Comparing the healing rate of different
bones measured with conventional
radiography in children from 0-17 years old:
A systematic review
Literature thesis
Otto, R. E. // 12315478
Master Forensic Science
5 ECTs
Wordcount: 7772
24-01-2020
Supervisor:
Drs. S. de Vries
Nederlands Forensisch Instituut
Examiner:
Prof. dr. R.R. van Rijn
Abstract
Child abuse is a common event which, in many cases, leads to fractured bones. A detailed timetable with radiological features can help estimating the age of a fracture. This information can be used by the police and can be important in criminal investigations to analyse and identify inconsistencies between the appearance of the fracture and the description of the timing of occurrence. However, studies that examined fracture dating and healing rates found contradicting results and based on the current research, it is now only possible to make an age estimation in weeks. In addition, little is known about the healing rate of different bones and whether or not their healing times vary. This suggests that perhaps the influence of different bones on the healing rate is insufficiently included in the general timetable. When it appears that there are differences, radiologists should move away from a general timetable and new ones should be created that consider the different bone categories separately, such as long and short bones, or when necessary, even look at every bone individually. This would allow for more accurate results and will prevent errors. The aim of this thesis is therefore to provide a systematic review of the studies that investigated the healing rate of different bones with conventional radiography in children from 0-17 years old. Relevant words were used for the search through PubMed. Seventeen articles were found that met the inclusion criteria. Although the results of the relevant literature are somewhat contradictory it suggest that arm fractures heal faster than leg fractures, but this result should be handled with care. It also demonstrates that besides fracture location the age of the child, type of fracture and whether the fracture is the result of abuse or an accident influences the healing rate. Future studies conducted in this area should aim for a standard way of reporting healing characteristics and study different bones and age groups separately.
Table of contents
1. Introduction... 4
2. Stages of bone healing...5
2.1 Inflammatory phase...5 2.2 Reparative phase... 5 2.3 Remodelling phase...6 3. Bone classification... 6 3.1 Long bones... 6 3.2 Short bones... 6 3.3 Flat bones... 6 3.4 Sesamoid bones... 6 3.5 Irregular bones... 6 4. Methods... 7 5. Results... 7 5.1 Search results... 7
5.2 Description of included studies...8
5.3 Healing rate ... 10
6. Discussion ... 12
6.1 Comparison of the results...12
6.1.1 Broad ranges...12
6.1.2 Arm versus leg fractures...13
6.1.3 No distinction between long bones...13
6.1.4 Clavicle fractures...14
6.1.5 Scaphoid fractures...14
6.1.6 Bone categories compared...14
6.2 Influencing factors...14
6.2.1 Age... 15
6.2.2 Accidental versus non-accidental...15
6.3 Research limitations...16
6.3.1 Radiographic healing features...16
6.3.2 Retrospective studies...17 6.3.3 Sample size... 17 7. Conclusion... 17 8. Future research... 18 9. References... 19 10. Appendix... 21 1. Introduction
According to the third study of Nationale Preventiestudie Mishandeling (NPM) in 2017 approximately 3% of the children in the Netherlands are exposed to child abuse and neglect. This number is even larger for young children from 0-3 years in comparison with children from 4-17 years of age (Alink et al., 2018). Especially physical abuse is common at a young age, frequently resulting in fractured bones. Moreover, at least half of the non-accidental injuries (NAI) occur in children younger than 1 year old (Warner et al., 2017). In some cases where NAI are present even multiple fractures can be found throughout the body. It is also possible that these fractures are in different stages of the healing process and are thus inflicted at different times. Radiologists are often asked to estimate the age of a fracture in cases of suspected child abuse. This information can be used by the police and can be important in criminal investigations to analyse and identify inconsistencies between the appearance of the fracture and the description of the timing of occurrence. This can help investigations by in- or excluding potential perpetrators and optimizing child protection services (Prosser et al., 2012).
Conventional radiography (CR) is the most commonly used method to demonstrate or rule out bone fractures and to make an estimation about the degree of healing of fractures in living children and adults. Computed tomography (CT) is predominantly used for the evaluation of trauma victims and has therefore generally no place in child abuse cases. An exception is when the vertebrae are fractured and an CT-scan is necessary to establish fracture stability. Besides CT, also magnetic resonance imaging (MRI) is mostly not used for initial diagnosis of child abuse. It is used in later stages in order to determine whether neurological trauma or spinal injuries are present (Bilo et al., 2010).
In spite of the high importance of correctly estimating the age of a fracture, since it can be used as evidence in possible child abuse cases, the estimation is still mostly based on the experience of the radiologist (Fadell et al., 2017; Halliday et al., 2011; Prosser et al., 2012; Walters et al., 2014; Warner et al., 2017). Besides the use of their clinical background, they refer to the textbook chapter of O’Connor and Cohen and to the textbook of Offiah & Hall (O’Connor,
Cohen, 1998; Offiah & Hall, 2009). These textbooks give an overview of the different radiographic healing stages and the duration of every phase. However, also these results are based on clinical experience. As this is not very reliable, studies have been conducted to find a good method which can make fracture dating a more objective field. The three primary studies in this field also unfortunately reported different results and used different radiographic features which makes it hard to compare their outcomes (Cumming, 1979; Islam et al., 2000; Yeo & Reed, 1994).
Throughout the past years more studies have been carried out and in 2018 the Royal College of Paediatrics and Child Health (RCPCH) wrote the Child Protection Evidence Systematic review on Fractures (Child Protection Evidence Systematic review on Fractures, 2018). In this review they created an overview of seven studies in order to combine their results and try to detect similarities which can be used to estimate the age of fractures more correctly. In the end they mentioned that it is hard to compare the results as the studies chose different radiographic features and because of the variable time interval between the radiographs. They also state that comparison is hampered since different bones are examined. Six of the seven studies looked at long bones such as the clavicle, humerus, femur and ulna, while only one study looked at ribs which belongs to the category flat bones. It may be possible that long bones heal differently than flat and short bones, or moreover that different long bones have various healing rates. It is of importance to gather more information on the healing rates of different bones as it can influence outcome of fracture age estimation.
