Systematic Review and Dosage Analysis: Hyperbaric Oxygen Therapy Efficacy in Mild Traumatic Brain Injury Persistent Postconcussion Syndrome

Background Mild traumatic brain injury results in over 15% of patients progressing to Persistent Postconcussion Syndrome, a condition with significant consequences and limited treatment options. Hyperbaric oxygen therapy has been applied to Persistent Postconcussion Syndrome with conflicting results based on its historical understanding/definition as a disease-specific therapy. This is a systematic review of the evidence for hyperbaric oxygen therapy (HBOT) in Persistent Postconcussion Syndrome using a dose-analysis that is based on the scientific definition of hyperbaric oxygen therapy as a dual-component drug composed of increased barometric pressure and hyperoxia. Methods In this review, PubMed, CINAHL, and the Cochrane Systematic Review Database were searched from August 8–22, 2021 for all adult clinical studies published in English on hyperbaric oxygen therapy in mild traumatic brain injury Persistent Postconcussion Syndrome (symptoms present at least 3 months). Randomized trials and studies with symptomatic and/or cognitive outcomes were selected for final analysis. Randomized trials included those with no-treatment control groups or control groups defined by either the historical or scientific definition. Studies were analyzed according to the dose of oxygen and barometric pressure and classified as Levels 1–5 based on significant immediate post-treatment symptoms or cognitive outcomes compared to control groups. Levels of evidence classifications were made according to the Centre for Evidence-Based Medicine and a practice recommendation according to the American Society of Plastic Surgeons. Methodologic quality and bias were assessed according to the PEDro Scale. Results Eleven studies were included: six randomized trials, one case-controlled study, one case series, and three case reports. Whether analyzed by oxygen, pressure, or composite oxygen and pressure dose of hyperbaric therapy statistically significant symptomatic and cognitive improvements or cognitive improvements alone were achieved for patients treated with 40 HBOTS at 1.5 atmospheres absolute (ATA) (four randomized trials). Symptoms were also improved with 30 treatments at 1.3 ATA air (one study), positive and negative results were obtained at 1.2 ATA air (one positive and one negative study), and negative results in one study at 2.4 ATA oxygen. All studies involved <75 subjects/study. Minimal bias was present in four randomized trials and greater bias in 2. Conclusion In multiple randomized and randomized controlled studies HBOT at 1.5 ATA oxygen demonstrated statistically significant symptomatic and cognitive or cognitive improvements alone in patients with mild traumatic brain injury Persistent Postconcussion Syndrome. Positive and negative results occurred at lower and higher doses of oxygen and pressure. Increased pressure within a narrow range appears to be the more important effect than increased oxygen which is effective over a broad range. Improvements were greater when patients had comorbid Post Traumatic Stress Disorder. Despite small sample sizes, the 1.5 ATA HBOT studies meet the Centre for Evidence-Based Medicine Level 1 criteria and an American Society of Plastic Surgeons Class A Recommendation for HBOT treatment of mild traumatic brain injury persistent postconcussion syndrome.

Background: Mild traumatic brain injury results in over 15% of patients progressing to Persistent Postconcussion Syndrome, a condition with significant consequences and limited treatment options. Hyperbaric oxygen therapy has been applied to Persistent Postconcussion Syndrome with conflicting results based on its historical understanding/definition as a disease-specific therapy. This is a systematic review of the evidence for hyperbaric oxygen therapy (HBOT) in Persistent Postconcussion Syndrome using a dose-analysis that is based on the scientific definition of hyperbaric oxygen therapy as a dual-component drug composed of increased barometric pressure and hyperoxia.
Methods: In this review, PubMed, CINAHL, and the Cochrane Systematic Review Database were searched from August 8-22, 2021 for all adult clinical studies published in English on hyperbaric oxygen therapy in mild traumatic brain injury Persistent Postconcussion Syndrome (symptoms present at least 3 months). Randomized trials and studies with symptomatic and/or cognitive outcomes were selected for final analysis. Randomized trials included those with no-treatment control groups or control groups defined by either the historical or scientific definition. Studies were analyzed according to the dose of oxygen and barometric pressure and classified as Levels 1-5 based on significant immediate post-treatment symptoms or cognitive outcomes compared to control groups. Levels of evidence classifications were made according to the Centre for Evidence-Based Medicine and a practice recommendation according to the American Society of Plastic Surgeons. Methodologic quality and bias were assessed according to the PEDro Scale.
Results: Eleven studies were included: six randomized trials, one case-controlled study, one case series, and three case reports. Whether analyzed by oxygen, pressure, or composite oxygen and pressure dose of hyperbaric therapy statistically significant symptomatic and cognitive improvements or cognitive improvements alone were achieved for patients treated with 40 HBOTS at 1.5 atmospheres absolute (ATA) (four randomized trials). Symptoms were also improved with 30 treatments at 1.3

INTRODUCTION
This is a scientific review of the evidence for hyperbaric oxygen therapy treatment of non-penetrating mild traumatic brain injury (mTBI) (1) Persistent PostConcussion Syndrome (PPCS) (2). The conclusions and recommendations are based on the Centre for Evidence-Based Medicine (CEBM) Levels of Evidence (3) and the American Society of Plastic Surgeons Grade Practice Recommendation guidelines for clinical treatment (4).
Hyperbaric oxygen therapy (HBOT) has been applied to chronic TBI wounds in animals and humans since 1989 (46, with apparently conflicting results for mTBI (46,(99)(100)(101)(102)(103)(104)(105). Various researchers have attributed the different results to mischaracterized sham groups/the effects of different doses of HBOT (46,80,(99)(100)(101)(102)(103)(104)(105), design differences (106), (small sample size, dissimilar outcome measures/populations/sites/protocol adherence, non-equivalence of groups, selection bias) (73,81), ritual experience (75), and placebo/Hawthorne effects (107). The conflict stems from control group selection based on the historical definition of HBOT (55) where the bioactivity of barometric pressure and <100% oxygen were dismissed. This systematic review of the science and literature on HBOT in mTBI PPCS analyzes the studies/data based on the scientific definition of HBOT (58) and the U.S. FDA's classification of HBOT as a dual-component drug (57). Recognizing that the HBOT studies consist of different doses of increased pressure and hyperoxia, the evidence is ranked according to the Evidence-Based Medicine hierarchy of evidence for treatment in clinical studies (3). The

