Objectives Multiple systematic reviews have reported on the impact of rib fracture fixation in the presence of flail chest and multiple rib fractures, however this practice remains controversial. Our aim is to synthesise the effectiveness of surgical rib fracture fixation as evidenced by systematic reviews.
Design A systematic search identified systematic reviews comparing effectiveness of rib fracture fixation with non-operative management of adults with flail chest or unifocal non-flail rib fractures. MEDLINE, EMBASE, Cochrane Database of Systematic Reviews and Science Citation Index were last searched 17 March 2017. Risk of bias was assessed using the Risk Of Bias In Systematic reviews (ROBIS) tool. The primary outcome was duration of mechanical ventilation.
Results Twelve systematic reviews were included, consisting of 3 unique randomised controlled trials and 19 non-randomised studies. Length of mechanical ventilation was shorter in the fixation group compared with the non–operative group in flail chest; pooled estimates ranged from −4.52 days, 95% CI (−5.54 to –3.5) to −7.5 days, 95% CI (−9.9 to –5.5). Pneumonia, length of hospital and intensive care unit stay all showed a statistically significant improvement in favour of fixation for flail chest; however, all outcomes in favour of fixation had substantial heterogeneity. There was no statistically significant difference between groups in mortality. Two systematic reviews included one non-randomised studies of unifocal non-flail rib fracture population; due to limited evidence the benefits with surgery are uncertain.
Conclusions Synthesis of the reviews has shown some potential improvement in patient outcomes with flail chest after fixation. For future review updates, meta-analysis for effectiveness may need to take into account indications and timing of surgery as a subgroup analysis to address clinical heterogeneity between primary studies. Further robust evidence is required before conclusions can be drawn of the effectiveness of surgical fixation for flail chest and in particular, unifocal non-flail rib fractures.
PROSPERO registration number CRD42016053494.
- rib fracture
- flail chest
- multiple rib fractures
- internal fixation
- systematic review
- mechanical ventilation
- length of hospital stay
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
- rib fracture
- flail chest
- multiple rib fractures
- internal fixation
- systematic review
- mechanical ventilation
- length of hospital stay
Strengths and limitations of this study
Multiple databases were searched for studies and study selection was undertaken by two researchers, reducing the risk of error.
Risk of bias of studies was assessed using the ROBIS tool by two researchers.
Primary research within the reviews was mapped to identify if the reviews were based on the same primary evidence.
Reviews were included regardless of risk of bias.
Multiple rib fractures (adjacent unifocal fractures) and flail chest (three or more adjacent bifocal rib fractures with paradoxical chest wall movement) have high mortality (18.7%) due to the associated complications of acute respiratory distress syndrome, pneumonia and haemorrhage.1 Chest trauma accounts for 15% of all trauma admissions2 most commonly high-energy transfer injury as a result of a road traffic accidents (57.01%) but also from low energy falls (22.96%).1 Flail chest in particular has a high mortality rate as chest wall disruption causes an increase in the work of breathing for patients who are often in significant pain.
Current treatment options for severe chest injury are mainly supportive, including multimodal analgesia and anaesthesia, as well as non-invasive and invasive ventilation.3 Surgical fixation is thought to be beneficial to patients with respiratory failure,4 intractable pain5 or if failing to wean from invasive ventilation secondary to chest trauma.6 7 Fixation has potential to restore chest wall biomechanics and reduce the complications associated with poor ventilation and secretion clearance.7 Due to the rapidly, although heterogeneous, growing evidence base from multiple systematic reviews it is essential to synthesise evidence for this intervention to ascertain safety and efficacy.
This report is part of a wider systematic review to (1) identify and synthesise the evidence of the effectiveness of surgical rib fracture fixation, (2) evaluate the evidence for indications and timing of fixation and (3) identify the outcomes reported in the literature. Systematic reviews and primary studies were eligible for inclusion. This paper maps and synthesises this evidence from systematic reviews assessing the effectiveness of fixation of rib fractures.
