BMJ Open: first published as 10.1136/bmjopen-2016-013029 on 13 January 2017. Downloaded from

PEER REVIEW HISTORY

BMJ Open publishes all reviews undertaken for accepted manuscripts. Reviewers are asked to complete a checklist review form (http://bmjopen.bmj.com/site/about/resources/checklist.pdf) and are provided with free text boxes to elaborate on their assessment. These free text comments are reproduced below.

ARTICLE DETAILS

TITLE (PROVISIONAL) Risk of in patients with isolated minor rib fractures: A nationwide cohort study AUTHORS Ho, Sai-Wai; Teng, Ying-Hock; Yang, Shun-Fa; Yeh, Han-Wei; Wang, Yu-Hsun; Chou, Ming-Chih; Yeh, Chao-Bin

VERSION 1 - REVIEW

REVIEWER Edward Baker Lecturer Practitioner King's College London UK REVIEW RETURNED 03-Aug-2016

GENERAL COMMENTS Thank you for sending me this study to review. I have the following comments that you should consider in any for development of this study and associated write up.

There is a small sample size for a „nation wide‟ cohort study. Can you justify this in your text? This may be amplified by the fact that you are reporting chest trauma rates from the USA which will be

significantly to Taiwan. http://bmjopen.bmj.com/

Your brief introduction reports demographics of more severe chest trauma and you only introduce minor rib fractures in the final paragraph. The introduction would benefit from being more specific to your study aims.

The data that you are reporting on your relatively old now (2009- 2011). Can you explain the delay in publication or why more up-to on September 28, 2021 by guest. Protected copyright. date data has not been utilised? You may find that treatment and diagnostics (particularly access to diagnostics) has increased in the last 5 years.

I remain a little confused about where the data was accessed and how this was done. I feel that you need to explain the NHIRD process more clearly for readers who may not be based in a country that involves health insurance.

Your discussion follows a very bio-chemical process for the development for pneumonia and pharmokinetic functions of NSAIDS in the management of pain. I don‟t think this adds a great deal to the overall output of the study and doesn‟t add anything to the knowledge base. The discussion is very descriptive generally.

Remember also that you are referring to studies undertaken on patients with significantly more rib fractures and therefore much higher risk of developing complications. These may not be BMJ Open: first published as 10.1136/bmjopen-2016-013029 on 13 January 2017. Downloaded from

comparable studies?

You report the NHIRD to a major strength of the study. In reality the retrospective methodology is a weakness that you should have accounted for in your limitations.

In your tables you have inconsistent presentation of p-values. Can you review this and standardise your presentation of data?

Can you review the use of English? Of the most part this is good but there are areas where there is confusion with grammar and sentence structure.

Overall I have to ask myself what this study adds to the knowledge base. There is sufficient evidence available to link the cause and effect of rib fractures and subsequent complications. There is also a large base of retrospective evidence available identifying predictive variables. The addition to the knowledge base here, lies in the focus on patients with 1 or 2 fractured ribs. There is currently limited data available in this area but without further specific data including position of rib fractures and more specific information the benefit is limited.

REVIEWER Krista L Kaups MD, MSc, FACS University of California San Francisco Fresno CRMC Department of Surgery, First Floor 2823 Fresno Street Fresno CA 93721 United States REVIEW RETURNED 17-Aug-2016

http://bmjopen.bmj.com/

GENERAL COMMENTS Thank you for the opportunity to review this manuscript. This study is a large, database analysis using propensity matching to compare patients sustaining isolated rib fractures to matched patients in a national health service database. The significant strength of this study is its comprehensive nature, with the inclusion of essentially all in-patient and ambulatory patients in a large sample (1 million