Classification of Evidence/Analysis of Studies
Studies were classified as Level 1-5 according to CEBM guidelines (3). Classifications were based on significant immediate post-treatment symptom and/or cognitive outcomes in treatment groups vs. no-treatment control groups or lower doses of HBOT and compared according to the individual pressure, total oxygen, combined hyperbaric pressure and total oxygen doses of HBOT. The pressure was measured in atmospheres absolute (ATA) and oxygen in atmosphere-minutes (AMs). AM oxygen dose was the total cumulative oxygen dose over the entire treatment course per group for oxygen in excess of room air. It was the sum of the product of the atmospheric pressure times the FiO 2 times the number of minutes for each phase of every hyperbaric treatment (compression, at depth, and decompression phase), multiplied times the total number of treatments. Constant compression and decompression rates were assumed. The average pressure from the surface to depth (compression) and depth to the surface (decompression) was multiplied by the FiO 2 of the breathing gas for these phases. A linear increase in FiO 2 and ∼90% FiO 2 was assumed by 8 min of compression (98) for protocols in monoplace (single-person) chambers that compressed with 100% oxygen. Practice Grade Recommendation was determined from the levels of evidence, according to the American Society of Plastic Surgeons grading system in Burns et al. (4).

Methodologic Quality and Risk of Bias Assessment
The physiotherapy evidence database (PEDro) scale (109) was used to assess the methodologic quality/risk of bias of the randomized trials included in the final analysis. Randomized trials are given a cumulative score of 0-10 based on individual scoring of 10/11 items (the first, eligibility criteria, is omitted from the score because it is an external validity item), using 1 point for "present, " and 0 points for "absent." The 10 items beginning with #2 are: (2) random allocation, (3) concealed allocation, (4) groups similar at baseline, (5) subject blinding, (6) therapist blinding, (7) assessor blinding, (8) 1 key outcome for > 85% of subjects, (9) 1 key outcome: intention-totreat analysis, (10) 1 key outcome between-group statistical comparison, and (11) 1 key outcome point measurements and variability. Scoring was performed by the single author based on stated scoring items in the text of each article. Additional reporting bias and conflict of interest were noted for investigators who were employees of the funding source. All items were scored as 0 if they were not mentioned. Attempts were made to contact the authors of the studies to resolve the scoring on allocation concealment. Quality of studies ("poor, fair, good, excellent") was judged according to the scoring legend of Cashin et al. (110).

