All adult patients with an acute stroke will be eligible to be considered for study participation. There are no definite guidelines for oxygen treatment after acute stroke, and there is uncertainty among stroke physicians about who should be given oxygen and for how long. The eligibility criteria for inclusion in the trial reflect this uncertainty, and allow for randomization of all acute stroke patients who do not have definite indications for or definite contraindications to oxygen treatment.
Hence, adult patients will be eligible for trial inclusion if they were admitted with symptoms of an acute stroke within the preceding 24 hours, and in the doctor’s opinion there is no clear indication for, and no clear contraindication to oxygen treatment.
The diagnosis of stroke will be made by history and clinical examination and is at the discretion of the admitting doctor. It will based on the World Health Organization (WHO) criteria (rapidly developing clinical signs of focal or global disturbance of cerebral function, with symptoms lasting 24 hours or longer, or leading to death, with no apparent cause other than of vascular origin) . Within the first 24 hours of symptom onset a definite distinction between a stroke and a transient ischemic attack cannot be made. However, most patients who still have persistent symptoms after one hour will be confirmed to be having a stroke. Since waiting for 24 hours for confirmation would unnecessarily delay treatment, we omitted the time element from the definition of stroke for the purposes of trial inclusion.
Patients will be excluded from the trial if the responsible doctor considers the patient to have definite indications for, or contraindications to, oxygen treatment at a rate of 2 to 3 L/min. The decision will be left to the responsible clinician. This exclusion criterion has been chosen to ensure that all patients are treated according to best medical practice.
Potential indications for oxygen treatment could be: oxygen saturation on air <90%, hypoxia associated with acute left ventricular failure, severe pneumonia, pulmonary embolus, and chronic respiratory failure treated with long term oxygen at home.
Potential contraindications to fixed dose oxygen treatment could be type two respiratory failure and very severe hypoxia.
Patients will also be excluded if the stroke is not the main clinical problem, or if they have another serious life-threatening illness likely to lead to death within the subsequent few months. This group of patients is excluded because it is unlikely that they are going to derive any benefit from the trial treatment.
The one-week assessment will confirm the diagnosis, and will document deaths and neurological status (NIHSS), compliance with the intervention, and complications. It will be performed by a member of the local research team trained in the assessment tools seven days (± one day to allow for weekends and holidays) after enrolment. In the case of patients who are discharged before the end of one week, or who cannot be followed at seven days, the patient will be assessed at discharge and if appropriate the follow-up completed at this time point. Wherever possible we will strive to assess the patient at day seven after randomization in hospital or, if discharged, in clinic or in their place of residence as the patient specifies. If the one-week assessment is not possible then the discharge assessment is acceptable as the one-week outcome. Data will be entered online or sent to the trial center via fax.
Three months, six months and twelve months
The main follow-up will be performed centrally at three months, a standard procedure for most acute stroke trials. Assessments will be based on a questionnaire sent to the patient’s preferred follow-up address by the central team, after checking with the GP that the patient is still alive, unless the patient has specified a different preference at the one-week assessment. Central follow-up will ensure blinding of the assessors to the intervention. For non-responders the address will be checked via the GP and the local researcher and then resent. If there is no response, patients will be contacted by phone to see if they would prefer a personal follow-up or are happy to reply to questions via the telephone. Missing or inconsistent data will be cross-checked with the medical notes, with the GP, or by personal contact with the patient. If patients are not contactable using these methods, we will determine if they have died and, if so, what the cause of death was by requesting information from the Office of National Statistics or the NHS Strategic Tracing Service. The latter will also be contacted if the patient is no longer resident at the address given and the GP has not seen the patient recently and does not know the patient’s new address and new GP.
The follow-up questionnaire will contain the modified Rankin score (mRS) , Barthel Index of Activities of Daily Living Score (ADL) , Nottingham Extended Activities of Daily Living (NEADL) index , EQ-5DTM[50–52], and questions regarding memory, sleep, speech and discharge status.
The primary outcome is measured using the mRS score at three months . Secondary outcomes are measured at one week using: the number of patients with neurological improvement (≥4 point decrease in the NIHSS) , the number of deaths, the highest oxygen saturation during the first 72 hours, and the lowest oxygen saturation during the first 72 hours. Further secondary outcomes are measured at three months using: the mortality rate, the percentage of patients living at home, the Barthel ADL score, the EuroQol score and the Nottingham Extended Activities of Daily Living scale (NEADL). Further explanatory analyses include: antibiotic and sedative use during week one, the highest heart rate during the intervention >100, the highest systolic blood pressure during the intervention >200 mm Hg, the highest diastolic blood pressure during the intervention >100 mm Hg, oxygen saturation during the intervention, the percentage of patients describing their memory as ‘as good as before the stroke’, the percentage of patients describing their sleep as ‘as good as before the stroke’, the percentage of patients without significant speech problems, and the change in outcomes over time (three month assessments repeated at six and twelve months).
