHD medication. Of the children who filled an ADHD medication prescription, 86.2% filled >1. Discrepancies between ADHD diagnoses and medication use have been reported, with discrepancy rates ranging from 20% to 38%.16,17 Specifically, not all youth diagnosed with ADHD receive medication to treat ADHD16 and not all youth treated with ADHD medications such as stimulants receive clinical diagnoses of ADHD. Of the children in this birth cohort from Texas and New York, 48,173 children had a diagnosis for ADHD, of which 43.4% (n = 20,900) did not fill >1 prescription for an ADHD medication. Of the 32,127 children who filled >1 prescription for ADHD medication, 15.1% (n = 4854) did not have a clinical diagnosis for ADHD.

Matching Children Exposed to Surgical Procedures With Unexposed Children

Forty-two thousand six hundred eighty-seven exposed children were identified, with 25,844 from Texas and 16,843 from New York. After performing propensity score matching, exposed cases and unexposed controls were found to have similar distributions of covariates used to calculate the propensity scores, with summary data for a few select variables displayed in the Table. Standardized differences for all variables used in propensity score matching were calculated, with 99.6% of the variables having standardized differences of <0.1, confirming good balance between exposed cases and controls (Supplemental Digital Content, Figures 1–8, http://links.lww.com/AA/C999, absolute standardized differences for covariates comparing exposed and unexposed children). The remaining 0.4% of the variables had standardized differences <0.15 and were found in variables for children receiving T&As during infancy and children receiving inguinal hernia repair between ages 4 and 5 years old, which are the patient populations with the smallest sample sizes.

