Influence of light conditions (colour temperature and illuminance) on the evaluation of root translucency for the application of Lamendin’s age‑at‑death estimation technique

Vol.:(0123456789) 1 3 International Journal of Legal Medicine https://doi.org/10.1007/s00414-022-02902-1 ORIGINAL ARTICLE Influence oflight conditions (colour temperature andilluminance) ontheevaluation ofroot translucency fortheapplication ofLamendin’s age‑at‑death estimation technique JoanViciano1 · IuriIcaro2 · CarmenTanga2 · DomenicoTripodi3 Received: 23 May 2022 / Accepted: 12 October 2022 © The Author(s) 2022 Abstract Estimation of age-at-death represents a central focus in forensic human identification, as it is a key parameter used in the identification of unidentified bodies. In 1992, Lamendin etal. published a simple technique for estimating the age-at-death of adult skeletal remains based on two dental criteria: the gingival regression and the extent of dentine translucency. Although Lamendin’s technique is widely used in forensic contexts and the evaluation of root translucency is a key element in the technique, the light conditions for measuring this parameter have not been adequately established. The aim of the present study is to analyse the influence of colour temperature and illuminance level of a LED light source when root translucency is evaluated to optimize the use of Lamendin’s technique for age-at-death estimation. The results describe how light settings may affect the visual perception of root translucency by different examiners and, therefore, affect the accuracy of the ageat-death estimation methods and techniques based on this parameter. Keywords Forensic anthropology· Lamendin’s technique· Dental age estimation· Root translucency· Light conditions Introduction The estimation of age-at-death from skeletal remains is one of the main challenges for forensic anthropologists and represents a key parameter used to identify unknown individuals [1]. In 1992, Lamendin etal. [2] published a simple technique for estimating the age-at-death of adult skeletal remains based on two dental criteria: the gingival regression and the extent of dentine translucency. Both variables were measured on the labial surface of single-rooted teeth and expressed in relation to the total root length. Lamendin’s technique and its derived modifications (e.g. [3, 4]) attracted considerable attention for forensic purposes because its accuracy, speed and simplicity due to its application was based on simple observations from intact teeth, and it did not require prior training or special equipment, only requiring a caliper and an adequate light source to record the root translucency. Since the initial development of Lamendin’s technique, several types of variability conditions were evaluated to optimize the use of this technique, such as (i) target teeth to which this technique is applied (i.e. differences between the diverse classes of single-rootedteeth (incisors, canines, premolars) or between maxillary and mandibular teeth); (ii) dental surface analysed (i.e. differences between labial/buccal or palatal/lingual tooth surfaces); (iii) biological profile of the analysed individual (i.e. differences between the sexes or between different populations); (iv) statistical strength for the development of the methodology; (v) environmental conditions (i.e. taphonomic impact and postmortem interval in dental tissue diagenesis); (vi) discrepancies in the application of the technique by different examiners (i.e. repeatability and reproducibility of the collected measurements) and (vii) other factors other than age that influence gingival regression and root translucency (e.g. dental pathologies, direct (on sectioned teeth) or indirect (on intact teeth) techniques) (see Parra etal. [5] for more details on these * Joan Viciano [email protected] 1 Department ofMedicine andAging Sciences, ‘G. d’Annunzio’ University ofChieti-Pescara, Chieti, Italy 2 Department ofLegal Medicine, Toxicology andPhysical Anthropology, University ofGranada, Granada, Spain 3 Department ofMedical andOral Sciences andBiotechnologies, ‘G. d’Annunzio’ University ofChieti-Pescara, Chieti, Italy International Journal of Legal Medicine 1 3 variability conditions for the application of Lamendin’s technique). Although Lamendin’s technique is widely used in forensic contexts and the evaluation of root translucency is a key element in the technique, light conditions for measuring this parameter