RESULTS
In total, 681 articles were searched, 11 studies were included in the final analysis (Figure 1). The largest exclusion (57%) was for     duplicate articles and the second-largest exclusion was by title (38%). An example of the third search was the Ma et al. study (111) on HBOT for firefighters with lifetime TBI and chronic emotional distress. The study was excluded due to the primary outcome of regional cerebral blood flow and the absence of symptomatic or cognitive outcomes. The 11 studies that met the criteria for inclusion are listed in Table 1 (along with companion articles on non-primary outcomes). The 11 studies are: three case reports (66,67,89), a prospective case-controlled series reported in two publications (68,80), a prospective case series (84), and six randomized trials (46, 69-71, 75, 81) reported in 19 publications (46, 69-79, 81-83, 85-88). Three of the randomized trials contained control groups (46,70,75), two with a crossover design (46,70), and the other three were comparative dosing studies (69,71,81). Four of the randomized trials were performed by the U.S. Department of Defense (DoD) (69,71,75,81) (85). Analysis was conducted on the three case reports (66,67,89), the case-controlled series (80), the case series (84), and six randomized trials (46, 69-71, 75, 81).
The In the prospective case-controlled series, Harch et al. (68,80) reported the first 16 (68) and then all 30 (80) military subjects with mTBI/PPCS and PTSD. The first 16 subjects (68) which were not case-controlled experienced significant improvement in selfreported symptoms, PPCS and PTSD symptom questionnaires (RPQ total, post-treatment minus pre-treatment: −15.6 ± 12.8, CI: −22.7 to −8.5, p = 0.0002), abnormal neurological exam, 7/13 neuropsychologist-administered cognitive tests, two neuropsychologist-administered affective instruments, and two quality of life measures after 40 HBOTs, and quantitative texture and statistical parametric mapping analysis of SPECT brain blood flow imaging after the first and 40th HBOT at 1.5 ATA oxygen (3,335 AMs oxygen). Statistical parametric mapping analysis revealed significant increases in brain blood flow in 85 regions of the brain, almost exclusively white matter, and in the hippocampi, that were consistent with the statistically and clinically significant increases in neuropsychologically measured working and delayed memory. The SPECT findings were different from and inconsistent with placebo SPECT brain blood flow studies. This non-controlled portion of the casecontrolled study concluded that the significant symptomatic, cognitive, and SPECT improvements could not be explained by placebo effects.
The study design was changed at the midpoint of Harch et al. (68) to include a matched group of imaging controls for all 30 subjects and the full cohort of 30 subjects, 27 with comorbid PTSD, was reported as a case-controlled study in Harch et al. (80). Subjects had sustained 3.5 previous TBIs and were treated with 40 HBOTs at 1.5 ATA oxygen (3,420 AMs oxygen) 3.35 years post index TBI. Their PTSD symptom scores were 32% higher (63.4)  to −9.6, p < 0.001 (actual p value was less, but the cutoff of 0.001 was used for this publication)], a significant reduction in suicidal ideation, and reduction in psychoactive medication use. SPECT brain imaging was abnormal in subjects compared to a matched cohort of population normals and normalized after HBOT in 75% of abnormal regions.
The prospective case series, Mozayeni et al. (84), reported outcomes on 32 military and civilian subjects treated 9.5 years after mTBI with 40-82 treatments (average 55.8) at 1.5 ATA oxygen (4,326 AMs oxygen). Mood changes and two computerized neurocognitive test batteries (ANAM4 and CNS Vital Signs) were administered; TBI symptom questionnaires were not used. Significant improvement was demonstrated on 8/8 Automated Neuropsychological Assessment Metrics (ANAM) mood, 6/7 ANAM neurocognitive, and 7/10 CNS Vital Signs neurocognitive measures. Improved outcomes were associated with shorter delays to treatment, younger age, military status, and increased numbers of HBOTs. In summary, both the prospective case-controlled and case series demonstrated multi-domain statistically significant outcomes/benefits of HBOT at a dose of 1.5 ATA oxygen with one study (80) showing simultaneous and correlative improvements in brain blood flow deficits compared to a group of untreated population normals.
The first of the comparative dosing studies was conducted by Wolf et al. (69) and randomized 50 military subjects 3-71 months after an average of 3.4 mTBIs to 30 1.3 ATA air or 2.4 ATA oxygen HBOTs (1,002 and 6,900 AMs oxygen, respectively).
The authors found 6-week post-treatment symptom reductions (improvements) of 13% in the HBOT group and 41% in the 1.3 ATA air group with 10 subtests improved on the Immediate Post-Concussion Assessment and Cognitive Testing (ImPACT) symptom testing for the 1.3 ATA air group vs. two subsets in the 2.4 ATA oxygen group, but no within or between-group change score statistics were performed on this data. For the combined ImPACT symptom and cognitive total score 6-weeks post-treatment there were statistically significant reductions (improvements) in both groups [12% in the HBO 2 group and 31% in the air group: control group (t = 3.76, p = 0.001) and the HBO 2 group (t = 3.9, p = 0.001), but no statistical analysis was performed on the between-group improvements. Similarly, there were statistically significant within-group improvements in 6-weeks post-treatment PTSD symptoms, but no change score comparison between the two groups. Rather, it was stated that "Difference testing between the sham-control and HBO 2 groups did not reveal any significant differences on the PCL-M composite mean score (t = −0.205, p = 0.84) or on the ImPACT total mean score (t = −0.943, p = 0.35) at any time (Figures 1, 2), including at 6 weeks post-exposure (Table 1)." In a subset analysis of this study Scorza et al. (113) reported that the mTBI/PPCS subjects without PTSD demonstrated a "trend toward harm" (worsening of scores) on the ImPACT symptom score in the 2.4 ATA oxygen group. This indicates a larger positive effect on 2.4 ATA subjects with comorbid PTSD and a negative effect on PPCS (the target of the study) without PTSD. Amplification of HBOTinduced PPCS benefit by comorbid PTSD was found in multiple studies (68,80,81,87). The negative finding of Scorza (113) and corresponding 2.4 ATA oxygen group composite (symptoms and cognition) ImPACT score trajectory [ Figure 2 in Wolf et al. (69) and Figure 4 in Harch et al. (46)] have been argued as a toxic/overdose effect of oxygen in the 2.4 ATA oxygen group (46,80,99).
The resulting cognitive outcomes in the study of Wolf et al. (69) were published in Wolf et al. (76) and showed significant within-group improvement for both groups, but no statistically significant between-group improvements for cognitive measures as well as PTSD symptoms. Curiously, there are no data presented in the paper. Rather, a derived metric, the "relative risk of improvement" (RROI) on subset historical features (e.g., 3 concussions, single event <2 years old, <1-year post-injury, blast injury, etc.) was reported. The article concluded that "Subgroup analysis of cognitive changes and Post-Traumatic CheckList Military results regarding PTSD demonstrated a relative risk of improvement using 2.4 atm abs hyperbaric oxygen." While true for highly specific subsets of the 2.4 ATA oxygen group this conclusion is confusing and misleading, implying that HBOT at 2.4 is efficacious for PPCS. The "data" tables actually show that 60% of the RROI entries favored the 1.3 ATA air group and only 40% favored the 2.4 ATA oxygen group, consistent with the symptom data. The companion article (79), reporting cognitive improvements and stem cell mobilization, also used derived correlational data and is similarly misleading. A positive correlation was claimed between stem cell mobilization and cognitive improvements in the 2.4 ATA oxygen group, implying that stem cell mobilization was responsible for cognitive improvements in subsets of this group. This claim and implication are inconsistent with the symptom data/analysis of Scorza et al. (113) and the 60/40% air group subset cognitive dominance, and lack evidence of stem cell migration/implantation in the brain. Wolf et al. (76) acknowledge the finding of Scorza et al. (113) of significant improvement of subjects with co-morbid PTSD but omit the major finding that the 2.4 ATA oxygen dose demonstrated a "trend toward harm" in PPCS without PTSD (113). At the same time, it is known that stem cells are mobilized into the circulation of humans at 2.0 and 2.5 ATA oxygen (114) in the absence of PPCS and PTSD and that stem cells are stimulated to proliferate and differentiate at sites of injury in the brain due to HBOT (115)(116)(117), but there is no evidence of deposition of circulating stem cells in brain tissue in the Wolf et al. study (69,76,79). More likely, the circulating stem cells had nothing to do with the improvement in cognition in the PPCS PTSD group since cognitive and symptom improvements were favored in the air pressurization group that had no evidence of stem cell mobilization. In summary, these cognitive outcome reports (76,79) do not report actual data, appear to be misleading, and the conclusions of the study, cognitive improvements in the 2.4 ATA oxygen group vs. the air group, are internally inconsistent and inconsistent with their own analysis (113) and the PPCS symptom data in Wolf et al. (69).
The second comparative dosing study (71) was reported in four publications (71)(72)(73)(74) Wolf et al. (69), no between-group change scores were statistically analyzed. Rather, one-way ANOVAs were performed on the pre-treatment mean RPQs for the three groups and then on the post-treatment mean RPQs. Identical analyses were performed on the PCL-M for PTSD symptoms; the 2 ATA oxygen group experienced the only significant finding in the study, a significant decrease in PTSD symptoms. The second publication (72) reported cognitive and psychomotor outcomes 1-week posttreatment and found "no immediate postintervention beneficial effect of exposure to either 1.5 ATA or 2.0 ATA oxygen compared with the Sham Air intervention." Scrutiny of the data suggests varying effects of the different doses of HBOT on the 55 outcomes, especially for four cognitive tests (California Verbal Learning Test: Long-Delay Free Recall, Index for Recognition, Short Delay Cue Recall, and Recognition Total Hits), but large standard deviations of the mean scores and separate one-way ANOVAs of the baseline and post-treatment outcomes make exact comparisons between the groups difficult. Identical to the first publication on this study no change score analysis was performed on any outcome.
