Eligibility criteria
Study types
Controlled, uncontrolled randomized trials and cohort studies will be included in which HFNO is used as the intervention and low flow oxygen and/or NIV are used as comparators.
Types of participants
The participants will be adult population, ≥18 years old, with AT2RF that have been managed by HFNO. AT2RF will be defined as acute respiratory failure with PaCO2 of >6 Kpa (2). The study setting will be limited to where the treatment is applied in an acute care setting (emergency department, respiratory ward or critical care units). Studies where HFNO is used as a weaning modality or in peri-operative settings will be excluded or where HFNO is used for patients with reversible conditions like, toxicity or drug overdose will be excluded.
Types of interventions
HFNO will be defined as application of a flow rate of ≥20 L/min, delivered by nasal cannula. Interventions included will be trials involving HFNO with at least one comparator that can be low flow oxygen (flow rate < 20 L/min), NIV or both.
Outcomes
The primary outcomes for this review is the Change in PaCO2 after HFNO application (measured at time points reported by authors).
The secondary outcomes, where available will be
- Respiratory parameters including, pH, PaO2, dyspnoea score, tidal volume and minute volume (measured at time points reported by authors).
- Mucous clearance (patients’ needing mucous clearance before, during or after HFNO application).
- Level of consciousness.
- Patient comfort (patients’ tolerance of HFNO compared to other intervention).
- Intubation rate (intubation after HFNO).
- Length of stay in hospital.
- Post-discharge COPD exacerbation rate.
- Readmission rate secondary to AECOPD.
Search strategy
We will identify trials that examined the efficacy of HFNO in adult acute AT2RF patients. Searches for relevant studies will be conducted on the electronic databases MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials, using no language restrictions. In addition, we will search google scholar and references of all articles for any pertinent studies that were not found in our initial search. With the assistance of a professional librarian, we developed a systematic search strategy using appropriate keywords and MeSH term (see Additional file 1). Medline search strategy will form the basis for the other databases search.
Data collection and analysis
Selection of studies
A systematic review software (Covidence) will be used to store citations and removing duplicates (29). Study screening will undergo different stages by two reviewers independently. First, title and abstract review will be conducted and the studies which does not meet the inclusion criteria will be removed. Any disagreement will be resolved by discussion and referred to a third reviewer. Second, potentially eligible studies will undergo full text review to confirm inclusion. Any disagreement will be resolved through discussion with a third reviewer.
Data extraction
Two reviewers using a bespoke data extraction form will extract data independently (additional file 2). From eligible studies, information will be extracted on the type and setting of the study, recruitment information, participant characteristics (including age and underlying conditions), inclusion criteria, nature of the intervention(s) in each group (e.g. flow rate and method of delivery), time-points of measurements and outcomes as described above. Any disagreements will be resolved through discussion within the review team. If any of the required data are not available or insufficient from the study publications, they will be requested from the authors of the trial.
Assessment of risk of bias
For included studies, two reviewers will independently assess risk of bias using the Cochrane Risk of Bias tool for randomized trials and Newcastle-Ottawa scale for cohort studies (see Additional file 2) (30,31). Risk of bias for randomized trials will be assessed for the following domains:
- Random sequence generation.
- Allocation concealment.
- Blinding of participants and personnel.
- Blinding of outcome assessment.
- Incomplete outcome data.
- Selective outcome reporting.
- Other biases.
Risk of bias for cohort studies will be assessed for the following domains:
Each potential source of bias will be marked as high, low or unclear with justification. We will construct a ‘Risk of bias’ table in RevMan 5.3 to present the results (32). We will use the assessment of risk of bias to perform sensitivity analyses based on methodological quality, as necessary. To avoid bias within this systematic review, the review will be conducted using this protocol and any variations will be recorded in a specific section of the final published review.
Measure of effect
For binary variables, data will be calculated as odds ratios (ORs) with 95% confidence intervals (CIs). Data for continuous variables will be calculated as mean differences (MDs or as standardised mean differences (SMDs) with 95% CIs. Where the trial outcomes are sufficiently similar, meta-analysis will be performed and statistical heterogeneity examined between these similar trials. Heterogeneity will be evaluated using the I2 statistic, with a value of >50% possibly implying substantial heterogeneity (31). The pooled estimate will be calculated using the fixed effects model unless there is significant heterogeneity when a random effects model will be used. The outcomes will be presented in forest plot figures. We will undertake two comparisons: trials that compare HFNO versus; low flow rate oxygen and trials that compare HFNO versus NIV. Trials with NIV comparators will be analysed separately due to the established efficacy of NIV in reducing PCO2. When data cannot be meta-analysed, we will present a narrative synthesis and present data in tables.
Unit of analysis
The unit of analysis will be the patient/participant. We anticipate that all trials will have a parallel group design and thus no adjustment will be necessary for crossover or clustering.
Subgroup analysis and sensitivity analysis
If sufficient studies are available, we will undertake subgroup analyses for all outcomes against specific patient conditions such as COPD, neuromuscular disorders, interstitial lung disease. We will perform a sensitivity analysis to investigate the effect on the primary outcome of excluding trials with high risk of bias.
Summary of finding tables
We will assess the quality of the evidence associated with HFNO for AT2RF by using the principles of the GRADE system (31,33). Evidence we be will assessed for methodological quality, directness of evidence, data heterogeneity, precision of effect estimates, and risk of publication bias. We will present the findings using a ’Summary of findings’ (SoF) table constructed in RevMan 5.3 (34).
Standards
Reporting will conform to systematic review and meta-analysis PRISMA checklist preferred reporting items (see Additional file 3) (35). This systematic review has been registered with PROSPERO, an international prospective register of systematic reviews (https://www.crd.york.ac.uk/prospero/).