Authors: Sian Roberts-Walsh, Prasanth Sukumar, Valerie Twomey, Áine Carroll
Categories: Original Research, Anoxic Brain Injury, Brain injury, Cardiac arrest, Disability, Hypoxic Ischemic Encephalopathy, Outcome, Overdose, Rehabilitation, Retrospective Cohort Study
Source: Archives of Rehabilitation Research and Clinical Translation
Authors: Sian Roberts-Walsh, Prasanth Sukumar, Valerie Twomey, Áine Carroll
To investigate the etiology and outcome of patients with HIE admitted to a complex specialist rehabilitation facility over a 10-year period, to assess if numbers had changed, and to assess the effectiveness of rehabilitation for these patients. Hypoxic ischemic encephalopathy (HIE) is a leading cause of long-term neurologic disability.
An observational retrospective cohort study.
A National Rehabilitation Hospital.
All health care records with an ICD-9 code (348.1) or ICD-10 code (G93.1) for HIE from 2008 to 2017 totaling 104 participants.
Not applicable
Modified Barthel Index (MBI), Disability Rating Scale (DRS), Sensory Modality Assessment and Rehabilitation Technique scale, and discharge destination. Data were obtained from the patient administrative system of the brain injury program and were abstracted using a specially designed data abstraction tool. For statistical analysis, SPSS version 26.0 was used. Comparisons across etiologic groups were calculated using analysis of variance.
A total of 572 episodes were recorded under the code G93.1; 468 did not meet the inclusion criteria and 104 records were reviewed systematically using a standardized data extraction proforma. Sixty-nine (66%) were male and 35 (33%) were female. Cardiovascular causes of HIE were most common (35.6%), followed by overdose (22.1%). Most had moderate to severe disability on admission (MBI and DRS). Severe disability was associated with respiratory arrest, overdose, and neurologic causes, whereas independence was associated with cardiovascular causes. MBI and DRS improved in the majority with the greatest improvements seen with cardiovascular etiology. In this cohort study, 45.2% of patients were discharged home, 18.3% to a nursing home, and 18.3% to an acute hospital.
Improvement in functional outcomes after HIE was correlated to etiology. This may have implications in helping to predict patient outcomes post-HIE. Study limitations include incomplete recordings in charts and varying sample sizes within etiologic groupings.
The brain has one of the highest metabolic demands of all the organs in the body and as such, is very susceptible to a reduction in oxygen supply. Hypoxic ischemic encephalopathy (HIE) is a brain injury that results from an insufficient oxygen supply to the brain because of a lack of oxygen and/or blood flow to the neural tissues.^1^^,^^2^ The severity of injury is generally related to the duration of oxygen deprivation. Many events can lead to a lack of oxygen or disruption of cerebral blood flow, such as cardiac or respiratory arrest, strangulation, carbon monoxide poisoning, near-drowning, trauma, or anaphylaxis. The pathophysiology of HIE is unclear, but it would appear that a primary insult precipitates a cascade of cellular and subcellular responses that culminate in secondary brain injury and neuronal cell death.3, 4, 5, 6 Certain areas of the brain are more vulnerable to hypoxia including the gray matter and areas supplied by the distal branches of deep and superficial penetrating blood vessels including the upper brainstem, cerebellum, and subcortical structures.^7^ Improvements in trauma systems and prehospital emergency care have led to a significant increase in survival rates after anoxic events,^6^^,^^8^ and thus, the number of HIE cases is increasing. HIE accounts for between 10% and 15% of causes of nontraumatic brain injury in emergency departments and acute hospitals.^2^ It is a major cause of admission to rehabilitation services and a leading cause of long-term disability and death.^1^^,^^2^^,^^9^
There is a broad range of outcomes that can result from HIE ranging from complete resolution of symptoms to coma or death.^8^ Poor outcomes experienced by patients with HIE are well-documented and include cognitive, physical, psychosocial, and psychological problems.^1^^,^^2^^,^^8^^,^10, 11, 12 Access to rehabilitation has been shown to improve function;^13^ however, data are limited. Most publications on HIE relate to neonates with far fewer studies published on adults. Previous studies in this area have focused on comparing outcomes in patients with anoxic brain injury with those with traumatic brain injury or outcomes between cardiac and noncardiac causes of anoxic brain injury.^1^^,^^2^^,^^7^^,^14, 15, 16, 17 However, as far as we know, different causes of hypoxic brain injury have not so far been compared, and although some papers have studied rehabilitation outcomes, these have not examined how outcomes may relate to etiology.^11^^,^^18^ With the implementation of systems of care (trauma, cardiac, stroke), more people will be surviving with life-altering anoxic brain injuries, and it is crucial that there is a better understanding of the determining factors of optimal outcomes for these patients.
