SEARCH

SEARCH BY CITATION

Keywords:

  • amputation;
  • antemortem;
  • fracture;
  • histology;
  • human dry bone;
  • radiology;
  • survival

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theory and practice of dating traumatic lesions
  5. Materials and methods
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgements
  10. Conflict of Interest
  11. References

The post-traumatic status of antemortem fractures in human dry bone remains is currently defined as being either ‘healing’ or ‘healed’. However, detailed ‘dating’ of the related post-traumatic time interval would be desirable, since it would aid in assessing individual medical status and care at the time of death. Within forensic pathology practice, fresh tissue healing phases are routinely used as an intrinsic parameter for the length of the post-traumatic time interval. Unfortunately, the direct application of such a method is hampered when applied to dry bone skeletal material.

This study explores the possibility of applying a fracture dating system, drawn forth from the traditional forensic pathology method, on dry bone remains. More specifically, the aim is to establish the extent to which various histomorphological features indicative of specific time intervals of healing are consistently detectable. Human dry bones with fractures and amputations in various phases of healing were studied.

Results show that the complementary use of radiological and histological investigation techniques improves the differentiation between various healing phases and thus allow for a more detailed dating of lesions. For future use, healing features that have proven to be consistently detectable and their related post traumatic time intervals are listed. The system aids in demarcating a considerably more ‘narrow’ post-traumatic time interval than usual. Copyright © 2012 John Wiley & Sons, Ltd.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theory and practice of dating traumatic lesions
  5. Materials and methods
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgements
  10. Conflict of Interest
  11. References

It is generally agreed upon that studying trauma in human remains can yield interesting individual, cultural and environmental information (Krogman and Iscan, 1986; Aufderheide and Rodríguez-Martin, 1998; Ortner, 2003). Indeed, a large pool of literature exists describing and interpreting gross anatomical traumatic lesions observed in skeletons such as fractures, amputations and dislocations (e.g. Berryman and Haun, 1996; Grauer and Roberts, 1996; Mays, 1996; Lovell, 1997; Alvrus, 1999; Galloway, 1999; Judd and Roberts, 1999; Nakai et al., 2001; Anderson, 2002; Holt et al., 2002; Djuriç, 2006; Mitchell, 2006; Lovell, 2008; Redfern, 2009; Van der Merwe et al., 2009).

Notwithstanding its acceptance, investigating traumatic lesions in dry bone tissue is still challenging, since palaeopathologists, and sometimes forensic pathologists and anthropologists, are confronted with the investigation of material that lacks soft tissue (Ubelaker and Adams, 1995; Grauer and Roberts, 1996; Lovell, 1997; Sauer, 1998; Ortner, 2003; Rodríguez-Martin, 2006). Traditionally, the moment that lesions are inflicted is defined as being either ‘antemortem’ (prior to death), ‘perimortem’ (occurring around the time of death) or ‘postmortem’ (after death). Antemortem lesions are defined by the presence of healing features (e.g. callus formation), whereas postmortem lesions are defined by the lack of any healing features. For postmortem lesions that occurred during excavation, the colour difference between fracture surface and periosteal surface may also aid in identifying postmortem lesions (Van der Merwe et al., 2009). Perimortem lesions are generally those that cannot be reliably assigned to either the ante- or postmortem group. A more detailed discussion of the differentiation between ante- peri- and postmortem lesions is beyond the scope of this publication, but can be found in various comprehensive publications (e.g. Quatrehomme and Iscan, 1997; Wheatley, 2008; Wieberg and Wescott, 2008).

Describing lesions as being either, antemortem, perimortem or postmortem has severe limitations as it gives no estimation of the length of the time period between the moment the traumatic insult occurred and the time of death (the post-traumatic interval). Usually, the status of antemortem lesions is expressed as being ‘healing’ or ‘healed’ (see for instance Brickley, 2006). However, a more detailed temporal specification of antemortem time would be desirable, since a more precise ‘dating’ would aid palaeopathologists and forensic anthropologists in interpreting facets such as medical status and medical care at the time of death. More specific information on differences in the timing of multiple injuries in a specific individual may also be essential in evaluating evidence for child abuse and torture (e.g. Maat, 2008). Also, it would assist in determining the sequence of multiple traumata observed in a single individual. The latter could especially be of value to the growing body of research on the occurrence of trauma recidivism (e.g. Judd, 2002; Martin et al., 2010).

Theory and practice of dating traumatic lesions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theory and practice of dating traumatic lesions
  5. Materials and methods
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgements
  10. Conflict of Interest
  11. References

In forensic pathological investigations, estimation of the post-traumatic interval is usually done by analyzing the local soft tissue response (healing), since healing is supposed to be an intrinsic and reliable time indicator (e.g. Oehmichen, 2004). Although in forensic pathology practice dating is mostly applied to soft tissue lesions (e.g. cutaneous wounds), the principle can also be utilized for bone tissue injuries, since bone tissue response also follows a strict time-dependent developmental sequence irrespective of complex variables such as the type of lesion, location, age at death and health status (Todd and Iler, 1927; Frost, 1989; Frost, 1989; Tosounidis et al., 2009; Vigorita, 2009).

The usefulness of healing as an indicator of post-traumatic survival time was, for instance, illustrated in an investigation of war crimes. Following an extensive literature review on fracture healing, a time table was constructed and applied that linked the elapsed time after bone tissue injury to radiographic and histological healing features (Maat, 2008). The approach proved to be adequate for cases regarding adult and paediatric individuals (Maat and Huls, 2010).

Despite the fact that the approach was designed for fracture dating, it should also be suitable for amputations, since stages and timing of bone healing are supposed to be similar (Barber, 1930). Comparisons between Maat's timetable (2008), based on a literature review of microscopic and radiologic observations and the work by Barber remains (1929; 1930; 1934), based on gross anatomical and radiological observations in macerated, supports this statement.