The third publication (73) reported 1-week and 3-month symptom, functional, cognitive, and psychomotor outcomes and found "No significant time by intervention interaction"  (72), strict pre/post-change scores (treatment effects) for each group were not calculated, and similar to both the first (71) and second (72) publications the pre/post means of these differences were not compared between groups. Regardless, the treatment effects for each individual HBOT does/group appear small and insignificant. The net result is a suggestion of small differing effects of the three different doses of HBOT with statistically significant change only seen for reduction of PTSD symptoms in the 2 ATA oxygen group and some cognitive domains in the combined three groups.
The third comparative dosing study (81) randomized 71 military subjects 25.6 months after an average of 3.6 TBIs to receive either forty 1.2 ATA air or 1.5 ATA oxygen treatments (600 and 3120 AMs oxygen) in a 12-week period. Symptom, quality of life, neuropsychological, neurological, electroencephalography, sleep, auditory, vestibular, autonomic, visual, neuroimaging, and laboratory testing were performed before and immediately, 3 and 9 months post-treatment. Baseline characteristics suggested "worse brain injury" in the HBOT group due to significantly more frequent diffuse/traumatic axonal injury on MRIs, greater TBI symptoms (RPQ-13 and RPQtotal), PTSD symptoms (PCL-C hyperarousal score), number of combat deployments, and worse anger control. Despite this difference in injury, the compressed air group showed a deterioration in postconcussive (NSI and RPQ) and PTSD symptoms (PCL-C) during the treatment period while the HBOT group experienced significant improvement on the NSI, RPQ-3, and PCL-C compared to the compressed air group (difference in mean change score between groups: NSI: −7.6, CI = −14.4 to −0.7, p = 0.03; RPQ-3: −1.5, CI = −2.7 to −0.3, p = 0.01; RPQ-13: −5, CI = −10.7 to 0.6, p = 0.08; PCL-C: −7.3, CI = −13.5 to −1, p = 0.02), statistically moreso on the NSI and PCL-C for those subjects with PTSD. In addition, the HBOT group experienced significant improvement in some cognitive processing speed and sleep measures, and in those with PTSD improved functional balance and reduced vestibular complaints, all compared to the pressurized air treatment.
Companion publications (82,83,86,88) presented components of Weaver et al. (81) in greater detail. Walker et al. (82) reported markedly disrupted sleep quality in study subjects and significant improvement in self-reports of Pittsburgh Sleep Quality Index measures (five out of eight at 13-weeks and two out of eight at 6 months) in the oxygen group compared to the air pressure group. Meehan et al. (83) found worse balance/gait measures and more affective symptoms in mTBI subjects compared to healthy controls, some within-group improvements in these domains favoring the HBOT group at 13 weeks and 6 months post-treatment, significant (but "minimal") improvements in the HBOT group compared to the air group on balance measures, and improvements on postural control, generally favoring HBO 2 . Those with affective symptoms, particularly PTSD, who were treated with HBOT had the most improvement in postural control and otolith function. Hart et al. (86) was an extended outcome study and not pertinent to this review. Lastly, Wetzel et al. (88) demonstrated significant abnormalities in ocular metrics in the overall subject population compared to population normals. Multiple abnormalities improved in both treatment arms (pressurized air and HBOT) that persisted by 3 months post-treatment at which time the normals and study subjects were statistically indistinguishable.
In summary, the comparative dosing studies (69, 71, 81) demonstrated significant improvements in symptoms of mTBI/PPCS at 1.5 ATA oxygen and 1.3 ATA air, deterioration at 1.2 ATA air and 2.4 ATA oxygen, and no significant change at 2 ATA of pressure with 0.21, 1.5, or 2.0 ATA oxygen. In addition, significant improvements in cognition, balance, and gait measures were obtained with the 1.5 ATA oxygen dose.
The first randomized controlled trial (RCT) (70) randomized 67 civilians 33 months after a single mTBI to 40 HBOTs at 1.5 ATA oxygen or no treatment control (3,720 vs. 0 AMs oxygen) in 2 months. Mindstreams computerized testing (organized into four cognitive domains), single-photon emission computed tomography (SPECT) brain blood flow imaging, and a quality-of-life evaluation were performed pre-, within 1-3 weeks after control and treatment periods for the control crossover group, and after treatment for the HBOT group. Significant improvements were demonstrated in cognitive function and quality of life in both groups following HBOT but no significant improvement was observed following the control period. SPECT imaging revealed elevated brain activity in good agreement with the cognitive improvements. Cognitive function improvements in the HBOT group consisted of Information Processing Speed [t (31) (71)(72)(73), the outcome instrument change scores after the control period and HBOT were not statistically compared. Rather, the groups were compared at baseline and after both groups had completed HBOT. The groups were statistically indistinguishable at both comparisons and both groups had experienced significant within-in group improvements after HBOT on all instruments with medium to large effect sizes.
The second RCT (75) randomized 72 military subjects 23 months after an average of 3.1 mTBIs to one of three groups: 40 HBOTs at 1.5 ATA oxygen, 1.2 ATA air, or no HBOT (3,120, 600, or 0 AMs oxygen) during 10 weeks. Symptom questionnaires, computerized neuropsychological testing, and traditional neuropsychological testing were administered pre-, at the midpoint of treatment and after treatment for symptom questionnaires and the computerized cognitive testing, and at two time points (pre/post full treatment) for the traditional cognitive tests. There were no significant two-point change score differences between the three groups on the abbreviated RPQ (p = 0.24), yet a seemingly large difference between the HBOT and no-treatment group (25% of the no-treatment group, 52% of the HBOT group, and 33% of the air-pressure group); it does not appear that either treatment group was compared individually to the no-treatment group. For the total RPQ score both chamber groups demonstrated significant improvement compared to the no-treatment control group (5.4, 95% CI, −0.5 to 11.3, P = 0.008 HBO group; 7, 95% CI, 1.0 to 12.9, P = 0.02 air group) and no difference between the two-chamber groups (P = 0.70). Significant improvements were achieved despite many subjects not receiving the full 40 treatments (48% of the HBOT group and 59% of the "sham" group received 40 HBOTs), while symptom improvement was greater for those who completed all 40 chamber sessions. For the secondary symptom outcome score, Neurobehavioral Symptom Inventory (NSI), the no-treatment group showed a slight worsening, while both treatment groups showed improvement [Pre minus post scores: −1.1 (CI: −7.3 to 5.2, no treatment) vs. 3.7 (CI: −3.7 to 11.2, HBOT) and 6.9 (CI: 1.4 to 12.4, "sham" air). Positive scores are improvement]. Similar to the RPQ two-point change metric, it appears that neither treatment group was statistically compared to the no-treatment group. Rather, it was stated that "these change scores were not statistically different." For PTSD symptoms, depression, generalized anxiety, pain, sleep, quality of life, and computerized cognitive testing improvements occurred in both groups, generally favoring the air treatment, but there was no statistical difference between treatment groups. Again, none of these data appear to be compared to the no-treatment control.
The third RCT (46) randomized 60 civilian and military subjects 4.6 years after an average of 3.9 mTBIs to 40 HBOTs at 1.5 ATA oxygen (3,420 AM's oxygen) in 8-9 weeks vs. no treatment (0 AM's). The no-treatment control group was then crossed over to 40 HBOTs similar to Boussi-Gross et al. (70). Subjects completed symptom and quality of life questionnaires, and neuropsychologist-administered neuropsychological, psychological, and sleep testing or questionnaires. The HBOT group experienced significant improvements compared to the control group in one of the two co-primary outcomes (NSI: mean difference in change scores between groups: −23.9 ± 9.2, CI = −29.2 to −18.6, p = 0.0001) with the greatest improvement in the cognitive domain. This was the largest percent improvement (52%) in symptomatic outcome for a treatment group in all of the studies reviewed. The authors noted that this outcome was likely positively biased by the daily interaction of the P.I. with the subjects during treatment (an IRB safety requirement) and completion of the symptom questionnaire at the treatment site. The 52% figure, however, was in the range of one other study outcome figure (37%) for the same symptom questionnaire and same treatment dose (75) where the P.I. and subjects were blinded to treatment. Non-significant improvement was obtained in the other coprimary outcome (Working Memory: mean difference in change scores between groups: 1.5 ± 6.5, CI = −2.23 to 5.13, p = 0.431. Overall, the HBOT group achieved significant improvement in 11/14 outcome instruments compared to 13/14 after HBOT in the crossed-over Control Group. Twelve of the fourteen outcome instruments were administered by the blinded neuropsychologist at the non-treatment testing site. The other two outcomes, TBI and PTSD symptom questionnaires, were collected at the treatment site by the unblinded hyperbaric technician.