The purpose of this study was to investigate the number of admissions, etiology, and outcomes of patients with HIE admitted to a Complex Specialist Rehabilitation Centre over a 10-year period, to determine whether etiology affected functional outcomes and discharge destination, and to assess the effectiveness of rehabilitation for these patients. We aim to understand differences across etiologies without exploring causes for these differences in our analysis.
A retrospective cohort of all patients discharged from a complex specialist rehabilitation hospital over a 10-year period (2008-2017) was identified using the International Classification of Diseases (ICD) codes. The ICD is a globally used diagnostic tool for epidemiology, health management, and clinical purposes.^19^ The specific codes chosen, based on the recommendations from our clinical coders, were the ICD-10 code G93.1 (post-2015) and the equivalent ICD-9 code 348.1 (pre-2015) corresponding to all causes of anoxic/ischemic/hypoxic brain injury excluding neonatal hypoxia.
Data were obtained from the patient administrative system of the brain injury program and abstracted in accordance with the REporting of studies Conducted using Observational Routinely collected health Data (RECORD) guidance.^20^ Inclusion criteria included (1) a brain injury secondary to anoxia or hypoxia; (2) 18 years of age or older; and (3) first admission for a full program of rehabilitation. Exclusion criteria were as (1) any cause of brain injury not secondary to anoxia or hypoxia; (2) age under 18 years; (3) readmission with sequelae of the index injury; and (4) admission outside of time window.
Data were abstracted using a specially designed data abstraction tool (password-protected Excel spreadsheet) that included the following data fields to answer our research nonidentifiable descriptive data (age and sex), cause of HIE, level of disability (as measured by the Modified Barthel Index [MBI] and Disability Rating Scale [DRS]), the Sensory Modality Assessment and Rehabilitation Technique (SMART) assessment score, documented downtime if cardiac arrest, results of electroencephalogram (EEG) and computed tomography (CT)/magnetic resonance imaging (MRI), and discharge destination. The data abstraction was executed in accordance with best practice guidance for the performance of medical record abstraction.21, 22, 23, 24, 25, 26, 27 The data abstractor (lead author) was trained on the variables and the use of the data abstraction form. A pilot abstraction of 10 charts was completed by the data abstractor and verified by the coresearcher (coauthor). Any errors or points of clarification were discussed and resolved. The research team met on a weekly basis to check the abstraction accuracy and monitor progress. Convenience sampling was used with all cases of HIE selected over the 10-year period.
The University's Research Ethics Committee granted exemption from full ethical review because this was a retrospective review of health care records, so there was no need to gain informed consent.
The MBI is a measure of the activities of daily living, which indicates the level of a patient's independence.^28^ It covers 10 domains of functional activities such as feeding, grooming, or bathing. Each activity is given a score ranging from 0 (unable to perform a task) to a maximum of 5, 10, or 15 (fully independent) depending on the activity. The maximum total score is 100.
The DRS was developed as an outcome measure in brain injury to overcome the poor precision of the Glasgow Outcome Scale. It measures the level of disability in 4 categories (arousability; awareness and responsivity; cognitive ability for self-care activities; dependence on others; and psychosocial adaptability). Patients are scored from 0 to 30, where 0 is no disability and 30 is death. Although developed for traumatic brain injury, it has been used as an outcome measure for patients with nontraumatic causes of brain injury.^29^ These measures are routinely collected for all patients admitted to the brain injury program.
The SMART is a behavioral assessment scale of consciousness that assesses responses to multimodal sensory stimulation in disorders of consciousness.^30^ This outcome measure is used in the National Rehabilitation Hospital for the disorder of consciousness service that is a subset of the brain injury program.
Changes in the level of disability between admission and discharge, as evidence of functional improvements, were calculated and compared.