Application in palaeopathology and forensic anthropology

It goes without saying that forensic pathological trauma dating has great potential for implementation in palaeopathology. This was for instance demonstrated by Mays (1996) who applied Barber's gross anatomical approach in a case study regarding healed amputations (Barber, 1929, 1930, 1934; Todd and Barber, 1934). Recently, Cattaneo et al. (2010) also showed that useful histological markers for the dating of traumatic lesions were even traceable in macerated bone. However, macerated bone material, like fresh forensic material, is usually better preserved than archaeological dry bone material. It therefore allows for ‘better’ analysis. As a result, the consistency in the detectability of healing features as used in the approaches of Maat (2008) and Barber (1929, 1930, 1934) are still left to be tested in case of dry bone material.

This study will therefore specifically attempt to: (i) evaluate which features (as described by Barber (1929, 1930, 1934)) and Maat (2008) are adequately assessable in traumatic lesions in dry bone material, and (ii) to determine to what extent they allow for estimation of the ‘age’ of an injury.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theory and practice of dating traumatic lesions
  5. Materials and methods
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgements
  10. Conflict of Interest
  11. References

Material

Dry bone specimens from various collections were assembled in order to ensure a sufficiently large sample for investigation. Only traumata from presumably healthy individuals were included. Because of the destructive nature of histological methods, we were not permitted to sample cranial bone material. Therefore, only post-cranial material was included in the study. There is however no indication that the healing process, and its timing would be different in cranial bones from that in post-cranial ones (Frost, 1989a, 1989b; Vigorita, 2009).

A total of 22 specimen with fractures and nine amputations in various phases of healing were included in this study. They were excised from 21 individuals. Twelve fractures and seven amputations were obtained from the Gladstone skeletal collection from Kimberley, South Africa (Van der Merwe et al., 2009). One fracture originated from a 17th century Dutch male whaler exhumed on Spitsbergen, Norway (Maat, 1981), and two fractures from a 19th century population from Bloemendaal, The Netherlands. The sample was further extended with six fractures and one amputation from the dissection hall collection of the Leiden University Medical Centre (LUMC). Three intentionally sawn/broken specimens from the Bloemendaal collection were used as controls.

Throughout our assemblage, wherever possible, a (contra lateral) control sample was included for comparison of bone density and histological architecture. General information on the studied bone material is listed in Table 1.

Table 1. Skeletal material
 Individual No.aOriginBoneGross anatomical datingbPreservation phasec
  1. Abbreviations: SA = South Africa; NO = Norway; NL = The Netherlands; Hu = Humerus; Fe = Femur; Ra = Radius; Ul = Ulna; Ri = Rib; Ti = Tibia; Fi = Fibula; MCP = Metacarpal; St = Sternum.

  2. a

    Numbering used in prior published articles.

  3. b

    According to Van der Merwe et al. (2009).

  4. c

    According to Gordon and Buikstra (1981).

FracturesN74.5Kimberly, SAHuAntemortem1
 N38.1Kimberly, SARiPerimortem1
 N38.2Kimberly, SAFeAntemortem1
 N38.2Kimberly, SATiAntemortem1
 N38.2Kimberly, SA3th MCPAntemortem1
 N38.3Kimberly, SARaAntemortem1
 N74.4Kimberly, SAUlAntemortem1
 S2.3Kimberly, SAFePerimortem2
 S2.3Kimberly, SARaAntemortem2
 S2.9Kimberly, SA8th RiAntemortem1
 S3.5Kimberly, SAFePerimortem2
 Sk. 1Unknown, NLFiAntemortem1
 Sk. 2Unknown, NLTiAntemortem1
 Sk. 2Unknown, NLFiAntemortem1
 Sk. 4Unknown, NLFeAntemortem1
 Sk. 5Unknown, NLHuAntemortem1
 Sk. 5Unknown, NLRaAntemortem1
 541Spitsbergen, NOFeAntemortem1
 1033Bloemendaal, NLRiAntemortem2
 1033Bloemendaal, NLRiAntemortem2
 H 1Bloemendaal, NLStPostmortem2
 H 2Bloemendaal, NLFePostmortem2
AmputationsN34.3Kimberly, SATiAntemortem2
 N38.2Kimberly, SAFePerimortem1
 N8.1Kimberly, SATiAntemortem2
 S2.6Kimberly, SATiAntemortem1
 S2.7Kimberly, SAHuPerimortem2
 S2.7Kimberly, SAHuAntemortem2
 S2.7Kimberly, SARa + UlPerimortem2
 Sk. 3Unknown, NLFeAntemortem1
 H 2Bloemendaal, NLFePostmortem2

Methods

All lesions were photographed and studied gross anatomically, to define whether they are either ante-, peri- or post-mortem. For this, we used standard anthropological methods (Van der Merwe et al., 2009). Subsequently, conventional plain radiographs in an anterior-posterior and a medio-lateral direction were taken. The lesion plus a representative length of unaffected bone was imaged together with its contra lateral control specimen (whenever available). To prevent bias by specimen recognition, each image was anonymized by random numbering.

After radiological imaging, histology samples, so called ‘thick slices’, were excised from the centre of each lesion and its control (i.e. perpendicular to the fracture/amputation plane). A new random numbering blinded the identity of these ‘thick slices’. From each thick slice, two thin sections were produced with a minor adaptation according to the method of De Boer et al. (2011): i.e. the embedding medium was also used for cover slipping. One of the thin sections remained unstained; the other was stained with haematoxylin to enhance the visibility of tissue architecture. (De Boer et al. 2012). All sections were microscopically investigated using bright and polarized light.

By combining the work of Barber (1929, 1930, 1934); Maat (2008; pers. comm.) and Maat and Huls (2010), a table was constructed in which healing features were linked to time intervals (Table 2). Features that are impossible to assess in archaeological material, such as soft tissue changes, were not taken into account. In the few cases in which descriptions given by Maat and Barber did not agree (some minor gross anatomical changes), preference was given to the publication of Maat, since it included a larger quantity of and more recent data. The merged data were then used to develop a multiple choice questionnaire that addressed the consistency in detectability by examiners of each of the designated healing features.

