Establishing an accurate PMI estimation has important legal implications, as well as identifying the forensic relevance of a case. However, dating time of death of human skeletal remains constitutes a major challenge in forensic sciences, even when the aim is merely to differentiate those with forensic interest from archaeological remains, due to the countless number of factors that can affect the decomposition process. The developed methods do not offer precise estimations or require expensive, complex, and time-consuming applications. Therefore, numerous techniques have been recently studied in order to determine an easier and more suitable method for completing this task. This research aims to thoroughly explore the potential of the luminol chemiluminescence technique for estimating time since death in a robust sample with a large range of PMI intervals, using different forensically relevant cutoff values.

The analysis of the luminol chemiluminescence results across different postmortem intervals reveals this method as a suitable tool, as a presumptive test, for time since death estimation. Although some unexpected results exposed positive reactions in samples with dates of death older than 100 years, suggesting that the luminol technique can detect hemoglobin in historic skeletal human remains, significant differences were found in the chemiluminescence intensity through the different time intervals, pointing out the decreasing intensity of the chemiluminescence reactions with time. According to the 5-level scale analysis, intense chemiluminescence reactions, described as “strong positive” results, were solely observed in the first three intervals (5 to 20 years PMI). It is also worth noticing that, even in these lower PMI intervals, the most common reaction is the “positive” reaction [69.23% (5–10 years); 49.95% (10–15 years); and 51.16% (15–20 years) versus 30.77%; 19.82%, and 6.98% for the “strong positive”, respectively]. The decrease in “positive” reactions as the PMI intervals increase is clearly noticeable, being replaced by a majority of “weak positive” reactions, achieving 100% of the results in the 25–30 years interval. From the 30–40 years interval to more than 100 years PMI, a very faint light-reaction (“barely positive”) appeared in 13 of 21 clavicles, whereas in 3 (30–50 year PMI) a “weak positive” was observed, and in the remaining 5 no reaction was detected.

These results are mainly consistent with the existing literature. Although the great number of false positives in PMI over 100 years was not expected (> 50% of false positives), it was previously described in other studies. Ramsthaler et al. (2009) described an occurrence of false positive results in 7.5% of the cases. In previous research (2011), Ramsthaler and team reported a percentage of false positive results of 30%. Later, Capella et al. (2018) recorded an even higher number of ancient bones testing positive for luminol (41.1%), closer to our obtained results. Still, in most of these studies, including the present one, the small sample size for PMIs over 100 years should be considered cautiously. The primary issue associated with a high percentage of false positives, usually related to low forensic relevance cutoff values, is the pursuit of cases either irrelevant or with no forensic interest, wasting valuable time in examinations and further techniques. On the contrary, Introna and colleagues’ (1999) experiment disclosed an almost complete absence of chemiluminescence in samples older than 50 years (1 out of 10 samples showed a slightly faint reaction), concluding that no chemiluminescence reaction occurs on bone samples with PMI of 80 or more years. The observed differences in the results within studies may be explained by numerous reasons, such as the burial context storage conditions of the samples, or even the variations in the experimental design.

The proportion of false negative results constitutes a variance between our study and the previous investigations mentioned above. Out of the 230 samples, 2 were false negatives, considering the cutoff < 100. This value drops to 1 false negative result for the cutoff < 50 and 0 for the cutoff < 30. Even though false positive and false negative results should be given equal importance, from a scientific perspective, from a forensic context point of view, false negatives carry the potential danger of erroneously ruling out a forensically relevant case, closing the investigation.

The diagnostic performance of luminol chemiluminescence was evaluated using different cutoff values, reflecting different forensic relevance thresholds, as the diagnostic utility of a technique should always consider how forensic relevance is defined. The high TPR and PPV across all cutoffs (< 100, < 50, <30 years) indicate the test’s strong ability to correctly identify positive results. However, the low TNR values suggest a lower specificity due to the significant number of false positives. These results express that, although the luminol technique is effective at identifying recent cases, its reliability decreases for older remains due to lower specificity. DOR value for the cutoff < 100 showed reasonable diagnostic accuracy, being improved for the cutoff < 50. This value was not calculated to the cutoff < 30 due to the absence of false negatives, indicating robust performance. These findings suggest that luminol chemiluminescence is a valuable tool for estimating postmortem interval, particularly differentiating between recent and archaeological remains. Its high sensitivity makes it suitable for initial screenings, but the relatively low specificity advises the need for complementary tests to confirm results.

The usage of a 5-level scale for the PMI estimation through the luminol technique revealed substantial differences between PMI interval groups. According to our results, “barely positive” results were consistently exclusively recorded for PMI values over 30 years. On the other hand, “strong positive” and “positive” reactions were only observed in samples with no more than 20 or 25 years since death, respectively. These results highlight the potential of this technique when an intense or nearly faint chemiluminescence reaction is observed, as these results are never coincident in the same PMI range. However, a “weak positive” result was observed in a wide range of times since death (from 10 to 50 years), therefore highlighting the need for other techniques to corroborate the result and narrow the time range.