Table. - Percentages of Exposed and Unexposed Children With Selected Characteristics in Combined Texas and New York Medicaid Cohort
>Birth to ≤1 y Exposure >1 y to ≤2 y Exposure >2 y to ≤3 y Exposure >3 y to ≤4 y Exposure >4 y to ≤5 y Exposure
Yes
n = 14,172
No
n = 70,860
Yes
n = 8222
No
n = 41,110
Yes
n = 7587
No
n = 37,935
Yes
n = 7100
No
n = 35,500
Yes
n = 5606
No
n = 28,030
Sex F 11.2% 11.0% 19.7% 19.5% 30.8% 30.5% 37.1% 37.2% 41.1% 41.3%
M 88.8% 89.0% 80.3% 80.5% 69.2% 69.5% 62.9% 62.8% 58.9% 58.7%
Race African American 22.0% 22.9% 23.6% 23.3% 17.4% 16.6% 14.6% 14.5% 14.1% 13.9%
Hispanic 39.6% 39.3% 36.1% 36.2% 44.8% 45.6% 50.8% 51.0% 51.4% 51.5%
Other 10.5% 9.9% 10.2% 10.2% 8.3% 8.1% 7.6% 7.7% 6.6% 6.4%
Caucasian 27.9% 28.0% 30.1% 30.2% 29.4% 29.8% 27.0% 26.8% 27.9% 28.1%
Income-based on ZIP code of residence (Quartile) 1 26.3% 26.5% 27.8% 27.9% 29.0% 29.5% 29.9% 29.9% 30.0% 29.8%
2 25.3% 25.0% 26.2% 26.5% 26.5% 26.5% 27.5% 27.4% 27.6% 28.0%
3 23.9% 24.2% 24.0% 24.0% 24.1% 24.1% 22.8% 22.8% 24.1% 24.2%
4 24.4% 24.3% 21.9% 21.6% 20.4% 19.9% 19.9% 19.9% 18.2% 18.0%
Number of outpatient visits (Quartile) 1 5.3% 5.6% 8.5% 8.2% 7.6% 7.5% 8.7% 8.6% 8.8% 8.5%
2 24.6% 25.4% 18.6% 18.7% 16.9% 16.9% 17.9% 17.7% 17.6% 17.5%
3 26.5% 26.3% 24.3% 24.3% 25.0% 24.8% 25.9% 26.0% 30.0% 29.8%
4 43.6% 42.7% 48.6% 48.8% 50.4% 50.8% 47.5% 47.7% 43.6% 44.1%
Total number of inpatient stays including the birth admission (n) 1 89.0% 90.0% 75.5% 76.3% 71.8% 72.6% 69.5% 70.2% 68.6% 69.5%
2 8.8% 8.1% 17.1% 16.7% 19.3% 18.7% 20.3% 19.8% 20.6% 20.1%
≥3 2.2% 1.9% 7.4% 7.1% 8.9% 8.7% 10.3% 10.0% 10.8% 10.4%
Preterm No 88.0% 88.2% 90.2% 90.7% 93.8% 94.0% 94.4% 94.6% 95.7% 95.9%
Yes 12.0% 11.8% 9.8% 9.3% 6.2% 6.0% 5.6% 5.4% 4.3% 4.1%
Low birth weight No 90.6% 90.6% 92.3% 92.6% 94.4% 94.6% 94.7% 95.1% 95.1% 95.3%
Yes 9.4% 9.4% 7.7% 7.4% 5.6% 5.4% 5.3% 4.9% 4.9% 4.7%
Very low birth weight No 97.0% 97.3% 98.4% 98.5% 98.9% 99.0% 99.0% 99.2% 99.0% 99.1%
Yes 3.0% 2.7% 1.6% 1.5% 1.1% 1.0% 1.0% 0.8% 1.0% 0.9%
Extremely low birth weight No 98.7% 98.9% 99.2% 99.3% 99.6% 99.6% 99.5% 99.5% 99.6% 99.6%
Yes 1.3% 1.1% 0.8% 0.7% 0.4% 0.4% 0.5% 0.5% 0.4% 0.4%
Intrauterine hypoxia No 98.7% 98.8% 98.6% 98.6% 98.6% 98.6% 98.7% 98.8% 98.6% 98.7%
Yes 1.3% 1.2% 1.4% 1.4% 1.4% 1.4% 1.3% 1.2% 1.4% 1.3%
Diseases of the respiratory system (acute) No 65.6% 66.5% 16.0% 15.3% 8.5% 8.3% 6.8% 6.5% 5.0% 4.7%
Yes 34.4% 33.5% 84.0% 84.7% 91.5% 91.7% 93.2% 93.5% 95.0% 95.3%
Diseases of the digestive system (acute) No 82.7% 83.2% 58.7% 59.1% 49.7% 49.7% 43.4% 43.2% 39.6% 39.7%
Yes 17.3% 16.8% 41.3% 40.9% 50.3% 50.3% 56.6% 56.8% 60.4% 60.3%
Congenital anomalies No 85.7% 86.6% 80.1% 80.9% 80.0% 80.7% 79.5% 80.0% 80.2% 80.6%
Yes 14.3% 13.4% 19.9% 19.1% 20.0% 19.3% 20.5% 20.0% 19.8% 19.4%
Endocrine, metabolic, and immunity disorders (chronic) No 97.1% 97.3% 95.3% 95.5% 94.3% 94.5% 93.8% 93.9% 93.3% 93.4%
Yes 2.9% 2.7% 4.7% 4.5% 5.7% 5.5% 6.2% 6.1% 6.7% 6.6%
Diseases of blood and blood-forming organs (chronic) No 98.5% 98.7% 96.2% 96.3% 95.1% 95.5% 94.8% 94.9% 94.6% 94.8%
Yes 1.5% 1.3% 3.8% 3.7% 4.9% 4.5% 5.2% 5.1% 5.4% 5.2%
Diseases of the nervous system and sense organs (chronic) No 93.7% 94.1% 87.1% 87.9% 83.6% 83.9% 81.4% 82.3% 79.7% 79.8%
Yes 6.3% 5.9% 12.9% 12.1% 16.4% 16.1% 18.6% 17.7% 20.3% 20.2%
Diseases of the circulatory system (chronic) No 97.4% 97.6% 96.2% 96.3% 96.2% 96.2% 95.7% 95.9% 95.2% 95.5%
Yes 2.6% 2.4% 3.8% 3.7% 3.8% 3.8% 4.3% 4.1% 4.8% 4.5%
Diseases of the respiratory system (chronic) No 92.6% 93.0% 63.1% 63.3% 49.1% 49.1% 42.5% 42.2% 39.2% 39.1%
Yes 7.4% 7.0% 36.9% 36.7% 50.9% 50.9% 57.5% 57.8% 60.8% 60.9%
Diseases of the digestive system (chronic) No 89.0% 89.5% 84.2% 85.0% 85.4% 86.0% 86.3% 86.4% 87.6% 87.9%
Yes 11.0% 10.5% 15.8% 15.0% 14.6% 14.0% 13.7% 13.6% 12.4% 12.1%
Enrolled in Medicaid due to disability No 99.7% 99.9% 98.8% 99.0% 98.3% 98.4% 96.9% 97.2% 96.0% 96.1%
Yes 0.3% 0.1% 1.2% 1.0% 1.7% 1.6% 3.1% 2.8% 4.0% 3.9%
Abbreviations: F, female; M, male.