Those outcomes concerning memory, sleep and speech have been highlighted as important on the public consultation , but are not part of standard assessment scales for stroke outcome.
All study measures will be performed by research staff appropriately trained in the use of the assessment tools.
Data management and evaluation
Patients will be randomized using minimized randomization stratified by well validated prognostic factors (age, sex, living alone, normal verbal component of the GCS, ability to lift both arms and ability to walk) [56, 57], routine oxygen treatment during ambulance transfer, and baseline oxygen saturation, via a web-based randomization system. Randomization will not be stratified by the study center as this may result in unacceptably high rates of allocation prediction and selection bias . However, retrospective analysis by center will be performed to investigate any heterogeneity of treatment effect by center.
The local investigators, their research assistants, and data monitors will have access to the patient records. Personalized data (address, telephone number, email and fax) will be kept in the trial office, as electronic copies within each of the centers, and at the coordinating center to allow patients to be contacted for the three-, six- and twelve-month follow-up and to allow data checks and validation.
For all other purposes patient identifiable data will be converted to an alphanumeric code, using a specific code number for each patient. The principal investigators, members of the trial steering group, the trial managers, data managers, data monitors, the data analysts/programmers, and the trial statisticians will have access to the anonymized data.
Data will be stored on password protected office computers and on Zip discs, flash drives or CD-ROMs. At least three back-up copies will be made of all data. These will be kept in locked cupboards in separate buildings.
Data will be transmitted via fax, email or the web from each local center to the trial coordinator and data queries will be transmitted via the same route from the trial coordinator to local centers. Anonymized data may be made available to other researchers for meta-analysis and publication in media such as the Cochrane Database.
The analysis will be by intention to treat. The primary outcome is the mRS score, which has an ordinal range of 0 (best outcome) to 5 (worst outcome). This will be measured at three months (or at the last rating). Deaths will be allocated an arbitrary score of 6 . A later primary outcome assessment at six months was considered, but rejected because there is a risk of diluting treatment effects by newly occurring health problems unrelated to the trial intervention .
The trial tests two research hypotheses: 1. oxygen supplementation results in better (lower) mRS scores at three months than no oxygen supplementation; 2. Oxygen administered at night results in better mRS scores than oxygen given over a 24 hour period.
It cannot be assumed that any benefits from oxygen will be dose dependent. Oxygen supplementation throughout the 24-hour period may expose a significant number of patients to oxygen concentrations higher than normal. Being attached to oxygen may also limit mobility and hinder early mobilization. Oxygen given at night only will provide supplementation at a time when oxygen saturation is lowest in stroke patients, and will not interfere with rehabilitation. A prior hypothesis is that oxygen at night will have all the potential advantages without the disadvantages associated with daytime oxygen use.
The mRS will be compared between two groups using ordinal logistic regression . Both an unadjusted and an adjusted analysis will be performed. The latter will incorporate the following covariates: age, sex, baseline NIHSS score, and the ‘six simple variables’ (SSV) prognostic index for independence at six months. The SSV score is based on the following variables: age, living alone before the stroke, independent pre-stroke, normal verbal response to questions, able to lift the affected arm against gravity, and able to walk unaided [56, 57]. Planned subgroup analyses for the primary outcome are: baseline SSV prognostic index (<0.2, 0.2 to 0.35, 0.36 to 0.7, >0.7), NIHSS score at baseline (<5, ≥5), baseline oxygen saturation (<95, ≥95), treatment with oxygen prior to randomization (yes or no), time since onset of stroke in hours (<4, 4 to 6, 7 to 12, 13 to 24, >24), type of stroke (hemorrhage, infarction). Mortality will be assessed using survival analysis, and other secondary outcomes will be analyzed using appropriate statistical methods. All tests will be two-tailed and 95% confidence intervals will be calculated for all estimates of effect.
Many acute stroke studies have been underpowered because the expected treatment effect was unrealistically large . While thrombolysis within three hours of acute stroke has been shown to lead to moderate clinical benefits (0.5 points on the mRS) , neuroprotectant treatments may well achieve lesser (for example, 0.2 point on the mRS) treatment effects . As stroke is such a common condition and oxygen supplementation is inexpensive and universally available, even relatively small differences in outcome could have a major impact on the burden of the disease. For example, treating five patients with an average improvement of 0.2 mRS would improve one patient by one mRS category (for example from moderate disability to slight disability). However though important, small differences do require very large trials in order to show effectiveness.