patients) of a national database. on September 28, 2021 by guest. Protected copyright. Overall, the study is well-constructed and the authours have clearly reported the methodology employed for inclusion and matching criteria, as well as descriptions of statistical analyses used. The stated objective was investigation of the association between isolated single or two rib fractures and the development of pneumonia. The authors concluded that the occurrence of two rib fractures is associated with an increased incidence of pneumonia in the subsequent 30 days following diagnosis (1.6% vs 0.18%) with age greater than 65 years and the presence of COPD adding significant risk. The manuscript raises a number of questions: 1. Data is not provided regarding mechanism of . Rib fractures resulting from high kinetic energy trauma (e.g. motor vehicle crash) would be expected to have a higher incidence of and other pulmonary trauma, and potentially greater risk for pneumonia, than fractures sustained from a low-energy trauma (such as a ground-level fall). Could mechanism of injury be incorporated into the analysis? BMJ Open: first published as 10.1136/bmjopen-2016-013029 on 13 January 2017. Downloaded from

2. The presence of or hemothorax was not included in the study. Since either of these can both indicate the presence of a more severe overall pulmonary injury and also contribute to the later development pneumonia, can the authours comment as to why this was omitted? 3. Additionally, it was reported that only 0.6% of patients had pulmonary contusion, yet since a relatively small proportion of patients had CT scans performed, the actual incidence of contusion is unknown (as the authors report). Since the presence of contusion may significantly contribute to the development of pneumonia, a comparison of patients who had rib fractures with and without contusion would be helpful (perhaps an analysis of only patients who had CT scans performed). 4. The incidence of COPD and other comorbidities was much higher in the patients with rib fractures, so is it possible that these patients are simply more frail and at increased risk for injury? COPD is clearly a risk for pneumonia so this increase is relatively unsurprising. 5. Of the patients sustaining isolated rib fractures, a relatively small percentage (29%) were admitted to the hospital. As this may represent a group of patients who were more significantly injured or at greater risk to develop pneumonia (i.e. because of co- morbidities), an analysis of these patients would allow a better understanding of risk factors for developing pneumonia in comparison to the overall group. 6. The authours note the difference in incidence of pneumonia between patients with 1 versus 2 rib fractures. Was there a further difference in patients sustaining more than 2 fractures? Why were 1 and 2 rib fractures selected for review? 7. The statement is made, in the introduction, that „pneumonia is an inflammatory process „which is true. However, for the purposes of this study, it would seem that the infection-related component should not be ignored.

8. A number of the references do not appear to directly relate to the http://bmjopen.bmj.com/ manuscript content (including 21- 24, 30 – 32 which discuss inflammatory response in pneumonia and general discussions of COPD) and could be omitted.

VERSION 1 – AUTHOR RESPONSE

Reviewer #1: on September 28, 2021 by guest. Protected copyright. Comments 1. There is a small sample size for a „nation wide‟ cohort study. Can you justify this in your text? This may be amplified by the fact that you are reporting chest trauma rates from the USA which will be significantly to Taiwan.

Reply: Thanks for your valuable comments and to improve this aspect. The database contained 1 million patients that randomly selected from 23 million Taiwanese population in year 2010. After excluding non-adult patients and patients who ever have trauma in previous 1 year, only 3602 chest trauma patients were included for analysis. Then we select patient only with isolated single or two ribs fracture in these 3602 patients. We further excluded patients who had pneumonia in previous one year before chest trauma. After that, only 709 patients left for final analysis in study group. We added the following sentences in the result section “After excluding non-adult patients and patients who ever have trauma in previous 1 year, only 3602 chest trauma patients were included” [Page 11]. We also described the Taiwan condition of chest trauma by adding reference 8 and the following sentences in the introduction section “In Taiwan, chest trauma causes 18,856 hospitalizations in between year 2002 and 2004 [8]” [Page 5]. We hope that these changes and BMJ Open: first published as 10.1136/bmjopen-2016-013029 on 13 January 2017. Downloaded from replies may meet your requirement for being published.

2. Your brief introduction reports demographics of more severe chest trauma and you only introduce minor rib fractures in the final paragraph. The introduction would benefit from being more specific to your study aims.