have not been adequately established. Light settings used for measuring root translucency have not always been reported. Some authors only reported that root translucency was measured by placing the tooth against a bright light source, such as a light box or natural light, but did not provide further details (e.g. [3, 6–10]). Other authors provided more details of the light source such as the power of the lamp (in watts) (e.g. a 16-W negatoscope by Lamendin etal. [2] and Foti etal. [11]; a 40-W negatoscope by Vilacapoma Guerra [12]) or the type of light bulb (e.g. a LED X-ray viewer by Garizoain etal. [13]). Despite the different light sources used to evaluate root translucency, no indepth studies have been conducted on the influence of the light source on the evaluation of this parameter. Recently, Adserias-Garriga etal. [14] carried out a study on the light conditions used to acquire the root translucency measurements. They evaluated three diverse types of lights with different luminous intensities, 6500lx (equivalent to microscopic light), 3000lx (equivalent to negatoscope light) and 1600lx (equivalent to daily sunlight), and concluded that lighting should be considered to obtain a reliable estimation of age-at-death. Together with illuminance level, colour temperature is the other important characteristic of light to be considered regarding human visual perception [15]. However, this characteristic has not been investigated in previous studies. The colour temperature of a light source is defined as the colour of light emitted of an opaque and non-reflective blackbody whose colour is closest to that of the light source (i.e. it is a way to describe the light appearance provided by a light source) [16]. By convention, the colour temperature is expressed by the unit of absolute temperature, the Kelvin (K). Light from warm white light sources appear yellow white and has a colour temperature between about 2700 and 3500K. Cool white light is seen as blue white with colour temperatures ranging from 4500 to 7500K. Light sources with colour temperatures in the middle range (3500–4500K) are described as neutral white. Currently, the lighting industry formally refers to warm white (3000K), neutral white (3500K), cool white (4000–4500K) and daylight (6500K), based on the ANSI standard [16, 17]. Today, there are many LED light tables on the market that are used to evaluate the root translucency (e.g. [13, 18]). Their main advantage is their low cost and versatility due to their light stability and digital control. These LED light tables have a light illuminance level and colour temperature that can be adjusted through LED drivers, allowing digital control of the light emission and mixing of the different light parameters to adapt it to the requirements of different scenarios. Thus, the LED light spectrum will not only influence the chromatic aspect of the emitted light but also the colour perception of the objects illuminated by this light source. The aim of the present study is to analyse the influence of illuminance and colour temperature of a LED light source when the root translucency is evaluated to optimize the use of Lamendin’s technique for age-at-death estimation. Materials andmethods Sample andmeasurement collection Fifty-one permanent teeth were clinically extracted at the Department of Medical and Oral Sciences and Biotechnologies of the ‘G. d’Annunzio’ University of Chieti-Pescara (Italy). Teeth were extracted due to periodontal reasons and made available for educational purposes without any identifying information. Teeth were extracted clinically from their sockets, washed with water, digested for 5min in a 0.05% solution of sodium hypochlorite, dried and placed in plastic bags. Only single-rooted teeth unaffected by restorations and pathological processes (e.g. root caries, internal root resorption) were included in the study. Thus, from the original sample of 51 permanent teeth and after the application of inclusion/exclusion criteria, the final study sample consisted of a total of 30 teeth (6 maxillary and 21 mandibular incisors, 2 mandibular canines, 1 mandibular premolar). A digital dental caliper (Masel Orthodontics Inc., USA) was used for the collection of measurements to an accuracy of 0.01mm. The measurements were taken following the technique outlined by Lamendin etal. [2], involving root height (RH), periodontal regression height (PH) and root translucency height (RTH). RH was defined as the maximum