The sample size calculation is based on an odds ratio of 0.83 for a more adverse outcome (higher mRS score) for the oxygen and control groups, observed in the first 200 patients in the Stroke Oxygen Pilot Study (ISRCTN12362720) [24, 25] (C Roffe, P Jones, S Sills, personal communication). The sample size allows for a 5% drop-out rate for such things as retrospective exclusions for a change of diagnosis (numbers based on the Stroke Oxygen Pilot study) plus a 5% rate of missing outcome data (this overall 10% loss to follow-up gives a safe margin; target would be less than 3%).
A sample size of 6,000 patients will therefore provide 95% power to detect the specified odds ratio between oxygen (continuous and night only groups combined) and no oxygen at p ≤ 0.05 (two-tailed), and 87% power at p ≤ 0.05 (two-tailed) to detect the same effect between continuous oxygen and oxygen at night only. Adjustment for a maximum 10% loss to follow-up gives a target sample size of 6,669.
Increase of sample size in October 2012
A sample size of 8,000 patients will be used, as increasing the recruitment target would give greater power to detect an interaction between subgroups (defined by severity) and the effect of oxygen versus control. The magnitude of the increase in power can be estimated by considering the increase in power to detect the pre-specified odds ratio of 0.83 in the subgroup of ‘moderate through to very severe’ patients (those with an NIHSS score greater than or equal to 10, who are more likely to benefit from the treatment). Using a revised loss to follow-up rate of 5%, power to detect the specified odds ratio between oxygen and control in this subgroup would rise from 43% to 50%. For the whole sample, power to detect this effect between oxygen treatment and no oxygen treatment will rise from 95% to approximately 98% and that for the comparison of continuous oxygen and oxygen at night will rise from 87% to approximately 94%.
Approval of the study by the research ethics committee
Multicenter ethical approval was granted for version 2 of the protocol by the North Staffordshire Research Ethics Committee on 25 June 2008 (COREC 06/Q2604/109).
Good clinical practice
The study will be performed in accordance with the principles stated in the Declaration of Helsinki . Study procedures will be guided by the standards outlined in the MRC Guidelines for Good Clinical Practice in Clinical Trials .
Patient information and consent
Consent will be obtained according to the requirements of the Multicentre Research Ethics Committee and the Local Research Ethics Committee before the start of recruitment.
The patient, and where appropriate the next of kin, will be given full and adequate oral and written information about the nature and purpose of the study, possible risks and benefits. A copy of the patient information sheet and the signed consent/assent will be given to the patient. Patients will be informed that they are free to discontinue participation in the study at any time.
Fully informed consent will be sought from all competent subjects. In patients who are conscious, but not fully competent to understand the information to make a reasoned decision, we will provide a simple explanation of the trial and seek the patient’s agreement, and also seek assent from the next of kin or from an independent physician. If an incompetent individual has been included in the trial without giving fully informed consent we will strive to obtain fully informed consent as soon as the patient is able to do so. This will be documented on the one-week follow-up form.
The reason for including patients unable to give fully informed consent is that roughly one third of stroke patients will have problems with speech and with the understanding of spoken and written material as a consequence of their stroke. It is important to include these patients in the study since they are just as likely to benefit from the treatment as patients who are able to communicate.
It is further important to include as wide a spectrum of stroke patients as possible, in particular patients with severe strokes. Patients with severe strokes may be more likely to develop hypoxia, and may therefore be more likely to benefit from oxygen treatment than patients with mild strokes. However, patients with severe strokes are more likely to be confused, drowsy or dysphasic, and thus unable to give informed consent. Exclusion of subjects unable to give informed consent is thus likely to bias trial outcome. Furthermore, since the group of patients who are unable to consent has different clinical characteristics from patients who can give consent, the results of the study may not be applicable to patients with similar clinical presentations to the excluded patients.
The information sheets to be given to patients, relatives, or the independent clinician have been reviewed and edited by service users from Strokes R Us (Stoke-on-Trent, UK) and Different Strokes (Coventry, UK).
Monitoring of suspected unexpected serious adverse reactions
All suspected unexpected serious adverse reactions (SUSARs) that are believed to be due to the trial treatment will be reported as soon as possible within one working day of the clinician becoming aware of the event by phoning the study helpline or by emailing firstname.lastname@example.org. A SUSAR report form will be completed as fully as possible and sent, via fax, to the Chief Investigator and the sponsor (North Staffordshire Combined Healthcare Trust Research and Development). On this form the patient will be identified by a unique identifying number consisting of the trial identification number (ISRCTN) followed by the number the patient was allocated at randomization. The SUSAR form will be filed in the trial master file and in the case record form. A SUSAR follow-up form will be completed as soon as possible within five days of the event and submitted via fax to the coordinating center and sponsor. This will also be filed in the trial master file and the case record form. Unless the event has resolved or a decision has been taken that no further follow-up is required, further follow-up forms will be completed, faxed and filed as outlined above until the event has resolved. Relevant details of the SUSAR and its follow-up will also be recorded in the patient’s medical notes. The sponsor will inform the licensing authority, the competent authorities of any member state in which the trial is being conducted, and the relevant Research Ethics Committee (West Midlands – Staffordshire Research Ethics Committee, Barlow House, 3rd Floor, 4 Minshull Street, Manchester, M1 3DZ, Fax 01785 254 640) of the SUSAR as soon as possible, and no later than seven days after first becoming aware of the event. The sponsor will provide details of follow-up reports and resolution. The sponsor will also inform all the principal investigators of the trial of the SUSAR. At the end of each year from the start of the trial the sponsor will provide the licensing authority (Medicines and Healthcare products Regulatory Authority, UK) with a list of all SUSARs relating to the trial during that year and any other relevant new information relating to the investigational product which may affect the conduct of this trial.