Table 2. Healing features and associated post-traumatic time intervals, combined from Barber (1929, 1930, 1934), Maat et al. (2008, 2010) and Maat (pers. comm.)
Category of lesionHealing featureTime interval
  1. a

    Features visible by plain radiographic analysis.

  2. b

    Features visible by histological analysis.

Common• Frayed bone lamellae at the lesion marginsbBefore 48 hours
 • Absorption of the cortical bone adjacent to the lesionaAfter 4–7 days
 • First Howship's lacunae at the lesion marginsbAfter 4–7 days
 • Smoothening of the lesion marginsa, bAfter 4–7 days
 • Start of endosteal and periosteal osteogenesis separable from cortexa, bAfter 7 days
 • Periosteal osteogenesis at distance from the fracture siteAfter 7 days
 • Clearly visible endosteal callus formationa, bAfter 10–12 days
 • Aggregation of spiculae into woven bonea, bAfter 12–20 days
 • Primary bone tissue depositionbAfter 12–20 days
 • Osteoporosis of the cortexa, b

• Margin of the lesion appears more sclerotica

After 12 days

After 12–20 days

 • Start of the transition of primary woven bone into secondary lamellar bonebAfter 14 days
 • Cortical ‘cutting and closing cones’ orientated towards the lesionbAfter 14–21 days
 • Fields of calcified cartilage at sites of callus formationAfter 14 days
 • Clearly visible periosteal callus a, bAfter 15 days
 • Endosteal callus becomes indistinguishable from the cancellous bone in the marrow cavitya, bAfter 17 days
 • Periosteal callus becomes firmly attached (inseparable) to the cortex a, bAfter 6 weeks
Specific for

fractures

• First scattered bone tissue spiculae between the lesion endsa, bAfter 4–7 days
 • Union by bridging of the cortical bone discontinuitya, bAfter 21–28 days
 • Smoothening of the callus outlinebAfter 2–3 months
 • After inadequate immobilization: Pseudoarthrosis developmenta, bAfter 6–9 months
 • After adequate immobilization:Quiescent appearance indicating subsided healinga, bAfter 1–2 years
Specific for• Visibility of cut marks on the amputation surfaceaLess than 13 days
amputations• Start of ‘capping’ of the medullary cavityaAfter ‘not many weeks’
 • Complete capping of the medullary cavityaAfter ‘several months’

By means of this questionnaire, the radiographs and histological sections were assessed. Three examiners assessed the images and sections independently. In the questionnaire, each healing feature could be marked as: ‘yes’ (the described feature was present), ‘no’ (the described feature was not present), a question mark (the provided image/section did not allow for a conclusive answer) or as ‘NA’ (not applicable). Agreement between the three examiners, and thus consistency in the detection of healing features, was measured by calculating a one-way random intraclass correlation coefficient (ICC) for each healing feature for each modality. Prior established categories were used to interpret the ICC. According to Landis and Koch (1977), an ICC greater than 0.6 is considered to reflect substantial agreement between examiners.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theory and practice of dating traumatic lesions
  5. Materials and methods
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgements
  10. Conflict of Interest
  11. References

In the following text, all healing features will be printed in italics. Due to the nature of dry bone tissue, two features of healing described by Maat (scattered bone spiculae and fields of calcified cartilage at sites of callus formation) were not present in any of our samples.

Furthermore, as expected, healing features were absent in the postmortem control samples included in this study.

Results of radiographic analyses

In analysis of the plain radiographs, several common healing features were detected with substantial interobserver agreement (see Table 3). Agreement on the presence of clearly visible callus at the lesion site, both at the endosteal and periosteal aspect was substantial (ICC 0.756 and 0.770). The examiners also agreed upon the existence of remodeling of endosteal callus making it indistinguishable from the cancellous bone (ICC of 0.837). Internal and external callus formation at some distance from the lesion site produced low ICC (0.345 and 0.343). Also, the evaluation of absorption of the cortical bone adjacent to the lesion (ICC of 0. 275), more sclerotic lesion margins (ICC of 0.421), osteoporosis of the cortex (ICC of 0.341), the presence of smoothening of lesion margins (ICC of 0.416) and firm attachment of periosteal callus to the cortex (ICC of 0.154) did not produce substantial interobserver agreement.

Table 3. Intraclass correlation coefficients of healing features, for microscopic and radiographic analyses
Healing featureUnstained histologyStained histologyaX-ray analysis
  1. Abbreviations: ICC = Intraclass Correlation Coefficient. CI = Confidence Interval. NA = Not Applicable.

  2. a

    Haematoxylin stained, according to De Boer et al. (2012).

  3. b

    An ICC of 1.000 indicated no difference in observation between the examiners.

  4. c

    N = 22.

  5. d

    N = 9.

 ICC (95% CI)ICC (95% CI)ICC (95% CI)
Smoothening of the lesion margins.806 (.629–.917).863 (.726–943).416 (.125–.693)
Absorption of cortical bone adjacent to the lesionNANA.275 (.052–.632)
Presence of endosteal callus   
Distant from the lesion site.652 (.470–.798).393 (.178–.608).345 (.131–.580)
At the lesion site.736 (.583–.851).709 (.546–.834).756 (.611–.863)
Presence of periosteally situated callus   
Distant from the fracture site.678 (.505–.815).649 (.455–.799).343 (.126–.568)
At the lesion site.912 (.848–.953)1.000b.770 (.629–.871)
Firm attachement of periosteally situated callus.900 (.766–.967).739 (.484–.902).154 (.000–.532)
Local osteoporosisic of the cortex.664 (.484–806).706 (.540–.832).341 (.116–.574)
Margin of the lesion appears more scleroticNANA.421 (.204–632)
Endosteal callus becomes indistinguishable from the cancellous bone in the marrow cavity.845 (.740–.916).370 (152–.590).837 (.728–.911)
Features specific for fracturesc   
Union by bridging of the cortical bone discontinuity by primary woven bone.882 (.779–.945).629 (.403–.807).608 (.378–.794)
Union by bridging of the cortical bone discontinuity by secondary lamellar bone.443 (.191–.684).764 (.590–.884).939 (.882–.972)
Smoothening of the callus outline.510 (.118–.838).348 (.013–.715).838(.652–.942)
Features specific for amputationsd   
Visibility of cut marks on the amputation surface1.000b.733 (.409–.924)0.000
Start of ‘capping’ of the medullary cavity1.000b1.000b1.000b
Complete closure of the medullary cavity1.000b.857 (.350–.996)1.000b

In fractures, examiners agreed upon the detectability of the smoothening of the callus outline (ICC of 0.838). high ICC were also seen with regard to the union by bridging of the cortical bone discontinuity, irrespective of whether this union was constructed by means of primary woven or secondary lamellar bone (ICC of 0.608 and 0.939).