The review was undertaken systematically using the methods described by the Centre for Reviews and Dissemination.8
Adults (over 18 years) who have sustained one or more rib fractures following blunt chest trauma, with or without pulmonary contusion were eligible. Single rib fracture, multiple rib fractures and flail chest injuries were included but treated as separate injuries and therefore reported and analysed separately. Patients with penetrating injuries were excluded. Studies of mixed populations with penetrating and non-penetrating injuries were included only if data were presented separately for the two groups. Surgery for chronic non-union was excluded.
Any method of internal surgical fixation such as plate or strut fixation; metal or synthetic material including intramedullary splints and suture fixation were eligible for inclusion.
External surgical fixation (traction methods, splints and Hoffman style pin and bar fixation) and non-surgical management (such as supportive ventilation, epidural and regional anaesthesia).
All outcomes were eligible (eg, mortality, pain and pneumonia). The primary outcome of interest was duration of mechanical ventilation due to the close relationship with mortality and morbidity of ventilator associated complications.
Systematic reviews were included if they specified a search strategy in at least one literature database and included primary research. No restrictions were placed on the study design of the primary studies.
Literature reviews that did not have a defined research question, search strategy or defined process of selecting articles.
MEDLINE including PreMEDLINE, EMBASE, Cochrane Database of Systematic Reviews and Science Citation Index. Clinical guidance, policy documents and relevant databases such as National Institute for Health and Care Excellence (NICE) Evidence, the UK Department of Health policy content, National Clinical Guideline Centre and Scottish Intercollegiate Guidelines Network were searched on 14 December 2016 and updated on 13 March 2017. The Conference Proceedings Citation Index was also searched for unpublished literature.
The start date for the MEDLINE searches was 1976 as that was the year that Advance Trauma Life Support was introduced internationally, incorporating new methods of resuscitation which have significantly improved outcomes.
The search strategy, developed for MEDLINE, is provided in online supplementary file 1 and was adapted to run appropriately on other databases. To identify relevant further reviews reference lists of included studies were assessed for eligibility.
Searches were downloaded into Endnote X7 (Clarivate Analytics, V.7.1 release date 2 April 2014) and de-duplicated. Two researchers (HI and EC) independently screened titles and abstracts. Any paper classified as potentially eligible by either reviewer was ordered as a full text and independently screened by both reviewers. It was originally planned to have second screening of only 50% but resources allowed for full duplicate screening. A third researcher reviewed disagreements (CM) where a consensus could not be reached between the researchers.
Extracted data included study characteristics, patient characteristics, intervention, comparator, outcome measures, duration of follow-up, effect estimates, SE and CIs as available.
One researcher completed data extraction (HI); a second researcher cross-checked 50% (EC). Discrepancies were cross-checked by both researchers at a second review and a consensus reached.
Risk of bias
Quality assessment with the ROBIS tool9 was undertaken by one researcher (HI) and checked by a second (CM). Discrepancies were resolved by discussion.
All types of internal surgical fixation were synthesised as one group. Flail chest and multiple rib fractures are considered different injuries and were synthesised separately for each outcome extracted. Each outcome was narratively synthesised including number of reviews using the outcome and effect estimates with 95% CIs from the source review. Important numerical data was presented in tables for all outcomes measured. All outcomes that were reported in the reviews were included in the report to avoid reporting bias.10 Although not fully applicable, reporting was in accordance as much as possible with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement.11
The registered protocol encompasses a larger body of work which includes synthesis of primary research for effectiveness, indications for surgery, timing of surgery and mapping of outcome measures. Only the synthesis of systematic reviews is reported here. Although all outcomes were extracted and presented in tables only those that were measured in two or more studies were narratively synthesised.
Patients were not involved in the preparation or conduct of this review.