distance from the apex of the root to the cemento-enamel junction. PH was defined as the maximum distance from the cemento-enamel junction to the line of soft tissue attachment. RTH was measured as the maximum extent of the translucent zone from the apex of the root. All measurements were taken macroscopically on non-sectioned teeth from the labial surface along the longitudinal axis of the tooth, and all values were recorded in millimetres. All teeth were measured for RH and PH under a LED desk lamp, and then, they were placed on a LED light table (Ohuhu, USA) in a dark room to measure RTH. This LED light table is adjustable, allowing different colour temperatures and illuminance intensities to be selected. According to the specifications of the manufacturer, the three colour temperatures of this LED light table were defined as cool white (5000K), neutral white (4000K) and warm white (2700K). Illuminance intensity was not defined by the manufacturer. To establish the illuminance intensity of the three levels (maximum, medium, minimum) International Journal of Legal Medicine 1 3 of the LED light table, a KPS-LX30LED luxmeter (KPS, Spain) was used. Ten measurements were collected for each combination of illuminance level and colour temperature, and the mean value was calculated as the reference value for defining the illuminance of this LED light table (Table1). Examiners andtraining session indental measurements Dental measurements were collected by three examiners with different backgrounds in dental anthropology: (i) the first examiner has a PhD degree and is highly experienced in odontometrics and Lamendin’s technique (experienced examiner); (ii) the second examiner is a PhD student with extensive knowledge in dental morphology and trained in odontometrics but without prior training in Lamendin’s technique (intermediate examiner) and (iii) the third examiner is a PhD student without previous knowledge in dental morphology and no prior training in odontometrics and Lamendin’s technique (inexperienced examiner). To train the intermediate and inexperienced examiners, a 3-h training session was conducted by the experienced examiner to recognize dental features and to correctly collect the RH, PH and RTH measurements following the original technique of Lamendin etal. [2], as well as for the correct use and adjustment of light settings (colour temperature and illuminance) of the LED light table. For this training procedure, the intermediate and inexperienced examiners measured a sample of 12 teeth in two sessions with direct feedback from the experienced examiner. Study design All three examiners (experienced, intermediate and inexperienced) collected RH, PH and RTH measurements using the same set of calipers and LED light table. All measurements were repeated at separate times, with a minimum period of 2weeks and a maximum of 1month between the two measurements. RTH was measured applying different combinations of colour temperature and illuminance (Fig.1). Figure2 illustrates the study design that was followed. All measurements were directly collected into a preformatted Table 1 Light settings (colour temperature and illuminance) of the LED light table Abbreviations: N, number of teeth; Max, maximum value; Min, minimum value; Mean, mean value; SD, standard deviation Colour temperature Illuminance (in lux) Level NMax Min Mean SD Cool white (5000K) Maximum 10 2880 2840 2863 13.17 Medium 10 1780 1749 1764 9.62 Minimum 10 119 116 118 0.81 Neutral white (4000K) Maximum 10 2000 1961 1979 11.33 Medium 10 1213 1186 1200 9.82 Minimum 10 80 77 78 1.01 Warm white (2700K) Maximum 10 2230 2190 2210 14.14 Medium 10 1521 1494 1508 9.08 Minimum 10 99 97 98 0.59 Fig. 1 View of root translucency height (RTH) of the same tooth with the different light settings (i.e. different combinations of colour temperature and illuminance) as defined in Table1 International Journal of Legal Medicine 1 3 Fig. 2 Schematic representation of the data collection procedure for each measurement International Journal of Legal Medicine 1 3 Microsoft Excel worksheet. Both randomization and blinding procedures were conducted to guarantee higher-quality data collection by preventing any subjective bias. Randomization ensured that teeth were randomly assigned to different combinations of colour temperature and illuminance for each examiner to prevent systematic arrangement during measurement procedure and to avoid predictability. In addition, the examiner