This is why in this thesis an extensive literature search will be conducted to find studies on age estimation of fractures in children up to 17 years old and compare the results of different bones. This in order to answer the question: ‘Is it possible to detect a difference in healing rate between various bones when dating fractures with conventional radiography in children from 0 to 17 years old?’. If differences can be found it is important to think about the consequences and what future research is needed. For example, if the differences in healing rate are large, is it still possible to use a general timetable, or is it necessary to create new ones separately for bone categories. Future research should then focus on what bones have similar healing rates and which can thus be grouped together in a timetable.
This thesis will begin with a small introduction on the different stages of fracture healing as it is important to know these features in order to understand and interpret the radiographs. Then will be continued with an explanation about the different types of bones present in the human as this may influence the healing rate. Hereafter, the literature selection will be described in the methods and in the results the outcomes of the studies will be presented. In the discussion the results will be compared, influencing factors on bone healing rate will be discussed and limitations of the literature will be described. Finally,
a conclusion will be given and the thesis will be ended with future recommendations in the field of age estimation of bone fractures.
2. Stages of bone healing
Bone healing is a three phased process, which consists of an inflammatory phase, followed by a repair phase and ends with the remodelling phase. These stages are not strictly distinct, but may overlap. In these stages different radiographically features can be distinguished in order to estimate the age of a fracture (Wraighte & Scammell, 2007).
2.1 Inflammatory phase
When a bone fractures, the periosteum is torn away from the underlying cortical bone and soft-tissue and adjacent muscles rupture. This causes blood vessels to break and the formation of a fracture haematoma triggers the inflammatory phase (Mckinley, 2003). The ends of the fractured bone die and are resorbed by osteoclasts. During this phase the initial features of soft-tissue swelling are seen (Wraighte & Scammell, 2007). The haematoma will eventually be replaced with new bone. Radiographically, subperiosteal new bone formation (SPNBF) appears as a thin layer of bone. This new bone runs parallel to the original cortex and increases in thickness over time. First with a single-layered character and develops into a multi-layered characteristic (Walters et al., 2014). 2.2 Reparative phase
The second phase is the reparative phase in which bone-forming cells are recruited, fracture callus is developed and cells replicate in the periosteum. The callus in created at the ends of the bone and this what unites the bones in the end (Walters et al., 2014). First soft callus is created, which refers to a cartilaginous phase of early new bone before ossifying and calcifying. Hard callus is woven bone that is ossified and becomes nearly as dense as the cortex (Prosser et al., 2012). Some studies differentiate radiographically between soft and hard callus, while other studies claim that it is not possible to see the difference and that they are only histological terms (Halliday et al., 2011; Prosser et al., 2012). The studies that say you can see a difference describe it radiographically as; soft tissue has a fluffy appearance and hard callus is ticker with a more dense characteristic (Walters et al., 2014).
2.3 Remodelling phase
The remodelling phase can be the longest of the three, as this can continue long after the fracture has clinically healed. In early stages of callus formation too much is formed and the excessive callus is removed in this phase. This is also important, because even though hard callus provides enough strength as a new bone, it does not fully restore the characteristics of a bone. In order to reinforce the properties of the bone hard callus is removed by osteoclasts and lamellar bone is developed by osteoblasts (Marsell & Einhorn, 2011). In this way the
fracture site is smoothed and the bone is reshaped towards its original strength. In young children remodelling can be so effective that fractures heal without a trace (Prosser et al., 2012; Wraighte & Scammell, 2007).
3. Bone classification
The human baby has more than 300 bones at birth which eventually fuse together to 206 bones of the adult skeleton. These 206 bones are divided in 5 categories based on their shapes and additional function.
3.1 Long bones
The name long bones does not refer to their size but that they are longer than that they are wide. These bones are cylindrical and function as lever, they move when the muscles contract and thus facilitate movement. Examples of long bones are the ulna and radius in the arms, tibia and femur in the legs and the clavicle (Betts et al., 2013).
3.2 Short bones
Just like with the long bones the name does not refer to the size but short bones are almost equally in length, width and thickness. They help with stability and provide some movement. They are only present in carpals of the wrists and in the tarsals of the ankles (Betts et al., 2013).
3.3 Flat bones
Flat bones are thin and often curved. They function regularly as protection for organs such as the brain and heart. Besides their protective function they also serve as an attachment point for muscles. Ribs and cranial bones are examples of flat bones (Betts et al., 2013).
3.4 Sesamoid bones
These bones are small and roundly shaped and are embedded in the tendons. Their presence helps to protect the tendons from tension and wear. The patella (kneecap) is the largest sesamoid bone in the body (Betts et al., 2013).
3.5 Irregular bones
This last category is called irregular as the bones included have various shapes and therefore cannot fit in any other category. Because of the complex shapes these bones can protect some internal organs such as spinal cord and organs in the pelvic cavity. Examples of irregular bones are the vertebrae and multiple facial bones (Betts et al., 2013).
4. Methods
The PubMed database was searched for studies published in English or Dutch, with no specific date criteria, using the search words listed in Appendix A.
Studies were included if their subjects were human, under 18 years old and reported the fracture healing time in days/ weeks/ months. Animal studies or subjects older than 17 years old were excluded. Children with a gestational age of <34 weeks were also excluded as this can influence fracture healing. The participants were not allowed to have any known diseases that can influence bone remodelling or structure and were not allowed to take medicine that can influence bone structure. Fractures that were treated with internal or external fixation were also excluded. Only studies that examined fractures with x-ray were included in this review as CT-scans and MRI give different radiographic features which makes it difficult to compare their results. Studies were also excluded when they were reviews or case studies.
