Key Points
Question
Do children with previously normal neurocognitive function who survive an episode of acute respiratory failure requiring intensive care and invasive mechanical ventilation have worse long-term neurocognitive function than their matched siblings?
Findings
In this cohort study that included 121 sibling pairs, children discharged from intensive care hospitalization for respiratory failure without severe cognitive dysfunction compared with their matched siblings had a mean IQ score of 101.5 vs 104.3, a difference that was statistically significant.
Meaning
Acute respiratory failure in early childhood was associated with a slightly, but statistically significant, lower subsequent IQ score.
Abstract
Importance
Approximately 23 700 US children undergo invasive mechanical ventilation for acute respiratory failure annually, with unknown long-term effects on neurocognitive function.
Objective
To evaluate neurocognitive outcomes of children who survive pediatric intensive care unit (PICU) hospitalization for acute respiratory failure compared with their biological siblings.
Design, Setting, and Participants
Prospective sibling-matched cohort study conducted at 31 US PICUs and associated neuropsychology testing centers. Patients were 8 years or younger with a Pediatric Cerebral Performance Category score of 1 (normal) before PICU admission and less than or equal to 3 (no worse than moderate neurocognitive dysfunction) at PICU discharge, excluding patients with a history of neurocognitive deficits or who were readmitted and underwent mechanical ventilation. Biological siblings were aged 4 to 16 years at testing, with Pediatric Cerebral Performance Category score of 1 and no history of mechanical ventilation or general anesthesia. A total of 121 sibling pairs were enrolled from September 2, 2014, to December 13, 2017, and underwent neurocognitive testing starting March 14, 2015. The date of the final follow-up was November 6, 2018.
Exposures
Critical illness and PICU treatment for acute respiratory failure.
Main Outcomes and Measures
The primary outcome was IQ, estimated by the age-appropriate Vocabulary and Block Design subtests of the Wechsler Intelligence Scale. Secondary outcomes included measures of attention, processing speed, learning and memory, visuospatial skills, motor skills, language, and executive function. Evaluations occurred 3 to 8 years after hospital discharge.
Results
Patients (n = 121; 55 [45%] female patients) underwent PICU care at a median (IQR) age of 1.0 (0.2-3.2) years, received a median (IQR) of 5.5 (3.1-7.7) days of invasive mechanical ventilation, and were tested at a median (IQR) age of 6.6 (5.4-9.1) years. Matched siblings (n = 121; 72 [60%] female siblings) were tested at a median (IQR) age of 8.4 (7.0-10.2) years. Patients had a lower mean estimated IQ than matched siblings (101.5 vs 104.3; mean difference, –2.8 [95% CI, –5.4 to –0.2]). Among secondary outcomes, patients had significantly lower scores than matched siblings on nonverbal memory (mean difference, –0.9 [95% CI, –1.6 to –0.3]), visuospatial skills (mean difference, –0.9 [95% CI, –1.8 to –0.1]), and fine motor control (mean difference, –3.1 [95% CI, –4.9 to –1.4]) and significantly higher scores on processing speed (mean difference, 4.4 [95% CI, 0.2-8.5]). There were no significant differences in the remaining secondary outcomes, including attention, verbal memory, expressive language, and executive function.
Conclusions and Relevance
Among children, survival of PICU hospitalization for respiratory failure and discharge without severe cognitive dysfunction was associated with significantly lower subsequent IQ scores compared with matched siblings. However, the magnitude of the difference was small and of uncertain clinical importance.
This sibling-matched cohort study assesses neurocognitive function 3 to 8 years after hospital discharge in children who had acute respiratory failure requiring intensive care and invasive mechanical ventilation without a history of neurocognitive dysfunction and in matched siblings to determine the relationship between acute respiratory failure and treatment and long-term neurocognitive outcomes.
Introduction
In 2005, an estimated 23 700 infants and children outside of the neonatal period received critical care with invasive mechanical ventilation for acute respiratory failure in the US.1 Although most survive, long-term morbidity is increasingly recognized.2,3,4 Many neonates and adults surviving respiratory failure have long-term neurocognitive dysfunction.5,6,7,8 Neonates with respiratory failure are commonly born prematurely, at risk of intracranial hemorrhage, and with immature lungs requiring weeks or months of respiratory support, placing them at risk for subsequent neurodevelopmental impairment. Outcomes of adults with respiratory failure are affected by adult-onset comorbidities and age-related frailty and diminished cognitive capacity. In contrast, little data exist regarding long-term neurocognitive outcomes after respiratory failure in infants and children without perinatal problems or identified cognitive dysfunction.9 In addition, given concern about neurotoxic effects of critical illness and its treatment on the developing brain,10,11,12 infants and young children may be uniquely susceptible to adverse neurocognitive outcomes after invasive mechanical ventilation.
The Randomized Evaluation of Sedation Titration for Respiratory Failure (RESTORE) trial13 was a cluster-randomized clinical trial of pediatric patients receiving mechanical ventilation for acute respiratory failure. A sedation protocol at intervention sites vs usual care at control sites did not change the duration of mechanical ventilation. In this study, RESTORE-Cognition,14 multiple domains of neurocognitive function were assessed 3 to 8 years after hospital discharge in trial patients without a history of neurocognitive dysfunction and in matched, healthy siblings to determine the relationship between acute respiratory failure and treatment (sedation, mechanical ventilation, and other therapies) and long-term neurocognitive outcomes. We hypothesized that, compared with unexposed siblings, (1) patients exposed to acute respiratory failure and critical care therapies would have worse neurocognitive function and (2) neurocognitive function would be worse among patients hospitalized at a younger age.