Reply: Thanks for your valuable comments and to improve this aspect. We deleted the description about severe chest trauma and added more description about minor rib fractures in the introduction section. We also supplemented the description as “Minor thoracic injury was defined by the presence of chest abrasion, or chest contusion, or isolated minor (IMRF), defined as single or two ribs fractures represent up to 42% of ED visits for blunt chest trauma [12-13]. Most of the patients were direct discharged from ED after primary management. Unfortunately, complications such as delay pneumothorax, hemothorax, pneumonia and significant functional limitations have even been reported [11, 14-16]. Among all kind of minor thoracic injuries,” And “Actually, IMRF can produce significant pain which impair both coughing function and clearance of secretion, leading to and subsequent pneumonia [18].” And “This relationship is clinically important because delay pneumonia after rib fractures has been shown to be significantly associated with mortality [21]” [Page 5-6]. We hope that these changes and replies may meet your requirement for being published.

3. The data that you are reporting on your relatively old now (2009-2011). Can you explain the delay in publication or why more up-to date data has not been utilised? You may find that treatment and diagnostics (particularly access to diagnostics) has increased in the last 5 years.

Reply: Thanks for your valuable comments and to improve this aspect. The Taiwan National Health Research Institutes only release the insurance database between year 1995 and December 2011 for medical research when we applied at year 2015. Now, the most up-to date database is up-to December 2013. (http://nhird.nhri.org.tw/date_03.html). We hope that these replies may meet your requirement for being published.

4. I remain a little confused about where the data was accessed and how this was done. I feel that http://bmjopen.bmj.com/ you need to explain the NHIRD process more clearly for readers who may not be based in a country that involves health insurance.

Reply: Thanks for your valuable comments and to improve this aspect, we have added the red color words in the material and method section as following “The Taiwanese National Health Insurance system was established in 1995 and covers the medical expenses of approximately 98% of the

Taiwanese population (23 million people), making it one of world‟s largest population-based datasets. on September 28, 2021 by guest. Protected copyright. The LHID2010 had a longitudinal design which contains all ambulatory and inpatient claims data including disease diagnosis code, drugs prescription, diagnostic examinations and interventions on one million beneficiaries from 23 million populations who were randomly sampled from the 2010 registry of beneficiaries of the NHIRD.” [Page 7] We hope that these changes and replies may meet your requirement for being published.

5. Your discussion follows a very bio-chemical process for the development for pneumonia and pharmokinetic functions of NSAIDS in the management of pain. I don‟t think this adds a great deal to the overall output of the study and doesn‟t add anything to the knowledge base. The discussion is very descriptive generally.

Reply: Thanks for your valuable comments and to improve this aspect, we have deleted the following description “The mechanism of NSAIDs involves the inhibition of cyclooxygenase and reduced prostaglandin synthesis. Administering NSAIDs has been shown to limit inflammation and tissue damage in the last decade [21, 22]. However, Voiriot et al found that NSAIDs might alter the course of BMJ Open: first published as 10.1136/bmjopen-2016-013029 on 13 January 2017. Downloaded from infection and result in more invasive disease and complications [23]. The aforementioned studies have shown that NSAIDs can prevent pneumonia because of their analgesic effect but not their anti- inflammation effect. Additional clinical research studies are required to investigate these mechanisms.” and related references in the manuscript. We hope that these changes and replies may meet your requirement for being published.

6. Remember also that you are referring to studies undertaken on patients with significantly more rib fractures and therefore much higher risk of developing complications. These may not be comparable studies?

Reply: Thanks for your valuable comments and to improve this aspect. Most of previous studies were investigated the relationship of severe chest trauma and delay . However, IMRF after chest trauma are commonly observed in the ED but the disposition varies among different ED settings. Research which focused on the relationship between IMRF and subsequent pneumonia is limited with small sample size. This relationship is clinically important because delay pneumonia after rib fractures has been shown to be significantly associated with mortality. We hope that these replies may meet your requirement for being published.

7. You report the NHIRD to a major strength of the study. In reality the retrospective methodology is a weakness that you should have accounted for in your limitations.