5. Results
5.1 Search results
The initial search resulted in 1833 articles. After the removal of duplicates 1818 articles remained. Titles and abstract were screened with the use of Rayyan QCRI and all non-relevant studies were eliminated. In total 60 articles were read and 16 studies met the criteria. After screening the reference lists of the included literature an additional study was found (Mukhtar et al., 2018), see figure 1.
5.2 Description of included studies
The data represent children aged 0-17 years. The inclusion and exclusion criteria varied across the literature and are shown in table 1. Some studies included fractures as a result of abuse (Halliday et al., 2011; Sanchez et al., 2013) and others only included fractures caused by accidents (Davison et al., 2016; Fadell et al., 2017; Gholson et al., 2011; Hosokawa et al., 2017; Islam et al., 2000; Mukhtar et al., 2018; Prosser et al., 2012; Sherr-Lurie et al., 2011; Walters et al., 2014; Warner et al., 2017; Wulff & Schmidt, 1998) . Malone et al. (2011) state that they do not know whether the fractures are the result of abuse or an accident. Domzalski et al. (2006) and Huckstadt et al. (2007) mention that all fractures included in their study are caused by accidents except for one and three unknown causes.
It also varies what part of the bone the studies included. Some studies do not mention what specific part of the bone they include and thus probably do not differentiate between these fractures. However, Islam, et al. (2000) includes diaphyseal, diametaphyseal and metaphyseal fractures but excludes epiphyseal-physeal fractures. While Halliday, et al. (2011) excludes metaepiphyseal-physeal fractures, Domzalski, et al. (2006) only includes fractures at the distal metaphyseal structure and makes a distinction between oblique and transverse fractures. Song, et al. (2019) and Hosowaka et al. (2017) solely include the diaphyseal part and Sherr-Lurie, et al. (2011) includes proximal and distal epiphysis and diaphyseal fractures. Gholson, et al. (2011) makes a distinction between fracture occurrence in the distal and proximal pole and waist of the scaphoid. Naturally, also the inclusion criteria of the studies varied for the type of bone that is examined. Most studies looked at fractured long bones, however a large array of this category is displayed in the literature. Walters et al. (2014), Song et al. (2019), Fadell et al. (2017) and Mukhtar et al. (2018) solely included clavicle fractures, while Prosser et al. (2012), Halliday et al. (2011) and Warner et al. (2017) did not differentiate between different long bones. Islam et al. (2000) included the radius and ulna and Malone et al. (2011) examined the radius and tibia. Sherr-Lurie et al. (2011) only looked at humeral fractures, Hosowaka et al. (2017) only examined femoral fractures and Davidson et al. (2016) solely studied fractures of the metacarpal neck. The four remaining studies did not examine long bones but other categories. Sanchez, et al. (2013) is the only study that examined fractured ribs, which belongs to the category flat bones. Wulff, et al. (1998), Gholson, et al. (2011) and Huckstadt et al. (2007) looked at scaphoid fractures which is one of the carpal bones in the wrist and belongs the short bones.
In this review only studies that immobilized fractures with a cast were included and studies that treated fractures with internal or external fixation were excluded. Also some studies did not immobilize the fractures (Fadell et al., 2017; Sanchez et al., 2013; Walters et al., 2014). Sometimes it is necessary to set the bone into its original place, studies that did this with open reduction were excluded and only studies that used closed reduction were included.
Table 1, Description of studies Study (year) Fracture result of abuse Bones Bone structure Total number of patients Age (mean) Wulff et al. (1998) No Scaphoid Not mentioned 33 7.6-16.10 years (13.4 years) Islam et al. (2000) No Radius, ulna Diaphyseal, diameta-physeal, metaphyse al 141 1-17 years (8 years) Domzalski et al. (2006) No, except for one unknown case Tibia Metaphyse al 26 2-14 years (6 years) Huckstadt et al. (2007) No, except for 3 unknown cases
Scaphoid Distal pole, waist, proximal pole 17 9-17 years (14 years) Halliday et al. (2011) Yes Femur, fibula, humerus, radius, tibia, ulna Epiphyseal, diaphyseal 31 14 days - 44 months (5 months) Malone et al. (2011) No information available Radius, tibia Not mentioned 107 0-5 years
Sherr-Lurie et al. (2011) No Humerus Epiphysis, diaphyseal 7 1 day Gholson et al. (2011)
No Scaphoid Distal pole, waist, proximal pole 181 7-17 years (14.6 years) Prosser et al. (2012) No Clavicle, femur, fibula, humerus, radius, tibia, ulna Not mentioned 63 0-5 years (4.8 years) Sanchez et al. (2013)
Yes Ribs Not
mentioned 16 <12 months Walters et al. (2014) No Clavicle Not mentioned 131 0-3 months Davison et al. (2016) No Metacarpal neck Not mentioned 36 11-13 years Fadell et al. (2017) No Clavicle Not mentioned 61 0-6 months (24 days) Warner et al. (2017) No Long bones Not mentioned 40 9 days - 12 months Hosowaka et al. (2017)
No Femur Diaphyseal 7 1 day
Mukhtar et al. (2018)
No Clavicle Diaphyseal 7 8-16 years
(10.5 years) Song et al. (2019) Not mentioned Clavicle Diaphyseal 23 10.4-15.6 years (13.4 years) 5.3 Healing rate
The different studies used different radiographic healing features to indicate how long it takes for the bone to heal, which is why, multiple definitions will be
used here to describe the healing rate. These are the presence periosteal reaction, callus, union, bridging and remodelling. An overview of the results of the studies is given in table 2.
