Relative blood loss in forensic medicine—do we need a change in doctrine?

Blood loss is a frequent finding in routine forensic pathology case work. It may constitute a genuine cause of death (in 5 to 10% of forensic autopsies, according to [1]) or a contributing factor tovadjust another primary cause of death. Forensic pathology doctrine, as referenced in different textbooks, relates lethality and risk of death to the relative blood loss (blood lost out of total blood volume). A relative blood loss of more than 33% \(\left (\frac {1}{3}\right )\) has been described as life-threatening [2, 3], whereas 50% was seen as ‘usually lethal’. Other authors have described relative blood loss percentages for acute blood loss. Here, a loss of 40% and ‘more than 66%’\(\left (\frac {2}{3}\right )\) have been categorized as ‘lethal’ [4, 5].

In addition to absolute and relative loss, internal blood loss with collection of blood in cavities and defined spaces may cause compressive mechanical effects, such as in the skull, the pericardial sac or the thoracic cavity [6].

A basic understanding of blood loss assessment in emergency medicine and pre-clinical trauma scenarios is needed in any assessment of forensic cases.

In primary clinical care, the Advanced Trauma Life Support (ATLS®) classification of hypovolemic shock is used. Patients suffering from blood loss are classified depending on heart rate, systolic pressure, pulse pressure and other parameters. Relative blood loss is part of the classification system, where 7% of body weight is assumed for TBV [7]. The ATLSⓇ-classification has been criticized as a mere teaching tool rather than a practical guideline, as the postulated classes are rarely found in real-life scenarios [8].

The estimation of blood loss is part of routine autopsy in which free blood quantities are collected from the body cavities and measured. A variety of factors negatively influence the assessment of relative blood loss during autopsy (see Fig. 1). In many forensic cases, blood loss is both internal and external, with external blood loss rarely being precisely quantifiable. Internal blood loss is difficult to assess in cases of diffuse tissue haemorrhage and dilution with for example edema or intestinal contents. Also, the extent of blood loss may be limited by coexisting internal diseases and/or coexisting trauma leading to ‘premature’ decompensation as well as mechanical aspects of the bleeding itself such as in pericardial tamponade, intracranial haemorrhage and structural damage to the heart. Cardiopulmonary resuscitation (CPR) may prolong the extravasation of blood after cessation of circulation. Also, injuries as complications of resuscitation attempts, sometimes dramatic, have been reported frequently (for example [9, 10]). Classical autopsy findings of blood loss, besides a secured source of bleeding and possible pooling of blood, include sparse lividity, organ pallor, subendocardial haemorrhage, wrinkling of the spleen capsule and ‘shock kidneys’. They are however not present in every case.

TBV is often generalized as 5 to 6 l for most individuals. In addition, there are a variety of rules of thumb and dedicated formulas for TBV estimation. More recent body-mass-index(BMI)-adapted formulas have, to our knowledge, not been used in legal medicine. Rules of thumb for the estimation of TBV, as published in forensic textbooks and scientific articles, are simple correlations of TBV with body weight, either as ‘volume per body weight’ (‘reference book method’, ‘index based’), a percentage of body weight or a baseline volume combined with a percentage of body weight (see Table 1). For these rules of thumb, we could not find scientific evidence, and more variations may exist.

Formulas listed in Table 2 use non-linear BMI-/constitution-dependent scaling. Non-linear TBV dependency on weight and constitution has been acknowledged in clinical medicine for some time [11‐13]. The developed algorithms have been used clinically, especially in obesity surgery, but have not yet been used in the context of forensic pathology. Overweight persons constitute a significant proportion of the general population as well as forensic autopsy cases. For example, Sheikhazadi et al. [14] reported a mean BMI of 25.12 for autopsies in Tehran (n= 1222).

Remarkably, in allometry, blood volume in interspecific comparison is treated as constant and no power-laws apply: TBV = 70 ml/kg body weight [15] or TBV ≈ 6–7% of body volume (for all mammals except aquatic ones) [16].

Autopsy reports from the forensic medicine departments of the University of Saarland Medical School in Homburg/Saar and the Goethe University Frankfurt Medical School in Frankfurt/Main were retrospectively screened for ‘bleeding to death’, exsanguination, haemorrhagic shock and hypovolemic shock as cause of death. In some cases, classification as primarily due to exsanguination was retrospectively rejected. During restrospective analysis of autopsy reports, the cohort had to be restricted to cases of sudden, quantifiable and exclusively internal haemorrhage due to difficulties described in ‘Autopsy findings’. The inclusion criteria were defined as sudden, quantifiable (minimum 600 ml) blood loss into the thoracic cavities and/or the abdominal cavity. Following this procedure, 80 cases remained. For these, amount, source and location of internal blood loss were noted as well as sex, age, height, weight, case history including resuscitation efforts, and cause of death. In total, eight formulas were applied to each case: In addition, the autopsy reports were screened for the findings listed below. The gradual transition of findings could not be systematically reflected in this process, due to the differences in description by various examiners in the retrospective setting: In addition to descriptive statistics, confirmative analysis was performed using pearson’s correlation coefficient r with p value to analyze the degree of association between two or more variables as well as Levene’s test for comparing CPR-cases with non-CPR-cases (Python library Pandas [26]). Visualization was performed using the Python libraries Seaborn and Matplotlib [27].