Electronic searches identified 791 records; an additional 39 records were collected following reference checking. The full text screening identified 12 systematic reviews eligible for inclusion, there were 21 papers excluded because they were not classified as systematic reviews (figure 1). See online supplementary file 2 lists the excluded studies.
Eleven systematic reviews and one rapid evidence synthesis,12 published between 2010 and June 2016, met the inclusion criteria. Table 1 provides a summary of review characteristics. Three of the reviews13–15 were presented as best evidence topics16
Nine reviews12 13 15 17–22 evaluated the effectiveness of internal surgical fixation in patients with flail chest, two included patients with multiple rib fractures14 23 and one included all rib fractures but only reported outcomes for flail chest.24
Three reviews18–20 included only RCTs and eight included other study designs12–15 17 21–23 (two systematic reviews, 19 non-randomised studies, 11 case series and two case reports) (table 2). As would be expected, there was overlap across the reviews in the included primary studies. The total number of patients who had internal fixation in primary studies (excluding duplicate studies) was 1036, and there were 1187 controls.
The rapid evidence synthesis by NICE12 was the first review published in 2010, consisting of seven primary studies including one RCT published in 2001.25 Cataneo et al20 was the first meta-analysis published (in 2015) and included three RCTs.25–27 Two further systematic reviews published since then18 19 identified the same three RCTs and repeated the same meta-analyses for the same review question.
Risk of bias
Seven reviews rated as low risk of bias,12 15 18–22 three as unclear14 17 23 and two as high.13 24 (table 3)The high risk of bias rating was due to lack of detail in the search strategy, no attempts to minimise errors of data extraction and no quality assessment of included studies.
The only review for which a protocol was identified was the Cochrane review undertaken by Cataneo et al.20
All reviews undertook a narrative synthesis with six also including a meta-analysis.17–22Table 4 summarises the meta-analyses for flail chest, table 5 summarises the narrative syntheses for flail chest and table 6 the narrative syntheses for multiple rib fractures. Across all the reviews, 18 outcomes were reported and 11 outcomes were reported by more than one review.
Primary outcome: length of mechanical ventilation (days)
There was substantial variation across the reviews in the pooled estimates for this outcome, related to pooling different sets of studies. The largest reduction in duration of mechanical ventilation when surgery is compared with no surgery was reported by Slobogean et al21 pooling two RCTs25 26 and six non-randomised studies28–33 (mean difference [MD] −7.5 days, 95% CI −9.9 to –5.5) (table 4). The MD was 3 days greater than the pooled estimates from Leinicke et al22 (–4.52 days, 95% CI −5.54 to −3.50) and Swart et al17 (–4.57 days, SD 0.59).
There were differences in the data reported across the four meta-analyses17–20 that included the same three RCTs. Schuurmans et al18 extracted median duration from the Marasco et al RCT,27 producing an estimate MD −6.53 days, 95% CI −11.88 to –1.18. In contrast, Coughlin et al,19 Cataneo et al20 and Swart et al17 report the total mean time on mechanical ventilation which they state was obtained directly from the authors producing an estimate MD −6.30 days, 95% CI −12.16 to –0.43.
Variations also arose in relation to the extraction of data from the RCT by Granetzny et al26 who did not report SDs for mechanical ventilation. Slightly different SD values are found in all six meta-analyses17–22 which may have arisen from different methods of imputation and all give slightly different estimates.
Multiple rib fractures
Two reviews14 23 included one primary study33 that had matched non-operative controls33 (non-operative treatment not described) and reported a statistically significant reduction in postoperative ventilator days (p=0.02) in favour of the fixation group (table 6) but no statistically significant difference in total ventilator days (p=0.12).
Seven reviews reported mortality; six undertook a meta-analysis.15 17–22 Three reviews18–20 which pooled the same three RCTs25–27 showed a non-statistically significant reduction in mortality with surgery compared with no surgery (risk ratio [RR] 0.56, 95% CI [0.13 to 2.42]18 20 and RR 0.57, 95% CI [0.13 to 2.52]19 [table 4]).