Table 2, Radiographic features per study Study
(year)
Periosteal reaction
Callus Bridging Remodellin g
Union No distinction between long bones
Halliday et al. (2011) Callus: Peak: 11 days First seen from day 4 Endosteal callus: Peak: 3 weeks First seen from day 8 Prosser et al. (2012) Peak: 15-35 days First seen from day 5 Soft callus: Peak: 22-35 days First seen from day 12 Hard callus: Peak: 22 days First seen from day 19 Peak: 36 days First seen from day 19 Peak: 36 days First seen from day 45 Warner et al. (2017) Peak: 9-49 days First seen from day 7 Peak: 9-26 days First seen from day 9 Peak: 15-67 days First seen from day 15 Peak: 51-247 days First seen from day 51 Tibia, ulna, radius
Islam et al. (2000) Peak: 4-7 weeks First seen from week 2 Peak: 4-7 weeks First seen from week 2 Peak: 13 weeks First seen from week 3 Peak: 9 weeks First seen from week 4 Domzalski et al. (2006) Oblique fracture: 6 weeks Transverse fracture: 6.5 weeks
Malone et al. (2011) Range: 15-75 days Range: 24-93 days Range: 24-156 days Clavicle Walters et al. (2014) Peak: day 15 First seen from day 9 Fadell et al. (2017) Peak 1: day 11 Peak 2: day 42 First seen from day 7 Peak 1: day 12 Peak 2: day 61 First seen from day 11 Peak 1: day 22 Peak 2: day 63 First seen from day 20 Peak 1: day 49 Peak 2: day 59 First seen from day 35 Mukhtar et al. (2018) Range: 9-12 weeks Average: 10.4 weeks Song et al. (2019) Range: 6.1-23.4 weeks Average: 12 weeks Humerus Sherr-Lurie et al. (2011) 2 weeks Metacarpal neck Davison et al. (2016) 12 weeks Femur Hosowaka et al. (2017) Soft callus: First seen from day 10 Hard callus: First seen from day 15 Ribs
Sanchez et al. (2013)
1-3 weeks 3-5 weeks 7-9 weeks 9 weeks
Scaphoid Wulff et al. (1998) 7.7 weeks Huckstadt et al. (2007) 7.4 weeks Gholson et al. (2011) Distal pole: week 6 Proximal pole: week 15 Waist: week 9 6. Discussion
6.1 Comparison of the results
In order to answer the research question ‘Is it possible to detect a difference in healing rate between various bones when dating fractures with conventional radiography in children from 0 to 17 years old?’ the results of the studies will be compared in the next section.
6.1.1 Broad ranges
Overall, it can be seen that the studies find broad ranges when observing the radiographic healing features. For example, Malone et al. (2011) find a range for bridging from day 24 to day 93 and Warner et al. (2017) observe a peak between day 15 and day 67. Fadell et al. (2017) find a peak at day 22 and a second peak at day 63 and Prosser et al. (2012) state that a peak is observed from day 36. Besides the large ranges for bridging these are also found for periosteal reaction, callus and remodelling. However, this is not that odd for remodelling since this can take a very long time, as explained in section 2.3. Song et al. (2019) also find a very broad range for union, namely from week 6.1 to week 23.4. They state this was due to three patients who had delayed union at week 17.1, 17.6 and 23.4. These broad ranges make it difficult to precisely estimate the age of a fracture and still only makes it possible to estimate the age in weeks.
6.1.2 Arm versus leg fractures
Some of the studies included multiple long bones and do not differentiate between the healing rates of the bones and report an average time for all the participants. Only Malone et al. (2011) separate the healing time of the radius and tibia to see whether there are any differences. They report that the healing time of forearm fractures is shorter than healing time of leg fractures at stages 2 and 3, which they characterize as: ‘stage 2: Granulation: beginning of resorption along fracture line, “fluffy” callus formation, blurring of fracture line, absence of a complete mature callus’ and ‘stage 3: Callus: mature callus formation around fracture site; callus bulging over site and demonstrating a radiopaque appearance, fracture line visible but may be blurred’. They reason, since no differences are found in later stages, that fracture location is a particularly important factor in the beginning of fracture healing. Later stages are more influenced by other factors such as mobility, because movement is limited in the beginning because of discomfort.
The results of Malone et al. (2011) correspond with Sherr-Lurie et al. (2011) that find a shorter healing time for birth related humeral fractures in comparison with the longer healing time found in Hosowaka et al. (2017) for birth-related femur fractures. These results suggest that arm fractures heal faster than leg fractures and that different radiographic timetables should be created in order to estimate the time of the fracture correctly.
However, Islam et al. (2000) examined fractures in the ulna and radius in which you would expect a shorter healing time based on the analysis above, but a peak for bridging and remodelling was observed at week 13 and week 9, respectively. This longer healing time can be caused by the low number of patients that were 4 years old or younger.
It is also expected that Domzalski et al. (2006) find longer healing times, since they examine fractures in the tibia. Nevertheless, they observe union at an average from 6 weeks, which is a lot faster than Islam et al. (2000). They also examined older children, similar to Islam et al. (2000), so the differences are not due to age. It can be caused by that Domzalski et al. (2006) only look at fractures of the distal metaphysis and that Islam et al. (2000) included the whole diaphyseal, diameta-physeal, metaphyseal structures. This may suggest that the part of the bone that is fractured, besides fracture location, also influences the healing rate.