Nomographs for the BMI-dependent algorithms by Lemmens et al. (see Fig. 2) and nadler et al. (see Fig. 3) were designed using the PyNomo [28] package for Python. The approach by friesen produces results close to Lemmens et al. and was therefore omitted.

The absolute blood loss found at autopsy ranged from 660 ml (chosen cut-off value 600 ml) up to 3800 ml (2 outliers, see ‘Discussion’).

There was no statistically significant correlation between either age and total blood loss, weight and total blood loss or BMI and total blood loss. In particular, older age was not correlated with less blood loss, no higher susceptibility to blood loss could be found. Heavier persons did not lose higher amounts of blood. BMI-adapted formulas produced more moderate TBV estimations for overweight and obese individuals, apart from that none of the algorithms and formulas was particularly outstanding in terms of indication of a life-threatening condition. The amount of blood loss as well as relative blood loss (based on TBV estimation using various formulas) was seemingly randomly distributed across cases of all ages and body constitutions.

Relative blood loss of 30% (ATLSⓇ Class III, complicated haemorrhage) was not reached between 37% (Baseline2) and 72% of cases (Baseline1, maximum TBV) with average 53% of cases for all 8 algorithms.

Relative blood loss of 33% (one-third being described as ‘life-threatening’ in forensic literature) was not reached between 49% (Baseline2) and 79% of cases (Baseline 1, maximum TBV) with average 61% of cases for all 8 algorithms. A maximum of 20% of cases lost > 50% of their TBV when using the Baseline2 formula (see Fig. 4).

For 8 algorithms, 9 TBV estimations were performed, as Baseline1 formula differs for minimum and maximum estimations. The lowest relative blood loss was calculated as only 8.4% (P8), the highest as 81.0% (Baseline2). The mean relative blood loss ranged from 23.6% (Baseline1, maximum) to 35.9% (Baseline 2). The overall mean relative blood loss for all cases and algorithms was 30.7% (see Fig. 5).

Overall, the most frequent finding was organ pallor (80.7%), followed by shock kidneys (62.8%), subendocardial haemorrhage (58.9%), wrinkling of the spleen capsule (57.6%) and reduced lividity (51.2%).

However, only 4 cases (5.1%) showed a combination of all five findings, followed by two findings in 35.9%, three findings in 21.8%, four findings in 17.9% and one finding in 15.4 %.

For 3 cases (3.8%) none of the typical autopsy findings was described, despite individual absolute blood losses of 2100 ml (2 cases) and 2500 ml. All 3 cases involved rupture of the aorta due to internal causes.

Due to the inclusion criteria, the sources for bleeding cannot be generalized. In our limited cohort we found:

Trauma originated mainly from either deceleration injury in motor vehicle accidents or crushing of some sort. Due to the inclusion criteria of quantifiable/quantified blood loss volume, retroperitoneal haemorrhage following ruptured aneurysm of the abdominal aorta was underrepresented, since retroperitoneal blood accumulation was often not quantified absolutely but rather described or measured two dimensionally only.

For 66 out of 78 cases, it could be established whether CPR had been performed (n= 40) or not (n= 26). While cases with CPR had a higher mean absolute blood loss than cases without CPR (1.797 ml versus 1540 ml), this difference was not significant (levene’s test p= 0.22). In one case, all internal blood loss detected at autopsy could be directly attributed to chest compression during CPR (see below).

In one additional case, all blood loss present at autopsy could be attributed to CPR alone: a 73-year-old female patient (165 cm, 75 kg) suffered complications in the form of intraabdominal haemorrhage during gall bladder surgery. The bleeding was stopped, blood products were administered and the operation was concluded. After extubation and relocation to the recovery room however, the patient suffered cardiac arrest and died despite immediate CPR. Autopsy revealed multiple rib fractures due to CPR with 2 fragments protruding into the thoracic cavity. Both the left cardiac atrium and the right ventricle were ruptured as well as the pericardial sac with a total of 1500 ml of blood in the left thoracic cavity.