Exposed and unexposed children had similar lengths of follow-up, with a median duration of follow-up after birth of 1589 days (interquartile range [IQR]: 1320 days) in the exposed and 1565 days (IQR: 1301 days) in the unexposed children. Overall median postprocedure follow-up time was 767 days (IQR: 1026 days). For unexposed children who did not have a procedure, this time was calculated from the median age of exposure in the exposed children in each age-at-exposure category. Follow-up times varied based on age at exposure with the longest median follow-up time in children exposed at age ≤1 year at 886 days (IQR: 988 days) and the shortest in children exposed between age >4 and ≤5 years old at 696 days (IQR: 959 days). However, in each age-at-exposure category, follow-up was similar between exposed and unexposed children.

ADHD Medication Fills Associated With Exposure to Surgery and Anesthesia

A significantly increased risk of persistent ADHD medication use was found in children from Texas and New York exposed to surgery and anesthesia with all ages at exposure combined (P < .0001) (Figure 1) as well as when stratified by age-at-exposure category (Figure 2). Cox proportional hazards modeling revealed that children exposed to any of the 4 surgical procedures had a significantly increased risk of ADHD medication use with an HR for all ages combined of 1.37 (95% confidence interval [CI], 1.30–1.44) (Figure 3A). When evaluating different ages at exposure, HRs ranged from 1.22 to 1.47. In procedure-specific analysis, the increased risk associated with exposure to anesthesia for pyloromyotomy, inguinal hernia repair, and circumcision ranged from HRs of 1.12 to 1.35 (Figure 3B–D), with the lowest risk found after hernia repair. The risk associated with a T&A procedure, however, was higher than the other 3 procedures with an HR of 1.48 (95% CI, 1.39–1.58) (Figure 3E). When excluding all T&A patients, the HR for all ages combined in children with exposure for pyloromyotomy, inguinal hernia repair, or circumcision was 1.21 (95% CI, 1.12–1.32). When varying the definition of persistent ADHD medication use to ≥3 prescriptions, the HR for all ages and all procedures combined was 1.38 (95% CI, 1.31–1.46) and 1.26 (95% CI, 1.15–1.38) when excluding T&A patients. When using a definition of ≥4 ADHD medication prescriptions, the HR for all ages and all procedures combined was 1.41 (95% CI, 1.33–1.49) and 1.27 (95% CI, 1.16–1.39) when excluding T&A patients.

Figure 1.
Figure 1.:
Cumulative incidence of persistent ADHD medication use following a single exposure to surgery and anesthesia for all exposed and unexposed children. For exposed children, data show the length of time after a surgical procedure until initial ADHD medication prescription. For unexposed children, the data show the length of time from the median age at time of surgery (in the exposed children) until initial ADHD medication prescription. The number of children at risk in 3-year time intervals is also displayed below the x-axis. ADHD indicates attention deficit hyperactivity disorder.
Figure 2.
Figure 2.:
Cumulative incidence of persistent ADHD medication use following a single exposure to surgery and anesthesia for exposed and unexposed children by age-at-exposure category in panels A–E. For exposed and unexposed children, data show the length of time after birth until initial ADHD medication prescription. The number of children at risk in 3-year time intervals is also displayed below the x-axis. ADHD indicates attention deficit hyperactivity disorder.
Figure 3.
Figure 3.:
Hazard of persistent ADHD medication after a single exposure to surgery and anesthesia. HRs for persistent ADHD medication use after a single exposure to surgery and anesthesia for all exposed versus unexposed children are shown in A. B–E, HRs for children with each individual procedure type. ADHD indicates attention deficit hyperactivity disorder; CI, confidence interval; HR, hazard ratio.