6.1.3 No distinction between long bones
Halliday et al. (2011), Prosser et al. (2012) and Warner et al. (2017) examined multiple long bones without making a distinction between the healing times of the different bones. The three studies all examined the presence of callus and this can therefore be compared. Halliday et al. (2011) observe a peak from 11 days, while Prosser et al. (2012) state that soft callus peaks between day 22 and day 35 and that hard callus is observed from day 22. This means that the peak from Halliday et al. (2011) starts earlier than Prosser et al. (2012), which could be explained by the fact that the patient group of Halliday et al. (2011) is younger. Warner et al. (2017) observe a peak from day 9-26 which is more
similar to Halliday et al. (2011) than Prosser et al. (2012). They also have a younger patient group which may be the reason. It is also possible that Prosser et al. (2012) examined more bones that have a slower healing rate in comparison to Warner et al. (2017) and Halliday et al. (2011). However, since they do not give the healing rates of the bones separately, it is not possible to say whether one bone heals faster than another and thus whether this is the reason for the differences.
6.1.4 Clavicle fractures
The results of the healing times of clavicle fractures vary across the studies. Fadell et al. (2017) already see a peak in bridging from week 4 while Song et al. (2019) and Mukhtar et al. (2018) observe union from week 12 and 10, respectively. The patient group of Fadell et al. (2017) is much younger than the patients examined in Song et al. (2019) and Mukhtar et al. (2018) which can be the cause of the differences. However, it can also be due to the fact that Song et al. (2019) and Mukhtar et al. (2018) look at displaced midshaft clavicle fractures. Maybe the displacement of the fracture causes the fracture healing time to increase. If this really has an impact it should considered when setting up a new timetable.
The study of Walters et al. (2014) solely examined the presence of callus and can therefore only be compared to the results of Fadell et al. (2017). Walters et al. (2014) observe callus for the first time at day 9 and detect a peak from day 15, this is in line with the findings of Fadell et al. (2017) that observe callus for the first time at day 11 and see a peak at day 12. Their studies were also more equal in terms of age group compared to Song et al. (2019) and Mukhtar et al. (2018) which can cause the similarities.
6.1.5 Scaphoid fractures
Gholson et al. (2011) find union between 6-15 weeks while Wulff et al. (1998) and Huckstadt et al. (2007) see union after approximately 7.5 weeks for scaphoid fractures. This means that there is already a large difference in healing time of the same bone. This can be caused by the longer healing time of the proximal pole, since Gholson et al. (2011) state that the mean time to observe union of the proximal pole is 15 weeks, which is longer than the time to observe union in the distal pole and waist, which are 6 and 9 weeks respectively. In Huckstadt et al. (2007) it is mentioned that only 2 proximal pole fractures were included and 20 fractures of the distal pole or waist. This could lower the mean time to observe union in Huckstadt et al. (2007) to 7.4 weeks in comparison to a mean time of 10 weeks in Gholson et al. (2011). This suggests that even in the same bone, but in a different part of it, healing times differ. If this is really the case, general timetables cannot be used to estimate the age of a fracture and besides separating individual bones, fractures in different parts of the same bone should also be taken into account.
6.1.6 Bone categories compared
In order to say something about whether the different bone categories have different healing rates the results of the studies should be grouped together per category. Unfortunately, it is hard to compare the results, since Sanchez et al. (2013) looked at flat bones and reported the healing times of periosteal reaction, callus, bridging and remodelling while the studies that examined short bones only reported the time of union. The studies that looked at long bones differed in what radiographic features they analysed. This makes it also difficult to combine their results.
The result of Sanchez et al. (2013) bears a resemblance to the results of Wulff et al. (1998) and Huckstadt et al. (2007) as bridging is seen from week 7-9 in Sanchez et al. (2013) and union is observed around 7.5 weeks in Wulff et al. (1998) and Huckstadt et al. (2007). However the results of Gholson et al. (2011) differ, which makes difficult to say anything about the (dis)similarities between short and flat bones. The results of the long bones differ too much to combine them and compare them as a group to short and flat bones. This indicates that the healing rates of bones differ independently to what category they belong to.
6.2 Influencing factors
Despite the fact that bone fracture healing follows a predictable order as described in section 2, different healing rates are found when comparing the literature, even when the same bone is analysed. This can be due to many factors that influence the healing rate besides fracture location such as comparing different age groups, whether the fracture is a result of abuse or an accident and what structure of the bone is fractured. These factors will be discussed more extensively in the next section.
6.2.1 Age
One important factor that can influence the healing rate of fractures is age, as it is assumed that younger children heal faster than older children and adults (Pickett, 2015). This is thought because younger children have a thicker periosteum and their process of new-bone formation is faster (Bilo et al., 2010; Marsell & Einhorn, 2011; Pickett, 2015). Malone et al. (2011) looked into the differences of bone healing between age groups. They state that children from 0-1 year have a reduced stage 1 time compared to 4-5 year olds, which they described as ‘stage 1: No healing: sharp fracture lines, absence of bridging and callus formation’. As this stage does not include healing this rather means that the healing process is started quicker in younger children.
In Domzalski et al. (2006) it is found that the oblique fractures demonstrated union approximately 0.5 week faster than transverse fractures. They reason that oblique fractures probably heal faster than transverse fractures as a smaller fracture surface area is included. This can suggest that the type of fracture can also influence the healing rate. However, the oblique fracture group was on average 5.5 years old and the children that had a transverse fracture were on average 6.6 years old, which can mean that age has an impact
on healing rate. They mention that this is in line with the findings of Hansen et al. (1976) who reported longer healing times in older children regardless of the type of fracture. Besides the difference in oblique and transverse fractures Domzalski et al. (2006) also found that time to consolidation of the fracture was shorter for younger children than for older children, namely 5.5 weeks for children younger than 6 years and on average 8 weeks for children older than 6 years. This indicates that fractures of patients of younger age heal faster than when they are older.
It is also notable that the studies with the oldest age groups, namely Islam et al. (2000), Davidson et al. (2016), Song et al. (2019), Gholson et al. (2011) and Mukhtar et al. (2018) observe the longest healing times.
