Evidence Table 9. Quality assessment of observational studies

Author year (Quality)Study designExposure measurement (HBOT therapy same for all participants?)Stable baseline established?Confounding factors addressed? (severity of disease, other treatment); control for confounders?Masking (patients, outcome assessors)Timing of baseline measures stated and sufficiently close to intervention?Timing of followup measurements stated and adequte?Outcome measures stated and objective?CommentsExternal validity
Traumatic Brain Injury
Artru 197694(Fair)Before-afterYes, except one compressed at 2.2 atm instead of 2.5 because of a bad pulmonary condition.NoSeverity, diagnosis of each patient reported.Not reportedNot reported when or how clinical status measured, other measures taken in close proximity.CBF, metabolic rates for oxygen, glucose and lactate and CSF parameters measured 2 hours after exit from the chamber, clinical status seemed to be measured immediately after exit.No for clinical status, yes for other measures.Not reported how patients selected, no standard assessment of severity of illness, only 6 patients.
Hayakawa 197195(Fair)Time-seriesYes for dose, number of treatments varied from 10 to 15NoNoNoYesYesYesVery small sample from one institution, no info on patient selection.
Mogami 196996(Poor)Before-afterNoNoNoNot reportedNot reported, “before treatment”.YesSubjective, observation (e.g., increased awareness and responsiveness)Not clear, some patients comatose, some requiring ventilation, others with only mild deficits. No information on selection of patients, no description of patients except diagnosis and symptoms.
Ren 200199(Poor)Retrospective cohortSimilar HBOT treatments (not exactly the same). Control interventions actually used not stated.NoNoNoYesYesYesNo information on selection.
Uneven numbers in groups, reason not given. Mean GCS = 5.3 a& 5.1, not enough other details to assess.
Rockswold 200197 (Fair)Time-seriesDose and duration the same, number of treatments varied according to response.NoYes- time from injury to treatment, type of brain injury and types of multiple trauma.YesYesYesYesPatients were from one Level I trauma center, number screened not given, excluded patients placed in barbiturate-induced coma.
Sukoff 198298 (Poor)Before-afterNo- frequency changed based on ICP and clinical response.NoNoNoNot reportedYesSubjectiveVague reporting of results, other than ICP measurements.Only a series of patients in whom HBOT was effective in reducing ICP and showed improved neurological status. No information on how many other patients treated without improvements. Patients requiring vasopressors or with dilated and fixed pupils excluded.
Other Brain Injury
Chuba 1997107 (Poor)Before-afterNoNo- all patients presented with new or increasing neurologic deficits.Some confounding factors listed (e.g., tumor); not controlled for.Not reportedNot clearYesNo- symptoms, method not describedVery few outcome measures reported.Children with radiation-induced necrosis of CNS, mostly in supratentorial location. All patients had increasing neurologic deficits and had failed steroids.
Imai 1974108(Poor)Before-afterYesNo- not reportedNoNot reportedNot clear“After one or two treatments”YesFew details about how subjects selected or if/when baseline tests done.No information on selection of patients, no description of patients except diagnosis; very mixed group selected: patients with presenile dementia, chronic alcoholism, cerebral vascular disease, and CO intoxication.
Mathieu 1987109 (Poor)Before-afterDose the same, but number of treatments varied according to conditionNoNoNoBaseline information not reported except for GCS measure on admission, timing not clear in relation to HBOTTiming not reportedDeath, neurological sequelae described, “recovery without sequelae” not defined, not objective. Method of measuring outcomes not reported.All patients at one institution with attempted hanging, 71 (42%) had psychiatric illness.
Studies of Cerebral Palsy
Chavdarov 2002121 (Poor)Before-afterYesNoReports data grouped by level of severity.NoYesYesYes, but measured by the same assessor before and after treatmentNot clear how patients selected, all were from one hospital in Bulgaria, described inclusion criteria (history of seizures excluded, among other).
Machado 1989122 (Poor)Time-seriesNoNoNoNoNot clear (“had been reviewed before and after therapy”)YesSubjective- observationsFew details about intervention, or outcome measurementsRetrospective review of CP patients treated at one institution, states that report is observations based on clinical findings and without scientific documentation, followed up only 39% (only those that lived in Sao Paolo). Diagnosis not defined.