Reply: Thanks for your valuable comments and to improve this aspect. We added the following sentences in the limitation section. “At last, the weakness of retrospective methodology nature in this study should be taken into consideration.” [Page 17] We hope that these replies may meet your requirement for being published.

8. In your tables you have inconsistent presentation of p-values. Can you review this and standardise your presentation of data?

Reply: Thanks for your valuable comments and to improve this aspect. We changed all the p-values http://bmjopen.bmj.com/ to P-values. We hope that these replies may meet your requirement for being published.

9. Can you review the use of English? Of the most part this is good but there are areas where there is confusion with grammar and sentence structure.

Reply: Thanks for your valuable comments and to improve this aspect. The revised manuscript was edited for proper English language and grammar by the qualified native English speaking editors with on September 28, 2021 by guest. Protected copyright. a certification. We hope that these replies may meet your requirement for being published.

10. Overall I have to ask myself what this study adds to the knowledge base. There is sufficient evidence available to link the cause and effect of rib fractures and subsequent complications. There is also a large base of retrospective evidence available identifying predictive variables. The addition to the knowledge base here, lies in the focus on patients with 1 or 2 fractured ribs. There is currently limited data available in this area but without further specific data including position of rib fractures and more specific information the benefit is limited.

Reply: Thanks for your valuable comments and to improve this aspect. IMRF after chest trauma are commonly observed in the ED but the disposition varies among different ED settings. Our study demonstrated that patients with 2 isolated rib fractures are particularly susceptible to pneumonia. We recommended that physicians should pay attention to this complication especially in patients with over 65 years of age and COPD. We also recommended that even patients with single or two rib fractures should receive attentive follow-up care. BMJ Open: first published as 10.1136/bmjopen-2016-013029 on 13 January 2017. Downloaded from

We supplemented the following sentences in conclusion as “We recommended that even patients with single or two rib fractures should receive attentive follow-up care.” [Page 18]. We hope that these changes and replies may meet your requirement for being published.

Reviewer #2: 1. Data is not provided regarding mechanism of injury. Rib fractures resulting from high kinetic energy trauma (e.g. motor vehicle crash) would be expected to have a higher incidence of pulmonary contusion and other pulmonary trauma, and potentially greater risk for pneumonia, than fractures sustained from a low-energy trauma (such as a ground-level fall). Could mechanism of injury be incorporated into the analysis?

Reply: Thanks for your comments. The database did not include the mechanism of injury. We stated in the limitation that “Second, the NHIRD database did not provide the mechanism of injury or detailed clinical parameters such as the , , and laboratory data of the studied patients. Rib fractures resulting from high kinetic energy trauma would be expected to have a higher incidence of pulmonary contusion and other pulmonary trauma that potentially greater risk for pneumonia, than fractures sustained from a low-energy trauma” [Page 16]. We hope that these changes and replies may meet your requirement for being published.

2. The presence of pneumothorax or hemothorax was not included in the study. Since either of these injuries can both indicate the presence of a more severe overall pulmonary injury and also contribute to the later development pneumonia, can the authours comment as to why this was omitted?

Reply: Thanks for your comments. The aim of this study is to investigate patients with only isolated http://bmjopen.bmj.com/ minor rib fractures (only 1 or 2 ribs fracture) and risk of pneumonia. It is well known that chest trauma complicated with pneumothorax or hemothorax indicated a more severe pulmonary injury that may contribute to the later development pneumonia. So, we excluded patients with pneumothorax or hemothorax We further explained in the materials and methods section by adding “then excluded patients diagnosed with open chest trauma, traumatic pneumothorax and hemothorax (ICD-9-CM codes

860.1, 860.3, 860.5, 861.1, 861.3, 807.1, and 807.3) during the study period because this kind of on September 28, 2021 by guest. Protected copyright. pulmonary injury can contribute to the later development pneumonia” [Page 8]. We hope that these changes and replies may meet your requirement for being published.