Dose Analysis of Studies
Effects of composite pressure and oxygen doses of HBOT on immediate post-treatment symptoms are presented in Table 2 for Table 1 randomized trials and case-controlled series with symptom outcomes [data abstracted from Table 7 Table 7 was recalculated to the immediate post-treatment period using the ImPACT symptom data in Table 2 of Wolf et al. (69) and the total ImPACT score (symptoms and cognition) from Exposure Interval 6 in Figure 2 of Wolf et al. (69)]. The data show symptom improvements at 1.5 ATA oxygen and 1.3 ATA air, improvements and deteriorations at 1.2 ATA air, no effect at the normoxic oxygen control (0.21 ATA oxygen) and one dose of oxygen (1.5 ATA oxygen), both at 2 ATA pressure, an improvement at the second oxygen dose (2.0 ATA), and no change at 2.4 ATA oxygen. According to the Scorza subset analysis (113), however, and "trend toward harm" of the 2.4 ATA oxygen dose for subjects with mTBI PPCS without PTSD the actual effect of 2.4 ATA oxygen on mTBI/PPCS without PTSD is negative. The positive (75) and negative (81) outcomes at 1.2 ATA air may be explained by the 50% greater number of patients with comorbid PTSD in Miller et al. (75), a subject demographic associated with larger treatment effects in multiple studies (68,80,81,87), and the 70% of subjects at risk for sleep apnea and post-treatment testing at altitude in Weaver et al. (81), phenomena that would favor negative treatment effects.
Pressure dose effects on immediate post-treatment symptom outcomes for the studies in Table 2 are presented in Figure 2. An asymmetric bell-shaped dose-response curve shows maximum symptom improvement in the 1.3-1.5 ATA pressure range with much less improvement at lower and higher pressures. Oxygen dose effects on symptom outcomes for Table 2 studies are in Figure 3 and demonstrate a similar asymmetric bell-shaped doseresponse curve with maximal results in a broad range from 1,002 to 4,860 atmosphere-minutes of oxygen and decreased responses at the lowest and highest doses of oxygen. The only aberrant value (3,720) is possibly due to the 2 ATA pressure component of the 1.5 ATA O2/2 ATA dose in the Cifu et al. study (71); the identical oxygen dose was delivered in Boussi-Gross et al. (70) at 1.5 ATA pressure with positive cognitive, mood, and functional imaging changes. Analysis of Figures 2 and 3 suggests that barometric pressure in the narrow range of 1.3-1.5 ATA has a more important effect on mTBI PPCS symptoms than oxygen pressure where a near-equal oxygen effect occurs from 1,002 to 3,420-4,860 AMs.  Figure 2 [Wolf et al. (69)] to the Pre-exposure ImPACT symptom scores in Table 2 [Wolf et al. (69)]. Table 2 vs. pressure dose for HBOT mTBI PPCS studies. Symptom percent values represent average percent improvement in symptoms at a given pressure in Table 2, averaging the three different instrument (ImPACT, NSI, and RPQ) symptom outcomes.

