Mechanical or thermal damage: differentiating between underlying mechanisms as a cause of bone fractures

Thirty-eight fresh-frozen human cadaveric radii and ulnae (19 radii and 19 ulnae) were collected from the body donation program of the Department of Medical Biology of the Amsterdam University Medical Centre, location Academic Medical Centre (AMC), The Netherlands. The bodies, from which the specimens were taken, were donated to science in accordance with Dutch legislation and the regulations of the medical ethical committee of the AMC (see the section compliance with legal and ethical standards at the end of this manuscript). The forearms were collected and stored at − 20 °C prior to the study, thawed, and subsequently manually defleshed with a scalpel and stored at 4–8 °C throughout the course of the study. Population data (age, sex, and bone dimensions) were collected; 22 bones originated from females and 16 bones from males; for further details, see Table S4 in ESM Section 2. Since age and medical history of the cadavers are confounding factors [8], the selection of tissues was based on the following criteria; all donations fell in the “old” adult category (50 + years, based on Buikstra and Ubelaker, 1994, [22]), without evident signs of osteoporosis (i.e., notable brittleness or strongly deviating weight) or known, from the medical history of the deceased, bone cancer, local surgery, known pre-existing or healed trauma, or other bone defects. The age ranged from 64 to 88. Eight out of the 38 bones were used in pilot studies (least weighing, to decrease variance in weight for the actual experiment) targeted at determining the optimal conditions and factors for (i) generating the BFI-fractures and (ii) burning of the bones—more details on the pilot study and results can be found in ESM Section 3. The conditions for the main experiment were set based on the results of the pilot study (ESM Section 3). The remaining 30 bones were used for the main experiment. Nitrile gloves were worn at all times during the study to prevent contamination.

The 30 bones for the main experiment were split into 3 groups (Table 1) with 10 bones per group (see Table S8 in ESM Section 4 for specimens of each group). Sex, type of bone (radius/ulna), and dimensions were distributed as evenly as possible across the groups. In each group, 6 bones belonged to 6 different individuals whereas the remaining 4 bones (2 sets of forearms) belonged to 2 different donors (1 set from each cadaver) resulting in donations of 8 individuals to each group. The distribution of bones in every group is detailed in Fig. S9 and Table S10 in ESM Section 4.

Of the 30 thawed and defleshed bones, 20 bones (radii/ulnae) were fractured using a custom-made contraption resembling a pendulum apparatus (Fig. 1). The rod attached to a cylindrical impactor, which together weighted 3 kg and was 50 cm in length, mimicked the common weapon type involved in blunt force trauma cases such as a baseball bat [8]. The ideal angle and positioning of bone used on the contraption, for desired fracture production, were derived from the pilot study (see ESM Section 3). The rod was brought up to 80° and released at that angle, thereby creating a potential impact force of 2066 to 4650 N as a result from the kinetic energy (see ESM Section 3 (i) for more details).

Bild vergrößern

Bones were positioned in pronation on the contraption, in a constant manner to standardize fracture production—posterior side of bones facing the impact rod with the radial/ulnar tuberosity facing away from the rod, middle of diaphysis centered, and head of radius/olecranon of ulna placed on the right holder (when facing the contraption from the front) (see Fig. 1). After BFI, each bone was reconstructed carefully for photographic documentation, using a DSLR Nikon D700 with a 35-mm F2.0 AF-D lens, and analysis.