In effect, despite its wide use in forensic practice, the luminol chemiluminescence test usually requires further examination. It should be employed alongside other methods, like the evaluation of the inorganic components in bone tissue. If the results from these simple and faster methods align, it is likely that the estimation can be considered reasonably accurate. When there are no agreements in the estimated time interval, a posterior confirmation using more expensive methods that evaluate radioisotope levels in bone tissue (mainly radiocarbon dating) is needed. Combining methods represents a promising solution, especially for cases involving multiple fragments. In these cases, radiocarbon testing is unfeasible for economic reasons, so a preventive screening protocol should be applied in order to identify the elements worth testing. In forensic practice, due to the complexity of the decomposition process, a combination of methods is advisable, and the selection of these must rely on each case’s aspects.

The main disadvantage of the luminol technique, as with the vast majority of the methods intended for estimating the postmortem interval, relates to the number of factors that can influence the results obtained by this method. The primary limitations arise from the environmental burial conditions and diagenetic processes that can influence the decomposition rate, accelerating the degradation of the bone matrix, causing false negative results, or, on the opposite, slowing the decomposition process or mimicking positive reactions (interfering substances) even when no haemoglobin is present, revealing false positive results [9, 13, 15, 23]. Also, Caudullo et al. (2017) reported that the reliability of this technique in dating the time of death may be conditioned by the types of bone used, and its integrity (intact or more damaged). A previous work conducted by our team (Ermida et al. 2023a), aimed at accessing the influence of temperature, humidity, pH, and some soil types on the PMI estimation through the luminol test, concluded that these taphonomic factors can influence the results obtained through this technique. Therefore, this evaluation must be managed carefully, always considering the context in which a body is found. In the present study, the different origins of the collected samples did not allow the decomposition context to be considered in the analysis of the results, as it was unknown. Additionally, although the type of bone used was standardized, the level of bone preservation was not always the same, with some clavicles in the sample group being more damaged than others during the powder reduction process. Future research should focus on further investigating the impact of various factors on luminol chemiluminescence, such as environmental and taphonomic conditions to which the body has been exposed, its state of preservation, as well as intrinsic cadaver factors (e.g., cause of death or antemortem pathologies), which may affect haemoglobin degradation and consequently the luminol reaction [25]. Preparation procedures, the circumstances of the recovery, and storage conditions may also affect the outcome [8, 27]. Therefore, it is essential to normalize the entire procedure. Ideally, all samples should be stored uniformly until the method is applied, and the protocol should be thoroughly standardized.

The fact that a small portion of the sample (n = 7) came from autopsy cases may be considered a limitation of our research. Ideally, the entire sample would originate from the same location and include bones buried under the same conditions. However, we believe that this methodological difference did not affect the results, as they were similar to those obtained from buried bones with the same PMI.

Another limitation inherent to our study is the unequal distribution of the sample across the different postmortem interval ranges, along with the lack of individuals with PMI closer to the date of death. However, for ethical and judicial reasons, gathering human remains for scientific experiments can constitute a challenge. So, it was unattainable to increase the number of individuals for this sample. Hence, our team gathered all the possible clavicles with older/smaller postmortem intervals from the available collections, for which we obtained approval from all required entities, including ethics committees. In future research including a sample better distributed through the different PMI intervals would be valuable to yield even more robust results. Regardless of the outlined limitations, our work presents a robust sample, both in size and composition, with 239 individuals distributed for a wide range of PMI intervals.

Despite the mentioned limitations, luminol chemiluminescence presents many advantages supporting its widespread usage in forensic casework. Our research confirms it as a simple, fast, easy-to-interpret, and economical method. Technical aids, like digital image capture, are not mandatory since previous studies concerning the assessment of this technique intra and interobserver error by visual means, attest to its validity as a reliable and reproducible procedure, practically independent of observation bias, supporting its potential as a rapid diagnostic test [8, 17]. The new data presented in the current work adds important findings to the applicability of this technique. As many countries’ forensic thresholds to classify a case as of forensic relevance are lower than 30 years, the absence of false negative results for this cut-off value corroborates the pertinence of luminol chemiluminescence employment for distinguishing whether the remains are of forensic interest or not, further emphasizing the importance of the results arising from this investigation. Additionally, the noncoincidence between the “strong positive” and “positive” results up to 25 years PMI and the “barely positive” results from 30 years PMI onward represents an advancement in the understanding of this technique, as well as increased reliability in the application of the 5-level scale.

These important results bring this method closer to its acceptance and practical application in the legal setting. Still, some work remains for the luminol chemiluminescence technique to fully comply with the mandatory Daubert Standards [32], which ensure the quality, reliability, and replicability of the methods to be included in a scientific testimony in court. Despite being widely accepted for blood detection and meeting the criteria for testability and peer review, this technique application for PMI estimation is less established, needing further validation. However, studies like the present one provide valuable data, contributing to its recognition. According to our results, this methodology is reliable for PMI values up to 30 years, with clear trends in chemiluminescence intensity supporting its forensic application. It demonstrates high sensitivity but lower specificity, indicating the need for standardized protocols to reduce false positives. Thus, when evaluated against Daubert Standards, although this method meets several criteria, it still requires further refinement to enhance its diagnostic reliability.

This study corroborates what was previously reported in the literature, suggesting that the results achieved by this method must be interpreted with caution. Thus, according to our research, we cannot support the conclusion that luminol chemiluminescence is suitable for estimating time since death as a single method. Our findings demonstrate that a negative luminol test result, confirming a PMI of no forensic relevance, is more conclusive than a positive luminol test result. However, the almost absence of false negatives increases the interest in this technique as a presumptive test.