Types of studies

We included all randomised controlled trials (RCTs), regardless of publication status or language of publication.

Types of participants

Individuals undergoing abdominal or thoracic trauma surgery.

Types of interventions

The index intervention of cell salvage was compared with no cell salvage (standard care).

Types of outcome measures

Electronic searches

The Cochrane Injuries Group's Trials Search Co‐ordinator searched the following:

1. Cochrane Injuries Group Specialised Register (25 November 2014);

2. Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library) (issue 11 of 12, 2014);

3. Ovid MEDLINE, Ovid MEDLINE In‐Process & Other Non‐Indexed Citations, Ovid MEDLINE Daily and Ovid OLDMEDLINE) (1946 to 25 November 2014);

4. Embase Classic + Embase (OvidSP) (1947 to 25 November 2014);

5. PubMed (25 November 2014);

6. ISI Web of Science: Science Citation Index Expanded (SCI‐Expanded) (1970 to November 2014);

7. ISI Web of Science: Conference Proceedings Citation Index‐Science (CPCI‐S) (1990 to November 2014).

The authors searched the following:

1. The Chinese Bio‐medical Database (October 2014);

2. Clinicaltrials.gov (www.clinicaltrials.gov) (3 December 2014).

We report the search strategies used in (Appendix 1). We adapted the MEDLINE search strategy as necessary for the other databases. To the MEDLINE search strategy we added the Cochrane Highly Sensitive Search Strategy for identifying randomised trials (Lefebvre 2011) and to the Embase strategy we added the search strategy study design terms as used by the UK Cochrane Centre (Lefebvre 2011).

Searching other resources

We checked the reference lists of all relevant reviews and trials. We contacted authors of relevant trial reports in order to identify additional published or unpublished data.

Selection of studies

Two review authors (Li and Tian) independently screened the titles and abstracts of the citations identified by the search to determine which papers met the predetermined inclusion criteria. In cases of doubt or disagreement, we obtained a copy of the full article for inspection. We obtained the full text of all potentially relevant studies and independently assessed them to determine whether they met the inclusion criteria. In the event of a disagreement, we consulted a third author (Yang) to resolve the issue.

Data extraction and management

In keeping with the guidance of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a), we used a predesigned standardised study record form for data extraction. Two authors (Li and Liu) extracted data from the trial reports, consulting a third author (Yang) in the event of disagreement. We contacted investigators for missing data, where appropriate.

Our form collected the following information.

1. Administrative details ‐ titles; authors; publication details (year, volume number, issue number, and page numbers (where published) or titles, investigators, year in which the study was conducted (if not published)); and details of other relevant papers
2. Study details ‐ country, location and setting of the study; study design and details related to risk of bias within studies (e.g. randomisation, allocation concealment, blinding); inclusion and exclusion criteria; number of participants; characteristics of participants (including age, sex, type of trauma); dropouts; duration, frequency and completeness of follow up
3.Intervention details ‐ for both the cell salvage group and for that receiving standard or alternative care, with no cell salvage
4. Outcome data ‐ primary and secondary outcomes

Assessment of risk of bias in included studies

Three authors (Li, Liu and Sun) independently assessed the risk of bias of included studies according to methods suggested in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b), consulting a third author (Yang) when disagreements arose.

For the one study that met our inclusion criteria, we assessed the following items.

1.  Randomisation method (selection bias)

• Low risk of bias ‐ the method of randomisation allowed participants to have the same opportunity to receive either intervention; the investigators describe a random component in the sequence generation process, such as the use of random‐number tables, a computer random‐number generator, coin tossing or shuffling cards or envelopes, throwing dice, drawing of lots, etc.

• High risk of bias ‐ the investigators describe a non‐random component in the sequence generation process. Usually, the description involved some systematic, non‐random approach, such as by odd or even date of birth, some rule based on date (or day) of admission, hospital or clinic record number. Other non‐random approaches are used much less frequently than the systematic approaches mentioned above and tend to be obvious. They usually involve judgement or some method of non‐random categorisation of participants, such as allocation by judgement of the clinician, preference of the participant, the results of a laboratory test or a series of tests, or the availability of the intervention. If an open random allocation schedule (e.g. a list of random numbers) was used or assignment envelopes were used without appropriate safeguards (e.g. if envelopes were unsealed or non‐opaque, or not sequentially numbered), or any other explicitly unconcealed procedure, we classified the randomisation method as at high risk of bias

• Unclear risk of bias ‐ the investigators provide insufficient information about the sequence generation process to permit a judgement of low risk or high risk to be made (e.g. reporting the use of randomisation but providing no detailed information on the method used)

2. Allocation concealment (selection bias)

• Low risk of bias: participants could not foresee the randomisation method (e.g. central allocation including telephone, web‐based or pharmacy‐controlled randomisation; the use of sequentially numbered drug containers of identical appearance; sequentially numbered, opaque, sealed envelopes)