Methods
This study was approved by the University of Pennsylvania institutional review board (IRB), and local sites entered into a reliance agreement with the University of Pennsylvania, engaged in an individual investigator’s agreement with the University of Pennsylvania, or obtained local IRB approval. Parents or guardians provided in-person consent for neurocognitive testing at the time of testing, and children of the appropriate age (according to IRB requirements at each site) provided assent. The design, hypotheses, sample size, and power calculations for this study were all prespecified and created prior to availability of trial results. The study protocol (Supplement 1) has been previously published.14
We conducted a prospective cohort study that enrolled a subset of patients from the 87% of trial patients whose caregivers consented to follow-up 6 months after discharge and a control group of matched siblings. We contacted parents/guardians of trial patients via telephone after their trial participation was completed to administer screening questions for this study. If their child was eligible for this study and they agreed to participate, we screened for an eligible sibling, interviewed the parent/guardian, and scheduled the patient (and sibling if eligible) for neurocognitive testing. Inclusion criteria for this study were aged 8 years or younger at trial enrollment, Pediatric Cerebral Performance Category (PCPC) score of 1 (normal) prior to pediatric intensive care unit (PICU) admission, and PCPC score less than or equal to 3 (no worse than moderate neurocognitive dysfunction) at PICU discharge.15 We excluded patients with a history of acute or underlying disease associated with neurocognitive deficits, cardiac arrest, traumatic brain injury with loss of consciousness, genetic disorder associated with neurocognitive deficit, premature birth (<32 weeks’ gestational age), or birth weight less than 2500 g and those who, prior to testing, had a hospital readmission that included mechanical ventilation and sedation. Siblings of patients enrolled in this study were eligible if they were aged 4 to 16 years at the time of neurocognitive testing, had a PCPC score of 1, and had the same biological parents as and lived with the patient. Siblings were excluded using the same medical history exclusion criteria as patients, if they had been hospitalized and received sedation and mechanical ventilation, or if they ever received general anesthesia. Enrollment took place from September 2, 2014, to December 13, 2017, and neurocognitive testing was conducted from March 14, 2015, to November 6, 2018 (the final date of follow-up).
Per National Institutes of Health requirements and to assess the relationship between race and ethnicity on process and outcomes, race and ethnicity data were collected from the medical record during the trial hospitalization using investigator-defined categories. If data were not available in the medical record, research staff were instructed to ask families about race and ethnicity directly, using the same categories.
Exposures
As part of the trial, detailed data were collected on each patient’s medical history, acute illness, hospital course, receipt of critical care therapies and medications, and functional status (by the Pediatric Overall Performance Category and PCPC scores15). For this study, we obtained patient and sibling developmental histories, patient postdischarge medical problems, and socioeconomic status.
Outcomes
We performed developmentally appropriate, comprehensive, performance-based neurocognitive outcome assessments when patients and siblings were 4 years or older,14 allowing enough time after hospitalization for transient deficits to resolve and longer-lasting neurocognitive sequelae to manifest. The minimum testing age was chosen so comparable neurocognitive tests could be administered to all patients and siblings regardless of age at testing. Neurocognitive testing was conducted by either a licensed psychologist with specialized training in neuropsychology or a supervised neuropsychology technician. The primary outcome was IQ estimated by the age-appropriate Vocabulary and Block Design subtests of the Wechsler Intelligence Scale, which is highly correlated with full-scale IQ.16,17,18,19 Secondary outcomes included attention, processing speed, learning and memory, visuospatial skills, motor skills, language, and executive function, assessed using standardized tests (eTable 1 in Supplement 2).14 All test results were anchored with age-referenced scoring procedures. To ensure protocol fidelity across testing sites, we conducted training sessions with site technicians and supervisors. After each evaluation, the central neuropsychology team reviewed each test to ensure that data were complete and absent of errors. Testing scores were excluded if important administration- or child-related problems were noted (reliability code >2).
Statistical Analysis
Mean differences and 95% CIs were calculated using linear regression (intercept-only models) on paired differences of neurocognitive outcomes between matched patients and siblings. We also compared the estimated IQ of sibling pairs by using cutoffs of less than or equal to 70, less than or equal to 85, and greater than or equal to 115 using McNemar exact tests and by assessing Pearson correlation. The cutoff of 70 represents IQ scores 2 SDs below the population mean and generally denotes intellectual disability. The cutoffs of 85 and 115 represent 1 SD below and above the population mean, respectively, between which approximately two-thirds of the population typically scores. We explored adjusting for centered age at PICU admission, patient age at testing, sibling age at testing, and duration between hospital discharge and testing in the linear regression models. We centered these continuous variables (subtracted the mean from each individual value) to allow for meaningful interpretation of the intercept term as the mean difference between matched patients and siblings at the mean value of age or duration. We also calculated the mean difference restricting to sibling pairs with an age difference less than 4 years and to patients with PCPC score of 1 at the time of interview. We compared demographic, medical history, hospital course, and family characteristics of patients who scored substantially worse in estimated IQ (≥1 SD [15 points] lower) vs similar to or better than their siblings using linear, cumulative logit, logistic, multinomial logistic, and proportional hazards regression for continuous, ordinal, binary, nominal, and time-to-event variables, respectively. In linear regression analyses of patient characteristics, continuous variables were log-transformed as necessary. For the duration of mechanical ventilation and PICU stay, the proportional hazards assumption was assessed using a time-varying interaction between the logarithm of time and group. In both models, the interaction term was not statistically significant (P = .69 and P = .63); thus, the null hypothesis of proportionality of hazards was not rejected.
To examine the association of outcomes with categorical age at PICU admission (<1, 1-3.99, 4-8.99 y), sex, socioeconomic status (tertiles of Hollingshead Four-Factor Index of Socioeconomic Status score20), and trial treatment group, these variables were added as covariates to the linear regression models for the paired differences in testing scores between matched patients and siblings. We also explored the association of outcomes with categorical age at PICU admission restricting to patients with a PCPC score of 1. Due to the cluster-randomized design of the trial, all regression analyses accounted for PICU as a cluster variable using generalized estimating equations. Data were analyzed as available, without imputation for missing values. Because of the potential for type I error due to multiple comparisons, findings for analyses of secondary end points should be interpreted as exploratory. Analyses were performed with SAS, version 9.4 (SAS Institute), using 2-sided α = .05 level tests, except for pairwise covariate subgroup comparisons, which used 2-sided α = .01 level tests due to multiple comparisons.