The bones were burnt at the Fire Station Soest, in a room of an isolated concrete firehouse, which resembled a basic residential living room, mimicking a context frequently encountered in forensic casework (Fig. 2). The burning experiment was conducted on a sunny day with mild wind, an average ambient temperature of 26.1 °C, and a relative ambient humidity of 65% at the time of the burning experiment. The pyre consisted of a mixture of hardwood (± 8 kg)-softwood (± 3.5 kg) under two (120 × 80 cm) wood pallets. A foam (polyurethane) mattress covered with cotton mattress-cover (70% cotton) was placed on the pallet. The two groups of bones, control-burn (A) and BFI and burn (C), were encased into “packages” using mesh-wire (hexagonal-shaped, 13 mm perimeter), 100%-cotton shirts, and pig skin (Sus scrofa dom., 15 kg and approximately 1.33 cm of adipose tissue as substitute for human skin with subcutaneous adipose tissue which serves as a fuel). In each package, all bones were covered together, as a group, by the skin and cotton. The entire set-up (Fig. 2a–e) was ignited using 4 cardboard paraffin-blocks of approximately 4cm³ each and a lighter. The windows and doors of the firehouse were opened for ventilation. The burning duration and temperature were derived from the pilot study (ESM Section 3). The bones were burned for approximately 50 min. The approximated temperature of the fire ranged between 700 and 800 °C, based on the potential temperature of the used fuels (mainly wood, the mattress, and pig skin). The actual temperature range was estimated during the experiment based on measurements by means of an infrared thermometer (PCE-890U, measuring range − 50 to 1150 °C, accuracy when measuring above 500 °C was ± 2 °C), and flame color. The fire was extinguished by a professional firefighter with a soft to mild water spray (3 min of spraying). The bones were carefully collected from the debris after a cooling down time of approximately 1 h, photographed positioned as the bones were found in situ, with a Nikon D700 with a 35-mm F2.0 AF-D lens, documented, and placed in plastic buckets for transportation to the institute (location AMC). Nitrile gloves were worn during salvaging. The samples were handled with tweezers, for safety of researchers and to prevent contamination of discoloration after the burning experiment. The residual water from the bones was removed by carefully blotting the bones with tissue paper. The bones were left to further cool and dry for 24 h before subsequent analysis in the laboratory.

Bild vergrößern

All bones were again photographically documented after transportation from the firehouse to the institute and after each data collection step, to assess any damage to bone from handling throughout the fracture analysis process. Fractures (denoted as Fx in data) and their features (Table 2 and Fig. 3) were analyzed according to the data collection sheets derived from former studies (Fig. S3a and S3b in ESM Section 1). The descriptions of the features were made more specific to avoid ambiguity (see Glossary Table S26 in ESM Section 11). Macroscopic and microscopic observations of these bone fractures were made according to this data collection sheet.

Bild vergrößern

The prevalence of features was evaluated, intra-group and inter-group, and the main comparisons are denoted in Table 3. Microscopic observations were made using the Olympus SXZ9 stereomicroscope, under × 8 to × 10 magnification. ImageJ software was used to analyze the features and determine angle of fractures, as described in the next sub-section.

For the angle of fracture, apart from microscopic observation, the photographs of the bones were used to draw a negative slope (downhill) across the proximal end of the fracture, spanning from the compression to tension side of the bone. Thereafter, the angle was measured as in Fig. 4, thrice (blindly) at different times, to account for observational errors. The mean angle for each bone was then determined. The standard error of measurement and the angle of fracture were calculated for each bone.

Bild vergrößern

The burned bones were photographed in the laboratory, 14 days after the burning experiment. Bones from groups control-burn (A) and BFI and burn (C) were arranged on 4 white sheets of paper and photographed from top view. Photographs were taken using a DSLR, Nikon D700 with Nikon 35 mm AF-D F2.0 and 24 mm AF-D F2.8 lenses, then adjusted for major white balance and converted to L*a*b color space (ImageJ – color transformer plugin) to determine the L (lightness) and B-coordinate. Lighting was equalized and the camera color calibrated with an X-rite ColorChecker Classic [13]. Due to brittleness of bones, the anterior portion of the bones was photographed. For each bone, 4 sections (Fig. 5) were manually selected to calculate the average L* and B* values for subsequent temperature estimation by means of the published model by Krap et al. (2019) [13]. The correlation between color changes and exposure temperatures from Krap et al. (2019) study was used to determine the phase of burning each bone was in (95% overall accuracy) [13]. The temperature ranges observed at each of the 4 regions were then used to calculate the average temperature of burning at the intermediate regions and epiphyseal (ends) regions of the bone (see Fig. 5). This is to account for possible temperature variations that might occur at these regions, caused by the differential exposure to heat during the burning. The fractures, at either region, were analyzed in terms of quantity and classification of fracture: post-mortem pre-fire fractures, HIBF, and/or iHIBF. This approach was aimed at identifying patterns occurring between exposure temperature and the types of fractures (see Table S11 in ESM Section 5) and their features at each region and across the bone.

Bild vergrößern

Statistical analysis was performed in Microsoft® Excel and IBM SPSS Statistics 25. The performed group comparisons are shown in Table 3. Descriptive statistics was done for all groups.