In state-specific analyses, children in Texas used ADHD medications at nearly twice the rate found in children in New York (HR, 1.83; 95% CI, 1.74–1.92). However, the association between exposure to surgery and anesthesia and ADHD medication use was consistent in both states (Texas: HR, 1.35; 95% CI, 1.27–1.44 and New York: HR, 1.41; 95% CI, 1.29–1.55) (Supplemental Digital Content, Figure 9, http://links.lww.com/AA/C999). Interaction was evaluated with no significant difference between Texas and New York (P value for interaction = .46).

Effect of Unmeasured Confounding

The estimated HR for all procedures and all ages combined was 1.37 (95% CI, 1.30–1.44). In order for a single unmeasured confounder to account for the increased risk of ADHD medication use after anesthetic exposure (ie, to shift the lower 95% CI from 1.30 to 1.00), assuming that this confounder was present in 10% of the unexposed group and 20% or 40% of the exposed group, an associated HR of 6.4 or 2.3, respectively, would be required. Assuming that the unmeasured confounder was present in 20% of the unexposed group and 30% or 50% of the exposed group, an associated HR of 12.7 or 2.5, respectively, on the unmeasured confounder would be required to negate our results.

Claims for Other Psychotropic and Nonpsychotropic Medications After Exposure to Surgery and Anesthesia

Figure 4.
Figure 4.:
Hazard of psychotropic and nonpsychotropic medication use after a single exposure to surgery and anesthesia. HRs for persistent ADHD medication use, use of nonpsychotropic medications, and persistent use of psychotropic medications for conditions other than ADHD are displayed. ADHD indicates attention deficit hyperactivity disorder; CI, confidence interval; HR, hazard ratio.

When evaluating the use of other psychotropic medication claims in this cohort of patients, the HR was 1.37 (95% CI, 1.25–1.51) for sedative/anxiolytics, 1.40 (95% CI, 1.25–1.58) for antidepressants, 1.31 (95% CI, 1.20–1.44) for antipsychotics, and 1.24 (95% CI, 1.11–1.40) for mood stabilizers (Figure 4). In a negative control sensitivity analysis evaluating nonpsychotropic medication claims in children exposed to surgery and anesthesia, the HR for all ages and all procedures combined was 1.06 (95% CI, 1.04–1.07) for amoxicillin, 1.10 (95% CI, 1.08–1.12) for azithromycin, and 1.08 (95% CI, 1.05–1.11) for diphenhydramine. After excluding T&A patients, a lower risk of amoxicillin (HR, 1.04; 95% CI, 1.02–1.06) and azithromycin (HR, 1.08; 95% CI, 1.02–1.14) use was seen, but diphenhydramine use did not change.

Evaluation of Potentially Vulnerable Subgroups

Girls had a lower overall risk of ADHD medication use than boys (HR, 0.59; 95% CI, 0.55–0.62); however, evaluation of sex-based differences following exposure did not find a significant difference between boys and girls (P value for interaction = .06). In an analysis stratifying our existing matched cohort by sex, exposed girls were compared to matched unexposed girls, and exposed boys compared to matched unexposed boys, with circumcision procedures excluded. Balance between exposed and unexposed girls and boys was confirmed (Supplemental Digital Content, Figures 10–11, http://links.lww.com/AA/C999). In all ages and procedures combined, for girls, the HR was 1.56 (95% CI, 1.38–1.76), while for boys, the HR was 1.37 (95% CI, 1.28–1.47).