When considering amputations, no differences in observation were seen between examiners regarding the start of ‘capping’ of the medullary cavity with newly formed bone, or the complete capping of the medullary cavity (both an ICC of 1.000). In contrast to histological analysis, there was no substantial agreement on the detectability of cut marks on the amputation surface in radiological analyses.

Results of histological analysis

On a microscopic level, a number of healing features were detected with substantial interobserver agreement, irrespective of the use of histochemical staining (see Tables 3 and 4). The presence of frayed bone lamellae at the lesion margins (ICC of 0.881 and stained 0.763), smoothening of the lesion margins (ICC of 0.806 and stained 0.863), and the presence of clearly visible callus formation at the lesion site, both endosteally (ICC of 0.736 and stained 0.709) and periosteally (ICC of 0.912 and stained 1.000) showed substantial agreement. Also, firm attachment of callus on the periosteal surface (ICC of 0.900 and stained 0.739) and quiescent appearance of the lesion site after healing has concluded (ICC of 1.000 and stained 0.741) were generally agreed upon in both unstained and stained sections. Agreement regarding early indications of periosteal osteogenesis at distance from the fracture margins were considered ‘borderline’ (ICC of 0.678 and stained 0.649). The examiners did not agree convincingly on the detectability of endosteal osteogenesis at some distance from the fracture site (ICC of 0.625 and stained 0.393).

Table 4. Intraclass correlation coefficients for healing features from histology
Healing featureUnstained histologyStained histologya
  1. Abbreviations: ICC = Intraclass Correlation Coefficient. CI = Confidence Interval.

  2. a

    Haematoxylin stained, according to De Boer et al. (2012).

  3. b

    An ICC of 1.000 indicates no difference in observation between the examiners.

 ICC (95% CI)ICC (95% CI)
Frayed bone lamellae at the lesion margins.881 (.797–.936).763 (.620–.868)
Howship's lacunae at the lesion cleft.700 (.533–.828).575 (.376–.745)
Howship's lacunae at the periosteal surface.814 (.695–.989).599 (.405–.762)
Howship's lacunae at the endosteal surface.903 (.832–948).503 (.290–.695)
Cortical cutting and closing cones oriented towards the lesion.965 (.922–.977).577 (.376–.747)
Remodeling of endosteal callus into secondary lamellar bone.491 (.205–.742)1.000b
Remodeling of periosteal callus into secondary lamellar bone.242 (.000–.627)1.000b
Quiescent histomorphological appearance, indicating ended healing1.000b.741 (.583–.856)

Agreement on the detectability of some common healing features (fractures and amputations) was lower in haematoxylin stained sections (see Tables 3 and 4). In stained sections, agreement was lower on the presence of Howship's lacunae (ICC of 0.700 vs. stained 0.575) and cutting and closing cones (ICC of 0.965 vs. stained 0.577). Also, the final stage of endosteal callus remodeling, when endosteal callus becomes indistinguishable from the cancellous bone, was better detectable in unstained sections (ICC of 0.845 vs. stained 0.370). The agreement on detectability of transition of primary woven bone into secondary lamellar bone in callus was higher in haematoxylin stained sections, both in endosteally (ICC of 0.491 vs. stained 1.000) and periosteally situated callus (ICC of 0.242 vs. stained 1.000).

In fractures, differences existed in the ICCs between observations done in unstained and stained sections. There was, however, substantial agreement upon the detectability of union by bridging of the cortical bone discontinuity by primary woven bone in both unstained and stained sections (ICC of 0.882 and stained 0.629). In contrast, agreement on the presence of union by bridging of the cortical bone discontinuity by secondary lamellar bone was reduced in unstained if compared to stained sections (ICC of 0.443 vs. stained 0.764). The smoothening of callus outline was inconsistently assessed in fractures, irrespective of the application of haematoxylin (ICC of 0.510 and stained 0.348).

When considering unstained and stained sections of amputations, features such as the presence of cut marks (ICC of 1.000 and 0.825), the start of ‘capping’ the medullary cavity with newly formed bone (both an ICC of 1.000) and eventual complete ‘capping’ of the medullary cavity (ICC of 1.000 and 0.875) were all detected with substantial interobserver agreement.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theory and practice of dating traumatic lesions
  5. Materials and methods
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgements
  10. Conflict of Interest
  11. References

The results show that a considerable amount of healing features are still reliably detectable in dry bone tissue. Throughout the results, features indicating the same level of healing were in agreement with each other, supporting the statement that healing happens in an orderly, sequential fashion (e.g. Vigorita, 2009). The results will now be discussed in the time sequence in which they are reported to appear during the healing process (Barber, 1929, 1930, 1934; Maat, 2008; Maat and Huls, 2010). This sequence is also described in Tables 2 and 5.

Table 5. Consistently detected healing features visible in dry bone material
Healing featureTime intervalUnstained histologyStained histologyaPlain radiography
  1. a

    Haematoxylin stained, according to De Boer et al. (2012).