Fracture features correspond to categorical (color, stage of burning, fracture outline, etc.) and numerical data (dimensional changes); therefore, Kruskal–Wallis test and one-way ANOVA analysis were done to test for significant statistical differences (p < 0.05) on features between groups. All the fractures in every bone, in all three groups, were analyzed and included in the statistical comparisons performed on the groups. These analyses enabled the identification of BFI/HI-specific features and/or any overlapping features. Due to the limited sample size of n = 10 per group and presence of repetitive values for some features, Q-Q plots were made to determine normality of data and a Levene’s test was done to test for homogeneity of variance, both being requirements for one-way ANOVA analysis. ANOVA was chosen over a t-test as it provides more information about the partitioning of variances and is more flexible with analyzing the data [23]. Kruskal–Wallis (KW) test was done to analyze the categorical data. The categorical data was numerically labeled/scored to aid in this analysis, as tabulated in Table S11 in ESM Section 5. Mann–Whitney (MW) test was performed to determine differences in changes in lengths of bones and ratios of lengths between unburned and burned group BFI and burn (C) bones. This test was also used to determine differences between number of fractures at the epiphyseal ends and those at the intermediate region. Kruskal–Wallis test was used for analyzing differences in fracture class (post-mortem pre-fire, HIBF, or iHIBF) at the epiphyseal and intermediate regions. Finally, the observed features were cross-compared with current literature, in terms of feature description and occurrence in each fracture class, to assess and explain similar and conflicting results.

After exposure to BFI, all 20 bones from group control-BFI (B) and group BFI and burn (C) (pre-burning state) showed at least one fracture and 15 out of these 20 bones showed fractures at the intermediate part of the bone (see Fig. S13–S14 in ESM Section 6 and Tables S22–S23 in ESM Section 10). Twelve out of 20 bones showed fragmentation varying in size. Out of the 20 bones, there were 11 bones with complete-comminute fractures and 9 bones with complete-simple fractures. All fractured bones showed a smooth fracture surface morphology. Helical/curved was the most common fracture outline (9 out of 20 bones), followed by transverse (5 out of 20 bones), diagonal (4 out of 20 bones), columnar (1 out of 20 bones), and longitudinal-transverse (1 out of 20 bones). Twelve out of 20 bones were oblique fractures, followed by spiral, transverse, comminute, and one segmental fracture. Group B resulted in greater fracture lengths (predominantly 10–15 cm) on the proximal side than the distal side of the bone, and vice versa was observed for group C. The ratios of fracture length of tension to compression side were greater on the proximal side than on the distal side of the bones for group B while the ratios were the same for either sides of group C. As expected, these ratios were nearly the same for proximal and distal sides of the bones with transverse fractures (see bone sample UB7 and UB9 in Table S22 in ESM Section 10). Only one bone (see bone sample RC7 in Table S23 in ESM Section 10) showed a segmental fracture, and thus, this bone was not used for statistical analysis in order to avoid large deviations in fracture types.

The fire reached high temperatures, with a maximum of 800 ± 20 °C, about 25 min after ignition, when it spread and grew. It was stable for about 15 min, showed mainly bright red flames, before dying down for the last 10 min with consumption of fuel at the outskirts of the pyre, while the center still had fuel.

Group control-burn (A) bones were evaluated as a group, with new specimen/sample numbers, due to the cominglement of bones during the burning experiment (Fig. S12 in ESM Section 6 and Table S24 in ESM Section 10). Group BFI and burn (C) bones were individualized using the fracture margins, to original specimen numbers, thus being easily comparable to pre-burning results (Fig. S14 in ESM Section 6, and Tables S23 and S25 in ESM Section 10).

Both groups showed fragmentation, but this feature was not statistically evaluated since some fragments were missing and some broken during post-fire recovery or further broken during transport to laboratory. Fifteen out of 20 bones showed large areas of partial calcination and small areas of carbonization; the remaining bones were predominantly carbonized with small regions of partial calcination. Four out of the remaining five bones were partly burnt and partly still relative fresh and the last one was carbonized. Moreover, the bone excluded for evaluation after post-BFI fracture analysis was not examined post-burning (see bone sample RC7 in Table S23 in ESM Section 10). The discoloration and state of burning of both groups did not show any distinct trend based on visual color assessment, although it seemed that calcination began from the intermediate portion of the bone and proximal ends were more carbonized than the rest of bone. Group A bones showed transverse fracture outlines (4 out of 5 bones which showed fractures post-burning), while group C bones mostly showed helical/curved (5 out of 20 bones) and diagonal fracture outlines (4 out of 20) comparable to the findings before burning (see previous section). Both groups exhibited fracture margins corresponding to margins observed after BFI; however, group A had smooth fracture surface whereas group C largely had rough fracture surface and some smoothness in curved/sloped regions of the fracture surface near the fracture margins (Fig. 6).