Results
A total of 121 sibling pairs were tested (Figure 1), of which 116 were included in the primary outcome analysis and 66 to 119 were included in the analyses of secondary outcomes. Estimated IQ was not available for 5 sibling pairs, and secondary outcomes were not available for 1 to 6 sibling pairs because of problems with test administration or the child not completing the test. Sibling pairs were most frequently non-Hispanic White (67%), from families with at least 1 parent working full-time (47%) or part-time (17%), and had a median (IQR) Hollingshead score of 40 (30.5-50.3) (primarily middle- and upper–middle-class families) (Table 1). Overall, 27 patients (22%) were older than their matched siblings, 10 sibling pairs (8%) were twins, and 84 patients (69%) were younger than their matched siblings. The median (IQR) age at testing was 6.6 (5.4-9.1) years for patients and 8.4 (7.0-10.2) years for siblings (mean difference, –1.5 years [95% CI, –2.0 to –1.0]). Parent interviews were conducted a median (IQR) of 3.8 (3.2-5.2) years after hospitalization, while patients were tested a median (IQR) of 5.2 (4.3-6.1) years after hospitalization. A total of 55 patients (45%) and 72 siblings (60%) were females; 56 (46%) sibling pairs were of the same sex. At the time of the parent interview, 92% of patients and 100% of siblings had a PCPC score of 1.
Figure 1. Flow of Participants in a Study of the Association of Acute Respiratory Failure in Early Childhood With Long-term Neurocognitive Outcomes.
PCPC indicates Pediatric Cerebral Performance Category; PICU, pediatric intensive care unit.
aSixteen patients were unable to undergo testing due to behavioral or other factors on the day of testing.
Table 1. Characteristics of Sibling Pairs in a Study of the Association of Acute Respiratory Failure in Early Childhood With Long-term Neurocognitive Outcomes.
| Characteristics | No. (%) | |
|---|---|---|
| Patients (n = 121) | Siblings (n = 121) | |
| Demographic characteristics | ||
| Age at PICU admission, y | ||
| Median (IQR) | 1.0 (0.2-3.2) | |
| <1 | 62 (51) | |
| 1-3.99 | 34 (28) | |
| 4-8.99 | 25 (21) | |
| Age at testing, y | ||
| Median (IQR) | 6.6 (5.4-9.1) | 8.4 (7.0-10.2) |
| 4-4.99 | 25 (21) | 9 (7) |
| 5-5.99 | 21 (17) | 7 (6) |
| 6-16.99 | 75 (62) | 105 (87) |
| Birth order | ||
| Older sibling | 27 (22) | 84 (69) |
| Twin | 10 (8) | 10 (8) |
| Younger sibling | 84 (69) | 27 (22) |
| Sex | ||
| Female | 55 (45) | 72 (60) |
| Male | 66 (55) | 49 (40) |
| Race and ethnicity | ||
| African American/Black | 22 (18) | 22 (18) |
| Asian | 1 (<1) | 1 (<1) |
| Hispanic/Latinx | 14 (12) | 14 (12) |
| White | 81 (67) | 81 (67) |
| Othera | 3 (2) | 3 (2) |
| Baseline medical characteristics | ||
| PRISM III-12 score, median (IQR)b | 6 (2-9) | |
| Risk of mortality based on PRISM III-12 score, median (IQR) | 2.2 (0.9-4.9) | |
| Medical history | ||
| Asthma (prescribed bronchodilators or steroids) | 9 (7) | |
| Cancer (current or previous diagnosis) | 2 (2) | |
| Primary diagnosis for PICU admission | ||
| Bronchiolitis or asthma (or reactive airway disease) | 53 (44) | |
| Pneumonia or aspiration pneumonia | 45 (37) | |
| Acute respiratory failure related to sepsis | 15 (12) | |
| Other acute diagnosesc | 8 (7) | |
| RESTORE intervention groupd | 73 (60) | |
| Hospital course | ||
| Duration of mechanical ventilation, median (IQR), d | 5.5 (3.1-7.7) | |
| Length of PICU stay, median (IQR), d | 8.0 (5.1-12.0) | |
| Rescue therapy (ECMO and/or HFOV) | 16 (13) | |
| SpO2 ever <90% | 15 (12) | |
| Pediatric acute respiratory distress syndrome based on worst oxygenation index/oxygen saturation index during PICU stay | ||
| At-risk or mild (OI <8.0 or OSI <7.5) | 44 (36) | |
| Moderate (OI of 8.0-15.9 or OSI of 7.5-12.2) | 41 (34) | |
| Severe (OI ≥16.0 or OSI ≥12.3) | 36 (30) | |
| Multiple organ dysfunction syndrome | 87 (72) | |
| New | 16 (13) | |
| Concurrent | 71 (59) | |
| Respiration dysfunction only | 34 (28) | |
| Maximal number of organ dysfunctions, median (IQR) | 2 (1-3) | |
| Ever received vasoactive medications | 53 (44) | |
| Pediatric cerebral performance category at hospital discharge | ||
| 1 (normal) | 114 (94) | |
| 2 (mild disability) | 5 (4) | |
| 3 (moderate disability) | 2 (2) | |
| Interview | n = 116 | |
| Duration between hospital discharge and interview, y | ||
| Median (IQR) | 3.8 (3.2-5.2) | |
| <3.5 | 38 (33) | |
| 3.5-5.0 | 46 (40) | |
| ≥5.0 | 32 (28) | |
| Parent education | n = 115 | |
| High school graduate/GED or less | 15 (13) | |
| Some college or technical school | 32 (28) | |
| College graduate/postgraduate | 68 (59) | |
| Parent work status | n = 115 | |
| Raising children full-time or student | 42 (37) | |
| Working | ||
| Part-time | 19 (17) | |
| Full-time | 54 (47) | |
| Hollingshead Four-Factor Index of Socioeconomic Status score, median (IQR)e | 40 (30.5-50.3) | |
| Duration between hospital discharge and testing, y | ||
| Median (IQR) | 5.2 (4.3-6.1) | |
| <4.0 | 17 (14) | |
| 4.0-5.99 | 71 (59) | |
| ≥6.0 | 33 (27) | |
Abbreviations: ECMO, extracorporeal membrane oxygenation; GED, general educational development; HFOV, high-frequency oscillatory ventilation; OI, oxygenation index; OSI, oxygen saturation index; PICU, pediatric intensive care unit; PRISM III-12, Pediatric Risk of Mortality III score within first 12 hours in the PICU.