• Frayed bone lamellae at the lesion marginsBefore 48 hoursxx 
• First Howship's lacunae at the lesion marginsAfter 4–7 daysx  
• Smoothening of the lesions marginsAfter 4–7 daysxx 
• Start of periosteal callus formation, distant from the lesion margins, separable from the cortex.After 7 daysxx 
• Endosteal callus formation clearly visibleAfter 10–12 daysxxx
• Osteoporotic appearance of the cortexAfter 12 daysxxx
• Start of the transition of primary woven bone into secondary lamellar boneAfter 14 days x 
• Cortical cutting and closing cones orientated towards the lesionAfter 14–21 daysx  
• Clearly visible periosteally situated callusAfter 15 daysxxx
• Endosteal callus becomes indistinguishable from the cancellous bone in the marrow cavityAfter 17 daysx x
• Periosteal callus becomes firmly attached (inseparable) to the cortexAfter 6 weeksxx 
Features specific for fractures    
• Union by bridging of the cortical bone discontinuityAfter 21–28 daysxxx
○ by primary woven bone xx 
○ by secondary lamellar bone  x 
• Smoothening of the callus outlineAfter 2–3 months  x
• After adequate immobilization: quiescent appearance indicating subsided healing.After 1–2 yearsxx 
Features specific for amputations    
• Visibility of cut marks on the amputation surfaceBefore 13 daysxx 
• Start of ‘capping’ of the medullary cavity‘After not many weeks’xxx
• Complete capping of the medullary cavity‘After several months’xxx

Frayed bone lamellae at the lesion margins are reported to only be observable within the first 48 h after injury. Due to the morphological character and dimensions of this feature, it could only be viewed by histological analyses. The use of histochemical staining did not affect the consistency of the detection. However, it was noted that the use of polarized light dramatically enhanced the visibility of frayed margins (Figure 1).

image

Figure 1. Micrographs of the frayed lesion margin of an amputated femur (S38.2). Undecalcified and unstained section. Bar indicates scale. Figure 1a) The cutting margin at the amputation end shows a frayed, brush-like appearance due to the fanning of bone lamellae. The yellow discolouring at the margin is caused by taphonomical processes. Figure 1b) Same section as 1a, now viewed with polarized light. The use of polarized light clearly enhances the visibility of the frayed margin. This figure is available in colour online at wileyonlinelibrary.com/journal/oa.

Download figure to PowerPoint

After 4 to 7 days, lesion margins will start to become eroded by osteoclasts. Their activity is indicated by the presence of Howship's lacunae. These are only visible on a microscopic level. The use of histochemical staining impeded interobserver agreement. This might be due to confusion in microscopic image interpretation from the histotechnique of section surface staining. This leaves deeper situated tissue within the section unstained.

Osteoclastic activity on the lesion margins results in a smoothening of the lesion margins, at the earliest reported to be visible after 4 days. Good interobserver agreement was obtained irrespective of section staining. The unsatisfactory agreement in case of radiological analysis might be explained by difference in magnification and resolution (ICC of 0.416).

Due to a combination of taphonomic processes and excavation tissue damage, the next step in healing sequence, the formation of bone spiculae between the lesion margins (becoming visible after four to 7 days) can not be found in dry bone specimens.

It was expected that by natural absorption of cortical bone adjacent to the lesion margins, margins would appear less opaque (see Table 2). However, this feature did not perform well in terms of interobserver agreement (ICC of 0.275). This might be explained by the difficulty to differentiate between antemortem healing changes and postmortem taphonomic alterations of the cortex.

After initial osteogenesis in the form of bone spiculae, callus formation starts both on the endosteal and periosteal aspect of the fractured bone. Minor periosteal osteogenesis (callus formation, visible after 7 days) usually starts at some distance from the lesion site (Maat, pers. comm.), i.e. in the ‘corner’ where the periosteum is being lifted from the bone by the subperiosteal hematoma, resulting in an appearance similar to the radiological ‘Codman's triangle’ visible in osteosarcomas. From there it progresses towards and unites with callus formation at the lesion site. Since this phenomenon only exists for a short period of time, it was unfortunately not observed in our sample. As a result, we examined periosteal callus at some distance as a part of the callus formation at the lesion site. This most likely caused the borderline interobserver agreement of the feature (ICC of 0.678 and stained 0.649). However, it is expected that if cases of early periosteal callus formation had been included in our collection, this could very well have shown to be consistently detectable.

Endosteal osteogenesis (callus formation) may start after 7 days, and is usually clearly visible after 10 to 12 days. It was consistently detected in both stained and unstained histological sections as well as during radiological analysis. In addition, we analyzed whether endosteally situated callus, like periosteal callus, also started at some distance from the site of lesion. Related ICC were, however, low with broad 95% confidence intervals, both histologically (unstained ICC of 0.625 and stained ICC of 0.393) and radiologically (ICC of 0.345). This is probably due to the difficulty to differentiate between early endosteal callus formation and the naturally present surrounding cancellous bone.

As healing progresses, local osteoporosis of the cortex may be observable. This is usually not visible before 12 days. Osteoporosis within our sample was only consistently detected in stained and unstained histological sections, and not during radiological investigation (ICC of 0.341). The latter might be due to differences in technical resolution and magnification, resulting in more debatable findings in plain radiographs. This seems to especially hold true for early stages of local osteoporosis. In advanced phases, local osteoporosis becomes clearly visible on plain radiographs (Figure 7).

As new bone formation progresses, the margin of the lesion appears more sclerotic (after 12–20 days, see Tab. 2). The related ICC, however, did not present substantial interobserver agreement (ICC of 0.421). As in the case of local osteoporosis of the cortex, this may be caused by differences in resolution and magnification.

In amputations, the presence of cut marks can be seen up till about 13 days after the traumatic event (see Table 2). This feature was only consistently reported in unstained and stained histological analysis and was not consistently detectable with radiography (ICC of 0.000).

As stated earlier, loose microscopic bone tissue spiculae are undetectable due to postmortem decomposition. The same holds for the fields of calcified cartilage that only will be seen in fresh material after 14 days.

After 14 days, the newly formed endosteal and periosteal callus starts its transition from primary woven bone into secondary lamellar bone. The histological detectability hereof was primarily analyzed with the use of polarized light, both in stained an unstained sections. However, our results suggest that staining increased consistent detection above the 0.6 threshold (endosteal: ICC of 0.491 vs. 1.000 and periosteal: ICC of 0.242 vs 1.000). The phenomenon of improved visibility of separate bone lamellae by the use of haematoxylin was noted before by De Boer et al. (2011, 2012). The improvement in the ICC between stained and unstained sections may therefore be the result of a cumulative effect of staining and polarized light.