Montgomery 1999123(Fair)Before-afterNo, varied by centerYes. Inclusion criteria stated functional plateau in rehabilitation for 12 months (defined as no measurable functional changes in gross motor performance as documented by their physical therapists.Potential effect of age is addressed by type of test given.Not clear: outcome assessors had “no contact” with children during HBOT; but not necessarily masked.Pre and post evaluations separated by 37.2 +/- 8 daysPre and post evaluations separated by 37.2 +/- 8 daysStated, some subjective, some objectiveThis study randomized subjects to two weeks of HBOT or 4 weeks of HBOT, with no control group.Excluded were those with recent rhizotomy, thoracic surgery, epilepsy, cancer, asthma, VP shunts, previous HBOT, anti-spasticity meds, or behavior problems.
Studies of Stroke
Hart 1973149 (Poor)Time-seriesNo, same pressure and time, but different drug therapies and a second course given if improvement seenNo- criteria were 3 weeks stable and at least 4 weeks from onset of ischemic attack.No- all patients received physical therapy, gait training, and speech therapy. Only age and other medications given in different recovery groups were reported, but not controlled for.NoNot reportedNot clear (period of 7 days reported for improved patients, but duration of follow-up not clear).Stated, but not clearly objective (neurologic exam, EEG, radioisotope scans)Defined as middle cerebral artery ischemia (proven by angiography or radioisotope scan), neurologic state stable for at least 3 weeks, and at least 4 weeks from onset. No information on patient selection.
Holbach 1977a141 (Poor)Time-seriesYes, except TIA patients had only 1 treatment (“test treatment”), but results are reported separately.No States 38 patients had complete stroke, 36 were chronic stroke patients, but timing of HBOT since onset of stroke not reported.NoNot reported“Before HBOT treatment”“During and at conclusion of HBOT” sessionsStates neurological exam, but not describedOutcome measurements unclear.Diagnosis vague, no information on patient selection.
Holbach 1977b142 (Poor)Time-seriesYes for dose, number of treatments varied from 10 to 15.NoPatients selected for surgery based on initial response to HBOT.Not reportedYesYesNeurological exam and neurological deficits, not specifiedAll patients were judged suitable to undergo surgery if necessary, no other information on patient selection.
Imai 1974108 (Poor)Before-afterYesNo- not reported.NoNot reportedNot clear“After one or two treatments”YesFew details about how subjects selected or if/when baseline tests done.No information on patient selection, no description except diagnosis; mixed group selected: patients with presenile dementia, chronic alcoholism, cerebral vascular disease & CO intoxication.
Jain 1989134 (Poor)Before-afterYesNo- time since onset ranged from 3 months to 5 years, all were undergoing stroke rehabilitation.No, all patients had simultaneous physical therapy.NoYesYesYesResults on spasticity are very difficult to interpret (stratification of results not explained), also subjects continued to receive PT after HBOT during followup.21/50 patients already undergoing rehabilitation with PT in HBOT chamber, only those with clinically assessable spasticity are included. Diagnosis vague.
Jain 1990143 (Poor)Time-seriesYesNo One patient was 5 years post-stroke, all others less than 12 months (range 3 weeks-11 months); no day-to-day neurological changes during first week of admission to clinic.No, all patients had simultaneous physical therapy.Not reportedYesYesYesOnly patients seen by senior authors included, only 10 of 25 had spasticity measurements recorded.Almost no description of patients given, except that they had occlusive cerebrovascular disease, 24 had hemiplegia. Spasticity grade at baseline ranged from 2–5 (scale 0–5), mean 3.4 (only 10 patients with this data).
Data for grip strength in table not discernable (e.g. before = 0/32 and after = 8). A footnote refers to stronger/weaker hand strength. First 8 patients eceived PT immediately after HBOT, and noticed that this significantly helped with spasticity, so the rest received PT during HBOT treatments.
Kapp 1981144 (Poor)Time-seriesYes.No- HBOT started on day of diagnosis or after obtaining consent from family.Yes, stratified results but no attempt to control statistically.Not reportedYesYesSome measures objective and some subjectivePurpose of study was to use HBOT to identify subjects who are candidates for revascularizationInclusion criteria described, all patients seen by author in 2-year period who fit criteria enrolled, number screened and eligible not given.