Classification of Evidence, Methodologic Quality, and Risk of Bias Assessment
The CEBM Levels of Evidence (3) Table 2 vs. oxygen dose for HBOT mTBI PPCS studies. Symptom percent values represent the average percent improvement in symptoms for each oxygen dose in Table 2, averaging the three different instrument (ImPACT, NSI, and RPQ) symptom outcomes. Case series (and poor quality cohort and case-control study 5 Expert opinion without explicit critical appraisal or based on physiology bench research or "first principles" wide confidence intervals. Wide confidence intervals, despite significant p-values and moderate to large treatment effects, are likely due to TBI heterogeneity factors identified in Section Limitations (below) and small sample size. The other four randomized trials to not fit cleanly in the CEBM hierarchy. Two of the rigorously performed higher quality/low bias studies published as randomized controlled trials, Wolf et al. (69) and Cifu et al. (71), that were previously assigned Level 1 status in an mTBI/PPCS review article (100) no longer meet the definition of a randomized controlled trial due to the scientific definition of HBOT used in the present review and the recharacterization of the control groups as treatment groups. They are classified Level 2. For the same reason the Weaver et al. (81) study, a rigorously performed high quality/low bias study published as an RCT, is now a Level 2 study. The fourth of these randomized Level 2 studies, Boussi-Gross et al. (70), is a randomized and true controlled crossover trial but did not have published confidence intervals, a CEBM (3) requirement for Level 1 classification. All four of these studies are given a 2B designation with a plus sign to denote, a higher level than true Level 2 cohort studies. There is one Level 3 study (80) PEDro analysis of methodological quality/bias is presented in Table 6. According to PEDro scoring statistics (119), studies scoring ≥6/10 on the PEDro scale are considered "moderate to high quality." Using the qualitative assessment of Cashin et al. (110) for the entire scale of PEDro Scale scores, studies were rated as: poor (<4), fair (4, 5), good (6)(7)(8), and excellent (9, 10). Four (69,71,75,81) of the six randomized and randomized controlled studies were good to excellent quality/low bias with PEDro scores of 8 (69), 9 (71), 9 (75), and 8 (81). Two of these four studies (75,81) were at 1.5 ATA oxygen. The other two randomized controlled studies, both at 1.5 ATA oxygen, were fair quality/greater bias with PEDro scores of 5 (46) and 4 (70). One point of the score differential between the higher and lower quality studies is due to design differences based on the presumption by the investigators of the four higher quality/low bias studies that a low dose of pressure and oxygen were sham   Scores of: <4 are considered "poor," 4 to 5 "fair," 6 to 8 "good," and 9 to 10 "excellent." *Harch et al. (68) is not included in this table due to the inclusion of its 16 subjects in Harch et al. (80).
controls. The pseudo-shams were pressurized treatments that allowed the blinding of patients to a dose of HBOT. The two crossovers lower quality/greater bias studies (46,70) (46,80). Many of the investigators in the low-bias studies (69,71,75,81) are employed by the funding source.

