cardiopulmonary resuscitation (CPR) related rib fractures in adults

It is well-known that rib fractures occur as a consequence of chest compressions during cardiopulmonary resuscitation (CPR), but when a patient dies, how can a pathologist determine whether those rib fractures are due to CPR, or another traumatic event such as an assault?

Key questions for the pathologist to address

In order to assess whether rib fractures in adults were caused by CPR or some other injurious insult, the pathologist can turn to the literature for assistance.

Dr Andrew Davison (Senior Lecturer in Forensic Pathology, Wales Institute of Forensic Medicine, UK) considers the following questions when he finds rib fractures at autopsy, which may have been caused by CPR:

1. Where do CPR-related rib fractures occur (i.e. the rib number and location/ site within a rib)?
2. Are the site and likelihood of fracture related to: age, sex, length of resuscitation attempt, type of resuscitation, and level of life support training of the person(s) carrying out chest compressions?

His annotated review of the current literature illustrates how the practice of forensic pathology can be informed by examining the evidence base in the literature.

synopsis of the literature

Himmelhoch et al (Closed-chest cardiac resuscitation. A prospective clinical and pathological study 1964) authored a prospective post-mortem study looking at in-hospital cardiac arrest. He stated that the autopsy technique involved a "special effort to detect rib fractures".

There were 65 patients of whom 61 died (and there were 52 autopsies). Rib fractures were detected in 24 of the 52 (47%) and the majority of patients (31% of the 52) had 2 to 8 fractures.

There was no information on which ribs were involved or the location of fractures within each rib. There was no apparent correlation between the number of fractures and the age of the patient.

Henriksen (Rib fractures following external cardiac massage 1967) looked at the results from 37 autopsies. The stated aim was to assess the incidence and topography of thoracic cage fractures. CPR was performed by various medical personnel. Diagrams of the ribcage for many (but not all) of the cases indicate the site of rib fracture.

It was stated that lesions were "scattered over the anterior surface of the chest" and that the "lateral and posterior surfaces were not involved". There was no definition of anterior, lateral or posterior, and somewhat confusingly the fractures on the anterior surface are divided into parasternal and "lateral".

Despite saying that the lateral surfaces were not involved, fractures were seen beyond the anterior axillary line in one case; a series of fractures in the mid-axillary line.

It was stated that rib fractures were observed in 97% of patients over the age of 20 years and that the average number of fractures in each case increased with age. The number of fractures seem unrelated to the duration of external cardiac massage. The author was of the opinion that fractures may occur during the first few compressions. A major problem with this study is that there was no exclusion of trauma cases.

Krischer et al (Complications of cardiac resuscitation 1987) reviewed 705 autopsies. No information is provided with regard to age of the deceased and trauma cases were not excluded. There was no information on which ribs were fractured or fracture location.

Rib fractures were found in 31.6% and in nearly two thirds of those cases there were 5 or more rib fractures. Bilateral rib fractures occurred in 79%.

The author concluded that 20% of rib fractures were "avoidable" i.e. they did not occur in rib numbers 3 to 5 and/ or they were "lateral".

There was an increased number of fractures with increasing age and length of resuscitation. Previous series were quoted indicating an incidence of 17 to 75% rib fractures following CPR.

Hoke and Chamberlain (Skeletal chest injuries secondary to cardiopulmonary resuscitation 2004) reviewed evidence on thoracic fractures after conventional CPR and active compression decompression (ACD)-CPR, and produced a review of the literature.

The author concluded that sound methodological studies do not exist and it is not possible to compare studies.

Prospective post-mortem studies directed specifically at rib fractures are required.

Problems with previous studies include:

• small sample size;
• various means of diagnosis (autopsy, x-ray, clinical);
• retrospective nature;
• differing inclusion criteria (e.g. some studies only look at the survivors, others only non-survivors); and
• absence of exclusion criteria, particularly trauma and age (children/ adults).

Prospective post-mortem studies following CPR in adults show a rib fracture incidence of 32-60%; looking at all studies the incidence is 13-97%. After conventional CPR the average number of fractures is 5 to 8 and about two thirds of those with fractures have multiple fractures.

Fractures are more commonly seen on the left side of the chest than the right.

The author quotes two other studies which state that 65-70% of rib fractures occur in rib numbers 3 to 6 (Henriksen 1967) and 89% occur in numbers 2 to 7 (German language reference).