Other includes Native Hawaiian or Other Pacific Islander and multiracial/more than 1 race.
The scale for the PRISM III-12 score ranges from 0 to 74, with higher scores indicating a higher risk of death. Scores were collected in the first 12 hours in the PICU.
Other acute diagnoses include laryngotracheobronchitis, pulmonary edema, pertussis, pneumothorax (nontrauma), and pulmonary hemorrhage.
Patients enrolled at intervention sites in the RESTORE trial were treated using the sedation protocol vs patients enrolled at control sites who were treated with usual care.
The scale for the Hollingshead Four-Factor Index of Socioeconomic Status score ranges from 8 to 66, with higher scores indicating higher socioeconomic status based on parent education and occupation. A Hollingshead score of 40 is generally considered to represent middle class.
Patients underwent critical care at a median (IQR) age of 1.0 (0.2-3.2) years, 112 (91%) were previously healthy, and all had a PCPC score of 1 at hospital admission (Table 1). The most common etiologies of respiratory failure were bronchiolitis or asthma (53 patients [44%]) and pneumonia (45 patients [37%]). Seventy-three patients (60%) had been treated at intervention sites. Nearly two-thirds of patients had moderate (41 patients [34%]) or severe (36 patients [30%]) pediatric acute respiratory distress syndrome. The median (IQR) duration of mechanical ventilation was 5.5 (3.1-7.7) days, 15 patients (12%) received high-frequency oscillatory ventilation, and 3 (2%) received extracorporeal membrane oxygenation support. All patients had respiratory failure and most (87 patients [72%]) had multiple organ dysfunction syndrome. The median (IQR) PICU length of stay was 8.0 (5.1-12.0) days, and 114 patients (94%) had a PCPC score of 1 at hospital discharge.
Primary Outcome
Three to 8 years after PICU hospitalization for acute respiratory failure, patients had a lower mean estimated IQ than their siblings (101.5 vs 104.3; mean difference, –2.8 [95% CI, –5.4 to –0.2]) (Table 2). Twenty patients (17%) had estimated IQ scores at least 15 points lower than their siblings (eTable 2 in Supplement 2) and 9 (8%) had scores at least 15 points higher. More patients had an estimated IQ of less than or equal to 85 than their siblings (19 [16%] patients vs 8 [7%] siblings; P = .03) but not of less than or equal to 70 (2 [2%] patients vs 3 [3%] siblings; P > .99). There were no significant differences in frequency of estimated IQ greater than or equal to 115 (18 [16%] patients vs 22 [19%] siblings; P = .52). The Pearson correlation between patient and sibling estimated IQ scores was 0.43 (95% CI, 0.27-0.57).
Table 2. Neuropsychological Testing Results in a Study of the Association of Acute Respiratory Failure in Early Childhood With Long-term Neurocognitive Outcomesa.
| Neurocognitive outcomes and tests | No. of sibling pairs tested/eligible pairs | Score range | Mean (SD) score | Mean difference (95% CI)b | P valueb | |
|---|---|---|---|---|---|---|
| Patients | Siblings | |||||
| Global cognitive function | ||||||
| Estimated IQ: vocabulary and block design | 116/121 | 40 to 160 | 101.5 (12.6) | 104.3 (12.2) | −2.8 (−5.4 to −0.2) | .03 |
| Expressive language | ||||||
| Vocabulary | 117/121 | 1 to 19 | 10.6 (2.9) | 10.9 (2.7) | −0.3 (−0.9 to 0.3) | .36 |
| Processing speed | ||||||
| Coding | 88/90 | 1 to 19 | 10.0 (3.2) | 9.9 (2.9) | 0.1 (−0.5 to 0.8) | .71 |
| Processing speed index | 66/67 | 40 to 160 | 105.4 (13.6) | 101.1 (13.5) | 4.4 (0.2 to 8.5) | .04 |
| Verbal memory | ||||||
| Short delay free recall | 84/90 | −4 to 4 | 0.22 (0.87) | 0.35 (0.94) | −0.13 (−0.35 to 0.10) | .28 |
| Long delay free recall | 84/90 | −4 to 4 | 0.13 (0.93) | 0.33 (0.98) | −0.20 (−0.48 to 0.08) | .16 |
| Memory for stories | 84/90 | 1 to 20 | 10.4 (2.7) | 10.7 (3.1) | −0.3 (−0.9 to 0.3) | .36 |
| Memory for stories delayed | 84/90 | 1 to 20 | 9.8 (2.5) | 10.4 (3.2) | −0.6 (−1.3 to 0.0) | .06 |
| Nonverbal memory | ||||||
| Facial memory | 87/90 | 1 to 20 | 9.5 (2.8) | 10.4 (3.1) | −0.9 (−1.6 to −0.3) | .007 |
| Memory for location | 85/90 | 1 to 20 | 9.7 (3.1) | 9.4 (2.5) | 0.3 (−0.4 to 1.1) | .39 |
| Visuospatial | ||||||
| Block design | 119/121 | 1 to 19 | 9.7 (3.2) | 10.7 (3.0) | −0.9 (−1.8 to −0.1) | .03 |
| Visual-motor integration | 119/121 | 44 to 155 | 95.2 (9.8) | 94.9 (11.8) | 0.3 (−1.9 to 2.4) | .81 |
| Attention | ||||||
| Digits forward | 86/90 | 1 to 20 | 9.5 (2.6) | 9.1 (2.5) | 0.5 (−0.2 to 1.1) | .14 |
| Fine motor control | ||||||
| Fine manual control index | 115/121 | 20 to 80 | 46.3 (9.6) | 49.5 (9.3) | −3.1 (−4.9 to −1.4) | <.001 |
| Working memory | ||||||
| Digits backward | 85/90 | 1 to 20 | 10.7 (2.8) | 10.8 (2.7) | −0.1 (−0.8 to 0.7) | .83 |
Higher scores denote better performance in all tests. See eTable 1 in Supplement 2 for the details on all neuropsychological tests.