3. Additionally, it was reported that only 0.6% of patients had pulmonary contusion, yet since a relatively small proportion of patients had CT scans performed, the actual incidence of contusion is unknown (as the authors report). Since the presence of contusion may significantly contribute to the development of pneumonia, a comparison of patients who had rib fractures with and without contusion would be helpful (perhaps an analysis of only patients who had CT scans performed).

Reply: Thanks for your comments. We further investigated the IMRF patients with and without pulmonary contusion that confirmed by CT. There are only 4 IMRF patients associated pulmonary contusion that confirmed by CT. None of these patients developed delay pneumonia. We also supplemented as “However, the rate of pulmonary contusion which was confirmed by CT in IMRF patients was only 0.6% (4/704) in our study. None of these patients developed delay pneumonia.” in the discussion section[Page 14]. Moreover, we supplemented the table below showing the analysis of BMJ Open: first published as 10.1136/bmjopen-2016-013029 on 13 January 2017. Downloaded from

IMRF patients with and without pulmonary contusion confirmed by computerized tomography. Because there is no pneumonia in patient with pulmonary contusion (confirmed by CT), we decided not to include this table in the manuscript. We hope that these changes and replies may meet your requirement for being published.

4. The incidence of COPD and other comorbidities was much higher in the patients with rib fractures, so is it possible that these patients are simply more frail and at increased risk for injury? COPD is clearly a risk for pneumonia so this increase is relatively unsurprising.

Reply: Thanks for your valuable comments and to improve this aspect. As COPD is clearly a risk for pneumonia, we used a 1:8 propensity score to match the control group patients according to age, sex, diabetes mellitus, hypertension, cardiovascular disease, , and chronic obstructive pulmonary disease. After that, the incidence of COPD and all above comorbidities did not have statistical different between rib fracture group and control group [Table 1]. We hope that these replies may meet your requirement for being published.

5. Of the patients sustaining isolated rib fractures, a relatively small percentage (29%) were admitted to the hospital. As this may represent a group of patients who were more significantly injured or at greater risk to develop pneumonia (i.e. because of co-morbidities), an analysis of these patients would allow a better understanding of risk factors for developing pneumonia in comparison to the overall group.

Reply: Thanks for your valuable comments and to improve this aspect, we further investigated IMRF patients with and without hospitalization in new Table 4 below.

Moreover, we added the following sentences in result section: “We further investigated IMRF patients with and without hospitalization (Table 4). IMRF patients who had been admitted to hospital were significantly have underlying co-morbidities such as hypertension (p=0.022) and cardiovascular disease (p=0.015). However, there was no significant different in risk of delay pneumonia in patients http://bmjopen.bmj.com/ who were admitted to hospital or not (p=0.313)”. We hope that these changes and replies may meet your requirement for being published.

6. The authors note the difference in incidence of pneumonia between patients with 1 versus 2 rib fractures. Was there a further difference in patients sustaining more than 2 fractures? Why were 1 and 2 rib fractures selected for review?

on September 28, 2021 by guest. Protected copyright. Reply: Thanks for your comments. Delay pneumonia complication was common after multiple ribs fractures. However, research which focused on the relationship between isolated minor and subsequent pneumonia is limited or with only small sample size. IMRF after chest trauma are commonly observed in the ED but the disposition varies among different ED settings. Actually, IMRF can produce significant pain which impair both coughing function and clearance of secretion, leading to atelectasis and subsequent pneumonia. The aim of this study is to investigate patients with only isolated minor rib fractures (only 1 or 2 ribs fracture) and risk of pneumonia. We stated more clearly in the introduction section.[Page 5-6]. We hope that these changes and replies may meet your requirement for being published.

7. The statement is made, in the introduction, that „pneumonia is an inflammatory process „which is true. However, for the purposes of this study, it would seem that the infection-related component should not be ignored.