The remodeling process is accompanied by an increased number of cortical cutting and closing cones, observable 2 to 3 weeks after injury. Our results suggest a better detectability in unstained sections (ICC of 0.965) if compared to stained sections (ICC of 0.577). As cortical cutting and closing cones are recognized by their characteristic two end conical shape with Howship's lacunae at their ‘cutting’ end, the previously noted negative effect of staining on the visibility on Howship's lacunae might have hampered detectability.

As remodeling progresses, the endosteal callus becomes indistinguishable from the cancellous bone in the medullary cavity (first visible after 17 days). In radiological analysis, this was consistently detected. On a microscopic level, interobserver agreement was much higher in unstained than in stained sections (ICC of 0.845 vs. 0.370). This might be due to the above discussed confusing effect that histochemical staining may have on the visibility of overall bone tissue architecture.

According to Table 2, periosteal callus becomes clearly visible after about 15 days, observable both in stained and unstained histology, and radiological analysis. As callus formation progresses, the cortical defect in fractures is eventually bridged by callus after 21–28 days. This union by bridging of the cortical bone discontinuity was consistently detected in both (un)stained histological and radiological analyses.

Subsequently, the periosteal callus becomes firmly attached (inseparable) to the cortex. This is observable after 6 weeks and was only consistently detected in histological analysis, irrespective of the use of staining. Due to the earlier mentioned low resolution and magnification in plain radiographic images, the ICC was very low (ICC of 0.015). The smoothening of the callus outline in fractures (after two to three months) was only consistently detected during radiological investigation (ICC of unstained 0.510 and stained 0.348 vs. radiology 0.838). The low magnification of routine radiographs now proved to be in favour of tissue overview if compared to histology.

In amputations, healing eventually progresses towards the start of ‘capping’ of the medullary cavity and the eventual complete ‘capping’ thereof. This was consistently detected in unstained and stained sections, as well as in radiological analysis. In fractures, healing eventually subsides, leaving a quiescent histomorphological appearance after 1–2 years. Both stained and unstained sections proved to be consistent.

The authors used interobserver agreement, assessed by means of ICC calculation, as an indicator for the consistency in detection of a healing feature. Cohen's Kappa is the most used statistical calculation for assessing interobserver agreement, but for comparisons between more than two examiners, ICC is regarded superior (Berk, 1979). The ICC is defined as the proportion of true variance relative to total variance. In other words, a high ICC indicates that the method of analysis does not add variance to the total variance between subjects. Since calculation of the ICC uses total sample variation, its value is sensitive to the extensiveness of values in the sample. As a result, calculated ICC values are only ‘reliable’ in a heterogeneous population. As this research uses bone material with sufficient variability between healing phases, interobserver agreement on the presence of a parameter can reliably calculated. When the ICC is used as an indicator for consistency in observation, its calculated degree of agreement does not clarify whether a feature is detectable or undetectable. It only expressed the degree in consistency between examiners with respect to the detection of healing features.

We are aware of the limited sample size with regard to the number of amputations. Amputations are not frequently found in archaeological populations, and if so, not in varying phases of healing. Nevertheless, since fractures and amputations generally share a similar healing process (Barber, 1930), it is believed that the combined assembly allows for meaningful interpretation. For those features in which logical inferences resulted in a limited number of examined specimens (e.g. complete closure of the medullary cavity in amputations), ICC must be interpreted with caution, due to sample homogeneity.

Overall, our results suggest that not every healing feature as described by Barber (1929, 1930, 1934); Maat (2008) and Maat and Huls (2010) was consistently detected. Nevertheless, the results from this study suggest that if only those with substantial interobserver agreement are used, a fair estimation of the minimal and maximal time lapsed after an injury can still be made. This is a substantial improvement in diagnosis if compared to conclusions such as ‘healing’ or ‘healed’. For those who do not routinely ‘date’ fractures or amputations, a table was made showing only the consistently detected features together with its modality (Table 5). Furthermore, some reference figures are given (Figures 2-7).

image

Figure 2. The fracture margin of a third metacarpal (N38.2) showing Howship's lacunae as numerous bite-like indentations (arrows). They are caused by resorption of bone tissue by osteoclasts. Undecalcified and unstained section. Bright light. Bar indicates scale. This figure is available in colour online at wileyonlinelibrary.com/journal/oa.

Download figure to PowerPoint

image

Figure 3. Micrograph taken near the fracture site of a fractured radius (N38.3). In the centre, a ‘cutting cone’ running/drilling from left to right is indicated (asterisk). The cutting cone is characterized by Howship's lacunae. The closing cone is situated to the left, outside the field of photography. Undecalcified, unstained section. Bright light. Bar indicates scale. This figure is available in colour online at wileyonlinelibrary.com/journal/oa.

Download figure to PowerPoint

image

Figure 4. Micrographs of so-called 'separable periosteal callus' in a fractured third metacarpal (N.38.2). Undecalcified section, stained with haematoxilin. Bar indicates scale. Figure 4a) Section viewed with bright light. The seperable periosteal callus is only connected to the periosteal aspect of the cortex by small pillars of bone tissue. The sloping aspect of the callus illustrates the build-up of callus towards the fracture site. Figure 4b) Same section as shown in Figure 4a, now viewed with polarized light. Extensive taphonomic alteration of the periosteal callus hampers the visibility of the microarchitecture of the callus. The callus in this healing phase constitutes mainly of primary woven bone, in contrast to the secondary lamellar bone of the underlying cortex. This figure is available in colour online at wileyonlinelibrary.com/journal/oa.