Data will be stored and analyzed in accordance with national data legislation. Personalized data (address, telephone number, email and fax) will be kept in the trial office and as electronic copies within each of the centers and at the coordinating center to allow patients to be contacted for follow-up, and to allow data checks and validation. For all other purposes patient identifiable data will be converted to an alphanumeric code, using a specific code number for each patient.
North Staffordshire Combined Healthcare NHS Trust, Trust Headquarters, Bellringer Road, Trentham, ST4 8HH.
Person authorized by the sponsor to act on behalf of the sponsor
R&D Director, North Staffordshire Combined Healthcare NHS Trust, Trust Headquarters, Bellringer Road, Trentham, ST4 8HH. Tel: 01782 441651.
Professor C Roffe, North Staffordshire Combined Healthcare NHS Trust, Holly Lodge, 62 Queens Road, Hartshill, Stoke-on-Trent ST4 7LH. Tel 0300 123 0891 Fax 0300 123 0894, email: email@example.com.
Trial management committee
The trial management committee (TMC) is responsible for the overall design and conduct of the study, analysis of the data, reporting and dissemination of results. It will act on advice of the trial steering committee, the data safety and management committee, the advisory groups and the international advisory committee.
Membership: Professor C Roffe (chair, stroke physician, clinical lead of the West Midlands Local Stroke Research Network); Professor P Crome (geriatrician, clinical trialist and pharmacologist), Primary Care & Population Health, University College London; Professor R Gray (expertise in large clinical trials), University of Oxford; Professor J Sim (statistician), Keele University; Professor P Jones (statistician), Keele University; Mr Peter and Mrs Linda Handy (patient representatives), Strokes R Us, Stoke-on-Trent.
Trial steering committee
The trial steering committee (TSC) will oversee the study. Professor M Dennis (stroke physician, clinical trialist), University of Edinburgh will act as independent chairman. Other members are: Professor L Kalra (stroke physician, clinical trialist), King’s College, London; Professor S Maslin-Prothero (nursing, policy and practice in the NHS); J Daniels (clinical trialist), Birmingham Clinical Trials Unit; Mrs P Bell (patient representative, dysphasia support); Professor R Lindley (international advisor, stroke physician, clinical trialist); and members of the TMC.
Data monitoring and safety committee
The remit of the Data Monitoring and Safety Committee (DMSC) will be to ensure that patients are not exposed to unnecessary risks, by performing interim safety analyses, and to maintain patient safety. If oxygen treatment really provides substantial benefit or harm with respect to the primary endpoints, then this may become apparent before the target recruitment has been reached. Alternatively, new evidence might emerge from other sources that oxygen is definitely effective, ineffective, or adverse. To protect against this, during the period of recruitment to the study, interim analyses of major endpoints will be supplied, in strict confidence, to an independent Data Monitoring and Safety Committee along with updates on results of other related studies, and any other analyses that the DMSC may request. The DMSC will advise the chair of the TSC if, in their view, the randomized comparisons in the trial have provided both (a) ‘proof beyond reasonable doubt’ that for all, or for some, types of patient one particular treatment is definitely indicated or definitely contraindicated in terms of a net difference in the major endpoints; and (b) evidence that might reasonably be expected to influence the patient management of many clinicians who are already aware of the other main trial results. Appropriate criteria of proof beyond reasonable doubt cannot be specified precisely, but a difference of at least three standard deviations in an interim analysis of a major endpoint may be needed to justify halting, or modifying, the study prematurely. If this criterion were to be adopted, the exact number of interim analyses is of little importance, so no fixed schedule is proposed. The TSC can then decide whether to close or modify any part of the trial. Unless this happens, however, the TMC, the TSC, the investigators and all of the central administrative staff (except the statisticians who supply the confidential analyses) will remain unaware of the interim results.
The DMSC is chaired by Professor S Jackson (pharmacology, prescribing, clinical trials), King’s College London. Other members will be Professor T Robinson (stroke physician, clinical lead of the Trent Stroke Local Research Network), University of Leicester, and Dr M Lewis (statistician), Keele University.