Mean differences (95% CI) and P values were calculated using linear regression (intercept-only models) on paired differences of scores between matched patients and siblings accounting for pediatric intensive care unit as a cluster variable using generalized estimating equations.
When adjusting for patient age at PICU admission, patient age at testing, sibling age at testing, and duration between hospital discharge and testing, the difference in estimated IQ between patients and siblings remained statistically significantly different (eTable 3 in Supplement 2). When restricting to the 102 sibling pairs with an age difference of less than 4 years, the difference in mean estimated IQ remained statistically significant (101.5 vs 104.5; mean difference, –3.0 [95% CI, –5.9 to –0.1]). When restricting analyses to the 103 sibling pairs in which the patient’s PCPC score was 1 (normal) at the time of the postdischarge interview, the difference in mean estimated IQ was not statistically significant (102.3 in patients vs 104.2 in siblings; mean difference, –1.9 [95% CI, –4.7 to 0.9]).
Comparing patients who scored substantially lower in estimated IQ (≥1 SD) vs patients who scored similar to or higher than their siblings (eTable 2 in Supplement 2), patients with lower scores were younger at PICU admission and at testing. Median (IQR) age at PICU admission was 0.2 (0.1-1.1) years for patients who scored substantially lower than their siblings vs 1.4 (0.2-3.9) years for patients who scored the same as or higher than their siblings (mean difference, –1.3 years [95% CI, –2.3 to –0.2]). Median (IQR) age at testing was 5.4 (4.4-6.7) years for patients who scored substantially lower than their siblings vs 6.9 (5.7-9.5) years for patients who scored the same as or higher than their siblings (mean difference, –1.6 years [95% CI, –2.8 to –0.4]). There were no significant differences between groups in admission diagnosis or severity of illness at admission or during hospitalization, including duration of mechanical ventilation, severity of pediatric acute respiratory distress syndrome, receipt of rescue therapy with high-frequency oscillatory ventilation or extracorporeal membrane oxygenation, or occurrence of multiple organ dysfunction syndrome.
The difference in estimated IQ between patients and matched siblings was greater among patients admitted to the hospital at younger categorical age (Figure 2; eTables 4 and 5 in Supplement 2). Estimated IQ scores of patients hospitalized at younger than 1 year were a mean of 4.6 points (95% CI, –8.1 to –1.1) lower than their siblings vs patients hospitalized at 4 to 8 years of age. For sibling pairs in which the sibling was older than the patient (n = 80), the older sibling scored higher than the younger patient (mean estimated IQ, 105.1 vs 100.7; mean difference, 4.4 [95% CI, 0.9-7.9]). The difference in estimated IQ between siblings was not significantly different by sex, Hollingshead score tertile, or trial treatment group (all P > .10) (eTables 6-8 in Supplement 2).
Figure 2. Estimated IQ of Patients and Matched Siblings in a Study of the Association of Acute Respiratory Failure in Early Childhood With Long-term Neurocognitive Outcomes.
A. Within each box, solid circles represent mean values and horizontal lines denote median values; boxes extend from the 25th to the 75th percentile. Open circles denote individual observations. The horizontal line at an estimated IQ score of 100 indicates the population mean, and the dashed horizontal lines at estimated IQ scores of 85 and 115 indicate cutoffs representing 1 standard deviation below and above this value, between which roughly two-thirds of the population typically scores. E, Each vertical line represents 1 sibling pair; the vertical line plots the difference between the patient estimated IQ (black line) and the sibling estimated IQ (end of vertical line). Vertical lines that extend above the black line represent pairs in which the sibling scored higher than the patient, while vertical lines that extend below the black line represent pairs in which the sibling scored lower than the patient. PICU indicates pediatric intensive care unit.
Secondary Outcomes
Among secondary outcomes, patients scored lower than their siblings on visuospatial skills (Block Design mean score, 9.7 vs 10.7; mean difference, –0.9 [95% CI, –1.8 to –0.1]), nonverbal memory (Facial Memory mean score, 9.5 vs 10.4; mean difference, –0.9 [95% CI, –1.6 to –0.3]), and fine motor control (Fine Manual Control Index mean score, 46.3 vs 49.5; mean difference, –3.1 [95% CI, –4.9 to –1.4]) (Table 2). In contrast, the Processing Speed Index score was higher in patients among the 66 sibling pairs old enough to complete that test (mean, 105.4 vs 101.1; mean difference, 4.4 [95% CI, 0.2-8.5]).
Differences in scores between patients and siblings varied significantly by age at hospitalization for a number of tests (eTable 4 in Supplement 2). Block Design scores of patients were a mean of 1.8 points lower (95% CI, –3.0 to –0.5) compared with their siblings in patients hospitalized at younger than 1 year (9.2 vs 10.7; mean difference, –1.6 [95% CI, –2.7 to –0.4]) vs those hospitalized at age 4 to 8 years (10.4 vs 10.2; mean difference, 0.2 [95% CI, –0.9 to 1.3]) (eTable 4 in Supplement 2). In contrast, processing speed (assessed by Coding) was higher among patients exposed at younger compared with older ages (eTable 4 in Supplement 2). The only sex-related difference (P = .009) was in the Digits Backward test of working memory, in which female patients scored a mean of 1.3 points (95% CI, –2.2 to –0.3) lower than their siblings (10.4 vs 11.6), whereas male patients scored a mean of 1.0 point (95% CI, –0.2 to 2.2) higher than their siblings (11.0 vs 10.0) (eTable 6 in Supplement 2). Secondary outcomes were not affected by socioeconomic status or trial treatment group (eTables 7 and 8 in Supplement 2).