entering the data onto the preformatted worksheet was different from the examiner taking the measurements to prevent selection bias and increase objectivity among the examiners. After taking the measurements, the Prince and Ubelaker formulae [3] to estimate age-at-death were applied. Due to the fact that the Prince and Ubelaker formulae are differentiated by sex and that this biological parameter is unknown in our specimens, in order to apply the age-at-death estimation formulae, sex was randomly assigned using the Excel RAND function. Statistical analysis All statistical analyses were conducted using the statistical package for social sciences software IBM SPSS Statistics 25.0 for Windows [19]. Initially, the differences between the means of the repeated measurements collected at the two separate times by the three examiners were analysed to evaluate possible intraexaminer and interexaminer error. The intraclass correlation coefficient (ICC) was calculated to determine the level of agreement between the repeated measurements collected by the same examiner and by the different examiners. The ICC calculations were conducted using the “two-way mixed-effects absolute-agreement” model for both the intraexaminer and interexaminer errors. As the measure of the intraexaminer and interexaminer agreement, the ICC and its 95% confidence interval (95% CI) were calculated. To determine the degree of agreement, the ICC calculated was compared to the criteria proposed by Koo and Li [20], which establishes four levels of qualitative assessment: ICC > 0.9 indicates “excellent” reliability; ICC from 0.75 to 0.9 indicates “good” reliability; ICC from 0.5 to 0.75 indicates “moderate” reliability and ICC < 0.5 indicates “poor” reliability. Next, after the verification of the potential intraexaminer and interexaminer error, the average of the repeated measurements collected at the two separate times was used to adjust the values for the subsequent analyses. Then, the data were assessed for normality using the Shapiro–Wilk test, with p < 0.05 defining statistical significance. A twoway repeated measures ANOVA was run to determine if the differences in the RTH measurements collected by the different examiners (within subjects) were influenced by the colour temperature and illuminance of the LED light source. Finally, after the application of the Prince and Ubelaker formulae [3] for age-at-death estimation, a three-way ANOVA was run to examine the effect of the examiner’s perception, colour temperature and illuminance on the estimated age-at death. For both the two-way and three-way repeated measures ANOVA, Mauchly’s test of sphericity was used to verify if the sample covariance matrix violated the assumptions of the repeated measures ANOVA. If the assumption of sphericity was violated, to avoid the risk of increasing the Type I error, corrections were applied to produce a valid F-value. For this purpose, the degrees of freedom for the effect were adjusted (when estimated epsilon [ε] was greater than 0.75, then the Huynh–Feldt correction was used; when ε was less than 0.75, then the Greenhouse–Geisser correction was used). Because multiple pairwise comparisons were performed, the Bonferroni correction was applied to adjust the significance level (p-value) to keep the type I error at 5% overall. To interpret the measure of the effect size provided by the partial eta-squared (ηp2), threshold values of the effect size were interpreted as a small (0.01), medium (0.06) or large effect (> 0.14) [21]. Results Intraexaminer andinterexaminer error analyses In the intraexaminer error analysis (Table2), the experienced examiner showed similar results for all the measurements (RH, PH and RTH), with ICC values ranging from 0.980 to 0.998. For the intermediate examiner, the ICC values for the RH and PH measurements ranged from 0.976 to 0.988, with slightly lower ICC values of 0.868–0.986 for the RTH measurements. For the inexperienced examiner, the ICC values for the RH and PH measurements were high, ranging from 0.971 to 0.975, with low ICC values of 0.080–0.975 for the RTH measurements. In the interexaminer error analysis (Table3), the comparison between the experienced vs. intermediate examiner showed a high agreement for the RH and PH measurements, ranging from 0.981 to 0.984, with slightly lower ICC values of 0.882–0.955 for the RTH measurements. The comparison of the inexperienced examiner with the other two examiners (i.e. experienced