Side Effects
Side effects and adverse events were minor, but some were more frequent than expected and others were both notable and unusual; no treatment-induced serious adverse events were reported. Harch

DISCUSSION
Mild TBI is a heterogenous physical injury to the brain (7-27) that causes a wide range of signs, symptoms, and outcomes that have been defined in the chronic state as the Persistent PostConcussion Syndrome (2). TBI heterogeneity is reflected in the reviewed studies on HBOT in mTBI/PPCS in different subject populations (civilian and military, active duty and veteran), a wide range of ages (18-60 years), different etiologies (blunt and blast), varying times to injury [3 months (71) to 46 years (84)], PTSD or no PTSD, varied dosing protocols of pressure and oxygen, and an extensive list of outcome instruments. HBOT is classified as a medical gas by the U.S. Centers for Drug Evaluation and Research of the FDA and consists of two components, increased barometric pressure and hyperoxia (57,58,112), which have been argued to have varying effects on mTBI/PPCS depending upon the doses of pressure and hyperoxia employed (46,80,(99)(100)(101)112). Both have demonstrated a wide range of bioactivity (56,99,112,(121)(122)(123)(124)(125), including independent, overlapping, and interactive gene expression and suppression effects (123,124). This bioactivity is translated into wound repair and improved symptoms in diverse, mostly wound, conditions (56). This scientific understanding of hyperbaric oxygen therapy has eluded the hyperbaric medicine field for 359 years, particularly in the last 60 years, and has been confused and thwarted by the arbitrary definition of HBOT at a minimum pressure of 1.4 ATA of 100% oxygen (55,56). To rectify the confusion, help resolve the controversy of the mTBI/PPCS studies, and enable this systemic review, the arbitrary definition was supplanted by the scientific definition (58).
Symptom outcomes were chosen as the primary outcome in this review due to their broad reflection of the heterogeneity of TBI, their applicability to medical practice, and a recommendation by the U.S. FDA to prioritize a PPCS symptom questionnaire, the NSI, as a primary outcome for an RCT on HBOT (46) in mTBI PPCS (57). This choice and the FDA's recommendation were reinforced by the results of the Weaver et al. (81) study where investigators concluded that the "NSI would be a reasonable, simple primary outcome measure in future studies" after an extensive battery of tests in all of the Department of Defense HBOT mTBI/PPCS studies revealed that "more resource-intensive measures. . . did not prove useful for measuring the change in this population." The FDA also recommended that a series of clinical investigations with varying doses of pressure and hyperbaric hyperoxia should be performed (57). Inadvertently, the combination of civilian and DoD HBOT mTBI studies above have addressed these criteria with broad-based symptom outcomes (ImPACT, NSI, RPQ, or both NSI and RPQ) and a variety of pressures and levels of hyperbaric hyperoxia.
This symptom-based outcome review shows that, regardless of the heterogeneity of TBI and the studies herein, as well as negative reporting biases in studies (69,71,75,81) and positive biases in studies (46,80), analysis of the studies by outcomes, pressure dose, oxygen dose, or the composite dose of HBOT reveals that all symptom-based outcome studies performed at 1.5 ATA pressure and pressure of oxygen (46,75,80,81) show symptom reductions in mTBI PPCS. Two of these 1.5 ATA HBOT studies (75,81) are rated as CEBM Level 1 and 2 with good to excellent methodologic quality/low bias PEDro scores of 9 and 8, substantially higher than the average 5.1 scores of 37,417 PEDro scored randomized controlled physiotherapy trials (119), a category of clinical trials that share blinding challenges similar to hyperbaric medicine and surgery. The third of these 1.5 ATA oxygen trials, the RCT of Harch et al. (46) was a Level 1 study with a fair quality/greater bias PEDro Scale score of 5. The fourth study (80) is a Level 3 study. In addition, non-significant symptom reductions occurred in a single Level 2 excellent quality/low bias study at 2.0 ATA oxygen (71), and significant symptom reductions in single Level 2 and Level 1 moderate and excellent quality/low bias studies at 1.3 ATA pressure of air (69) and 1.2 ATA pressure of air (75), respectively. Effect sizes for all of the Level 1-4 studies in the present review are at least moderate and in excess of placebo effects for the oxygen groups compared to compressed air or no-treatment groups with symptom effect sizes greater than cognitive effect sizes (126).
The results achieved with the 1.5 ATA oxygen dose are reinforced when cognitive outcomes are added from these 1.5 ATA oxygen symptom outcome studies and the Boussi-Gross et al. (70) 1.5 ATA Level 2 oxygen study, the case series study of Mozayeni et al. (84), and the case reports of Wright et al. (67), and Shytle et al. (89). The symptom and cognitive outcomes are strongly supported by the functional imaging (17, 127) changes in two controlled studies (70,80). Both studies revealed significant improvements in brain blood flow. In Harch et al. (80) the improvements in blood flow were almost exclusively in the white matter, the primary site of injury in mTBI (9, 10, 15, 22-24, 30, 32-35, 47), but significant improvements in blood flow were also demonstrated in the hippocampi consistent with the improvements in memory function of the subjects. Near identical findings, improvements in memory, and simultaneous increase in blood vessel density were demonstrated at 1.5 ATA with oxygen in the Harch et al. (65) animal model of HBOTtreated mTBI. In Boussi-Gross et al. (70) improvements in cognitive domains matched improvements in HBOT-induced brain blood flow to the anatomic areas of injury responsible for cognitive deficits seen in the subjects. In addition, the increases and decreases in brain blood flow to different regions of the brain measured in Boussi-Gross et al. (70) were an independent alternative replication of the mathematical texture analysis and visual pattern shift of brain blood flow measured in Harch et al. (80). The widespread improvements on SPECT seen in these two studies are contrary to those seen in SPECT-studied placebo drug trials (128,129) and are consistent with the symptom and cognitive improvements seen in all of the studies at 1.5 ATA. Equally important, the beneficial outcomes with the 1.5 ATA dose were achieved with minor side effects or complications.
The outcomes achieved in the reviewed studies compare favorably with outcomes achieved in a non-controlled U.S. Department of Defense National Intrepid Center of Excellence treatment program for veterans with TBI with or without psychological health problems. DeGraba et al. (130) reported a 7.5-year experience with 1,456 veterans who were a minimum of 6 months post-TBI (average 5.09 years) and had sustained an average of 7 TBIs. Patients were treated in a 4-week intensive residential program that included over 20 different therapies, at least 20 different therapists, and multiple physicians. Results from 474 to 1,174 of the most severely symptomatic (numbers of subjects varied per outcome instrument reported) showed a decrease in NSI of 44%, Post Traumatic CheckList of 23%, General Anxiety Disorder scale-7 of 50%, and Personal Health Questionnaire-8 of 50%. Headache, a marker of TBI (131), had the least improvement (4 points, 6.5% reduction), half the amount that is considered a clinically significant change (130). The only study in this review with the same therapy benefit was the Harch et al. study (80), also a non-controlled study, whose subjects were a minimum of 6 months post-TBI (average 3.35 years) with an average of 3.5 TBIs. After 4 weeks of treatment with a single biological therapy, one technician, one physician, and a neuropsychologist subjects' RPQ decreased by 36%, Post-traumatic CheckList by 26%, General Anxiety Disorder-7 by 40%, and Personal Health Questionnaire-9 by 48%. In contrast to DeGraba et al. (130), headache, the second most responsive symptom next to dizziness (80), was reduced in 93% of subjects, indicating treatment of the underlying TBI The positive findings for the 1.5 ATA oxygen dose in the mTBI PPCS studies are further supported by positive outcomes in multiple studies on HBOT at 1.5 ATA in acute severe TBI (136)(137)(138)(139)(140)(141)(142)(143), and one study in moderatesevere subacute TBI (144), suggesting a shared sensitivity of TBI to 1.5 ATA oxygen regardless of acuity or severity. An additional RCT performed at 2.5 ATA in acute severe TBI (145) proved to be toxic, very similar to the 2.4 ATA oxygen dose in Wolf et al. (69,113). The acute treatment course in that study (145) was interrupted in multiple subjects due to pulmonary oxygen toxicity. Similar to Wolf et al. (69), the neurological results were minimal at this high oxygen dose: no difference in mortality or duration of coma was demonstrated between groups (145). The sole positive finding was an interim outcome, increased recovery of consciousness at 30 days, in a subgroup of HBOT-treated subjects with brainstem contusion.