Discussion
In this study of children without a history of cognitive dysfunction surviving hospitalization for acute respiratory failure at 8 years or younger, patients had a significantly lower subsequent IQ score compared with matched, healthy biological siblings living in the same household. The magnitude of the difference was small and of uncertain clinical importance. This difference was found despite not enrolling children who had severe neurocognitive impairment at hospital discharge.
Although the mean difference between patients and siblings would be small at an individual level, even small changes in mean IQ can have important implications depending on the distribution of scores. For example, population studies of lead exposure in childhood reported a mean of 6 IQ points lower,21,22 which is near the upper end of the 95% CI of the difference in the current study. The much-greater rates of patients with estimated IQ less than or equal to 85 and of estimated IQ at least 15 points below their siblings demonstrates an overall downward shift in estimated IQ among patients. These findings may have important academic, social, and economic implications for young children surviving acute respiratory failure,23,24 and are consistent with a 2020 study that found school problems among 13% of all PICU survivors in Finland.24
Multiple aspects of critical illness have been associated with impaired neurocognitive function in neonates and adults, who have markedly different pathophysiology and underlying diseases than children outside the neonatal period.6,25,26,27 Few studies examined children outside of the neonatal period,28 and even fewer included detailed neuropsychological evaluations or well-matched control groups.9,29 Elison et al evaluated 16 patients in the PICU compared with age- and sex-matched controls and found deficits in spatial and verbal memory and attention, with patients surviving sepsis having particular difficulty with pattern recognition.29 After a trial of tight glucose control in patients in the PICU (75% with congenital heart disease), patients scored lower on IQ and several other neuropsychological tests compared with mixed community and sibling controls. However, important underlying differences between the trial patients and controls limited the conclusions about the effect of the episode of critical illness and care due to chronic conditions associated with neurocognitive deficits in many trial patients.30
No specific illness or treatment factors were associated with patients having scores substantially lower (≥1 SD lower) than their unexposed siblings. Factors previously reported to be associated with poor functional status and quality of life after discharge31,32 were not significantly different in former patients who scored lower than matched siblings and in those who scored similar to or higher than matched siblings. No differences by socioeconomic status of the sibling pairs were found, which is consistent with them living together and being exposed to a similar environment.
Younger children who survived acute respiratory failure were enrolled specifically due to concerns related to increased vulnerability of the developing brain to critical illness and critical care–related insults, particularly given the development of functional brain architecture, refined thalamocortical connections, myelination, and programmed cell death, all of which can be disrupted by ischemia and inflammation.11,12 In addition, the maximum synaptic excess occurring in the second half of the first year followed by decline of low-usage circuits may be affected by sedative and anesthetic medications, which show greater toxicity in very young animals and in observational studies in young children.10,33,34 Recent clinical studies found no detrimental effects of transient anesthesia exposure in young children.35,36 In contrast, when young children were exposed to ICU therapies and sedation for days in this study (in contrast to hours of exposure to anesthesia), the difference in estimated IQ between siblings was greater for patients hospitalized in the youngest age group.
Patients in this study were more commonly younger than their siblings at the time of testing. The distribution of ages is consistent with the finding that half of the patients were hospitalized as infants, and siblings had to be 4 years or older to be tested, so more older siblings were eligible to participate than younger siblings. Because the tests chosen are designed to consider the developmental epoch of the child and are age-standardized, they are stable across the age range (eg, IQs do not increase with age and they do not decrease in the absence of congenital or acquired brain injury). Thus, the IQ obtained from testing a child aged 6 years will be comparable to that obtained from a child aged 12 years, and comparisons of results of children undergoing comparable tests at different ages are valid,37 which is consistent with the findings that the difference in estimated IQ between patients and siblings remained significant even when adjusting for patient and sibling age at testing.
Limitations
This study has several limitations. First, the unpredictable nature of acute respiratory failure (and critical illness in general) precluded baseline measurement of neurocognitive function among patients. Therefore, the extent to which the results of neurocognitive testing are a reflection of underlying ability vs effects of the episode of illness cannot be determined. However, by comparing patients’ biological siblings, long-term environmental exposures and genetic/biological influence were controlled. Nonetheless, although sibling IQs correlate with each other, they do not match exactly.38,39 Second, this study was focused specifically on a group without known neurocognitive deficits or medical conditions associated with deficits, so these results are not generalizable to children with underlying neurocognitive or developmental impairment. Third, to enhance feasibility of data collection at multiple sites across the US, IQ was estimated using 2 Wechsler Intelligence Scale subscales rather than measuring full-scale IQ, which are highly, but not perfectly, correlated.18,19 Fourth, little is known about sibling outcomes after critical illness, nor about whether parenting of siblings or child development differs based on birth order or on relationship between patient critical illness and the birth of siblings; those factors could affect the control group findings in this study. If siblings also incur negative effects related to the critical illness, differences between critically ill children and the control siblings would be blunted. Fifth, when the study was planned, 248 sibling pairs were estimated to undergo testing. Ultimately, fewer patients and siblings were eligible than anticipated, and some eligible patients did not consent or return for testing years after hospitalization, decreasing the power of the study.
Conclusions
Among children, survival of intensive care hospitalization for respiratory failure and discharge without severe cognitive dysfunction was associated with significantly lower subsequent IQ score compared with matched siblings. However, the magnitude of the difference was small and of uncertain clinical importance.
Section Editor: Christopher Seymour, MD, Associate Editor, JAMA (christopher.seymour@jamanetwork.org).