vs. inexperienced examiner and intermediate vs. inexperienced examiner) showed similar ICC values. For the RH and PH measurements, the ICC values were 0.973–0.974 for the comparison of experienced vs. inexperienced and 0.982–0.991 for the comparison of intermediate vs. inexperienced, with low values for the RTH measurements, from 0.132 to 0.793 for the comparison of experienced vs. inexperienced and from 0.170 to 0.929 for the comparison of intermediate vs. inexperienced. The ICC values showed high reproducibility in the intraexaminer error analyses (i.e. with “good” to “excellent” agreements) for both the experienced and intermediate International Journal of Legal Medicine 1 3 examiners, which indicated that the repeated measurements collected by them were particularly reliable. The overall data for the inexperienced examiner showed lower ICC values (which ranged from “poor” to “excellent” agreement). In addition, the overall data for the inexperienced examiner in comparison with the experienced and intermediate examiner (interexaminer error analyses) showed extremely low ICC values (which ranged from “poor” to “excellent” agreement). For this reason, measurements collected by the inexperienced examiner were excluded from the subsequent analyses and only the data of the experienced and the intermediate examiner were considered for performing the two-way and three-way repeated measures ANOVA analyses. Influence ofilluminance andcolour temperature ofaLED light source whentooth translucency isevaluated The Shapiro–Wilk test showed that the RTH measurements for all combinations of colour temperature and illuminance Table 2 Comparison of differences in the means for RH, PH and RTH measurements between repeated measurements within the examiners (i.e. intraexaminer error analysis) Abbreviations: RH, root height; PH, periodontal regression height; RTH, root translucency height; N, number of teeth; ICC, intraclass correlation coefficient; 95% CI, 95% confidence interval; F, F-statistic; p, p-value 95% CI Examiner Measurement Temperature Illuminance NICC Lower Upper F p Strength of agreement Experienced RH 30 0.998 0.996 0.999 496.087 0.000 Excellent PH 30 0.995 0.988 0.997 176.421 0.000 Excellent RTH Cool white Maximum 30 0.993 0.986 0.997 145.474 0.000 Excellent Medium 30 0.993 0.985 0.997 139.718 0.000 Excellent Minimum 30 0.980 0.954 0.991 54.893 0.000 Excellent Neutral white Maximum 30 0.992 0.962 0.997 190.905 0.000 Excellent Medium 30 0.996 0.991 0.998 227.142 0.000 Excellent Minimum 30 0.981 0.953 0.992 62.075 0.000 Excellent Warm white Maximum 30 0.988 0.976 0.994 83.865 0.000 Excellent Medium 30 0.996 0.991 0.998 224.969 0.000 Excellent Minimum 30 0.988 0.975 0.994 81.241 0.000 Excellent Intermediate RH 30 0.985 0.969 0.993 66.213 0.000 Excellent PH 30 0.976 0.949 0.989 44.833 0.000 Excellent RTH Cool white Maximum 30 0.903 0.781 0.955 11.672 0.000 Excellent Medium 30 0.986 0.970 0.993 71.495 0.000 Excellent Minimum 30 0.919 0.800 0.964 14.922 0.000 Excellent Neutral white Maximum 30 0.928 0.849 0.966 13.605 0.000 Excellent Medium 30 0.971 0.939 0.986 36.363 0.000 Excellent Minimum 30 0.893 0.765 0.950 10.331 0.000 Good Warm white Maximum 30 0.901 0.794 0.953 10.160 0.000 Excellent Medium 30 0.885 0.668 0.952 11.447 0.000 Good Minimum 30 0.868 0.605 0.946 10.186 0.000 Good Inexperienced RH 30 0.975 0.913 0.990 56.274 0.000 Excellent PH 30 0.971 0.939 0.986 34.394 0.000 Excellent RTH Cool white Maximum 30 0.973 0.944 0.987 38.195 0.000 Excellent Medium 30 0.601 0.166 0.809 2.879 0.003 Moderate Minimum 30 0.880 0.749 0.943 8.563 0.000 Good Neutral white Maximum 30 0.862 0.709 0.934 7.047 0.000 Good Medium 30 0.791 0.499 0.907 5.829 0.000 Good Minimum 30 0.080 − 0.952 0.564 1.086 0.413 Poor Warm white Maximum 30 0.975 0.948 0.988 39.494 0.000 Excellent Medium 30 0.770 0.510 0.891 4.809 0.000 Good Minimum 30 0.439 − 0.167 0.732 1.787 0.062 Poor International Journal of Legal Medicine 1 3 were normally distributed (p > 0.05) for both experienced and intermediate examiners. A factorial ANOVA (two-way repeated measures ANOVA) was conducted to compare the main effects of colour temperature and illuminance as well as their interaction effects on the collection of the RTH measurements, separated by examiners. Table4 shows the results of the ANOVA for the RTH measurements with different light settings. For the experienced examiner, the main effect of colour temperature was not statistically significant (p > 0.05). The main effect