Search results

A total of 1284 records were retrieved (828 from EMBASE, 204 from Medline, 242 from Web of Science, and 10 from Cochrane Central; Table 1). Figure 1 shows details on reasons for exclusion of identified records. After removal of duplicate records (n = 331), 953 records remained for eligibility screening. Of these 953 records, 736 were excluded based on title and abstract screening, leaving 217 records for full text analysis. In total, 20 studies remained for inclusion. A manual search of the references of the included studies yielded no new records. The most common reasons for exclusion were that studies did not report a classification system (n = 512), were case reports or case series (n = 187), or included ≥ 20% patients aged < 18 years (n = 129).

Table 1 Search strategy per databaseFig. 1Quality assessment and evaluation of publication bias

The average MINORS score for the included comparative studies was 16 out of 24 (range 13–18) and 8 out of 16 (range 5–11) for non-comparative studies. Studies scored lowest on the items prospective data collection, unbiased assessment of the study endpoint, and prospective calculation of the study size, as well as baseline equivalent of groups for comparative studies. Table 2 provides a full overview of scores per MINORS item. As far as the sample sizes allow, the funnel plots did not raise concern for publication bias (supplementary data). Funnel plots of the No-PC group however had fewer studies, thus possibly not having sufficient power to detect possible publication bias.

Table 2 MINORS score per studyStudy characteristics

Detailed study characteristics are shown in Table 3. Six studies compared patients with pulmonary contusion (PC) to patients without PC (no-PC) [11,12,13,14,15,16]. The remaining 14 studies only included patients with PC [17,18,19,20,21,22,23,24,25,26,27,28,29,30]. The exact timing at which initial and potential repeat CT scans were made were not reported in most studies. Most studies mentioned simply that admission CT’s were used. Two studies reported on CT’s made within 6 h, and four studies reported on CT’s made within 24–72 h after injury. All studies were developmental in nature, i.e. they developed novel models of PC classification.

Table 3 Overview of included studies classifying pulmonary contusion

Three of the comparative studies subdivided the PC group. Deunk et al. described one group where PC was only visible on CT and one group where PC was visible on both CXR and CT [13]. Miller et al. subdivided PC patients based on severity of PC, where mild PC was defined as 1–19% contused lung volume and severe PC was defined as ≥ 20% contused lung volume [15]. Zingg et al. included one group where PC was categorized based on the Abbreviated Injury Scale (AIS) and one group where PC was categorized based on the Blunt Pulmonary Contusion Score 18 (BPC-18) [16].

The study by De Moya et al. had both a retro- and prospective part. In the retrospective part, the authors identified predictors for the need for mechanical ventilation. The prospective cohort was subsequently used to test these predictors [12].

Patient and injury characteristics

The total number of patients per study varied from 49 patients to 148,140 patients. The majority of patients with and without PC across all studies was male (66.2%, 95% CI 60.7–71.5% for no-PC patients and 75.4%, 95% CI 72.2–78.4% for PC patients), however the non-overlapping nature of the 95% CI’s is convincing evidence of difference between the two groups. The mean age ranged from 39.0 to 64.5 years for no-PC patients (50.3, 95% CI 35.1–65.5) and from 31.0 to 55.1 years for PC patients (41.9, 95% CI 38.1–45.6). Given that the 95% CI’s of the pooled estimates overlaps, but do not include both pooled estimates, there is inconclusive evidence of a difference in age between both groups. Between 40.8% and 76.7% of patients with PC had bilateral PC (Table 3).

Injury characteristics per study are detailed in Table 4. The mean ISS ranged from 8.3 to 29.9 for no-PC patients and from 13.3 to 34.0 for PC patients. Rib fractures and flail chest was reported in 37.5–100.0% and 0.0–3.8% for no-PC patients, and in 61.3–100.0% and 0.8–19.4% for PC patients. Pneumothorax was reported in 4.4–41.2% for no-PC patients and in 3.3–82.6% for PC patients. Hemothorax was reported in 0.0–28.9% for no-PC patients and in 3.4–90.3% for PC patients. Pooled estimates for each variable are shown in Table 4. In general, patients with PC had a higher ISS, more often a flail chest, pneumothorax, or hemothorax, and similar rates of rib fractures than patients without PC. The 95% CI’s of the pooled estimates for pneumothorax do not overlap, indicating a convincing evidence of a difference between both groups. However, for ISS, flail chest, and hemothorax, the 95% CI’s overlap but do not include the pooled estimate. This signifies that there is inconclusive evidence of a difference between the groups for these variables. For number of rib fractures, the 95%-CI’s overlapped and also included the pooled estimate, signifying no evidence of difference between groups. All pooled estimates were derived from Forrest plots, which are shown in the supplementary data.

Table 4 Injury characteristicsOutcome measures—classification systems and severity of contusion

Detailed outcomes on classification systems, in-hospital outcomes and complications can be found in Table 5. Fourteen studies calculated the percentage of contused lung volume based on a variety of different volumetric analysis methods [13,14,15, 17,18,19,20,21,22,23, 25, 27, 29, 30]. Sturmwasser et al. created a CT Volume Index (CTVI) based on pixel analysis of CT scans [28]. De Moya et al., Zingg et al., and Mommsen et al. all used, among others, BPC-18 [12, 16, 24]. Mommsen et al. also incorporated percentage lung volume, AIS, and the Thoracic Trauma Severity Score (TTS) [24]. Pal et al. calculated uninvolved lung volume [26], and Choi et al. looked at the presence or absence of unilateral and bilateral contusion [11].

Table 5 Details on classification and risk factors for in-hospital outcomes and complications

Four studies reported a cut-off value for severe contusion, three of these were based on percentage lung contusion and one was based on BPC-18 and chest-AIS. Severe contusion was defined as either ≥ 20% contusion volume or a chest-AIS ≥ 3 or BPC-18 ≥ 3.

Outcome measures—pulmonary complications and in-hospital outcomes

Patients with between ≥ 18% to ≥ 24% contused lung volume, chest-AIS ≥ 3 or BPC-18 ≥ 3 had increased HLOS compared to patients with less PC (Table 5) [13, 15,16,17, 28]. Similarly, ICU-LOS was longer in patients with ≥ 18–20% contused lung volume or with a chest-AIS ≥ 3, BPC-18 ≥ 2 or TTS ≥ 2 compared to patients with a lower contused lung volume [13, 16, 24, 28]. In most studies pneumonia was associated with between ≥ 18% to ≥ 24% contused lung volume [13, 17, 21, 28, 29]. One study found an association with pneumonia for patients with mild contusion, defined as 1–19% contused lung volume [15]. ARDS was associated with between > 8.1% to ≥ 24% contused lung volume, as well as BPC-18 ≥ 2 or TTS > 9 [13, 17, 22,23,24,25, 29]. Longer duration of mechanical ventilation was associated with between ≥ 19% to ≥ 24% contused lung volume, as well as BPC-18 ≥ 2 or TTS ≥ 2 [15,16,17, 22, 24, 28].

Four studies did not report specific thresholds related with outcomes [11, 14, 18, 27]. Li et al. found that presence of any contusion was associated with HLOS, ICU-LOS, ARDS, mechanical ventilation, and duration of mechanical ventilation [14]. Similarly, Choi et al. found that presence of any contusion was associated with HLOS, and that presence of bilateral contusion was associated with pneumonia and need for intubation [11, 31]. Sarkar et al. found a positive association between percentage contused volume and ARDS [27]. Similarly, Choi et al. found a positive association between contusion volume and increased HLOS and need for mechanical ventilation, where each higher quartile percentage contusion was associated with higher odds of these adverse outcomes [18].