DISCUSSION

Children exposed to anesthesia for a single common surgical procedure were found to have a 37% increased risk of persistent ADHD medication use compared to unexposed children. This increased risk was present even after excluding T&A patients who may have a higher risk for ADHD due to coexisting OSA.30 In the negative control analysis with nonpsychotropic medications serving as negative control outcomes, when compared to unexposed children, exposed children had a 6%–10% increased risk of using nonpsychotropic medications such as antibiotics and diphenhydramine. This suggests that despite our matching methodology, there is still some unmeasured confounding. However, because the increased use of ADHD medication is disproportionately higher than that of the nonpsychotropic medications, unmeasured confounding may not account for all of the increase in ADHD medication use. We also reported state-specific variability in ADHD treatment rates, which is consistent with prior analyses.32 The finding of a similar increased use of ADHD medication following exposure to anesthesia and surgery in both states despite the state-specific treatment variability further strengthens the finding of an association between exposure to surgery and anesthesia and ADHD medication use.

In procedure-specific analysis, the increased risk of ADHD medication use was highest in children receiving T&A. A possible explanation could be that T&A children have underlying comorbidities predisposing them to ADHD. OSA specifically is associated with need for T&A and can also result in symptoms similar to those found in ADHD.30 T&A is thought to reduce ADHD symptoms, but residual symptoms may persist.33 While our results may be influenced by underlying ADHD risk in T&A patients, a significant increased risk of ADHD medication use was seen even after excluding T&A patients. This does, however, highlight the fact that the risk of neurodevelopmental deficit after different procedures may vary, which is relevant for interpreting studies combining many types of surgical procedures.

In interpreting these results in the context of the recent published literature, there are 3 large-scale studies using prospectively collected neuropsychological outcomes, the Pediatric Anesthesia Neurodevelopment Assessment (PANDA), the Mayo Anesthesia Safety in Kids (MASK), and the general anaesthesia or awake-regional anaesthesia in infancy (GAS) studies.5–7 All 3 studies found no association between anesthesia exposure and the primary outcome of intelligence quotient (IQ) in addition to a variety of other secondary neurocognitive outcomes. The PANDA and MASK studies, however, found that children exposed to general anesthesia had worse scores in assessments of behavioral function, while the MASK and GAS study both reported worse scores in executive function.5–8 Deficits in behavioral and executive function in addition to reductions in processing speed and fine motor ability reported exclusively in the MASK study have been found in children with ADHD.6 The finding of an increased risk of ADHD diagnosis in children exposed to anesthesia has also been previously described, with our prior analysis finding that a single exposure to anesthesia for common surgical procedures was associated with a 31% increased risk of ADHD diagnosis.10 Others have looked at a broader range of surgical procedures with increased risk of ADHD diagnoses after single exposures of 33% by Hu et al11 and 18% by Sprung et al.12 These estimates are similar to ours, but the cohorts evaluated in these studies were significantly smaller and these findings were not found to be statistically significant. Two studies evaluated anesthesia exposure and ADHD diagnoses using a national health care database from Taiwan. One study found no association,34 while a second study with a broader definition of ADHD and longer follow-up found an association in children with multiple exposures.35

This cohort of children has been previously evaluated to study the association between ages at anesthetic exposure and subsequent mental disorder diagnoses.10 The results from the present study with the addition of new pharmacy data are significantly more robust than the prior work for 2 main reasons. First, a limitation of the prior study is the inaccuracy of psychiatric diagnoses in administrative claims. The pharmacy claims used in the present study, however, are highly accurate in reflecting chronic medication use.14,15 Second, despite propensity score matching on a wide range of patient characteristics, in any observational study, including this one, there will be some degree of residual confounding. In the present study, the availability of pharmacy claims allowed for the evaluation of nonpsychotropic medications as negative control outcomes. The disproportionately higher use of psychotropic medications compared to the negative control nonpsychotropic medications suggests that the increase in ADHD medication use cannot be completely accounted for by confounding. To further explore the issue of confounding, we quantified the strength of an unmeasured confounder that would be required to negate our results. To put these results into context, in our cohort, the calculated HR of male sex on ADHD medication use was 1.71. Because male sex had a prevalence of 75% in our exposed cohort and 50% in our unmatched unexposed cohort, an unmeasured confounder with a similar prevalence would require an HR of 5.42 to negate our results. The strength of a theoretical unmeasured confounder would therefore need to be several times greater than that of male sex.

