The effect of tertiary surveys on missed injuries in trauma: a systematic review

We considered any study assessing a TTS using randomized or quasi-randomized trials, observational studies such as cohort, case–control and before-and-after design studies. Subjects were trauma patients admitted to any hospital, with no limits regarding age, gender, or severity of trauma.

We included any study that used the TTS as an intervention alone or as part of a larger intervention (such as a change in hospital trauma system). The TTS was defined as a review of the admitted patient within 24 hours (or after regaining consciousness) and included at least a repeated full physical examination.

Relevant studies were identified using electronic searches of MEDLINE (1966 to December 2010) and OVID (1980 to December 2010) and the Cochrane Library Central Registry of controlled trials, without language restrictions. The following key words were used to conduct the search: tertiary survey, trauma survey, traumatology, diagnostic errors, delayed diagnosis, missed diagnosis, missed injury, prognosis and long-term outcomes. The full search strategy is contained in Additional file 1.

Two reviewers (GK and GG) independently assessed all titles and abstracts for potential relevant articles, with any disagreement adjudicated by a third reviewer (LG). We retrieved the full-text article of any reference that appeared to meet the inclusion criteria. The eligibility of the full-text articles was assessed against the criteria of a standardized form.

The following data were extracted from the studies: title, year of publication, country of study, study design, number of participants, age and gender of participants, injury severity score (median ISS, proportion with ISS>15), mechanism of trauma (blunt vs. penetrating), presence of an altered level of consciousness and admission to intensive care unit (ICU). The outcome parameters on missed injury rates and long-term outcomes were collected when available. Authors were contacted in order to obtain missing data. We attempted meta-analysis to quantify and summarize results, but due to the inherent bias of the studies and the extent of the heterogeneity [18, 19], meta-analysis was deemed invalid and hence not reported. Primary and secondary outcomes were all proportions. Results of studies were pooled using simple weighted averages. Chi-square test was used to test for differences in proportions.

Since we anticipated possible differences in outcomes for certain demographic groups, we defined potential subgroups for analysis a priori, which included; age, gender, ISS (ISS>15), injury mechanism (blunt vs. penetrating), altered level of consciousness and ICU admission.

Two of us (GK and GG) independently used the Newcastle-Ottawa Scale [20] to assess the quality of non-randomized observational studies (Additional file 2). We classified studies to either low, moderate or high risk of bias if there were respectively up to 1, 2–3 or >3 inadequate items.

This systematic review was conducted to conform to the PRISMA standard (http://​www.​prisma-statement.​org).

Nine [3, 12, 13, 21‐25] of the included studies used Type I missed injury as outcome (missed at primary and secondary survey and detected by TTS), with the remaining single other study [11] utilizing Type II missed injury (injury missed by both initial assessment and the TTS).

The overall Type I missed injury rate in cohorts with a TTS conducted was 4.3% (Table 2). The two largest studies [12, 23] had similar Type I missed injury rates (1.3% and 1.6%, respectively), while a medium-sized study [21] had a Type I missed injury rate of 6.2% (unpublished data provided by author). The remaining studies reported Type I missed injury rates varying from 9.3 to 19.3%, with one small study [8] being an outlier (65%). In this study, a dedicated trauma surgery fellow aware of the purpose of the study performed all tertiary surveys.

Only one study reported Type II missed injury, with a rate of 1.51% after TTS introduction [11] Table 3.

Two studies [3, 21] compared Type I missed injury and one study [11] compared Type II missed injury using before and after studies. Type I missed injuries increased as a result of the TTS (3% vs. 7%, P < 0.01), and type II missed injuries decreased (2.4% vs. 1.5%, P = 0.01).

Table 4 summarizes the anatomical areas of missed injury, change in management and resultant morbidity or mortality as reported by the individual studies. There was a large variation in anatomical distribution of missed injuries, with orthopaedic extremity injuries, spine related injuries and facial injuries most commonly reported. The proportion of patients with missed injury requiring surgical intervention varied between 10-30%, which equates to 0.1-5% of the total population of these studies [3, 8, 12, 13, 22, 23]. Only two deaths were reported specifically as a result of a missed injury [8].

