Despite the advancement of scientific methodological rigor and the growing emphasis on evidence-based practice in forensic medicine, the evidentiary value of the diatom test for the PM diagnosis of drowning remains difficult to assess. An increasing number of studies have recently sought to refine analysis methods [20,21,22,23,24,25,26,27] and to address quantitative and qualitative aspects [20, 28,29,30,31], as well as the issue of false-positive diatom findings in non-drowned controls [27, 32]. While some reviews have summarized the existing literature on the diatom test [5, 7, 18, 33,34,35], no systematic review on the topic exists.

In contrast to typical literature reviews that address a general topic or problem of interest, systematic reviews deal with clear research questions that can be answered by transparent methods enabling informed decision-making based on the best available evidence [36,37,38]. We therefore conducted a systematic review with the aim to examine “how the presence of PM diatoms may be interpreted in medico-legal investigations of fatal drowning”.

Results from the 17 selected studies (from the initial 372 records identified), demonstrate that diatom concentrations are typically greater in the lungs and internal organs of drowning victims than in non-drowned control groups, that are either found on land or immersed PM. Such results, when considered in isolation, suggest that the diatom test could serve as effective supportive evidence for diagnosing drowning. However, considerable inconsistencies exist in the concentrations of diatoms detected in the lungs and internal organs of definite drowning victims (Table 5). The diagnostic value of diatoms in lung tissue alone is also controversial, as data from control groups in the selected studies suggests that during PM immersion, diatoms can indeed penetrate the airways passively.

Variations in study design, sample population, laboratory procedures and methodologies for analyzing diatom samples hinder the comparisons of results among the selected studies; they also hamper establishment of any cut-off concentration value to enable a drowning diagnosis based on diatoms in lung and other organ tissues. Reported cut-off concentration values for diagnosing drowning in the literature differ, with some studies proposing thresholds of ≥ 20 diatoms / 10 g for lung tissue and 5 diatoms /10 g for other internal-organ tissues [20, 27, 39,40,41,42,43]. Another proposal is ≥ 20 diatoms /100 g for lung tissue and 13 diatoms for liver tissue [44]. However, control cases have also been described as overlapping these cut-off values, exhibiting between 5 and 25 diatoms /100 g in lung tissue and up to 10 diatoms in closed organs [45]. Another proposal is that a threshold of 10 diatoms / 10 g of lung tissue is sufficient to indicate drowning when using the MD-VF-Auto SEM method [27]. Problematically, the overshadowing restriction that limits identifying an accurate cut-off value is the misleading assumption that, during the drowning process, each drowning victim inhales the same volume of liquid.

Ultimately, the same factors that hamper the establishment of cut-off concentrations of diatoms in drowning, also stem from issues relating to false-negative and -positive cases. These issues include lack of diatoms in drowning cases and discovery of diatoms in PM tissues of corpses with a COD other than drowning. These are two of the main criticisms levelled against the diatom test. Although the selected studies do report false-negative and positive results, they collectively fall short in thoroughly assessing such results’ impact on the evidentiary value of the test.

A common interpretation following diatom testing is that the absence of, or low concentrations of diatoms identified in PM tissue samples may be attributed to similarly low concentrations of diatoms in the drowning media or even to their absence [32, 46]. However, this interpretation overlooks a number of alternative factors that may also result in false-negative results. First, only a small volume of liquid is necessary to penetrate the airways and cause death by drowning [47], and cardiorespiratory arrest may occur during a very early stage of the drowning process. Secondly, inadequate tissue sampling during autopsy, including insufficient volume and number of samples, can result in low diatom concentrations [28]. A similar outcome will occur from the (partial) loss of diatoms during laboratory procedures such as tissue digestion, centrifugation, and transfer of pellets to microscope slides. Whatever the mechanism, a low concentration of diatoms in PM tissue will reduce the taxonomic spectrum available for comparison with the diatom species in the putative drowning medium.