The present study has several limitations. First, because anesthesia and surgery are linked, the effects of the anesthetic cannot be distinguished from the surgical procedure, any inflammation from the procedure, or the potentially stressful perioperative experience. However, a recent study randomly assigned children undergoing inguinal hernia surgery to general anesthesia or regional anesthesia and found worse executive function scores in the general anesthetic–exposed children.7 While that outcome was only one of the secondary outcomes, all children received a similar surgical procedure and perioperative experience, with the only difference being the exposure to a general anesthetic. Second, because an administrative database was used, specific data on perioperative events and types and doses of drugs used intraoperatively were not available. Therefore, determining the impact of specific aspects of perioperative care is not possible. The purpose of this study, however, was to evaluate the association between surgery and anesthesia and subsequent use of ADHD medication use. If an association exists, further studies could evaluate specific perioperative factors and whether any of these factors are modifiable. Third, while ADHD medication use will not identify children with ADHD who are not treated, this outcome reflects the presence of adequate concern by the parent to fill multiple prescriptions. Finally, Medicaid children come from lower-income families, and children from lower-income families have been found to be uniquely vulnerable to neurotoxic exposures.36,37 In addition, because publicly insured children have higher rates of ADHD medication use than privately insured children,38 different findings might be observed in privately insured or uninsured children. However, because Medicaid covers approximately 38% of all children in the United States, these results are relevant to a significant proportion of the population.39,40

CONCLUSIONS

In this study, Medicaid-enrolled children exposed to anesthesia for a single common pediatric surgical procedure at under age 5 years were 37% more likely to require subsequent persistent use of ADHD medications than unexposed children. These results are consistent with other studies that evaluated similar behavioral outcomes and add to the evidence of the association between early anesthetic exposure and deficits in behavior and executive function, including an increased risk of ADHD. By evaluating Medicaid-enrolled children, this study also specifically assesses children who may be particularly vulnerable to neurotoxic exposures.

ACKNOWLEDGMENTS

We acknowledge David DeStephano, BS, Department of Anesthesiology, Columbia University College of Physicians and Surgeons, New York, NY, for his help with data analysis.

DISCLOSURES

Name: Caleb Ing, MD, MS.

Contribution: This author helped to conceive and design the study, analyze and interpret the data, draft and revise the manuscript, and approve the final manuscript.

Name: Xiaoyue Ma, MS.

Contribution: This author helped to design the study, analyze and interpret the data, revise the manuscript, and approve the final manuscript.

Name: Ming Sun, PhD.

Contribution: This author helped to design the study, analyze and interpret the data, revise the manuscript, and approve the final manuscript.

Name: Yewei Lu, MS.

Contribution: This author helped to analyze and interpret the data, revise the manuscript, and approve the final manuscript.

Name: Melanie M. Wall, PhD.

Contribution: This author helped to design the study, interpret the data, revise the manuscript, and approve the final manuscript.

Name: Mark Olfson, MD, MPH.

Contribution: This author helped to conceive and design the study, interpret the data, revise the manuscript, and approve the final manuscript.

Name: Guohua Li, MD, DrPH.

Contribution: This author helped to conceive and design the study, interpret the data, revise the manuscript, and approve the final manuscript.

This manuscript was handled by: James A. DiNardo, MD, FAAP.