Sherr-Lurie et al. (2011) is one of the studies that looked at 1 day old children and found the shortest healing time as union is already demonstrated in week 2. However, Hosowaka et al. (2017) also looked at birth related fractures, thus at 1 day old new-borns, and found the presence of soft and hard callus from week 4 and 5, respectively. These are very contradicting results and future studies need to examine this further.
6.2.2 Accidental versus non-accidental
Another very important factor that can influence the healing time is whether the injuries are from accidental or abusive origin. Accidental injuries are usually immobilized by for instance a cast, or when necessary, treated with internal or external fixation, quite fast after the bone has been fractured. Abusive injuries are normally not immobilized or in a later stage and are thus subject to repetitive trauma (Sanchez et al., 2013). This is why the studies of Walters et al. (2014), Sanchez et al. (2013) and Fadell et al. (2017) are of importance as they did not include immobilized fractures, which corresponds more to abusive fractures. In addition, cast-immobilization can also hamper detailed evaluation of fracture healing which makes the age estimation of a fracture less reliable. Gholson et al. (2011) differentiate between acute and chronic fractures. Acute fractures are defined as treated within six weeks after injury and chronic fractures as treated after six weeks of injury. They found that chronic fractures take approximately 9 weeks longer to heal than acute fractures. Since abused children will not be brought to the hospital right away it is more difficult to estimate the age of a fracture, because chronic fractures behave in a different way. In their study they found that when cast-immobilization was attempted with chronic fractures only in 23% of the cases union was demonstrated without also needing surgical treatment. The success-rate of cast-immobilization of acute fractures was 90%. This shows that acute and chronic fractures heal and behave in different ways and this must be taken into account when estimating the age of a fracture.
Since some studies included fractures caused by abuse or of unknown origin the precise time of injury may be questionable as caregivers can lie about the age of the fracture (Domzalski et al., 2006; Halliday et al., 2011; Huckstadt et al., 2007; Malone et al., 2011; Sanchez et al., 2013). As a result, the observations done in these studies can possibly be not reliable enough and have to be handled with care.
It is also important to take into account that abused children may suffer of multiple fractured bones or head trauma. This may also influence the healing rate of bones, and when examining abused children this must not be forgotten. Moreover, with abused children chances are high that they are also neglected and do not receive the right nutrition, which can also affect the healing process (Pickett, 2015).
6.2.3 Bone structure and type of fracture
The studies also had different inclusion criteria when it came to what structure of the bone was fractured. This may also be an influencing factor on the healing rate of a fracture. Halliday et al. (2011) mentioned in their article that they excluded metaphyseal fractures, because these are less symptomatic and it is more difficult to estimate the age of these fractures. Other studies mention what structures they in- or exclude but do not clarify why. Gholson et al. (2011) made a distinction between the distal and proximal pole and the waist of the scaphoid bone. The difference between the healing time (range 6-15 weeks) is quite large which suggests that the bone structure also has an influence on the healing rate of a fracture.
In Warner et al. (2017) a distinction is made between complete and incomplete fractures, which they categorize as: ‘complete: fracture with both cortices disrupted’ and ‘incomplete: fracture having only a single cortex disrupted’. They state that callus is present in most cases between day 9 and 26 when both fractures are included but a peak is seen in the presence of callus between days 9 and 36 when only complete fractures are examined. This indicates that besides the possible influence of what bone structure is fractured also the type of fracture can influence the healing rate. This is in line with the findings of Domzalski et al. (2006) when they looked at the different healing rates of oblique and transverse fractures. They discovered that oblique fractures heal quicker than transverse fractures. However, as already mentioned there was also a difference in age.
6.3 Research limitations
The main limitation of this systematic review is that the included studies used different radiographic healing features which makes it hard to compare their results. Other limitations are that most of the studies are retrospective and that they have small samples sizes.
6.3.1 Radiographic healing features
The fact that the studies used different radiographic features makes is very hard to compare their results. For example, some studies use the term ‘bridging’ while other studies try to observe ‘union’. Whether these two features mean complete bridging or union or just imply the first moment the separated bone segments touch each other is not always clear since not all studies defined the terms they used. The studies that did described the term union, had slightly different definitions. Song et al. (2019) defined union as ‘formation of a complete bridging callus with obliteration of the initial fracture lines’ while
Mukhtar et al. (2018) characterized it as: ‘healing of at least three cortices in both views’, (both views being anteroposterior and 45° cephalic tilt view) and Gholson et al. (2011) described union as ‘>50% bridging trabeculae on radiographs’. These are already three different definitions to determine when union is present. Some studies thus look for complete union and other already say union is present when three cortices have healed. Because of this, union is described earlier is some studies and later in others which implies that the healing rates differ, even though they could be the same when one definition is used.
Besides the different definitions for union also bridging was de described in various ways in the studies. Warner et al. (2017) and Fadell et al. (2017) had the same definition for bridging: ‘fracture site has been completely crossed by intact callus on both sides of the bone, regardless of whether the fracture line still remains visible centrally’. Prosser et al. (2012) already had a slight difference: ‘the loss of fracture line definition with complete bridging of the fracture gap by a soft or hard callus’ and Islam et al. (2000) defined bridging as: ‘bridging (partial or whole) at the fracture site, defined as a loss of fracture margins’. The first three definitions state that bridging should be complete while Islam et al. (2000) mentions that it can also just be partial. This changes the timing when bridging is observed and makes comparison not completely accurate.
The same problem occurs with the presence of callus as some studies make a distinction between soft and hard callus or look at endosteal callus while other studies just use the term callus. Halliday et al. (2011) deliberately did not separate soft and hard callus because according to them these are histological terms that do not have clear defined radiological terms, while Hosowaka et al. (2017) and Prosser et al. (2012) distinguish between soft and hard callus as previously described in section 2.2. However Halliday et al. (2011) tried to describe the presence of SPBNF and callus, but said that they could not distinguish between these two.