Three reviews pooled randomised and non-randomised studies.17 21 22 Estimates were RR 0.19, 95% CI 0.13 to 0.2621; (RR 0.43, 95% CI 0.28 to 0.6922 and RR 0.44, SD 0.09.17 Overall, statistical heterogeneity was low (I2=0%) for this outcome in all studies that presented this data.17–22
Length of intensive care unit stay (days)
Eight reviews12 15 17–22 assessed length of intensive care unit (ICU) stay; six undertook a meta-analysis.17–22 Pooled estimates ranged from −3.25 days (SD 1.29)17 to −6.46 days, 95% CI −9.73 to –3.1919 and were all in favour of surgical fixation compared with a variety of comparators (table 4). The range in pooled estimates may be partly explained by the pooling of different sets of studies. However, differences occurred as some pooled median length of ICU stay and others pooled the mean. Furthermore, some used postoperative time spent in ICU and others the total time spent in ICU.17–19
Variation also arose across reviews in the data extracted from a trial that did not report SDs in the primary publication.26 Values were imputed or the raw data obtained from the authors resulting in SD values ranging from 0.7 to 4.4 and 2.2 to 7.3 in the operative and non-operative groups respectively. There was also a substantial difference in the effect estimate for this trial.26 In one review,22 the effect estimate of −10 days, 95% CI −15.41 to –4.59, was 5 days greater than the estimate used from the same trial in other reviews. It is the same as the as length of mechanical ventilation effect estimate reported in the same study26 so is possibly a transcription error.
Statistical heterogeneity ranged from substantial to none34 (I2=74.9%,22 40%,18 35%19 and 0.1%.21 The narrative syntheses concluded that in patients with flail chest undergoing surgical fixation length of ICU stay was reduced compared with non-operative management.15 24
Length of hospital stay (days)
Nine reviews12 15 17–22 24 reported length of hospital stay, six undertook a meta-analysis.17–22 Two reviews17 20 pooled the same two trials25 26 and found a significantly shorter hospital length of stay in favour of surgery compared with non-operative management (MD −11.39 days 95% CI −12.39 to –10.38). When non-randomised studies were included in the meta-analysis the pooled effects were smaller −3.83 days, 95% CI −7.12 to –0.5422; –4 days, 95% CI −7.4 to –0.721 and −4.48 days, SD 1.917 in favour of fixation (table 4).
Ten reviews,12 15 17–22 35 reported the risk of developing pneumonia, six undertook a meta-analysis.13 17–22 24 Three RCTs25–27 were pooled in two of the reviews15 17 and they found a RR of 0.36, 95% CI 0.15 to 0.85, in favour of fixation compared with non-operative management. When non-randomised studies were combined the RR ranged from 0.31, 95% CI 0.21 to 0.4121 to 0.45, 95% CI 0.29 to 0.7036 in favour of fixation (table 4). Substantial heterogeneity was seen in meta-analyses for this outcome18–20 that included the three RCTs25–27 (I2=66% to 74%). In the reviews that pooled the RCTs alongside the non-randomised studies21 22 there were lower levels of heterogeneity (I2=4% and I2=31%, respectively).
Five reviews reported a meta-analysis for tracheostomy.17 18 20–22 Pooled RRs ranged from 0.25, 95% CI 0.13 to 0.47 to 0.40, 95% CI 0.2 to 0.7 (table 4). Moderate and substantial heterogeneity was seen in two reviews (I2=42%,17 I2=64%),20 low in two reviews21 22 (I2=0%) and one did not report heterogeneity.18
One review,21 pooling four non-randomised studies28 30 37 38 estimated a RR of 0.14, 95% CI 0.56 to 0.23 with I2=0% in favour of fixation compared with non-operative management for sepsis. The estimate RR reported is not possible given the CI does not include the estimated value, 0.14. The lower interval of 0.56 could possibly be −0.56 creating a wider CI and would suggest that the author’s conclusion was correct and there was a statistically significant difference in favour of fixation.