In regard to false-positive results, a range of AM and PM contamination may account for detection of diatoms in non-drowned cases (and for an overestimation of diatoms in actual drowning cases). Detection of diatoms either in lung or in closed-organ tissues of submerged corpses that did not die from drowning may be interpreted as PM contamination [20, 23], thus limiting the diatom test’s reliability for a diagnosis of drowning [28]. Direct exposure to diatom-rich water (bi-cellular cultures of Thalassiosira baltica and Thalassiosira levanderi) via tracheostomy of the lungs of non-drowned cadavers with advanced putrefaction changes revealed, however, no diatoms during autopsy in closed organs 3–4 days after infusion [30]. Those authors demonstrated under controlled conditions, simulating extreme PM submersion, that the risk of false-positive results in closed organs due to PM diatom penetration in corpses is less significant than previously assumed [30]. On the other hand, passive PM penetration of liquid into the lungs should always be considered when interpreting diatom findings in the tissue. This is further supported by the fact that all selected studies indicate that all PM-immersed groups had diatoms in their lungs, albeit at lower concentrations than that identified in drowned victims. Interestingly, in a single study exposing whole cat lungs to diatom-enriched media in vitro, without any infusion, no diatom penetration of occurred [48].

PM contamination can also occur during autopsy, via the transfer of diatoms from the surface of the body to internal organs, between internal organs and during tissue sampling and laboratory procedures [31, 32, 41]. A prerequisite for diatom analysis is therefore the use of sterile or single-use instruments [28, 49]. As is frequent glove-changes and the use of bi- or tri-distilled H2O during centrifugation cycles and testing for other potential contamination sources, such as reagents [27].

Detection of diatoms unrelated to drowning can also result from AM diatom penetration. This can occur when victims are recovered from aquatic environments they were previously exposed to [50]. For instance, swimmers or divers can inhale diatom-rich liquid during earlier activities unrelated to the fatal drowning episode; or those living near the shore may inhale diatom species that are aerophilic [14, 27, 51]. Consumption of seafood such as shellfish, containing a high quantity of diatoms [52] may be an additional source of AM diatom penetration. This is, however, controversial, as little evidence demonstrates that diatom frustules can enter the systemic circulation via the human gastrointestinal tract, although simulated gastric lavage in anesthetized experimental animals has recently suggested otherwise [53].

To address these gaps, studies must employ stricter, and standardized protocols encompassing tissue sampling, laboratory procedures, diatom counting, and taxonomic analysis. A similar call for standardization was recently made by Ludes et al. [35], and our findings further support this view, as the selected studies only partially adhered to standardized methods. Future studies should aim at optimizing the quantitative and qualitative assessment of diatoms to establish reference intervals for the lung and internal organ tissues applicable to drowning and non-drowning cases. This will also reduce the ambiguity surrounding false-positive and false-negative results.

Although lung tissue was analyzed in all selected studies, differences emerged concerning the type and number of closed organs sampled. Surprisingly, only four studies examined bone marrow. The volume of samples varied between studies examining the same tissue, with lower volumes of lung tissue typically sampled than the volumes of closed organs [23, 24, 26, 31, 32, 41]. It could be argued that when establishing reference intervals for drowning, smaller samples may yield fewer diatoms. On the other hand, when sampling tissue such as the lungs, high concentrations of diatoms and other potential contaminants may interfere with diatom counting. This may also hinder taxonomic analysis making dilution of the sediment left after tissue digestion and centrifugation necessary. Furthermore, the sampling site in each organ should also be considered, due to the possible unequal distribution of diatoms. Importantly, having more than a single sample of adequate volume for each organ allows for the repetition of the diatom analysis to verify results. Standardizing sample collection is crucial for all diatom testing to ensure comparable quantitative results in both routine casework and control cases.

As to laboratory procedures, digestion of tissue samples, by means of chemical or enzymatic methods may lead to, at least partial loss of diatoms. Most of our selected studies (65%) employed chemical digestion (acids) due to its effectiveness in removing organic material and yielding diatoms with clean surfaces. However, the acid also causes a selective destruction of diatoms, especially of those with a weaker siliceous frustule [54, 55]. This can, in turn, result in underestimation of diatom concentration and a reduction in their taxonomic spectrum. The wide variability of chemical digestion protocols adopted in the selected studies reduces reliable comparisons of quantitative and qualitative results. In an effort to compare digestion methods, Fucci et al. [22] showed that the digestion protocol outlined in the European Standard for the routine sampling and preparation of benthic diatoms from rivers and lakes (EN13946), which utilizes 40% H2O2 and 1 M HCl, yielded higher diatom concentrations in both lung and closed-organ tissues, when compared to 37% HCl alone.

Compared to chemical methods, enzymatic digestion methods [20, 21, 23, 56] offer the advantage of being environmentally safer, albeit at a higher cost and with weaker tissue-digestion effectiveness [55]. This may in principle, limit the volume of tissues to be studied and may result in suboptimal cleaning of diatom surfaces. Poor cleaning can, in turn, pose challenges to taxonomic diatom identification. Although few studies (18%) included in this review have considered enzymatic methods, comparisons with acid digestion demonstrated similar results in lung, liver, and kidney tissues [20].