It would be better if before further research standard definitions are generated that can be used by everyone in the field and that a standard protocol is produced to ensure that all radiologists define the radiological features in the same way.
6.3.2 Retrospective studies
All but three studies included in this review are retrospective studies (Islam et al., 2000; Mukhtar et al., 2018; Song et al., 2019). Because of this the radiographs are not taken explicit for estimating the age of a fracture, but the images are obtained for the evaluation of the patient’s clinical status. This means that the time between the radiographs varies and that also the amount of radiographs per patient differs. This makes it difficult to correctly observe when callus is present or when fracture union is complete, since some of the patients may not even have a follow up radiograph while in other patients the fracture already healed and the radiograph was taken days or weeks later. In addition, Davison et al. (2016) only made a radiograph 12 weeks after the first appointment as in the other appointments only grip strength and pain was
assessed. This makes it hard to interpret their results since bridging and union could already have happened earlier.
6.3.3 Sample size
A lot of the included studies mention in their articles that a limitation was the small sample size. It can also be seen in table 1 that some studies examined a low number of patients which can give an erroneous image of the reality. Another flaw of a small sample size is that it limits the power to detect differences in the healing rate. This is also the reason that case studies are not included in this review since healing rate may also be individual specific and no average can be taken from these studies.
7. Conclusion
Estimating the age of a fracture can be very helpful in detecting child abuse. In order to do this as precisely as possible, information about bone healing and healing rates is necessary. In this review is has been shown that the studies which have been conducted show different results and are also hard to compare. The main reason for this is that studies use different radiographic healing features and even when the same term is used multiple definitions exist. The aim of this thesis is to answer the research question: ‘Is it possible to detect a difference in healing rate between various bones when dating fractures with conventional radiography in children from 0 to 17 years old?’. This is possible, since different healing rates have been found when bones are compared. However, also when the same bone is examined in multiple studies, different healing rates have been found. It can be said with caution that arm fractures heal faster than leg fractures, but also contradicting results are found here. Furthermore, broad ranges are found in which radiographic features are observed which makes it difficult to estimate the age of a fracture in a precise manner and still only makes it possible to make an estimation in weeks. Besides the influence of fracture location also the age of the patient, the type of fracture such as oblique and transverse and the structure of the bone that is fractured such as the diaphyseal and metaphyseal part seem to influence the healing rate. In conclusion, it is hard to say whether the detected differences in healing rate are caused by fracture location or that other factors have equal or more influence. This underscores the need for more research in the field of fracture healing. Therefore, this review highlights that estimating the age of a fracture should be done very carefully, maybe even more so than is done nowadays. 8. Future research
Future research should focus on multiple things. First of all, future studies should use the same radiographic healing features and use the same definitions to make sure that radiologists assign the features at the same moment. Secondly, it is very important that the healing rates of different bones are not grouped together, but that they are mentioned separately. Besides separating the bones, it could be relevant to separate the bone structures and the type of
fracture since these factors can also influence the healing rate. Thirdly, adequate age groups should be created and these should be examined individually. A fourth factor that should be considered is that fractures as result of abuse can give different healing rates, because immobilization is often postponed and malnutrition can influence the healing rate of a fracture.
The effect of multiple fractures or head trauma should be examined since this review did not differentiate between these patients. Also greenstick fractures were not included in this review and because these mostly occur in children examining the fracture rate can be promising for future research.
When taking these factors into account, more accurate information can be found that can help by creating more detailed timetables which radiologists can use by estimating the age of a fracture more objectively.
9. References
Alink, L., Pannebakker, F., van Berkel, S., Liniting, M., & Klein Velderman, M. (2018). Kindermishandeling in Nederland Anno 2017: De Derde Nationale Prevalentiestudie Mishandeling van Kinderen en Jeugdigen (NPM-2017).
Tijdschrift Voor Gezondheidswetenschappen.
https://doi.org/10.1007/s12508-013-0137-x
Betts, J. G., College, T. J., Desaix, P., Johnson, J. E., Wise, J. A., Womble, M., & Young, K. A. (2013). Anatomy & Physiology. Retrieved from
https://d3bxy9euw4e147.cloudfront.net/oscms-prodcms/media/documents/AnatomyandPhysiology-OP_xxKIcSo.pdf Bilo, R. A. C., Robben, S. G. F., & Rijn, R. R. (2010). Forensic Aspects of
Pediatric Fractures. Forensic Aspects of Pediatric Fractures.
https://doi.org/10.1007/978-3-540-78716-7
Child Protection Evidence Systematic review on Fractures. (2018). Retrieved
from https://www.rcpch.ac.uk/sites/default/files/2019-02/child_protection_evidence_-_fractures.pdf
Cumming, W. A. (1979). Neonatal skeletal fractures. Birth trauma or child abuse? Canadian Association of Radiologists Journal.
Davison, P. G., Boudreau, N., Burrows, R., Wilson, K. L., & Bezuhly, M. (2016). Forearm-Based Ulnar Gutter versus Hand-Based Thermoplastic Splint for Pediatric Metacarpal Neck Fractures: A Blinded, Randomized Trial. In
Plastic and Reconstructive Surgery.
https://doi.org/10.1097/01.prs.0000479974.45051.78
Domzalski, M. E., Lipton, G. E., Lee, D., & Guille, J. T. (2006). Fractures of the distal tibial metaphysis in children: Patterns of injury and results of
treatment. Journal of Pediatric Orthopaedics.
https://doi.org/10.1097/01.bpo.0000218523.82996.76
Fadell, M., Miller, A., Trefan, L., Weinman, J., Stewart, J., Hayes, K., & Maguire, S. (2017). Radiological features of healing in newborn clavicular fractures.