FOOTNOTES

    REFERENCES

    1. Rabbitts JA, Groenewald CB, Moriarty JP, Flick REpidemiology of ambulatory anesthesia for children in the United States: 2006 and 1996. Anesth Analg. 2010;111:1011–1015.
    2. Tzong KY, Han S, Roh A, Ing CEpidemiology of pediatric surgical admissions in US children: data from the HCUP kids inpatient database. J Neurosurg Anesthesiol. 2012;24:391–395.
    3. O’Leary JD, Janus M, Duku E, et al.A population-based study evaluating the association between surgery in early life and child development at primary school entry. Anesthesiology. 2016;125:272–279.
    4. Glatz P, Sandin RH, Pedersen NL, Bonamy AK, Eriksson LI, Granath FAssociation of anesthesia and surgery during childhood with long-term academic performance. JAMA Pediatr. 2017;171:e163470.
    5. Sun LS, Li G, Miller TL, et al.Association between a single general anesthesia exposure before age 36 months and neurocognitive outcomes in later childhood. JAMA. 2016;315:2312–2320.
    6. Warner DO, Zaccariello MJ, Katusic SK, et al.Neuropsychological and behavioral outcomes after exposure of young children to procedures requiring general anesthesia: The Mayo Anesthesia Safety in Kids (MASK) Study. Anesthesiology. 2018;129:89–105.
    7. McCann ME, de Graaff JC, Dorris L, et alGAS Consortium. Neurodevelopmental outcome at 5 years of age after general anaesthesia or awake-regional anaesthesia in infancy (GAS): an international, multicentre, randomised, controlled equivalence trial. Lancet. 2019;393:664–677.
    8. Ing C, Brambrink AMMayo anesthesia safety in kids continued: two new studies and a potential redirection of the field. Br J Anaesth. 2019;122:716–719.
    9. Pasini A, Paloscia C, Alessandrelli R, Porfirio MC, Curatolo PAttention and executive functions profile in drug naive ADHD subtypes. Brain Dev. 2007;29:400–408.
    10. Ing C, Sun M, Olfson M, et al.Age at exposure to surgery and anesthesia in children and association with mental disorder diagnosis. Anesth Analg. 2017;125:1988–1998.
    11. Hu D, Flick RP, Zaccariello MJ, et al.Association between exposure of young children to procedures requiring general anesthesia and learning and behavioral outcomes in a population-based birth cohort. Anesthesiology. 2017;127:227–240.
    12. Sprung J, Flick RP, Katusic SK, et al.Attention-deficit/hyperactivity disorder after early exposure to procedures requiring general anesthesia. Mayo Clin Proc. 2012;87:120–129.
    13. Daley MF, Newton DA, DeBar L, et al.Accuracy of electronic health record-derived data for the identification of incident ADHD. J Atten Disord. 2017;21:416–425.
    14. Hennessy S, Palumbo Freeman C, Cunningham FStrom BL, Kimmel SE, Hennessy SUS Government claims databases. In: Pharmacoepidemiology. 2012:5th ed. Chichester, UK: Wiley-Blackwell, 209–223.
    15. Choo PW, Rand CS, Inui TS, et al.Validation of patient reports, automated pharmacy records, and pill counts with electronic monitoring of adherence to antihypertensive therapy. Med Care. 1999;37:846–857.
    16. Albert M, Rui P, Ashman JJPhysician office visits for attention-deficit/hyperactivity disorder in children and adolescents aged 4-17 years: United States, 2012-2013. NCHS Data Brief. 2017;269:1–8.
    17. Burcu M, Zito JM, Metcalfe L, Underwood H, Safer DJTrends in stimulant medication use in commercially insured youths and adults, 2010-2014. JAMA Psychiatry. 2016;73:992–993.
    18. Pescosolido BA, Perry BL, Martin JK, McLeod JD, Jensen PSStigmatizing attitudes and beliefs about treatment and psychiatric medications for children with mental illness. Psychiatr Serv. 2007;58:613–618.
    19. Pescosolido BA, Jensen PS, Martin JK, Perry BL, Olafsdottir S, Fettes DPublic knowledge and assessment of child mental health problems: findings from the National Stigma Study-Children. J Am Acad Child Adolesc Psychiatry. 2008;47:339–349.