Download figure to PowerPoint

image

Figure 5. Micrographs showing the remodeling of callus in a fractured radius (N38.3). Undecalcified, unstained section. Bar indicates scale. Figure 5a) Section viewed with bright light. Overall, the callus has a disorganized appearance and consists mainly of primary woven bone. The lesion site is indicated by an arrow. Figure 5b) Same section as shown in Figure 5a, now viewed with polarized light. The use of polarized light enhances the random orientation of the primary woven bone fibers. Also, the discontinuity of the cortex at the lesion site (arrow) is enhanced by polarized light Figure 5c) Detail of the callus as shown in 5a and 5b. Viewed with bright light. A higher magnification shows one of the Haversian canals (asterisk), indicating that advanced remodeling of the callus has started. Figure 5d) Same section as shown in Figure 5c, viewed with polarized light. The detectability of Haversian systems, organized around a Haversian canal (asterisk), becomes indisputed by the use of polarized light. This figure is available in colour online at wileyonlinelibrary.com/journal/oa.

Download figure to PowerPoint

image

Figure 6. Radiographs of remodeled calluses. Top: The X-ray of a fractured radius (N.38.3) shows both an endosteally and a periosteally situated callus. The periosteal callus has a smoothened outline. Continuity of the cortex is restored, although its bone density at the lesion site is still decreased. Below: This X-ray of a fractured ulna (N74.4) shows a smoothened periosteally situated callus. The continuity of the cortex is almost totally restored and is slightly less radiodense at the fracture site. The internal callus is almost indistinguishable from the surrounding cancellous bone of the marrow cavity, indicating advanced remodeling.

Download figure to PowerPoint

image

Figure 7. Radiograph of an amputated tibia (N34.3). When compared ti its contralateral control, The amputated stump shows diminished density of the cancellous bone and a thinned cortex. This indicates osteoporosis due to disuse. The medullary cavity at the amputation end is almost closed by new bone formation, indicating advanced capping. The mottled appearance of the tibial stump is partly due to dirt.

Download figure to PowerPoint

We recommend using plain radiographic and histological analysis. Histology yields the best results if both unstained and haematoxylin stained sections are used, in combination with polarized light. Moreover, as usual, the value of negative observations is limited, in contrast to positive observations. Time intervals should be adjusted to specific conditions (e.g. shortened in case of children (Maat, 2008; Maat and Huls, 2010). It goes without saying that our approach stays open for improvement and reliability will increase with experience.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theory and practice of dating traumatic lesions
  5. Materials and methods
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgements
  10. Conflict of Interest
  11. References

By using complementary radiological and (un)stained histological investigation methods, a differentiation can be made between features that indicate various stages of the healing process after trauma as may be observed in human dry bone material. The consistency in the detection of healing features indicates that the bone healing process can be used to estimate the post-traumatic time interval of fractures and amputations.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theory and practice of dating traumatic lesions
  5. Materials and methods
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgements
  10. Conflict of Interest
  11. References