Trial protocol
eTable 1. Neuropsychological Testing Instruments
eTable 2. Characteristics of Patients Exposed to Sedation in Early Childhood According to Estimated IQ Score Comparison to Unexposed Siblings
eTable 3. Adjusted Analyses of Estimated IQ
eTable 4. Results of Neuropsychological Testing By Patient Age at PICU Admission
eTable 5. Results of Neuropsychological Testing By Patient Age at PICU Admission for Patients with Pediatric Cerebral Performance Category of 1 at Time of Post-Discharge Interview
eTable 6. Results of Neuropsychological Testing By Patient Sex
eTable 7. Results of Neuropsychological Testing By Hollingshead Four-Factor Index of Socioeconomic Status Score
eTable 8. Results of Neuropsychological Testing By Trial Treatment Group
Nonauthor collaborators
References
- 1.Wunsch H, Linde-Zwirble WT, Angus DC, Hartman ME, Milbrandt EB, Kahn JM. The epidemiology of mechanical ventilation use in the United States. Crit Care Med. 2010;38(10):1947-1953. doi: 10.1097/CCM.0b013e3181ef4460 [DOI] [PubMed] [Google Scholar]
- 2.Kyösti E, Liisanantti JH, Peltoniemi O, et al. Five-year survival and causes of death in children after intensive care-a national registry study. Pediatr Crit Care Med. 2018;19(3):e145-e151. doi: 10.1097/PCC.0000000000001424 [DOI] [PubMed] [Google Scholar]
- 3.Pinto NP, Rhinesmith EW, Kim TY, Ladner PH, Pollack MM. Long-Term function after pediatric critical illness: results from the survivor outcomes study. Pediatr Crit Care Med. 2017;18(3):e122-e130. doi: 10.1097/PCC.0000000000001070 [DOI] [PubMed] [Google Scholar]
- 4.Keim G, Yehya N, Spear D, et al. ; Pediatric Acute Lung Injury and Sepsis Investigators (PALISI) Network . Development of persistent respiratory morbidity in previously healthy children after acute respiratory failure. Crit Care Med. 2020;48(8):1120-1128. doi: 10.1097/CCM.0000000000004380 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Latronico N, Peli E, Calza S, et al. Physical, cognitive and mental health outcomes in 1-year survivors of COVID-19-associated ARDS. Thorax. Published online September 29, 2021. doi: 10.1136/thoraxjnl-2021-218064. [DOI] [PubMed] [Google Scholar]
- 6.Sasannejad C, Ely EW, Lahiri S. Long-term cognitive impairment after acute respiratory distress syndrome: a review of clinical impact and pathophysiological mechanisms. Crit Care. 2019;23(1):352. doi: 10.1186/s13054-019-2626-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Hopkins RO, Wade D, Jackson JC. What’s new in cognitive function in ICU survivors. Intensive Care Med. 2017;43(2):223-225. doi: 10.1007/s00134-016-4550-x [DOI] [PubMed] [Google Scholar]
- 8.Eves R, Mendonça M, Baumann N, et al. Association of very preterm birth or very low birth weight with intelligence in adulthood: an individual participant data meta-analysis. JAMA Pediatr. 2021;175(8):e211058-e211058. doi: 10.1001/jamapediatrics.2021.1058 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Kachmar AG, Irving SY, Connolly CA, Curley MAQ. A systematic review of risk factors associated with cognitive impairment after pediatric critical illness. Pediatr Crit Care Med. 2018;19(3):e164-e171. doi: 10.1097/PCC.0000000000001430 [DOI] [PubMed] [Google Scholar]
- 10.Block RI, Thomas JJ, Bayman EO, Choi JY, Kimble KK, Todd MM. Are anesthesia and surgery during infancy associated with altered academic performance during childhood? Anesthesiology. 2012;117(3):494-503. doi: 10.1097/ALN.0b013e3182644684 [DOI] [PubMed] [Google Scholar]
- 11.Finder M, Boylan GB, Twomey D, Ahearne C, Murray DM, Hallberg B. Two-year neurodevelopmental outcomes after mild hypoxic ischemic encephalopathy in the era of therapeutic hypothermia. JAMA Pediatr. 2020;174(1):48-55. doi: 10.1001/jamapediatrics.2019.4011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Ofek-Shlomai N, Berger I. Inflammatory injury to the neonatal brain—what can we do? Front Pediatr. 2014;2:30. doi: 10.3389/fped.2014.00030 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Curley MA, Wypij D, Watson RS, et al. ; RESTORE Study Investigators and the Pediatric Acute Lung Injury and Sepsis Investigators Network . Protocolized sedation vs usual care in pediatric patients mechanically ventilated for acute respiratory failure: a randomized clinical trial. JAMA. 2015;313(4):379-389. doi: 10.1001/jama.2014.18399 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Curley MAQ, Watson RS, Cassidy AM, et al. ; RESTORE-cognition Study Investigators . Design and rationale of the “sedation strategy and cognitive outcome after critical illness in early childhood” study. Contemp Clin Trials. 2018;72:8-15. doi: 10.1016/j.cct.2018.07.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Fiser DH. Assessing the outcome of pediatric intensive care. J Pediatr. 1992;121(1):68-74. doi: 10.1016/S0022-3476(05)82544-2 [DOI] [PubMed] [Google Scholar]
- 16.Wechsler D. Wechsler Intelligence Scales for Children.4th ed. Psychological Corporation; 2003. [Google Scholar]
- 17.Wechsler D. Wechsler Preschool and Primary Scale of Intelligence. 4th ed. Pearson; 2012. [Google Scholar]
- 18.Hrabok M, Brooks BL, Fay-McClymont TB, Sherman EM. Wechsler Intelligence Scale for Children-fourth edition (WISC-IV) short-form validity: a comparison study in pediatric epilepsy. Child Neuropsychol. 2014;20(1):49-59. doi: 10.1080/09297049.2012.741225 [DOI] [PubMed] [Google Scholar]
- 19.Sattler JM. Assessment of Children: Cognitive Foundations. 5th ed. Jerome M. Sattler Publisher; 2008. [Google Scholar]