It is also stated that "posterior fractures are exceptional after CPR".

Age may have an important impact on the occurrence of fractures but the influence of sex is unclear. Training also "might influence rate of complications as there appear to be fewer when resuscitation was performed in ICU" (referencing Bedell 1986). Other authors, however, have shown no difference in the level of training of the people performing chest compressions (see Oschatz 2001).

It has been argued that bystanders without life support training may be more cautious when they perform chest compressions. Despite some assertions, there is no evidence to support the hypothesis that CPR training itself reduces the incidence of injury. The duration of CPR "may" influence complications. There is no compelling evidence to show an increased complication rate with ACD-CPR.

Darok (Injuries resulting from resuscitation 2004) states that the common "regions" of fractures are rib numbers 2 to 7 on the left and 2 to 6 on the right, quoting a German language reference.

The frequency varies from 19-80%. The frequency is greater with increasing age. The location is "mainly in the medioclavicular line, thus close to the sternal joint".

The author states that there is a higher frequency of fracture using ACD-CPR.

The main factors influencing frequency and severity of injuries are:

• length of resuscitation time;
• age of the patient; and
• degree of qualification of emergency personnel.

Fractures of rib numbers 1 and 8 to 12 are "very rare" after CPR. A less experienced bystander may be likely to use excessive force.

Black et al (Chest wall injuries following cardiopulmonary resuscitation 2004) looked at non-survivors in a retrospective study of 499 autopsies. Trauma cases were excluded. Rib fractures were found in 29%, there were more females than males, and the incidence increased with age.

The average was 4 rib fractures per case which is possibly an underestimate.

There was no information on which ribs were fractured or the location within each rib. Indeed, sometimes the fractures were just reported as "multiple". It was postulated that the increased number of fractures in females may have been due to osteoporosis.

Lederer et al (Frequency of rib and sternal fractures associated with out-of-hospital cardiopulmonary resuscitation is underestimated by conventional chest X-ray 2004) authored a prospective but small study (19 cases) of non-survivors following an out-of-hospital cardiac arrest and CPR. The aim was to correlate chest x-ray and post-mortem findings. The study was restricted to adults but did not exclude trauma (2 cases).

Although they classified the fractures as anterior (parasternal to anterior axillary line), lateral (anterior axillary line to posterior axillary line) and posterior (posterior axillary line to paravertebral line), there was no information on which ribs were involved.

They stated that fractures occurred more frequently in the middle third of the thoracic cage.

Fractures were diagnosed by x-ray in 9 cases and autopsy in 18 cases. In total, chest x-ray identified 12 rib fractures and autopsy identified 83 fractures.

Fractures were more frequent on the left side of the chest and the location was as follows:

• anterior (26%)
• lateral (74%)
• posterior (10%)

It was further stated that "the majority of rib fractures detected by autopsy were located in the mid-lateral part of the thoracic cage".

They suspect that bystander CPR could be expected to be associated with a higher frequency of fractures due to selection of non-ideal compression points and ill-adjusted force.

They concluded that chest x-ray was not an ideal method of detecting rib fractures which occur after CPR. The paper appeared confused with regard to the location of the rib fractures i.e. cases or fractures; the table provided and text don't match but it seems clear that the majority of rib fractures were "lateral".

Hashimoto et al (Forensic aspects of complications resulting from cardiopulmonary resuscitation 2007) produced a review/ literature search and some of their own figures. It is not clear with regard to inclusion and exclusion criteria or age of the patients.

96 autopsies revealed rib fractures in 52% which were bilateral in 39%. There was a mean number of 7.3 fractures per case and rib fractures were "frequently observed in 3 to 5th ribs". There were no fractures in rib numbers 11 or 12.

Apparently 56% of fractures occurred in the mid-clavicular line but there was no more detail with regard to location. Rib fractures were said to be infrequent below the age of 20 years and the incidence increased with increasing age. They were more common in females (62%) than males (47%).

Buschmann and Tsokos (Frequent and rare complications of resuscitation attempts 2009) produced a literature review and personal experience. Rib fractures were more often on the left than the right and in the mid-claviclular line. Ideally there should be 100 compressions per minute to a depth of 4-5cm. The authors stated that "complication rates are lower if CPR measures are executed according to guidelines".