Li 1998135(Poor)Before-afterDose and duration the same, but course of 15 treatments varied from 3–4, and “rarely” more.No- some patients were 3 years post-stroke but results are not reported separately by time since stroke. Range was < 1 month to over 3 years.No.Not reportedNot reportedNot reportedStated, not objective (assessment of clinical symptoms by neurologists)Too few details to assess. A subset was compared to “controls” - no details on how either gorup selectedNo information on selection of patients, limited description of patients.
Neubauer 1980139(Poor)Before-afterNo.Yes for some patients (time of HBOT treatment from onset ranged from 4 hours to 10 years).No- patients received simultaneous physical therapy “when indicated”. Severity reported, no statistical tests done. Results reported by timing of treatment from onset of stroke.NoNot reportedNot reportedSubjectiveSome information on patients given, no info on selection process. Because consecutive patients, may be good but baseline characteristics not reported.
Noguchi 1983136(Poor)Before-afterNo. Number of treatments per day and total varied based on condition of patient. Varied from one time to 60 times, average of 17.2 per case.No.No.Not reportedNot reportedNot reportedNo, clinical assessment, signs and symptoms, physical exam, not described.No information on patient selection, diagnosis vague
Pilotti 1991147 (Poor)Retrospective comparison of cohortsSimilar HBOT treatments (not exactly the same). Control interventions actually used not stated.NAPatients in control and treatment groups were treated at 2 different institutions. Small differences with respect to age, sex, clinical history. Respiratory insufficiency, vascular insufficiency of inferior limbs and arterial hypertension were higher in HBOT-treated group (p < 0.05).YesYesNA (retrospective cohort)No dropouts or missing data reportedOverall mortality rates at the 2 hospitals are not given for comparison.No information on patient selection
Saltzman 1965145 (Poor)Time-seriesNo In 18 patients, HBOT given within 7 hours of onset of symptoms, in 7 patients, given 7–30 days after onset. Myringotomies performed before in “most” patients. Initial studies used 3.04 atm for < 1 hour, in later treatments, pressure below 2.5 ata, permitting “prolonged” exposure.No- patients had treatment <7 hours to 30 days after onset of stroke.Describe separately patients who received HBOT soon after onset, but other factors not addressedNoNot reportedUnclear. One patient reported out to 10 days, others only immediately after HBOT.Subjective (observation)Very vague description of improvements seen during and shortly after HBOT.No information on selection of patients, no description of patients except diagnosis. Diagnosis vague
Steenblock 1998137 (Poor)Before-afterNo. Treatment plans and physical therapy adjusted as the patient's condition warranted.Yes for some patients (time of HBOT treatment from onset ranged from 1 month to 10 years, average 29 months).No- all patients received physical therapy, electrical stimulation, hot or cold packs, ultrasound, short wave diathermy, paraffin bath therapy, and biofeedback.NoEvaluation at the “beginning of the program and at the end”“At the end”Stated, subjectiveResults are presented as % improvement or improvement no/yes.No information on selection of patients, reports that 100% of patients improved on one or more functions, little description of patients. Baseline severity of disability not presented.
Tsuro138 1983 (Poor)Before-afterDose the same, but number of treatments varied.No.No.NoNot reportedNot reportedObservations, not objective, not stated how determinedSeries of cases treated with HBOT at one institution, no information on selection of patients for treatment. Only 79 cases of “over a hundred” treated are presented.
Wassman 1986146 (Poor)Time-seriesYes for dose, number of treatments varied from 10 to 15.No- patients were about 3 months post-stroke.No.Not reportedYesYes“Motor deficit” method of measurement not described.Only includes patients who showed a definite increase of electrical power equivalent on EEG after an HBOT series.
Zhou Shn-rong 1995110 (Poor)Before-afterNo, dose the same but number of treatments varied.No- HBOT given 2–3 days after stroke.No.NoNot reportedNot reportedYes, but not objectiveFew details about how subjects selected or if/when baseline tests done.No info on selection of patients, no way to compare patients, e.g. described as ‘deep coma’

HBOT=hyperbaric oxygen therapy; atm=atmospheres; ICP=intracranial pressure; TIA=transient ischemic attack; CBF=cerebral blood flow; CSF=cerebrospinal fluid; GCS=Glasgow Coma Scale; CO=carbon monoxide; CP=cerebral palsy; PT=physical therapy; EEG=electroencephalogram; CNS=central nervous system

From: Evidence Tables

Cover of Hyperbaric Oxygen Therapy for Brain Injury, Cerebral Palsy, and Stroke
Hyperbaric Oxygen Therapy for Brain Injury, Cerebral Palsy, and Stroke.
Evidence Reports/Technology Assessments, No. 85.
McDonagh MS, Carson S, Ash JS, et al.

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.