    20. Olfson M, He JP, Merikangas KRPsychotropic medication treatment of adolescents: results from the National Comorbidity Survey-Adolescent Supplement. J Am Acad Child Adolesc Psychiatry. 2013;52:378–388.
    21. Wiswell TE, Tencer HL, Welch CA, Chamberlain JLCircumcision in children beyond the neonatal period. Pediatrics. 1993;92:791–793.
    22. (HCUP) HCaUP. Surgery Flag Software. Agency for Healthcare Research and Quality. Available at: https://www.hcup-us.ahrq.gov/toolssoftware/surgflags/surgeryflags.jsp. Published 2014. Accessed August 1, 2014.
    23. Fortier MA, Del Rosario AM, Rosenbaum A, Kain ZNBeyond pain: predictors of postoperative maladaptive behavior change in children. Paediatr Anaesth. 2010;20:445–453.
    24. Healthcare Cost and Utilization Project (HCUP). Chronic Condition Indicator (CCI) for ICD-9-CM. Rockville, MD. Agency for Healthcare Research and Quality, Available at: https://www.hcup-us.ahrq.gov/toolssoftware/chronic/chronic.jsp. Accessed December 1, 2015.
    25. Olfson M, Druss BG, Marcus SCTrends in mental health care among children and adolescents. N Engl J Med. 2015;372:2029–2038.
    26. Austin PCOptimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. Pharm Stat. 2011;10:150–161.
    27. Austin PCBalance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Stat Med. 2009;28:3083–3107.
    28. Lin DY, Wei LJThe robust inference for the cox proportional hazards model. J Am Stat Assoc. 1989;84:1074–1078.
    29. Lin DY, Psaty BM, Kronmal RAAssessing the sensitivity of regression results to unmeasured confounders in observational studies. Biometrics. 1998;54:948–963.
    30. Youssef NA, Ege M, Angly SS, Strauss JL, Marx CEIs obstructive sleep apnea associated with ADHD? Ann Clin Psychiatry. 2011;23:213–224.
    31. Lipsitch M, Tchetgen Tchetgen E, Cohen TNegative controls: a tool for detecting confounding and bias in observational studies. Epidemiology. 2010;21:383–388.
    32. Visser SN, Lesesne CAMental Health in the United States: Prevalence of Diagnosis and Medication Treatment for Attention-Deficit/Hyperactivity Disorder---United States, 2003. CDC Morbidity and Mortality Weekly Report (MMWR). 2005;54:842–847.
    33. Chervin RD, Ruzicka DL, Giordani BJ, et al.Sleep-disordered breathing, behavior, and cognition in children before and after adenotonsillectomy. Pediatrics. 2006;117:e769–e778.
    34. Ko WR, Liaw YP, Huang JY, et al.Exposure to general anesthesia in early life and the risk of attention deficit/hyperactivity disorder development: a nationwide, retrospective matched-cohort study. Paediatr Anaesth. 2014;24:741–748.
    35. Tsai CJ, Lee CT, Liang SH, Tsai PJ, Chen VC, Gossop MRisk of ADHD after multiple exposures to general anesthesia: a nationwide retrospective cohort study. J Atten Disord. 2018;22:229–239.
    36. Bellinger D, Leviton A, Waternaux C, Needleman H, Rabinowitz MLow-level lead exposure, social class, and infant development. Neurotoxicol Teratol. 1988;10:497–503.
    37. Stein LJ, Gunier RB, Harley K, Kogut K, Bradman A, Eskenazi BEarly childhood adversity potentiates the adverse association between prenatal organophosphate pesticide exposure and child IQ: The CHAMACOS cohort. Neurotoxicology. 2016;56:180–187.
    38. Zuvekas SH, Vitiello BStimulant medication use in children: a 12-year perspective. Am J Psychiatry. 2012;169:160–166.
    39. Monthly Child Enrollment in Medicaid and CHIP. State Health Facts: The Henry J. Kaiser Family Foundation. Available at: https://www.kff.org/medicaid/state-indicator/total-medicaid-and-chip-child-enrollment/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D. Accessed October 30, 2017
    40. Where Are States Today? Medicaid and CHIP Eligibility Levels for Children, Pregnant Women, and Adults. State Heath Facts: The Henry J. Kaiser Family Foundation. Available at: https://www.kff.org/medicaid/fact-sheet/where-are-states-today-medicaid-and-chip/#table1. Accessed March 28, 2018

    Supplemental Digital Content

    Copyright © 2020 International Anesthesia Research Society