Association of Multiple Rib Fractures With Pneumonia After Successful Cardiopulmonary Resuscitation: A Retrospective Cohort Study Using Propensity Score Matching

Background: Few studies have examined the impact of chest wall injury on respiratory complications after cardiopulmonary resuscitation. This is due to many confounding factors for the development of complications after cardiopulmonary resuscitation. Accordingly, we investigated the association between multiple rib fractures and the incidence of pneumonia during the post-resuscitation period after adjusting for confounding factors using a propensity score. Methods: This single-centre, retrospective cohort study enrolled adult, non-traumatic, out-of-hospital, cardiac arrest patients who maintained circulation for >48 h between June 2015 and May 2019. Rib fractures were evaluated by computed tomography on the day of hospital admission, and the association with newly developed pneumonia within 7 days of hospitalisation was analysed using propensity score matching with adjustment for variables previously reported to be risk factors for the development of pneumonia. Results: Of the 683 out-of-hospital cardiac arrest patients treated during the study period, 87 eligible cases were enrolled for analysis. Thirty-two patients had multiple rib fractures identied by computed tomography and 35 patients developed pneumonia. The presence of multiple rib fractures was signicantly associated with a higher incidence of pneumonia (propensity score-adjusted hazard ratio: 3.51; 95% condence interval: 1.59–7.72; p=0.002). Consistently, after propensity score matching, the multiple rib fracture group showed signicantly shorter pneumonia-free survival than the non-multiple rib fracture group (p<0.01). Conclusion: Multiple rib fractures are independently associated with the development of pneumonia after successful resuscitation.

cases were enrolled for analysis. Thirty-two patients had multiple rib fractures identi ed by computed tomography and 35 patients developed pneumonia. The presence of multiple rib fractures was signi cantly associated with a higher incidence of pneumonia (propensity score-adjusted hazard ratio: 3.51; 95% con dence interval: 1.59-7.72; p=0.002). Consistently, after propensity score matching, the multiple rib fracture group showed signi cantly shorter pneumonia-free survival than the non-multiple rib fracture group (p<0.01).
Conclusion: Multiple rib fractures are independently associated with the development of pneumonia after successful resuscitation.

Background
Cardiac arrest (CA) is a leading cause of death in many countries. Successful resuscitation relies on immediate chest compression [1,2]. Cardiopulmonary resuscitation (CPR), involving a high-ow fraction of hard and fast chest compressions, is associated with favorable neurological outcomes [3,4]. However, several investigators pointed out that chest compressions can cause various complications [5,6]. The most common complication is rib fracture, and various studies have reported an increase in its frequency as guidelines have been revised and hard and fast chest compressions emphasised [6][7][8].
During the post-resuscitation period, patients develop a systemic in ammatory condition de ned as post-CA syndrome, during which time they are highly susceptible to infection [9][10][11] and generally develop pneumonia [9,12,13]. Previous studies indicated that pneumonia is common following out-of-hospital CA (OHCA) [10,12], with a reported frequency of 29-70% [9]. Several factors, including mechanical ventilation and therapeutic hypothermia, have been associated with the development of pneumonia among resuscitated CA patients [9][10][11][12][13][14][15]. It has been reported that rib fracture increases the frequency of pneumonia and worsens the prognosis, while surgical interventions on the chest wall improve prognosis among patients with blunt chest trauma [16]. However, the association of rib fracture and pneumonia in patients with CA during the post-resuscitation period has not been investigated.
We hypothesised that multiple rib fractures after resuscitation may be an independent risk factor for the development of pneumonia during the post-resuscitation period. However, it is di cult to accumulate samples because of the frequent updates (every 5 years) in the guideline recommendations regarding chest compression. Moreover, the incidence of rib fractures and pneumonia can vary between the guideline periods and institutions. This limits the variables that can be adjusted by multivariate analysis.
To solve this problem, we investigated its causality among post-resuscitated patients by using the propensity score as a representative of known factors associated with the development of pneumonia.

Study design
This single-centre, retrospective, observational study was conducted to investigate the association between CPR-related rib fractures and the development of pneumonia during the post-resuscitation period. The study was approved by the institutional ethics committee of Nara Medical University, Nara, Japan.