Summary of Main Findings
According to the CEBM evidence classification hierarchy (3)

Limitations
The main limitation is the heterogeneity of the patients inherent in the diagnosis of mTBI/PPCS (causes of injury, circumstances, non-uniformity of force/vector/head position, anatomy, symptoms, past injuries, co-morbidities, time from injury to treatment, etc.). Additional limitations are the exclusion of non-English language literature, the discrepancy in treatment and testing sites with respect to altitude (81), small sample sizes out of proportion to the heterogeneity (46, 68-71, 75, 80, 81, 84), variability in the design of the studies (46, 68-71, 75, 80, 81, 84), heterogeneity of outcome instruments (46, 68-71, 75, 80, 81, 84), failure to appreciate that a pressure experience is not a sham such that randomized "controlled" trials were actually randomized comparative dosing studies (69,71,75,81), difficulty in structuring a study with a true pressure control (46, 69-71, 75, 81), lack of statistical analysis of outcome changes of one group compared to another or control (69)(70)(71)75), and paucity of literature to review. Despite all of these limitations which would bias data and conclusions toward rejection of the null hypothesis and Type II Error due to small sample size, the studies' data support the conclusions drawn herein.

Conclusions
In multiple randomized and randomized controlled studies, HBOT at 1.5 ATA oxygen demonstrated statistically significant symptomatic and cognitive or cognitive improvements alone in patients with mild traumatic brain injury Persistent Postconcussion Syndrome. Positive and negative results occurred at lower and higher doses of oxygen and pressure. According to pressure and oxygen dosage analyses, increased pressure within a narrow range appears to be the more important effect than increased oxygen which is effective over a broad range. Improvements were greater when patients had comorbid Post Traumatic Stress Disorder. Large blinded controlled trials would be ideal to confirm the results of this review, but despite small sample sizes, the studies using 1.5 ATA oxygen satisfy the Centre for Evidence-Based Medicine (3) Level 1 criteria and merit a Class A Recommendation for treatment of mTBI PPCS at 1.5 ATA of oxygen, according to the American Society of Plastic Surgeons Grade Practice Recommendations (4).

CLASSIFICATION OF EVIDENCE
This review provides Centre for Evidence-Based Medicine (3) Class 1 evidence that 40 hyperbaric oxygen treatments at 1.5 ATA oxygen are effective and well-tolerated in mTBI Persistent PostConcussion Syndrome. According to the American Society of Plastic Surgeons Grade Practice Recommendations, this evidence meets the threshold for and is a Grade A Practice Recommendation (4).

CONTRIBUTION TO THE FIELD STATEMENT
Mild TBI has been traditionally considered to be innocuous. It is now appreciated to cause permanent symptoms [the Persistent PostConcussion Syndrome (PPCS)] in a substantial number of patients. Hyperbaric oxygen therapy is a treatment for mostly acute and chronic wound conditions that is scientifically defined as a dual-drug therapy that uses increased oxygen and increased pressure to treat disease pathophysiology/wounds. Multiple studies have reported confusing and seemingly conflicting results of HBOT treatment of PPCS, due to the use of a non-scientific historical definition of HBOT that led to inadvertent maldesign of some of the studies. Using the scientific definition of HBOT this review resolves the confusion and conflict by analyzing the studies on HBOT in mTBI PPCS according to individual and composite doses of pressure and oxygen. The results show that barometric pressure is the more important component of HBOT compared to the pressure of oxygen and that in multiple studies a composite dose of pressure and oxygen, 1.5 ATA oxygen, demonstrated significant improvement in symptoms and cognition in patients with PPCS. This finding has the potential to change the standard of practice in the treatment of PPCS and has substantial implications for millions of patients worldwide afflicted with PPCS, including military war veterans.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.