Yang et al ("Buckle" rib fracture: an artifact following cardio-pulmonary resuscitation detected on postmortem CT 2011) performed a post-mortem CT study of 42 adult cases. Trauma was excluded. The authors looked at "buckle" fractures - where there was a smooth outer cortex and a kink-like fracture to the inner cortex. These apparently are hard to detect even at post-mortem. They measured the length of each rib and classified the location of fracture in relation to its distance from the costochondral junction and its relative position within the bony part of the rib i.e. anterior, middle or posterior third.

Fractures were distributed in rib numbers 2 to 9 (96.2% in ribs 2 to 7) and were most frequent in the 4th rib. The 9th rib was rarely involved - and only as a "buckle" fracture. 97.8 % (364) occurred in the anterior third of the rib, 2.2% (8) in the mid-part (mostly the 3rd and 4th ribs) and none in the posterior part.

"Buckle" fractures are frequently found near the costochondral junction and "not infrequently" found in the 8th and 9th ribs.

None of the fracture positions exceeded a ratio of 0.5 i.e. behind the halfway point between the spine and the costochondral junction.

The authors state that one advantage of their classification is that it is reproducible on CT but the disadvantage is that the description applies to the ribs themselves rather than their position in the ribcage i.e. below the 7th rib the anterior third of the rib is lateral or posterolateral in the chest.

They made the point that using the standard classification of mid-clavicular line or anterior-axillary line is very subjective.

They state that their study is small but they conclude that if rib fractures are seen in the posterior half of the rib or below the 9th rib, or involve a fracture of the outer cortex but not the inner cortex (reverse buckle) then non-resuscitation trauma should be considered.

Kim et al (Multidetector CT findings of skeletal chest injuries secondary to cardiopulmonary resuscitation 2011) authored a retrospective multi-detector CT study of 40 survivors of CPR. Trauma was excluded and all patients were adults.

The location of the fractures was classified as anterior, lateral or posterior (similar to Lederer et al).

Rib fractures were observed in 26 cases (65%); 92% were within the anterior ribcage, the rest (8% - 13 cases) were lateral. There were 173 rib fractures in total and multiple fractures in 25 of the 26 cases. The range was 2-13 fractures with a mean of 7, and fractures were bilateral in 69%. They "often" involved the 2nd to 7th ribs (no further details). There was no correlation with age, sex or length of resuscitation.

Smekal et al (Comparison of computed tomography and autopsy in detection of injuries after unsuccessful cardiopulmonary resuscitation 2013) compared CT and autopsy data in a prospective study of 31 non-survivors. CT detected 197 rib fractures in 22 patients and autopsy detected 192 fractures in 24 patients. The "average" patient showed a median of 9 fractures.

Although there were low numbers, trauma was excluded.

In 12 cases autopsy revealed more fractures than CT and in 8 cases CT revealed more fractures than autopsy.

There was no data on which ribs were involved or the location of the fractures.

They concluded that there was a strong agreement between CT and autopsy findings of rib fractures and that CT was valuable complement to autopsy. Autopsy was better for detecting sternal fractures.

Kim et al (Chest injury following cardiopulmonary resuscitation: a prospective computed tomography evaluation 2013) aimed to assess the frequency of rib fractures after CPR as a prospective study on survivors using CT. There were 71 cases and rib fractures were detected in 22. Females were more susceptible, possibly due to osteoporosis and increasing age.

When non-physicians performed CPR more ribs were fractured.

They excluded patients below the age of 18 years and trauma cases.

14 of the 22 cases showed multiple fractures ranging from 2 to 9. The mean age was 65 years and the median CPR time was 15 minutes.

There was no detail with regard to location of the fractures. The cases involved only manual compression.

Sperry (Anterior thoracic wall trauma in elderly homicide victims. The "CPR defense" 1990) did not involve a detailed discussion of the pattern of rib fractures after CPR. There was no reference for the assertion that they occur in the "middle of the chest". Characteristically, rib numbers 3 to 7 are involved but other ribs can be involved depending on the position of the hands and the force used.

Sommers (Potential for injury: trauma after cardiopulmonary resuscitation 1991) stated that properly performed CPR can produce injuries but they may be limited compared to improperly performed CPR. Previous studies are quoted with a rib fracture incidence of 19-52%. Mention is made of Krischer et al's "avoidable" fractures.