One review19 reported a meta-analysis of spirometry data which included two RCTs26 27 with spirometry measured at two different time points (three and 2 months respectively). No statistically significant differences in any spirometry data were seen between surgery and no surgery (table 4).
Two reviews reported a meta-analysis of chest deformity,20 21 both reported a statistically significant difference in favour of surgery compared with no surgery (RR 0.30, 95% CI 0.00 to 0.60, I2=2.1% and RR 0.13, 95% CI 0.03 to 0.67, I2=0%).
One review21 pooled an RCT25 and two non-randomised studies.37 39 for dyspnoea (RR 0.15, 95% CI 0.09 to 0.39 in favour of fixation). Duration of follow-up was 1 year for two of the primary studies25 39 and unclear in the third.37 It was unclear how dyspnoea was measured or defined in the three primary studies.
Other reported outcomes
Several other outcomes were reported within the systematic reviews however no others have been pooled in a meta-analysis. A narrative synthesis was not completed on the outcomes: wound infection, pain-requiring removal of metalwork, return to work, socio-professional disability cost, pulmonary embolism, pneumothorax and haemothorax. In the reviews, data on these additional outcomes was minimal and presented as a narrative synthesis without presenting numerical data (tables 5 and 6).
Twelve systematic reviews on the effectiveness of surgery for flail chest and multiple rib fractures published between 2010 and 2016. This is the first systematic review of reviews and highlighted that there are a large number of reviews with same aims and including the same primary studies.
Six17–22 of the 12 systematic reviews presented meta-analyses for flail chest based on overlapping primary studies. They reported reductions in length of mechanical ventilation, length of stay, pneumonia and tracheostomy rates with surgery compared with non-surgical management and inconsistent results for mortality. Across many of the meta-analyses there was moderate to high levels of heterogeneity and variation in the effect estimates.
A single systematic review found reductions in sepsis, dyspnoea, chest deformity and chest pain with surgery compared with no surgery management. Nevertheless, as the outcome measures were not defined it is difficult to know whether the reductions are clinically significant. Reporting of adverse outcomes was infrequent across the reviews, which could reflect lack of measurement and/or reporting of adverse events in the primary studies or the systematic reviews. Therefore, the benefits of surgery could be overestimated in light of the potential risks not being considered. Synthesising multiple meta-analyses data that include overlapping primary studies has the potential to overestimate the strength of the findings therefore it is important to be mindful of the limited evidence on which our conclusions are based. In addition, significant heterogeneity for several of the outcomes that were pooled makes drawing firm conclusions difficult.
Multiple rib fractures
Evidence in support of multiple rib fracture fixation in the absence of flail chest is limited. Two systematic reviews14 23 reported on one non-randomised study33 that recruited between 1996 and 2000, four case series40–43 and two case reports.44 45 Hence, due to limited evidence no conclusive statements on effectiveness can be drawn.
A significant amount of effort and time is required to conduct a high quality systematic review and should only be undertaken when there is sufficient cause46 47 (eg, to incorporate the findings of a new RCT or to address an evidence gap). Eight of the systematic reviews were published within 18 months although none were registered on PROSPERO48 so it is possible the authors were unaware of each other’s research. Registering reviews allows transparency of methods and also reduces research waste.49 As similar search strategies and search dates were used in each systematic review, inevitably many of the included studies were the same across reviews.
Only two of the 12 systematic reviews formally appraised the quality of the included studies, therefore 10 of the reviews were not in a position to fully consider the impact of risk of bias on their conclusions. High or unclear risk of bias within reviews have affected the conclusions drawn from this evidence synthesis. In a systematic review of 106 emergency surgery systematic reviews, a low risk of bias was found in 53.8%, identifying a common problem of poor quality reviews conducted in emergency surgery.50
Heterogeneity and meta-analysis errors
The I2 value describes the percentage of total variation across studies that is due to heterogeneity rather than chance.51 Examining the meta-analyses including RCTs highlights moderate to high levels of statistical heterogeneity.