of illuminance indicated that 36.9% of the variance on the evaluation of the RTH measurement was explained by the illuminance levels (F[2, 58] = 16.955, p < 0.001). Multiple comparisons indicated that there were statistically significant differences between the maximum and medium illuminance when compared to the minimum illuminance (p < 0.001). Thus, when the experienced examiner measured the RTH, there were statistically significant differences depending on whether this parameter was evaluated with maximum (mean = 5.904, SD = 0.394) or medium Table 3 Comparison of differences in the means for RH, PH and RTH measurements between repeated measurements between the examiners (i.e. interexaminer error analysis) Abbreviations: RH, root height; PH, periodontal regression height; RTH, root translucency height; N, number of teeth; ICC, intraclass correlation coefficient; 95% CI, 95% confidence interval; F, F-statistic; p, p-value 95% CI Examiners Measurement Temperature Illuminance NICC Lower Upper F p Strength of agreement Experienced vs. intermediate RH 30 0.981 0.883 0.994 89.124 0.000 Excellent PH 30 0.984 0.966 0.992 61.202 0.000 Excellent RTH Cool white Maximum 30 0.939 0.799 0.976 22.804 0.000 Excellent Medium 30 0.912 0.805 0.959 12.669 0.000 Excellent Minimum 30 0.884 0.666 0.952 11.305 0.000 Good Neutral white Maximum 30 0.955 0.819 0.984 33.612 0.000 Excellent Medium 30 0.900 0.672 0.961 14.031 0.000 Excellent Minimum 30 0.890 0.747 0.950 10.489 0.000 Good Warm white Maximum 30 0.908 0.608 0.967 17.022 0.000 Excellent Medium 30 0.882 0.743 0.945 9.427 0.000 Good Minimum 30 0.905 0.715 0.961 14.029 0.000 Excellent Intermediate vs. inexperienced RH 30 0.991 0.977 0.996 129.645 0.000 Excellent PH 30 0.982 0.961 0.991 55.266 0.000 Excellent RTH Cool white Maximum 30 0.596 0.145 0.809 2.919 0.003 Moderate Medium 30 0.170 − 0.348 0.540 1.305 0.239 Poor Minimum 30 0.929 0.835 0.968 16.290 0.000 Excellent Neutral white Maximum 30 0.566 − 0.185 0.829 4.001 0.000 Moderate Medium 30 0.211 − 0.381 0.585 1.348 0.213 Poor Minimum 30 0.603 0.067 0.823 3.312 0.001 Moderate Warm white Maximum 30 0.590 − 0.136 0.837 4.008 0.000 Moderate Medium 30 0.311 − 0.214 0.640 1.687 0.082 Poor Minimum 30 0.564 0.034 0.800 2.928 0.003 Moderate Experienced vs. inexperienced RH 30 0.974 0.453 0.994 116.428 0.000 Excellent PH 30 0.973 0.944 0.987 36.321 0.000 Excellent RTH Cool white Maximum 30 0.549 − 0.092 0.805 3.222 0.001 Moderate Medium 30 0.132 − 0.268 0.478 1.295 0.245 Poor Minimum 30 0.793 0.264 0.922 7.530 0.000 Good Neutral white Maximum 30 0.566 − 0.185 0.829 4.001 0.000 Moderate Medium 30 0.198 − 0.248 0.543 1.461 0.156 Poor Minimum 30 0.482 − 0.178 0.773 2.950 0.002 Poor Warm white Maximum 30 0.590 − 0.136 0.837 4.008 0.000 Moderate Medium 30 0.233 − 0.224 0.572 1.614 0.102 Poor Minimum 30 0.473 − 0.206 0.773 3.072 0.002 Poor International Journal of Legal Medicine 1 3 (mean = 5.801, SD = 0.384) illuminance in comparison with minimum (mean = 5.634, SD = 0.365) illuminance. The interaction effect was statistically significant (F[4, 116] = 4.004, p < 0.01), indicating that there was a combined effect of colour temperature and illuminance level on the evaluation of the RTH measurement, yielding a 12.1% of the variance explained by these combined factors. Multiple comparisons indicated that there were statistically significant differences between the cool-white light and neutral-white light (p < 0.05) and between cool-white and warm-white light (p < 0.05) when RTH was measured under a minimum illuminance. Thus, when the experienced examiner measured the RTH, if teeth were evaluated with a minimum illuminance, there were statistically significant differences (p < 0.05) depending on whether this parameter was evaluated under a cool-white light (mean = 5.515, SD = 0.366), neutral-white light (mean = 5.692, SD = 0.366) or warmwhite light (mean = 5.696, SD = 0.369). Similar results were obtained by the intermediate examiner. The main effect of colour temperature was not statistically significant (p > 0.05). The main effect of illuminance indicated that 12.4% of the variance on the evaluation of the RTH measurement was explained by the illuminance levels (F[2, 58] = 4.089, p < 0.05). Multiple comparisons indicated that there were statistically significant differences between the maximum illuminance when compared to the minimum illuminance (p < 0.05). Thus, when the intermediate examiner measured the RTH, there were statistically significant