• High risk of bias: participants randomised through a method such as use of assignment envelopes without appropriate safeguards (e.g. if envelopes were unsealed or non‐opaque, or not sequentially numbered), by alternation or rotation, date of birth or case record number, or any other explicitly unconcealed procedure

• Unclear risk of bias: the investigators provide insufficient information about allocation concealment, such as alternation methods or unsealed envelopes; or studies in which there is any information indicating that the investigators or participants could have influenced the composition of the comparison groups

3. Blinding (performance bias)

• Low risk of bias: no blinding or incomplete blinding, but the review authors judge that the outcome is not likely to be influenced by lack of blinding; blinding of participants and key study personnel is ensured, and it is unlikely that the blinding could have been broken

• High risk of bias: no blinding or incomplete blinding of participants and people administering the treatment, and the outcome is likely to be influenced by lack of blinding

• Unclear risk: insufficient information is available to permit a judgement of low risk or high risk, or no useful information has been obtained from the authors

4. Incomplete outcome data (attrition bias)

• Low risk of bias: no missing outcome data; missing outcome data balanced in numbers across intervention groups; missing outcomes not enough to have a clinically relevant impact on the final results for dichotomous and continuous outcome data; missing data have been imputed using appropriate methods

• High risk of bias: reason for missing outcome data related to the true outcome, with either an imbalance in numbers or reasons across intervention groups; missing outcomes enough to induce clinically relevant bias in the results for dichotomous and continuous outcome data; inappropriate methods were used to deal with the missing data

• Unclear risk: insufficient information is available to permit a judgement of low risk or high risk

5. Selective outcome reporting (reporting bias)

• Low risk of bias: all outcomes were reported in the article (if the study protocol was available) or all expected outcomes were mentioned in the published reports (the study protocol was not available)

• High risk of bias: one or more outcomes failed to be included or were not reported

• Unclear risk: insufficient information is available to permit a judgement of low risk or high risk

6. Other biases

• Low risk of bias: no other sources of bias were identified which might be expected to affect results in any direction

• High risk of bias: sources of bias were identified and are likely to bias results

• Unclear risk: sources of potential bias were identified but it is unclear in which direction the bias might affect results

Dealing with missing data

For the one study included in the present version of this review, no data appeared to be missing (i.e. data are provided on all surviving participants (n = 15) and reasons for death of those who did not survive are given (n = 29)).

In future updates of this review, we will assess missing data and attrition rates for each included study, and the number of participants who are included in the final analysis will be reported as a proportion of all participants in the study. Reasons given for missing data will be provided in the narrative summaries and we will seek to ascertain the extent to which the results are altered by missing data. We will also assess the extent to which studies have conformed to intention‐to‐treat analysis.

Assessment of heterogeneity

Only one study was identified that met our inclusion criteria. For future updates of this review, should sufficient data become available, we will use the Chi² test to assess heterogeneity between trials and the I² statistic to assess the extent of inconsistency. We will use a fixed‐effect model for calculating summary estimates and their 95% confidence intervals (CIs) unless significant heterogeneity is present, in which case results will be calculated using a random‐effects model.

Assessment of reporting biases

Only one study was identified that met our inclusion criteria. For future updates of this review, should sufficient data become available, we plan to draw funnel plots when data from 10 or more studies are available by outcome. Funnel plots help to investigate any relationship between effect size and study precision (closely related to sample size) (Egger 1997). Such a relationship could be due to publication or related biases, or due to systematic differences between small and large studies. If a relationship is identified, we will further examine the clinical diversity of the studies as a possible explanation and described this in the text.

Data synthesis

We analysed the data available using Review Manager version 5.3. We expressed results for dichotomous outcomes as odds ratios (ORs) with 95% CIs and those for continuous outcomes as weighted mean differences (WMDs).

Subgroup analysis and investigation of heterogeneity

We planned to carry out the following subgroup analyses and these will be conducted in future updates of this review, should data become available. We intend to explore important clinical differences among trials that might alter the magnitude of the treatment effect such as:

• injury type (abdominal/thoracic trauma);

• injury severity;

• method used to wash the red blood cells;

• the use of transfusion protocols.

We have selected these factors as each has been identified as being important because they may influence a person's inclination or opportunity to receive, and possibly benefit from, cell salvage.

Sensitivity analysis

In order to assess the robustness of our conclusions in the future we plan to consider performing sensitivity analyses to assess the impact of missing data regarding important aspects of risk of bias (including allocation concealment) on reported treatment effect(s).

If significant heterogeneity still exists after subgroup and sensitivity analyses and reasons for heterogeneity cannot be found, we will report the results of the studies narratively (rather than pool data inappropriately).