Data collection
We investigated patients with OHCA admitted to our facility between June 2015 and May 2019. Data were obtained from hospital data sets recorded in the Utstein style [17]. The inclusion criteria were: age ≥ 18 years; diagnosis of non-traumatic CA; and achievement of sustained spontaneous circulation or extracorporeal circulation for > 48 h. The exclusion criteria were: con rmed pre-existing pneumonia by chest computed tomography (CT) on hospital arrival; lack of chest CT evaluation after hospital admission; and missing data regarding time events for CPR.
We collected characteristics and the potential risk factors associated with pneumonia after CPR, including age ≥ 75 years, sex, total CPR time (including bystander CPR), therapeutic hypothermia (TH), use of muscle relaxants, use of prophylactic antibiotics within 24 h of CA, use of mechanical ventilation, and number of rib fractures.

Diagnosis
Rib fracture was diagnosed through chest CT performed before admission to the intensive care unit. All CT images were acquired with a 64-row helical CT system (Optima CT660; GE Healthcare, Waukesla, WI, USA). The scan variables were as follows: 120 kVp; auto mA; rotation time, 0.5 s; helical pitch, 0.531; image noise, SD10. Coronal, sagittal, and three-dimensional reconstructed images were captured and used for diagnosis. We included all types of rib fractures: with or without displacement, incomplete, and Buckle [18] type. Previous studies reported that most rib fractures after resuscitation were bilateral and multiple [18][19][20][21]; thus, we de ned three or more rib fractures as multiple rib fractures.
We evaluated the development of pneumonia during the rst 7 days of hospitalisation [13]. According to commonly used diagnostic criteria [22], pneumonia was diagnosed by the presence of clinically relevant ndings at auscultation, decreased oxygenation that could not be explained by pulmonary oedema or atelectasis, new consolidation on a chest radiograph persistent for ≥ 48 h, and an increase in airway secretions.
Rib fractures and pneumonia were diagnosed by two independent emergency physicians with > 8 years of experience. Any diagnostic disagreements were resolved by consensus between the physicians.
In this study, the term ail chest was de ned as three or more consecutive rib fractures, in two or more locations, creating a ail segment [16].

Post-resuscitation care
Patients were managed with sedation, analgesia, and mechanical ventilator support according to our standardised protocol, which adheres to resuscitation guidelines [23].

Statistical methods
Categorical and continuous variables are expressed as n (%) and median (interquartile range), respectively. The diagnostic agreement rate for multiple rib fractures using CT examination was evaluated using the kappa value. Owing to the low number of enrolled patients with the outcome of interest, we conducted the analysis using propensity score matching to adjust for observed confounding factors. To calculate the propensity score, we constructed a logistic regression model including risk factors for the development of pneumonia as independent variables, such as age, sex, total CPR time [24], mechanical ventilation [11][12][13], TH [25], and use of muscle relaxants [26] and prophylactic antibiotics [13,27]. Age and total CPR time were categorised (≥ 75 years and CPR time duration over its median value or less). Initially, we evaluated the longitudinal association between the presence of multiple rib fractures and the subsequent incidence of pneumonia using hazard ratios (HRs) adjusted for the propensity score based on a proportional model.
Next, we performed one-to-one propensity score matching between the multiple rib fracture and nonmultiple rib fracture groups. Using a calliper matching method, each patient in the multiple rib fracture group was matched with a patient in the non-multiple rib fracture group without replacement, with the closest estimated propensity within a calliper (≤ 0.5 of the pooled standard deviation of propensity scores). The C-statistic was calculated to evaluate the goodness of t. We examined the balance in baseline variables between the propensity-matched pneumonia and pneumonia-free groups using standardised differences, with > 10% indicating imbalance. Furthermore, we compared the period until the onset of pneumonia between the propensity score-matched groups using the Kaplan-Meier method and estimated the p-value based on the strati ed log-rank test. A p-value < 0.05 denoted statistically signi cant difference. R version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria) was used for the statistical analysis.

Results
During the study period, 683 patients with OHCA were hospitalised after resuscitation. Initially, we excluded patients aged < 18 years and with CA following trauma. Next, patients who expired within 48 h and who had con rmed pre-existing pneumonia through chest CT at hospital arrival were excluded. We further excluded 12 patients with missing time data or who did not undergo chest CT. Finally, 87 patients were included in the study analysis (Fig. 1).
Risk factors for pneumonia at baseline in the study participants before and after propensity score matching are shown in After propensity score matching, the multiple rib fracture group showed signi cantly shorter pneumoniafree survival than the non-multiple rib fracture group (p < 0.01) (Fig. 2).