For statistical analysis, SPSS version 26.0 was used. Descriptive data were calculated for admission MBI, including mean, Standard deviation (SD), Standard error (SE), 95% confidence interval for mean, and minimum and maximum values. Comparisons across etiologic groups were calculated using analysis of variance (ANOVA). A chi-square test of independence was conducted to examine the relationship between sex and etiologies. A Pearson chi-square test was used to look at the association between etiology and discharge destination. A significant difference was considered as P<.05.
A total of 572 episodes were recorded under the codes 348.1 and G93.1. After the exclusion of duplicate entries (readmissions), the number of records analyzed was 139. A further 35 were excluded because they did not meet the inclusion criteria (misclassified or outside the timeframe); 104 records were reviewed systematically. The number of admissions per year is summarized in figure 1, with a least squares fit trend line.Fig 1Number of admissions with HIE per year with trendline.Fig 1
The characteristics of patients discharged with a diagnosis of HIE are summarized in table 1.Table 1Characteristics of patients, causes, and frequency of HIETable 1(n=104)CharacteristicsNumberPercentAge, y 18-252625.0 26-586461.5 >591413.5Sex Male6966.3 Female3533.7Cause of HIENumberPercentCardiovascular causes3735.6Deliberate self-harm54.8Metabolic43.8Neurologic76.7Overdose2322.1Respiratory arrest21.9Trauma109.6Other1514.4Missing11Total104100Abbreviation: HIE, hypoxic ischemic encephalopathy.
Among the patients discharged with HIE, 69 (66.3%) were male and 35 (33.7%) were female, with 26 (25%) under age 25 years, 64 (61.5%) aged 26-58 years and 14 (13.1%) aged over 59 years.
Table 1 summarizes the causes of HIE. Cardiovascular causes accounted for 35.6% of cases, overdose for 22.1%, trauma for 9.6%, neurologic causes for 6.7%, deliberate self-harm for 4.8%, metabolic for 3.8%, and respiratory arrest for 1.9%. Other causes accounted for 14.4% of etiologies. Etiologies included in the “other” category included 1 case of anaphylaxis during surgery, 4 other cases related to surgery, 1 case of sepsis, 1 case due to an arteriovenous malformation, 1 case secondary to clostridium enterocolitis, 2 cases related to drowning, and 4 to alcohol use or withdrawal, 2 of which also had seizures. All metabolic causes involved hypoglycemia. Neurologic causes included seizures (n=5), Guillain–Barré syndrome (n=1), and limbic encephalitis (n=1). All deliberate self-harm cases involved attempted suicide by hanging. Trauma etiologies included 2 cases secondary to falls, 4 related to road traffic accidents, 2 due to stabbing, 1 after an assault, and 1 after a gunshot injury.
The overdose category (n=23) included single and multiple drug overdose of legal and illegal substances, including 1 case involving ecstasy, 1 involving cannabis, 1 metformin with opiate analgesia, 2 involving heroin, 2 involving methadone, 2 involving antidepressants, 4 involving cocaine, and 5 involving sedative–hypnotics. All occurred out of the hospital. Ten of the 22 charts did not specify the drug the patient overdosed on. Two cases reported drug overdose in combination with alcohol use. Whether the overdose was deliberate or accidental was not uniformly reported nor was the administration of naloxone.
There was a specific mention of out-of-hospital cardiac arrest in 34 cases. Downtime (the time from the recognition of cardiac arrest to the time of sustained return of spontaneous circulation) was recorded in 19 cases of cardiac arrest. Mean downtime was 18 minutes (range 4-90mins).
Cardiovascular etiologies were 73% male, overdose was 61% male, trauma was 90% male, deliberate self-harm was 80% male, and neurologic causes were 57% male. Metabolic etiologies occurred 100% in females, although there were only 4 cases. Cardiovascular etiologies (mean age 48.3y) occurred in an older age group than other etiologies (overdose: 29.7y; 40.1y; 29y; deliberate self-harm: 30.6y; 41.25y). A chi-square test of independence was conducted to examine the relation between sex and etiologies. The relation between these variables was not found significant, χ^2^ (7, N=103) =12.45, P=.087.