Reply: Thanks for your valuable comments and to improve this aspect. We changed the sentences BMJ Open: first published as 10.1136/bmjopen-2016-013029 on 13 January 2017. Downloaded from

“Pneumonia is an inflammatory process of the alveolar portions of the lung” to “Pneumonia is an inflammatory process of the alveolar portions of the lung commonly due to infection by microorganisms” in the introduction section. [Page 5] and added a reference 2 that discussed infection and pneumonia. We hope that these changes and replies may meet your requirement for being published.

8. A number of the references do not appear to directly relate to the manuscript content (including 21- 24, 30 – 32 which discuss inflammatory response in pneumonia and general discussions of COPD) and could be omitted.

Reply: Thanks for your valuable comments and to improve this aspect. We deleted the content “Specific factors promoting to pneumonia in COPD had been studied. The increased relaxation of upper and lower esophageal sphincters due to hyperinflation could cause micro-aspiration [30]. Moreover, malnutrition and corticosteroid use were commonly noted in COPD patients that altered host immune defense [31, 32].” And “The mechanism of NSAIDs involves the inhibition of cyclooxygenase and reduced prostaglandin synthesis. Administering NSAIDs has been shown to limit inflammation and lung tissue damage in the last decade [21, 22]. However, Voiriot et al found that NSAIDs might alter the course of infection and result in more invasive disease and complications [23]. The aforementioned studies have shown that NSAIDs can prevent pneumonia because of their analgesic effect but not their anti-inflammation effect. Additional clinical research studies are required to investigate these mechanisms.” in the revised manuscript. We hope that these changes and replies may meet your requirement for being published.

VERSION 2 – REVIEW

REVIEWER Krista L Kaups MD, MSc UCSF Fresno Community Regional Medical Center Dept of Surgery, First Floor 2823 Fresno Street

Fresno CA 93730 http://bmjopen.bmj.com/ USA REVIEW RETURNED 11-Nov-2016

GENERAL COMMENTS The authours have carefully and thoughtfully addressed the concerns expressed by the reviewers, which is appreciated. Specifically, the authors have more clearly described the nature of

the database and the methodology utilized. on September 28, 2021 by guest. Protected copyright. Although some of the limitations of the study (eg inability to evaluate injury mechanism and specific locations of the fractures) remain a concerns, I believe that the strength of the study lies in the comprehensive nature of the population involved. Additionally, the analysis defining particular risk for elderly patients with underlying disease further supports the literature in this area. The section on pages 14 describing NSAID use, as well as the section on pages 14-15 referring to prostanoid release and subsequent inflammatory cascades are not really reflective of the data presented and could be omitted. Finally, I would recommend and encourage further review and editing for English language usage.

VERSION 2 – AUTHOR RESPONSE

Reviewer: 2 1. The authors have carefully and thoughtfully addressed the concerns expressed by the reviewers, BMJ Open: first published as 10.1136/bmjopen-2016-013029 on 13 January 2017. Downloaded from which is appreciated. Specifically, the authors have more clearly described the nature of the database and the methodology utilized. Although some of the limitations of the study (eg inability to evaluate injury mechanism and specific locations of the fractures) remain a concerns, I believe that the strength of the study lies in the comprehensive nature of the population involved. Additionally, the analysis defining particular risk for elderly patients with underlying disease further supports the literature in this area. The section on pages 14 describing NSAID use, as well as the section on pages 14-15 referring to prostanoid release and subsequent inflammatory cascades are not really reflective of the data presented and could be omitted.

Reply: Thanks for your valuable comments and to improve this aspect. We deleted the section on pages 14 describing NSAID use and the section on pages 14-15 referring to prostanoid release and subsequent inflammatory cascades. We hope that these changes and replies may meet your requirement for being published.

2. Finally, I would recommend and encourage further review and editing for English language usage.

Reply: Thanks for your valuable comments and to improve this aspect. The revised manuscript was edited again (November 22, 2016) for proper English language and grammar by the qualified native English speaking editors with a certification. We hope that these replies may meet your requirement for being published. http://bmjopen.bmj.com/ on September 28, 2021 by guest. Protected copyright.