Nagel et al (Complications of CPR 1981) performed a prospective study of 2228 cases but did not exclude trauma. Rib fractures were found in 34% and caution was advised in interpretation.

Baubin et al (Increased frequency of thorax injuries with ACD-CPR 1999) looked at ACD-CPR in a prospective out-of-hospital study which involved adults and excluded trauma. The authors concluded that there were more ribcage fractures following ACD-CPR than conventional CPR.

There were 35 autopsies; rib fractures were detected in 55% of cases that received standard CPR and 87% of cases receiving ACD-CPR.

It was stated that even after "routine" post-mortem examination it may be difficult to detect all CPR injuries, even rib fractures. In their experience fractures usually occur at the beginning of CPR. They also noted that the ACD was performed for longer than standard CPR, although they concluded that it cannot be proven beyond doubt that ACD-CPR causes more rib fractures than standard CPR.

Paaske (Complications of closed-chest cardiac massage in a forensic autopsy material 1968) authored a retrospective study involving 260 cases of all ages. Some trauma was excluded (road traffic collisions). Rib fractures were found in 44%; bilateral in 21%.

There was a lack of information on the site of the fracture but it was stated that none were detected behind the mid-axillary line. The incidence of fracture increased with age.

Oschatz et al (Cardiopulmonary resuscitation performed by bystanders does not increase adverse effects as assessed by chest radiography 2001) looked at bystander CPR compared with CPR performed by those with life support training. Trauma cases were excluded but they only looked at survivors. They used chest x-ray which picked up rib fractures in 8% of both groups. The mean duration of CPR was 17 minutes for bystanders and 9 minutes for others.

There may have been a selection bias due to exclusion of those patients who subsequently died. The important point is that x-ray is not reliable in detecting rib fractures so the conclusion that CPR administered by non-medical personnel does not increase the number of life support-related adverse effects in patients surviving cardiac arrest is not necessarily valid.

Bedell and Fulton (Unexpected findings and complications at autopsy after cardiopulmonary resuscitation (CPR) 1986) involved a retrospective autopsy study of 130 hospital cases. There were only 25 rib fractures in 14 cases (11%). There were no further details with regard to number of rib or location. The authors suggest that a more professional technique may decrease the morbidity because fewer complications were found in those who had CPR on ICU compared to those who had CPR on a general ward.

Rabl et al (Serious complications from active compression-decompression cardiopulmonary resuscitation 1996) compared standard CPR with ACD-CPR. Autopsy reports were reviewed in 56 cases: 25 patients had standard CPR, 15 patients had standard CPR then ACD-CPR, and 16 patients had ACD-CPR alone.

Of the standard CPR patients, 7 had rib fractures (28%). Of those who had ACD-CPR, 80% had fractures (11 out of 16 (68%) ACD alone; 14 out of 15 (93%) standard CPR followed by ACD).

Guidelines for CPR suggest a force of 300-500N to a depth of 4-5cm. The authors concluded that ACD-CPR caused rib fractures more often than standard CPR.

Pinto et al (Manual and automated cardiopulmonary resuscitation (CPR): a comparison of associated injury patterns 2013) reviewed 175 post-mortem records to compare 'auto-pulse' CPR with manual CPR. Device-assisted CPR is separated into two categories: manual and automated.

Manual (i.e. ACD) uses a suction cup - such as 'Cardiopump' or 'ResQ' pump - applied over the anterior chest. This requires constant effort from the attending personnel.

Automated CPR works autonomously, there are two types: piston (e.g. 'Thumper'/ 'Lucas'), or band (e.g. 'autopulse' - load distributing band). The latter involves a band encircling the chest to constrict it. It is said to improve haemodynamic effects but its true efficacy is difficult to assess.

'Autopulse' provides 80 compressions a minute and constricts the chest depth by 20%. The protocol for use of 'autopulse' means manual CPR is used prior to 'autopulse'. It is not appropriate for trauma cases, those under 18 years and those above 300 lbs weight.

In this study there were 87 cases of manual-only CPR and 88 cases of manual/ 'autopulse' ; although it is not quite clear, it appears that "manual" refers to manual chest compressions rather than 'manual device-assisted CPR'.

The median age of the patients was 50 years.

Rib fractures were classified as anterior, antero-lateral, lateral, postero-lateral or posterior. There was a statistically significant difference in the frequency of rib fractures between the two groups. The manual group had a higher frequency of anterior fractures, and the 'autopulse' group had a higher frequency of posterior fractures. There was no significant difference with antero-lateral, lateral or postero-lateral fractures.