Of 104 charts reviewed, 97 had MBI recorded, and 75 had DRS recorded on admission. Most had moderate to severe disability on admission as measured by the MBI and DRS. Severe disability was associated with respiratory arrest, overdose, and neurologic causes, whereas independence was most likely with cardiovascular causes. We compared the admission MBI across the different diagnostic categories because this was the most complete data set. The results are shown in table 2. These differences were statistically significant (ANOVA: P=.008).Table 2Admission MBITable 2NMeanSDSE95% Confidence Interval for MeanMinimumMaximumLower BoundUpper BoundCardiovascular causes3672.737.06.260.285.30100Deliberate self-harm554.235.415.810.398.1092Neurologic622.830.612.5−9.355.0076Overdose2147.536.07.731.163.90100Trauma950.439.813.319.981.00100Other2036.433.77.520.652.10105Total9753.738.73.945.761.40105Abbreviation: MBI, Modified Barthel Index; SD, standard deviation; SE, standard error.
Most of the patients improved in all categories after inpatient rehabilitation (66.2%). MBI improved in 60% of cardiovascular cases, 60% of deliberate self-harm cases, 75% of neurologic cases, 76.5% of overdose cases, 50% of trauma cases, and 55.6% of other cases. These changes are summarized in figure 2.Fig 2Change in MBI from admission to discharge.Fig 2
Table 3 shows the mean MBI score at admission and discharge for each injury type; 97 MBI scores were recorded on admission and 82 on discharge. Although the average MBI score improved for each injury category, a repeated measure ANOVA revealed that there were no statistically significant differences between MBI scores at admission and discharge for different injury categories (P=.15).Table 3MBI score at admission versus dischargeTable 3Injury TypeMBI AdmissionMBI DischargeNMean ScoreNMean ScoreCardiovascular causes3672.73080.5Deliberate self-harm554.2563.6Neurologic622.8443.3Overdose2147.51765.8Trauma950.4852.1Others2036.41842.9Total9753.78263.6Abbreviation: MBI, Modified Barthel Index.
Seventy-seven records did not have a DRS recorded on discharge; therefore, we did not compare admission and discharge DRS.
A SMART score was recorded on 11 out of 29 records, which could have had a SMART score recorded (based on the MBI scores of 0-20 to indicate “total” dependency and a proxy for the need for a SMART). Of these, 6 had a SMART score that indicated a minimally conscious state and 5 had a SMART score indicative of a persistent vegetative state. Of those who had a SMART score recorded, 7 had cardiovascular and respiratory causes, 3 other causes, and 1 had a metabolic cause. There was no change in the level of consciousness for these cases.
Documentary evidence of an EEG was available for 20 cases. These were all reported as abnormal with diffuse slowing consistent with a hypoxic brain injury.
Ten records had no neuroimaging recorded in the health care record.
The CT scan reports were recorded in 53 cases, and 1 record stated that a CT scan was done, but there was no report in the health care record. Eight CT scans were reported as normal. One reported a right parietal parenchymal hemorrhage, 1 an acute on chronic subdural hematoma, 1 a right temporal and left frontal infarct, and 2 recorded poor gray/white matter differentiation.
MRI reports were available for 42 patients, of which 3 were reported as normal and 1 inconclusive. The majority reported diffuse ischemic changes in keeping with a HIE. The main areas affected were the basal ganglia and deep white matter.
Both CT and MRI were recorded in 17 cases. Of these, in 6 cases, both CT and MRI were reported as normal. Three had the CT reported as normal and subsequent MRI was reported as showing diffuse hypoxic changes.