Discussion
The frequency of rib fractures in this study was 48.3%, and most were bilateral, multiple, and severe. By analysing the propensity score-matched cohort, we found that multiple rib fractures after successful resuscitation were associated with the development of pneumonia within 7 days after hospital admission independent of other risk factors for pneumonia. To the best of our knowledge, while numerous studies have reported on rib fracture caused by chest compression [5][6][7][8] and the development of pneumonia after resuscitation [9][10][11][12][13][14][15], none have ever addressed this issue.
As the result of a dedicated effort to improve the survival outcome of OHCA, a signi cant number of resuscitated OHCA patients may develop pneumonia following multiple rib fractures. The management of pneumonia during the post-resuscitation period is critical. Pneumonia can prolong the duration of mechanical ventilation and length of intensive care unit stay, and may worsen the prognosis [12]. Several studies have reported various factors associated with the development of pneumonia, such as mechanical ventilation, TH, use of muscle relaxants, and lack of prophylactic antibiotics [9][10][11][12][13][14][15]. However, these factors are commonly present in resuscitated patients, as in our study participants. When the study participants were compared according to the presence or absence of multiple rib fractures before propensity score matching, the aforementioned factors were more frequently observed in the multiple rib fracture group. Because the incidence of pneumonia was also higher in this group, it appears that these risk factors were related to the development of pneumonia. However, after adjustment for these factors using propensity scores, our analysis indicated that multiple rib fractures were independently associated with the development of pneumonia during the post-resuscitation period.
The mechanisms through which rib fractures cause pneumonia are generally explained by pain [28], decreased vital capacity [29], and changes in chest wall dynamics that distort the movement of chest wall muscle [30]. The frequency of pneumonia increases with the number of rib fractures, particularly in elderly patients [26] and those with frail chest injuries [27]. Most of the rib fractures observed in this study were bilateral and multiple, and these patients were likely to develop pneumonia within 7 days after hospitalisation. To optimise post-resuscitation care, serious chest wall injuries (e.g., multiple rib fractures) should be carefully evaluated in resuscitated CA patients.
Early surgical xation of multiple rib fractures or ail chest due to chest trauma improves survival by shortening the duration of mechanical ventilation [19,20,31]. Similarly, early surgical xation of fractured ribs can bene t post-resuscitated OHCA patients who experience pneumonia following CPR-related chest wall injuries. Because it is almost impossible to eliminate CPR-related chest wall injuries without compromising the survival of patients with sudden OHCA, physicians need to consider this CPRassociated complication when treating resuscitated patients with OHCA to optimise post-resuscitation care. Further investigation of this issue will reveal its impact on communities and assist in establishing a surgical indication for CPR-related rib fractures.
Our study has several limitations. Firstly, the study was conducted in a single institution over a short observation period to minimise the heterogeneity of the study population. The 2015 resuscitation guideline was applied to our study. In addition, the diagnostic criteria for rib fractures and pneumonia may differ among institutions or investigators. For this reason, we had to establish the diagnostic criteria for this study. Consequently, the sample size is small, and generalization of the present is di cult. Hence, a multicentre study under the same setting is warranted. Secondly, due to the small number of study participants, we were unable to adjust for many confounders using multivariate analysis. Therefore, these ndings were obtained through calculation of propensity scores using variables with a small number of risk factors. Although we examined as many risk factors for the development of pneumonia as possible, the possibility of unknown confounding factors cannot be completely ruled out. Differences in patient background between the two groups after propensity score matching were assessed using standardized differences. Although the factors for witnessed and initial shockable rhythm were not fully adjusted for, the lack of difference in the duration of CPR is not expected to have a signi cant impact on the results of this study. Thirdly, clinical ndings (e.g., pain) may not be re ected in the diagnosis of rib fractures because we studied sedated patients after resuscitation and evaluated only imaging ndings. However, with careful diagnosis, the frequency of rib fractures in the present study was similar to that observed in other studies [7].

Conclusion
This study using propensity score-matching showed that rib fractures caused by successful resuscitation increased the incidence of pneumonia after hospitalisation. Consideration of the development of pneumonia in OHCA patients with chest wall injury may optimise post-resuscitation care. Abbreviations CA Cardiac arrest; CI:Con dence interval; CPR:Cardiopulmonry resuscitation; HR:Hazard ratio; OHCA:Out-ofhospital cardiac arrest; TH:Therapeutic hypothermia Declarations Ethics approval and consent to participate The study protocol was approved by the research ethics committees of the Nara Medical University, Japan (decision no. 2328). Given the retrospective observational design of the study, the requirement for participant consent was waived.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author upon reasonable request.
Competing interests YK designed the study, collected and analysed the data, and prepared the manuscript. HF revised the manuscript. KS provided support for the analysis of the data. KT, KM, KY, and YT assisted with data collection and critically revised the manuscript. All other authors actively reviewed and approved the nal manuscript.