Although the authors do not say how many in each group sustained a rib fracture, they state it was a "high percentage".

The frequency of the location of the rib fractures in the manual group was:

• anterior 54%
• antero-lateral 26%
• lateral 19%
• postero-lateral 0%
• posterior 0.4%

The frequency in the 'autopulse' group was:

• anterior 29%
• antero-lateral 24%
• lateral 13%
• postero-lateral 2%
• posterior 33%

In both groups rib numbers 3 to 5 were most commonly fractured and the frequency dropped "substantially" at rib numbers 1 and 7 to 9.

There was no statistical difference between the groups regarding side of chest, body mass index or sex. With regard to age, in the manual group there were more anterior rib fractures in those over 45 years.

'Autopulse' was associated with fewer sternal fractures; 'autopulse' cases showed vertebral fractures (4.5%).

There were more skin abrasions in the 'autopulse' group (96%) compared to the manual group (24%). Those abrasions occurred in the axilla, shoulder, chest and arms.

The authors provide a reference suggesting that the anterior rib shaft is weaker than the posterior region.

The authors suspect that the relatively high frequency of anterior fractures in the 'autopulse' group was due to previous manual CPR. They postulate that posterior fractures occur in a higher frequency in 'autopulse' CPR because of a lever system working in the posterior region of the rib with the fulcrum at the costotransverse articulation and immobility of the posterior part of the rib as a result of the body being on a board. Unfortunately, no diagram is provided, which might have assisted with this interpretation.

problems with the studies in the literature

1. The retrospective nature of the studies - probably underestimates rib fractures even in post-mortem studies; many post-mortem reports do not comment on the musculoskeletal system and, even if they do, the rib number and location may not be provided.

2. Paucity/ absence of information regarding: rib number, location of fracture, and age of the patient.

3. Variable classification used (anterior/ lateral etc).

4. Exclusion criteria not always stated (children/ prior trauma etc).

5. Small numbers in the studies.

6. Variable means of diagnosis (clinical, radiology (x-ray/ CT), and autopsy).

7. Different study groups (survivors/ non-survivors, or both).

Reasonable conclusions to be drawn from the literature in relation to rib fractures in adults?

1. Sound methodological studies do not exist.

2. Rib fractures following CPR in adults are quite common - 30 to 60% of prospective post-mortem studies.

3. Post-CPR rib fractures are often bilateral.

4. The site of rib fracture is dependent on: the position of the hands; force used; and method of chest compression (manual or device-assisted).

5. The vast majority (90%+) of fractures occur in ribs 2 to 7; fractures in the bony parts of rib numbers 1 and 8 to 10 are possible but probably very rare; it is difficult to see how fractures can occur in rib numbers 11 and 12 following standard manual CPR.

6. The vast majority of fractures (90%+) occur in the anterior third of the bony part of the rib, some occur in the middle third but - following standard manual CPR - none in the posterior third of the bony part of the rib.

7. Posterior rib fractures occur following automated band-type CPR.

8. Lateral fractures i.e. those occurring between the anterior and posterior axillary lines, do occur after standard CPR. If a fracture is noted in rib numbers 10 to 12 or in the posterior third of the bony part of a rib, then non-resuscitation trauma should be suspected.

9. There is some evidence to indicate fractures are more common:

• on the left side of the chest (Hoke and Chamberlain; Lederer et al; Buschmann and Tsokos);
• with increasing age of the patient (Krischer et al; Hoke and Chamberlain; Darok; Black et al; Hashimoto et al; Paaske);
• in females (Black et al; Hashimoto et al; Kim et al);
• with increasing length of time of resuscitation attempts (Krischer et al; Hoke and Chamberlain; Darok);
• following the involvement of untrained persons (Bedell and Fulton; Darok; Kim et al); and
• with the use of ACD-CPR (Baubin et al; Rabl et al).

10. Chest x-ray is unreliable as a diagnostic tool for detecting rib fractures.

11. CT is better than x-ray and may complement the post-mortem detection of rib fractures.

additional resources

• Anatomy relevant to chest trauma (and here)
• Physiology of CPR
• Lucas chest compression system
• Lucas chest compression system in action
• Pinterest board on 'Trauma and resuscitation'