Analysis of freshwater diatoms (specifically within the Cyclotella and Cybella genera) spiked into 2 g of kidney, liver, and bone marrow tissue of rabbits suggests that proteinase K has a greater extracting effectiveness than either HNO3 + H2O2 or Soluene-350 digestion [55]. For seawater diatoms (within the Navicula genera), proteinase K again showed in the same comparative study greater reclaiming ratios [55]. Kjeldahl flasks, especially when re-used or damaged, is also a potential contributor to false-negative and -positive results [31]. The siliceous coating of the diatom frustule can adhere to the flask’s inner glass surface during the heating procedure, or alternatively, diatom fragments can be trapped in small defects of the flasks and subsequently be released when the flask is re-used [20, 30, 31].

The development of the Microwave Digestion - Vacuum Filtration - Automated Scanning Electron Microscopy (MD-VF-Auto SEM) method addresses the partial loss of diatoms during acid digestion and centrifugation [25, 27]. Moreover, by also using an automated SEM system to capture images, the method enables a more detailed and comprehensive quantitative and qualitative analysis of diatoms compared to analysis by traditional light microscopy [26, 57]. However, this approach is currently only employed at a limited number of institutions, and appears to yield considerably higher diatom quantities in the lungs of drowning victims when compared to either the acid or the enzymatic digestion protocols [24, 26, 27]. Zhao et al. [26], using the MD-VF-Auto SEM method, normalized diatom concentrations by calculating the lung-to-water-sample diatom concentration-rate ratio (L/D). Further studies are necessary to assess the practical utility of this ratio in distinguishing drowning from merely PM immersion and distinguishing COD other than drowning.

Once tissue digestion is performed, either by chemical or enzymatic methods, the digestion product must be decanted and transferred into dedicated tubes for centrifugation and washing cycles. Such cycles too, should be standardized since a selective loss of diatoms can in principle, occur during these steps [25, 49]. As highlighted in our results, none of the selected studies had included any positive control during tissue digestion and centrifugation (Table 3). To better understand the potential loss of diatoms during the tissue digestion and centrifugation steps, spiking samples during routine casework as a positive control, with a known monocellulate diatom species and concentration may help determine digestion and centrifugation effectiveness. Such a spiking-and-recovery test was conducted by Seo et al. [58] to evaluate the effectiveness of a DNA co-precipitation method for the detection of diatoms in heart blood. The spike, which was introduced prior to digestion using proteinase K in non-drowned control subjects, showed recovery rates ranging from 88.4 to 100%. By introducing a known concentration of a distinct diatom species to a tissue sample before laboratory processing, a recovery analysis can provide valuable insights into the robustness and effectiveness of laboratory procedures and eventually offer insights for better interpretation of casework with low diatom concentrations or false-negative cases.

As mentioned, quantitative studies that aim to establish cut-off values and address the issue of false-positive and -negative issues are further hampered by lack of standardization for the transfer of digested tissue aliquots onto microscope slides. The 17 selected studies only rarely mention whether the entire pellet is transferred after centrifugation and washing cycles into one or more microscope slides, nor do they mention whether the centrifugation tubes are washed to collect any residual diatoms that may remain on the bottom and on internal surfaces before being transferred.

Equally crucial are the criteria used for diatom counting, as the studies consider interchangeable terms such as whole diatoms, valves, or fragments of diatoms. Although the inclusion or exclusion of fragments are at times detailed in the studies, counting is vulnerable to subjectivity. Moreover, because only separate valves are detected under a microscope following tissue digestion, reporting findings as diatom frustules, is misleading.

Lastly, although taxonomic analysis of diatoms is frequently performed by experienced in-house diatom experts, this task may be inherently subjective, associated with an increased risk of bias, and a lack of consistency and transparency. To mitigate this, diatom analysis focusing on control series in non-drowned bodies and analysis addressing laboratory procedures, should be conducted by an independent taxonomist without ties to routine PM diatom analyses in drowning cases.

Automated systems using artificial intelligence (AI) are a promising tool to standardize the identification, count, taxonomic classification, and comparison of diatoms, which potentially could mitigate subjectivity and human error. Several studies hav explored the potential of machine learning to train automated diatom detection and have obtained promising results [43, 59,60,61]. Nevertheless, difficulties remain in training the AI to differentiate between