Of the 104 charts reviewed, 100 had a discharge destination record. Discharge destinations are summarized in table 4. Forty-six (45.5 %) patients were discharged to their own homes, 19 (18.6%) to a nursing home, and 19 (18.6%) to an acute hospital. Discharge destination was not recorded in 4 cases. There was no significant association between etiology and discharge destination (Pearson χ^2^=0.406).Table 4Discharge destinationTable 4(n=100)Discharge destinationNumberPercentHome4645.5Acute hospital1918.6Nursing home1918.6Acute hospital awaiting nursing home87.8Long-term care32.9Step-down facility32.9Community hospital11Acute hospital awaiting return to parents’ home11Parent's home11Total100100
HIE is an increasing cause of admission to rehabilitation facilities in most jurisdictions.^31^ In this study, admissions with HIE decreased over time. It is outside the scope of this study to conjecture why this might be the case, but it does raise the importance of data across the continuum of care. Referral guidelines did not change over this period of time, and as the only complex specialist rehabilitation hospital in the country, no cases would have gone to other hospitals. Cases of HIE constituted 29.8% of admissions to the hospital over the 10-year period. With developments in trauma systems of care and basic and advanced life support, the number of admission will likely increase.^31^^,^^32^ In this study, most discharges were male aged <65 years (65.4%). Cardiovascular causes were the most common cause of HIE (35.6%), followed by overdose (22.1%). Most cases had moderate to severe disability on admission (as measured by MBI and DRS). Severe disability was associated with respiratory arrest, overdose, and neurologic causes, whereas independence was most likely with cardiovascular causes. There was a statistically significant improvement in the level of disability in the majority (49%) with the greatest improvements seen with cardiovascular etiology. Previous cohort studies have not looked at the effect of etiology on rehabilitation outcomes. An unexpected finding was the prevalence of overdose as a major etiology of HIE in our cohort, which is higher than other reported cohort studies. This may have implications for the prehospital management of cases because the availability of opioid antagonists may have reversed respiratory depression. Only 34 cases had specific mention of out-of-hospital cardiac arrest with downtime being recorded in only 19 cases. Mean downtime was 18 minutes (range 4-90min), and although it would be expected that prolonged downtime would be associated with a more adverse outcome, this was not the case in all cases. However, the number of cases with documentation was low. The need for more detailed clinical information on referral to the rehabilitation facility is clear.
According to our study, age and sex differences in etiologies exist with males being more represented than females in all causes except metabolic (4 cases were metabolic). Cardiovascular causes occurred more frequently in older age groups with overdose occurring in younger age groups. Future work is needed to understand the causes of differences across etiologies, which was beyond the scope of this study.
Severe disability was associated with respiratory arrest, overdose, and neurologic causes, whereas independence was most likely with cardiovascular causes.
The EEG and neuroimaging data did not show clear differences across the diagnostic categories. However, only 20 records had an EEG report. Whether the other cases had an EEG performed was not apparent and again raised the question of information sharing on transfer of care. Fifty-three cases had a CT reported and 42 had an MRI report recorded. This is despite MR Spectroscopy being superior to conventional neuroimaging in HIE and correlated with prognosis.^33^ Although the majority reported diffuse ischemic changes in keeping with a HIE with the main areas affecting the basal ganglia and deep white matter, this number is insufficient to correlate with diagnosis. It would be interesting for future research to review the neuroimaging rather than reports, for a more detailed assessment of the cerebral changes, which will be easier with developments in the Irish National Integrated Medical Imaging System. It is possible that some imaging was done in acute hospitals where the patients initially presented and that we did not have access to imaging or reports.
Forty-six (45.5 %) patients were discharged to their own homes, 19 (18.6%) to a nursing home, and 19 (18.6%) to an acute hospital. This is a most unsatisfactory outcome for patients due to the lack of funding for care packages or home adaptations. It would be interesting to assess if the functional gains achieved were maintained after transfer.
This is a cohort study from one specialist rehabilitation center; therefore, how generalizable these findings are is questionable. However, the hospital is the only specialist rehabilitation center in the country and, therefore, receives the more severe cases from around the country. The retrospective analysis of case records involves the subjective interpretation of health care records made by others; therefore, data extraction may misinterpret entries. However, the data fields selected were clear and did not require a lot of interpretation, and the review was carried out in accordance with recommended best practices for the performance of a retrospective chart review.
HIE is a relatively common cause of admission to specialist rehabilitation services. Specialist rehabilitation results in statistically significant improvement in disability with many cases being discharged to home. However, the percent discharged home is less than that for other causes of acquired brain injury (80%). This may be due to a more severe disability. Our results show that more comprehensive data shared across the continuum of care would be helpful including access to full investigations and results. There is no brain injury database, which makes research on the long-term outcomes for these patients difficult. Although national audits exist for critical care, stroke, and trauma, there is a need to include rehabilitation data fields and for long-term data to be collected. This would help our understanding of the pathophysiology, response to treatment, and long-term outcomes. There is also a pressing need to invest in appropriate specialist rehabilitation services for these patients. The hyperacute management has been well developed in recent years; there is now a requirement for a parallel investment in rehabilitation services. As advocacy groups “Don't save me then leave me.”^34^
The authors declare they have no known conflicts of interest.