Bild vergrößern

Identified HIBFs were prominently longitudinal, followed by straight transverse and step HIBFs in both groups while iHIBFs were predominantly longitudinal. All bones were very brittle, with group C bones being more brittle than group A. There were less bones with iHIBFs in group A (6 out of 10) than group C (8 out of 10). iHIBFs showed more carbonization of the inner part of the bone and more calcination of the outer layers of the bone; this pattern was evenly spread across the fracture surface. The fracture margins of iHIBFs were mostly calcined, with an almost gradual pattern turning inwards from white to gray. The identified post-mortem (pre-fire) fractures showed less carbonization and more calcination of the inner layers of the fracture surface after burning. The fracture margins of the post-mortem pre-fire fractures displayed an uneven pattern of carbonization and calcination (see Fig. 7). The HIBFs were similar in discoloration pattern to the post-mortem, pre-fire fractures post-burning.

Bild vergrößern

The fractures of both the HIBFs and iHIBFs penetrated into the medullary cavity, but one of the fracture margins of iHIBFs was elevated while those of HIBFs were in the same plane (Table 4). iHIBF near the BFI site showed these one-sided elevations as well. After reconstruction of diaphyseal sections, it became clear that longitudinal fractures with one-sided elevation appeared on both fragments of the post-mortem pre-fire fracture (post burning), with the elevated margin on the same side. HIBFs near the BFI site did not show any elevation.

Heat borders were also observed close to the BFI sites of group C bones, denoting transition of burning phases of bone. Microstructural fractures were found in partially calcined areas of bones. Large, calcined bone fragments and corresponding smaller fragments could be pieced back together and reconstructed more easily than carbonized bone fragments with related smaller fragments. Since direct comparison of bones was not possible, the combined data of unburned bones was compared with the combined data of burned bones of group C. Most burned bones exhibited blunt fracture margins; thus, the sharp edges resulting from BFI had lost its sharpness. The burned group C bones mostly showed longitudinal and some stepped HIBFs. The iHIBFs observed in group C were largely longitudinal fractures with elevated fracture margins on one side. As for bone dimensions, most bones in group C exhibit shrinkage. Only one bone (UC10), however, showed increased lengths for both proximal and distal part of the bone, denoting an expansion across the bone (see Table S25 in ESM Section 10). The ratios of length on tension to length on compression for proximal and distal bone had mostly decreased.

From visual observation, the majority of the bones showed both carbonization and calcination (Fig. 8). Based on the colorimetric model published by Krap et al. (2019), the majority of bones in group control-burn (A) and BFI and burn (C) were estimated to have been exposed to a maximum temperature of 450 to 700 °C (Tables S15–S16 in ESM Section 7); this shows the equality in burning conditions [13]. There were more fractures in the intermediate region than at the epiphyseal ends for both groups. HIBFs were the most common class of fractures across bones in group A (Table S17 in ESM Section 7). HIBFs and iHIBFs were the most common classes of fractures across bones in group C (Table S18 in ESM Section 7). In group C, there were more HIBFs and iHIBFs seen at the epiphyseal ends and more post-mortem (pre-fire) fractures at the intermediate region. There were more fractures in the intermediate region than at the epiphyseal ends for both groups control-burn (A) and group BFI and burn (C) (see Fig. 9a–d) (for further details, see Fig. S19–S20 in ESM Section 7). There seemed to be more carbonization, than calcination, at the epiphyses of most bones amongst both groups. No distinctive trend has been found when comparing the temperature of burning at the intermediate and epiphyseal ends and the classes of fractures (post-mortem pre-fire, HIBFs, or iHIBFs) observed at these regions.

Bild vergrößern

Bild vergrößern

Numerical data met the requirements for ANOVA, Q-Q plots exhibited normality, and Levene’s test showed homogeneity of variances. The categorical data failed to meet the ANOVA requirements; since the observations were independent of one another, the KW-test was performed. Table 5 shows the results from all statistical analyses which are discussed in detail below. Data on the dimensional changes can be found in the ESM Section 8.