We are indebted to the Leiden University Fund (LUF)/Van Trigt and the South Africa Netherlands research Programme for Alternatives in Development (SANPAD) for partial funding. We also thank the Department of Radiology of the LUMC and Ron Wolterbeek of the Department of Medical Statistics of the LUMC. We are greatly indebted to Dr David Morris of the McGregor Museum in Kimberley, South Africa, for allowing us to use the Gladstone material.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theory and practice of dating traumatic lesions
  5. Materials and methods
  6. Results
  7. Discussion
  8. Conclusion
  9. Acknowledgements
  10. Conflict of Interest
  11. References
  • Alvrus A. 1999. Fracture patterns among the Nubians of Semna South, Sudanese Nubia. International Journal of Osteoarchaeology 9(6): 417429.
  • Anderson T. 2002. Healed trauma in an early Bronze Age human skeleton from Buckinghamshire, England. International Journal of Osteoarchaeology 12(3): 220. DOI: 10.1002/oa.599
  • Aufderheide AC, Rodríguez-Martin C. 1998. The Cambridge Encyclopedia of Human Palaeopathology. Cambridge University Press: Cambridge.
  • Barber CG. 1929. Immediate and eventual features of healing in amputated bones. Annals of Surgery 90(6): 985992.
  • Barber CG. 1930. The detailed changes characteristic of healing bone in amputation stumps. Journal of Bone and Joint Surgery 12: 353359.
  • Barber CG. 1934. Ultimate anatomical modifications in amputation stumps. Journal of Bone and Joint Surgery 16(2): 394400.
  • Berk RA. 1979. Generalizability of behavioral observations: A clarification of interobserver agreement and interobserver reliability. American Journal of Mental Retardation 83(5):460472.
  • Berryman HE, Haun SJ. 1996. Applying forensic techniques to interpret cranial fracture patterns in an archaeological specimen. International Journal of Osteoarchaeology 6: 29.
  • Brickley M. 2006. Rib fractures in the archaeological record: a useful source of sociocultural information? International Journal of Osteoarchaeology 16(1): 6175. DOI: 10.1002/oa.809
  • Cattaneo C, Andreola S, Marinelli E, Poppa P, Porta D, Grandi M. 2010. The detection of microscopic markers of hemorrhaging and wound age on dry bone: a pilot study. The American Journal of Forensic Medicine and Pathology 31(1): 22.
  • De Boer HH, Aarents MJ, Maat GJR. 2011. Manual for the preparation and staining of embedded natural dry bone tissue sections for microscopy. International Journal of Osteoarchaeology (in press) DOI: 10.1002/0a.1242
  • De Boer HH, Aarents MJ, Maat GJR. 2012. Staining ground sections of natural dry bone tissue for microscopy. International Journal of Osteoarchaeology 22: 379386. DOI: 10.1002/0a.1242
  • Djuriç MP. 2006. Fractures in late medieval skeletal populations from Serbia. American Journal of Physical Anthropology 130(2): 167178 DOI: 10.1002/ajpa.20270
  • Frost HM. 1989a. The biology of fracture healing: an overview for clinicians. Part I. Clinical Orthopedics and Related Research 248: 283293.
  • Frost HM. 1989b. The biology of fracture healing: an overview for clinicians. Part II. Clinical Orthopedics and Related Research 248: 294309
  • Galloway A. 1999. The Biomechanics of Fracture Production. Broken Bones: Anthropological analysis of blunt force trauma, Galloway A (ed.). Charles C. Thomas: Springfield; 3562.
  • Gordon CG, Buikstra JE. 1981. Soil pH, bone preservation, and sampling bias at mortuary sites. American Antiquity 46: 566571.
  • Grauer AL, Roberts CA. 1996. Paleoepidemiology, healing, and possible treatment of trauma in the medieval cemetery population of St. Helen-on-the-Walls, York, England. American Journal of Physical Anthropology 100(4): 531544.
  • Holt BM, Fornaciari G, Formicola V. 2002. Bone remodelling following a lower leg fracture in the 11, 000-year-old hunter-gatherer from Vado all' Arancio (Italy). International Journal of Osteoarchaeology 12(6): 402. DOI: 10.1002/oa.639
  • Judd MA. 2002. Ancient Injury Recidivism: An example from the Kerma Period of Ancient Nubia. International Journal of Osteoarchaeology 12: 89106.
  • Judd MA, Roberts CA. 1999. Fracture trauma in a medieval British farming village. American Journal of Physical Anthropology 109(2): 229243.
  • Krogman WM, Iscan MY. 1986. The human skeleton in forensic medicine. Charles C. Thomas Publisher: Springfield, USA.
  • Landis JR, Koch GG. 1977. The measurement of observer agreement for categorical data. Biometrics 33(1): 159174.
  • Lovell NC. 1997. Trauma analysis in paleopathology. Yearbook of Physical Anthropology 104: 139170.
  • Lovell NC. 2008. Analysis and Interpretation of Skeletal Trauma. Biological Anthropology of the Human Skeleton, second edn, Katzenberg MA, Saunders SR (eds). John Wiley & Sons, Inc.: Hoboken; 341386.
  • Maat GJR. 1981. Human remains at the Dutch whaling stations at Spitsbergen, a physical anthropological report., 153–201. Groningen: University of Groningen.
  • Maat GJR. 2008. Case study 5.3: Dating of fractures in human dry bone tissue - the Berisha case. Skeletal trauma, Kimmerle EH, Baraybar JP (eds). Taylor & Francis Group: Boca Raton, FL; 245254.
  • Maat GJR, Huls N. 2010. Histological fracture dating of fresh and dried bone. Forensic Aspects of Pediatric Fractures, Bilo RAC, Robben SGF, Van Rijn R (eds). Springer-Verlag: Berlin - Heidelberg; 194201.
  • Martin DL, Harrod RP, Fields M. 2010. Beaten Down and Worked to the Bone: Bioarchaeological investigations of women and violence in the ancient Southwest. Landscapes of Violence 1(1): 19. Article 3.
  • Mays SA. 1996. Healed limb amputations in human osteoarchaeology and their causes: a case study from Ipswich, UK. International Journal of Osteoarchaeology 6(1): 101113.
  • Van der Merwe AE, Steyn M, l'Abbé EN. 2009. Trauma and amputations in 19th century miners from Kimberley, South Africa. International Journal of Osteoarcheology. (in press) DOI: 10.1002/oa1035
  • Mitchell PD. 2006. Trauma in the Crusader period city of Caesarea: a major port in the medieval eastern Mediterranean. International Journal of Osteoarchaeology 16(6): 493505. DOI: 10.1002/oa.853
  • Nakai M, Koji I, Sinsuke H. 2001. Healed bone fractures in a Jomon skeletal population from the Yoshigo shell mound, Aichi Prefecture, Japan. International Journal of Osteoarchaeology 9(2): 7782.
  • Oehmichen M. 2004. Vitality and time course of wounds. Forensic Science International 144(2–3): 221231. DOI: 10.1016/j.forsciint.2004.04.057
  • Ortner DJ. 2003. Identification of pathological conditions in human skeletal remains. Academic Press: San Diego.
  • Quatrehomme G, Iscan MY. 1997. Postmortem skeletal lesions. Forensic Science International 89(3): 155165.
  • Redfern R. 2009. A regional examination of surgery and fracture treatment in Iron Age and Roman Britain. International Journal of Osteoarchaeology 20(4): 443471. DOI: 10.1002/oa.1067
  • Rodríguez-Martin C. 2006. Identification and differential diagnosis of traumatic lesions of the skeleton. Forensic Anthropology and Medicine: Complementary Sciences From Recovery to Cause of Death, Schmitt A, Cunha E, Pinheiro J (eds). Humana Press: Totowa, NJ; 197221.
  • Sauer NJ. 1998. The Timing of Injuries and Manner of Death: Distinguishing among antemortem, perimortem, and postmortem trauma. Forensic Osteology: Advances in the identification of human remains, 2nd edn. Reichs KJ (ed). Charles C. Thomas: Springfield; 321332.
  • Todd TW, Barber CG. 1934. The extend of skeletal change after amputation. Journal of Bone and Joint Surgery 16(1): 5364.
  • Todd TW, Iler DH. 1927. The phenomena of early stages in bone repair. Annals of Surgery 86(5): 715736.
  • Tosounidis T, Kontakis G, Nikolaou V, Papathanassopoulos A, Giannioudis PV. 2009. Fracture healing and bone repair: an update. Trauma 11(3): 145156.
  • Ubelaker DH, Adams BJ. 1995. Differentiation of Perimortem and Postmortem Trauma using Taphonomic Indicators. Journal of Forensic Sciences 40(3): 509512.
  • Vigorita VJ. 2009. Orthopaedic pathology. Lippincott, Williams and Wilkins: Philadelphia, New York, Baltimore.
  • Wheatley BP. 2008. Perimortem or Postmortem Bone Fractures? An Experimental Study of Fracture Patterns in Deer Femora*. Journal of Forensic Sciences 53(1): 69. DOI: 10.1111/j.1556–4029.2008.00593.x
  • Wieberg DA, Wescott DJ. 2008. Estimating the timing of long bone fractures: correlation between the postmortem interval, bone moisture content, and blunt force trauma fracture characteristics. Journal of Forensic Sciences 53: 10281034.