- 20.Hollingshead AB. Four factor index of social status. Yale Journal of Sociology. 2011;8(Fall):32. [Google Scholar]
- 21.Needleman HL, Bellinger D. Studies of lead exposure and the developing central nervous system: a reply to Kaufman. Arch Clin Neuropsychol. 2001;16(4):359-374. doi: 10.1093/arclin/16.4.359 [DOI] [PubMed] [Google Scholar]
- 22.Bellinger DC. A strategy for comparing the contributions of environmental chemicals and other risk factors to neurodevelopment of children. Environ Health Perspect. 2012;120(4):501-507. doi: 10.1289/ehp.1104170 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Nevin R, Jacobs DE, Berg M, Cohen J. Monetary benefits of preventing childhood lead poisoning with lead-safe window replacement. Environ Res. 2008;106(3):410-419. doi: 10.1016/j.envres.2007.09.003 [DOI] [PubMed] [Google Scholar]
- 24.Kyösti E, Peltoniemi O, Liisanantti JH, et al. School performance after pediatric intensive care-association of mental well-being, chronic illnesses, and family socioeconomic status. Pediatr Crit Care Med. 2020;21(12):e1099-e1105. doi: 10.1097/PCC.0000000000002564 [DOI] [PubMed] [Google Scholar]
- 25.Hopkins RO, Haaland KY. Neuropsychological and neuropathological effects of anoxic or ischemic induced brain injury. J Int Neuropsychol Soc. 2004;10(7):957-961. doi: 10.1017/S1355617704107108 [DOI] [PubMed] [Google Scholar]
- 26.Jones C, Griffiths RD, Slater T, Benjamin KS, Wilson S. Significant cognitive dysfunction in non-delirious patients identified during and persisting following critical illness. Intensive Care Med. 2006;32(6):923-926. doi: 10.1007/s00134-006-0112-y [DOI] [PubMed] [Google Scholar]
- 27.Ehlenbach WJ, Hough CL, Crane PK, et al. Association between acute care and critical illness hospitalization and cognitive function in older adults. JAMA. 2010;303(8):763-770. doi: 10.1001/jama.2010.167 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Alievi PT, Carvalho PR, Trotta EA, Mombelli Filho R. The impact of admission to a pediatric intensive care unit assessed by means of global and cognitive performance scales. J Pediatr (Rio J). 2007;83(6):505-511. doi: 10.1590/S0021-75572007000800005 [DOI] [PubMed] [Google Scholar]
- 29.Elison S, Shears D, Nadel S, Sahakian B, Garralda ME. Neuropsychological function in children following admission to paediatric intensive care: a pilot investigation. Intensive Care Med. 2008;34(7):1289-1293. doi: 10.1007/s00134-008-1093-9 [DOI] [PubMed] [Google Scholar]
- 30.Mesotten D, Gielen M, Sterken C, et al. Neurocognitive development of children 4 years after critical illness and treatment with tight glucose control: a randomized controlled trial. JAMA. 2012;308(16):1641-1650. doi: 10.1001/jama.2012.12424 [DOI] [PubMed] [Google Scholar]
- 31.Watson RS, Asaro LA, Hutchins L, et al. Risk factors for functional decline and impaired quality of life after pediatric respiratory failure. Am J Respir Crit Care Med. 2019;200(7):900-909. doi: 10.1164/rccm.201810-1881OC [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Tasker RC, Menon DK. Critical care and the brain. JAMA. 2016;315(8):749-750. doi: 10.1001/jama.2016.0701 [DOI] [PubMed] [Google Scholar]
- 33.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(2):227-240. doi: 10.1097/ALN.0000000000001735 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Lin EP, Lee JR, Lee CS, Deng M, Loepke AW. Do anesthetics harm the developing human brain? an integrative analysis of animal and human studies. Neurotoxicol Teratol. 2017;60:117-128. doi: 10.1016/j.ntt.2016.10.008 [DOI] [PubMed] [Google Scholar]
- 35.Davidson AJ, Disma N, de Graaff JC, et al. ; GAS consortium . Neurodevelopmental outcome at 2 years of age after general anaesthesia and awake-regional anaesthesia in infancy (GAS): an international multicentre, randomised controlled trial. Lancet. 2016;387(10015):239-250. doi: 10.1016/S0140-6736(15)00608-X [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.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(21):2312-2320. doi: 10.1001/jama.2016.6967 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Baughman FD, Thomas MSC, Anderson M, Reid C. Common mechanisms in intelligence and development: a study of ability profiles in mental age-matched primary school children. Intelligence. 2016;56:99-107. doi: 10.1016/j.intell.2016.01.010 [DOI] [Google Scholar]
- 38.Devlin B, Daniels M, Roeder K. The heritability of IQ. Nature. 1997;388(6641):468-471. doi: 10.1038/41319 [DOI] [PubMed] [Google Scholar]
- 39.Bouchard TJ Jr, McGue M. Familial studies of intelligence: a review. Science. 1981;212(4498):1055-1059. doi: 10.1126/science.7195071 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Trial protocol
eTable 1. Neuropsychological Testing Instruments
eTable 2. Characteristics of Patients Exposed to Sedation in Early Childhood According to Estimated IQ Score Comparison to Unexposed Siblings
eTable 3. Adjusted Analyses of Estimated IQ
eTable 4. Results of Neuropsychological Testing By Patient Age at PICU Admission
eTable 5. Results of Neuropsychological Testing By Patient Age at PICU Admission for Patients with Pediatric Cerebral Performance Category of 1 at Time of Post-Discharge Interview
eTable 6. Results of Neuropsychological Testing By Patient Sex
eTable 7. Results of Neuropsychological Testing By Hollingshead Four-Factor Index of Socioeconomic Status Score
eTable 8. Results of Neuropsychological Testing By Trial Treatment Group
Nonauthor collaborators