European Radiology. https://doi.org/10.1007/s00330-016-4569-y
Gholson, J. J., Bae, D. S., Zurakowski, D., & Waters, P. M. (2011). Scaphoid fractures in children and adolescents: Contemporary injury patterns and factors influencing time to union. Journal of Bone and Joint Surgery - Series
A. https://doi.org/10.2106/JBJS.J.01729
Halliday, K. E., Broderick, N. J., Somers, J. M., & Hawkes, R. (2011). Dating fractures in infants. Clinical Radiology.
https://doi.org/10.1016/j.crad.2011.06.001
Hansen, B. A., Greiff, J., & Bergmann, F. (1976). Fractures of the Tibia in Children. Acta Orthopaedica Scandinavica, 47(4), 448–453.
https://doi.org/10.3109/17453677608988718
Hosokawa, T., Yamada, Y., Sato, Y., Tanami, Y., & Oguma, E. (2017).
Subperiosteal new bone and callus formations in neonates with femoral shaft fracture at birth. Emergency Radiology.
https://doi.org/10.1007/s10140-016-1462-6
Huckstadt, T., Klitscher, D., Weltzien, A., Müller, L. P., Rommens, P. M., & Schier, F. (2007). Pediatric fractures of the carpal scaphoid: A retrospective clinical and radiological study. Journal of Pediatric Orthopaedics.
https://doi.org/10.1097/01.bpb.0000271309.05924.80
Islam, O., Soboleski, D., Symons, S., Davidson, L. K., Ashworth, M. A., & Babyn, P. (2000). Development and duration of radiographic signs of bone healing in children. American Journal of Roentgenology.
https://doi.org/10.2214/ajr.175.1.1750075
Malone, C. A., Sauer, N. J., & Fenton, T. W. (2011). A Radiographic Assessment of Pediatric Fracture Healing and Time Since Injury. Journal of Forensic
Sciences. https://doi.org/10.1111/j.1556-4029.2011.01820.x
Marsell, R., & Einhorn, T. A. (2011). The biology of fracture healing. Injury. https://doi.org/10.1016/j.injury.2011.03.031
Mckinley, T. (2003). Principles of Fracture Healing.
Mukhtar, I. A., Yaghmour, K. M., Ahmed, A. F., & Ibrahim, T. (2018). Flexible intramedullary nailing versus nonoperative treatment for paediatric displaced midshaft clavicle fractures. Journal of Children’s Orthopaedics. https://doi.org/10.1302/1863-2548.12.170185
O’Connor, J., Cohen, J., & Kleinman, P. K. (1998). Diagnostic imaging of child
abuse (2nd edn). Diagnostic Imaging of Child Abuse. Mosby.
https://doi.org/10.1017/CBO9780511862366
Offiah, A., & Hall, C. (2009). Radiological atlas of child abuse (1st edn). NY:
Radcliffe Medical Publishing. https://doi.org/10.1007/s00247-010-1819-z
Pickett, T. A. (2015). The challenges of accurately estimating time of long bone injury in children. Journal of Forensic and Legal Medicine.
https://doi.org/10.1016/j.jflm.2015.04.012
Prosser, I., Lawson, Z., Evans, A., Harrison, S., Morris, S., Maguire, S., & Kemp, A. M. (2012). A timetable for the radiologic features of fracture healing in young children. American Journal of Roentgenology.
https://doi.org/10.2214/AJR.11.6734
Sanchez, T. R., Nguyen, H., Palacios, W., Doherty, M., & Coulter, K. (2013). Retrospective evaluation and dating of non-accidental rib fractures in infants. Clinical Radiology. https://doi.org/10.1016/j.crad.2013.03.017 Sherr-Lurie, N., Bialik, G. M., Ganel, A., Schindler, A., & Givon, U. (2011).
Fractures of the humerus in the neonatal period. Israel Medical Association
Journal.
Song, M. H., Yun, Y. H., Kang, K., Hyun, M. J., & Choi, S. (2019). Nonoperative versus operative treatment for displaced midshaft clavicle fractures in adolescents: A comparative study. Journal of Pediatric Orthopaedics Part B. https://doi.org/10.1097/BPB.0000000000000535
Walters, M. M., Forbes, P. W., Buonomo, C., & Kleinman, P. K. (2014). Healing patterns of clavicular birth injuries as a guide to fracture dating in cases of possible infant abuse. Pediatric Radiology. https://doi.org/10.1007/s00247-014-2995-z
Warner, C., Maguire, S., Trefan, L., Miller, A., Weinman, J., & Fadell, M. (2017). A study of radiological features of healing in long bone fractures among infants less than a year. Skeletal Radiology. https://doi.org/10.1007/s00256-016-2563-8
Wraighte, P. J., & Scammell, B. E. (2007). Principles of fracture healing.
Foundation Years. https://doi.org/10.1016/j.mpfou.2007.09.005
Wulff, R. N., & Schmidt, T. L. (1998). Carpal fractures in children. Journal of
Pediatric Orthopaedics.
https://doi.org/10.1097/01241398-199807000-00011
Yeo, L. I., & Reed, M. H. (1994). Staging of healing of femoral fractures in children. Canadian Association of Radiologists Journal.
10. Appendix
Search strategy:
((Fracture healing [Mesh] OR fracture healing [tiab] OR bone regeneration [Mesh] OR bone regeneration [tiab] OR fracture dating [tiab] OR fracture timing [tiab] OR age estimation [tiab]) AND (infant [Mesh] OR infant* [tiab] OR neonat* [tiab] OR newborn [Mesh] OR newborn [tiab] OR child [Mesh] OR child* [tiab]) AND (radiology [Mesh] OR radiology [tiab] OR radiography [Mesh] OR radiography [tiab] OR radiologic* [tiab] OR Fractures, Bone/diagnostic imaging [MAJR]))