There was also clinical variation in the primary studies in terms of indications and timing of surgery and it is possible that these between study differences could be a source of the substantial heterogeneity. For example, in one RCT25 patients were randomised after 5 days of invasive ventilation, whereas another RCT26 randomised and fixed within 24 to 72 hours regardless of initial intubation state. Also, many reviews define the comparator as usual care or non-operative care but do not elaborate on what encompasses this care. Differences in how outcomes were measured may also have contributed to between study heterogeneity. It was unknown due to lack of reporting whether the outcomes were equivalent in the pooled primary studies or overall between systematic reviews.
In all systematic reviews with meta-analyses, they reported that two reviewers were involved in the data extraction to minimise errors.17 19–22 Despite attempts to minimise errors and therefore an apparent low risk of bias, some errors (up to an MD of 10 days in the measurement of length of intensive care stay) were identified across reviews. It is worth noting that there were no significant changes in the conclusions drawn from these analyses. Although there was substantial statistical and clinical heterogeneity and lack of consideration of risk of bias in many of the reviews, conclusions tended to be similar and in the direction of benefit with fixation suggesting that further high quality RCTs investigating the effectiveness (including adverse effects) of internal surgical fixation over non-operative management are warranted.
Multiple databases were searched for studies and study selection was undertaken by two researchers, reducing the risk of error and bias. Although only English language studies were included, some sources of unpublished studies were searched. A mapping of the studies included in the reviews was undertaken to take into account individual studies being included in multiple reviews and hence double counting studies.
All systematic reviews were included irrespective of their risk of bias scoring. It could be argued that several reviews were stretching the traditional definition of a systematic review however they did hold to the protocol definition with an electronic database search strategy and included primary evidence. Due to best evidence topics and rapid evidence synthesis being included it was then difficult to apply the ROBIS tool consistently. The ROBIS tool is not designed for rapid evidence synthesis and therefore this type of review showed high risk of bias as they were being assessed against a tool designed for full systematic reviews. Rapid evidence syntheses, by their nature address a trade-off between time and methodological rigour and comprehensiveness.52
The considerable duplication of work across reviews could be mitigated through protocol registration and greater attention to establishing whether a review is necessary by scoping the literature before commencing a new review. Despite this review identifying 12 systematic reviews they only included 37 unique primary studies, only three of which were RCTs. Synthesis of the reviews has shown some potential improvement in patient outcomes with flail chest after surgical intervention. However, there were differences in indications and timing of interventions in the primary studies and moderate to high levels of heterogeneity across reviews. For future review updates, meta-analysis for effectiveness may need to take into account indications and timing of surgery as a subgroup analysis to address clinical heterogeneity between primary studies. Further robust evidence is required before conclusions can be drawn of the effectiveness of surgical fixation for flail chest and in particular, multiple rib fractures.
Contributors HMAI contributed to conceptualisation, methodology, investigation, formal analysis, original draft preparation. EC contributed to investigation, validation, review and editing. WE contributed to conceptualisation, review and editing. AR contributed to funding acquisition, conceptualisation, review and editing. CH contributed to methodology, supervision, conceptualisation, review and editing. CM contributed to methodology, investigation, validation, conceptualisation, supervision, review and editing. All authors approve the final version of the manuscript and are accountable for all aspects of the work.
Funding This review was completed as part of an MD project which was funded by an educational grant from Orthopaedic Research UK. The research was undertaken at the BOA Orthopaedic Surgery Research Centre (BOSRC) at York Trials Unit.
Competing interests AR declares receiving research grants from NIHR; research and educational grants from DePuy Ltd outside the submitted work.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement All data used for the preparation of this review are reported within the manuscript or its supplementary files.
Patient consent for publication Not required.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.