Only one study investigated a difference in missed injury rates after introduction of a TTS for any of the pre-defined subgroups [11]. It reported a decrease in Type II missed injuries in patients admitted to a trauma intensive care unit (5.7% vs. 3.4%, P < 0.05). Another reported a paediatric trauma population (Type I missed injury rate: 15.8%), but did not specifically investigate age as a factor of interest [13]. A third study reported a lower mortality trend (5.4% vs. 4.1%, P = 0.17) associated with a higher detection of Type I missed injuries by TTS, introduced as part of a trauma service (3.6% vs. 6.2%, P < 0.01, via author communication) [21].

No other included study assessed differences in (any type of) missed injury rate after introduction of a TTS for the other pre-defined subgroups (age, gender, ISS, mechanism of injury or altered level of consciousness).

No studies that assessed the effect of the TTS on long-term health outcome were identified.

Data for the two studies included in the analysis of the systematic review (using missed injury rate at initial assessment, detected by TTS) were not complete in the original publications. One quoted a historical overall missed injury rate of 2% (without data on sample size) [3] and another did not report missed injury data in the published manuscript [21]: we obtained this information by writing to the authors directly.

Our findings were based on weak evidence: there were no relevant randomized studies, so only observational studies with their inherent risk of bias were available. Meta-analysis was attempted, but due to few studies being eligible and being prone to bias as well as substantial heterogeneity, we have not reported this. We were unable to assess the effect of TTSs on morbidity, so any improvement in patient outcomes consequent on reduction in missed injuries has to be inferred.

Other shortcomings included variation in the trauma patient populations (one study including paediatric trauma patients only [13]); geography (with one study conducted in Uganda [24] where trauma patterns and trauma care may be different from those of the other studies); and in the intervention (differently defined in two studies - one study [22] aimed to decrease missed injuries by formalizing the radiology review component of the TTS, while another [21] assessed the effect of implementing a complete trauma service, of which a TTS was a component).

The definition of missed injury varied between studies: one study defined missed injury as any injury that escaped detection at time of the TTS [11], (i.e. Type II). The other nine studies, including two before-and-after studies, used the more commonly used ‘any injury missed by primary and secondary survey, and detected as a result of the TTS’ (i.e. Type I), which really represents a delayed diagnosis (or increase in injury detection). For example, in one study [21] the TTS was associated with increased missed injury rate, but reduced mortality. This seems counter-intuitive until one realises this is related to Type I missed injuries, leading to more frequent detection at 24 hours by TTS.

This systematic review highlights several issues. Firstly missed injury needs a consistent, clear and expanded definition to facilitate future research and provide a tool for benchmarking. The use of these differences in definition precludes overall comparison of studies and need to be made explicit in order to legitimately compare studies. We propose a classification of missed injuries (Table 5). It is likely that the third group in this proposed classification (Type III: missed injury detected after hospital discharge) has been under-reported, since there are no published data.

Secondly, for the nine studies using the Type I missed injury definition, a mean injury detection rate of 4.3% was found. This can be used as a yardstick to compare future studies assessing Type I missed injuries. This may yet be an over-estimation of Type I missed injuries (or rather, delayed diagnoses), since the reported missed injury rate was approximately 1.5% in the two larger studies that together included more than 9,000 subjects and where investigator bias would have been minimal.

Furthermore, we summarized the anatomical distribution of missed injuries and how this changed management as reported by the individual studies (Table 4). This relates to the clinical relevance of these injuries, since it would be reasonable to expect at least delayed recovery or even prolonged morbidity without these interventions. Very few deaths as a result of a missed injury were reported and need for a change in management in the form of surgical intervention was variable or not reported. None of the studies included in this systematic review pre-defined clinical significance in the design of the study. Three studies discussed clinical relevance of missed injuries, with the common denominator being whether the missed injury would have lead to morbidity or mortality as judged by expert opinion (Table 6). Interestingly, one study defined a clinically significant missed injury as any change in management, including ordering further imaging [22]. We did not pre-define clinically relevant missed injury as an outcome of this systematic review. The available data suggests that the current literature does not have a widely agreed definition for clinically significant missed injury, evidenced by the variable reporting and outcomes. A more consistent and reproducible approach to this issue is warranted. Very few studies related the TTS to morbidity and mortality and as such we cannot comment on the effect the TTS had on patient outcomes.

Lastly, we found no reporting of long-term outcomes after TTS. Studies have reported long-term outcomes for the multiple injured patients [26] or subgroups of patients with specific injuries [27], but not the effect of the TTS per se.