differences depending on whether this parameter was evaluated with maximum (mean = 5.247, SD = 0.393) or minimum (mean = 4.997, SD = 0.362) illuminance. The interaction effect was statistically significant (F[4, 116] = 4.874, p = 0.001), indicating that there was a combined effect of colour temperature and illuminance level on the evaluation of the RTH measurement, yielding a 14.4% of the variance explained by these combined factors. Multiple comparisons indicated that there were statistically significant differences between the cool-white light and neutral-white light when RTH was measured under a medium or minimum illuminance (p < 0.05). Thus, when the intermediate examiner measured the RTH, if teeth were evaluated with a medium illuminance, there were statistically significant differences (p < 0.01) depending on whether this parameter was evaluated under a cool-white light (mean = 5.334, SD = 0.420) or neutral-white light (mean = 5.091, SD = 0.402). When teeth were evaluated under a minimum illuminance, there were also statistically significant differences (p < 0.05) in the RTH measurements depending on whether this parameter was evaluated under a cool-white light (mean = 4.821, SD = 0.346) or neutral-white light (mean = 5.150, SD = 0.375). Application ofthePrince andUbelaker formulae forage‑at‑death estimation Because root translucency is a key parameter for estimating the age-at-death from skeletal remains and in light of the results obtained previously, the Prince and Ubelaker formulae [3] to estimate age-at-death were applied to the sample (see section “Study design” for more details on the sex assignment for the specimens). Thus, a factorial ANOVA (three-way repeated measures ANOVA) was conducted to examine the main effects of examiner, colour temperature and illuminance as well as their interaction effects on the estimation of the age-at-death. The Shapiro–Wilk test showed the data of the estimated age-at-death were normally distributed (p > 0.05) for all combinations of colour temperature and illuminance. The results of the three-way repeated measures ANOVA (Table5) revealed that the main effect of examiner indicated that 23.2% of the variance on the estimated age-at-death was explained by differences between the examiners when the different measurements (RH, PH and RTH) were collected for the application of the Prince and Ubelaker formulae (F[1, 29] = 8.763, p < 0.01). A comparison between examiners indicated that there were statistically significant differences (p < 0.01) between the experienced (mean = 48.250, SD = 1.236) and the intermediate (mean = 46.675, SD = 1.271) examiner. The main effect of illuminance indicated that 30% of the variance on the evaluation of the estimated age-at-death was explained by the illuminance levels (F[2, 58] = 12.424, p < 0.001). Multiple comparisons Table 4 ANOVA for the RTH measurements, depending on the light settings Abbreviations: SS, sum of squares; df, degrees of freedom; MS, mean squares, F, F-statistic; p, p-value; ηp2, partial eta square Examiner Factor SS df MS F p ηp2 Experienced Temperature 0.374 1.692 0.221 2.567 0.095 0.081 Illuminance 3.324 2 1.662 16.955 0.000 0.369 Temperature × illuminance 0.619 4 0.155 4.004 0.004 0.121 Intermediate Temperature 0.522 2 0.261 0.947 0.394 0.032 Illuminance 3.256 2 1.628 4.089 0.022 0.124 Temperature × illuminance 3.522 4 0.881 4.874 0.001 0.144 International Journal of Legal Medicine 1 3 indicated that there were statistically significant differences (p < 0.001) between the maximum (mean = 47.764, SD = 1.257) and medium illuminance (mean = 47.590, SD = 1.240) when compared to the minimum illuminance (mean = 47.034, SD = 1.186). No statistically significant main effect on colour temperature was observed (p > 0.05). The interaction effect of temperature × illuminance was statistically significant (F[4,116] = 6.574, p < 0.001), indicating that there was a combined effect of colour temperature and illuminance level on the estimated age-at-death, yielding an 18.5% of the variance explained by these combined factors. Multiple comparisons indicated that there were statistically significant differences between the coolwhite light and neutral-white light (p < 0.001) and between cool-white light and warm-white light (p < 0.05) when age-at-death was estimated using a minimum illuminance. Thus, under a minimum illuminance of the light source, there were statistically significant differences depending on whether this parameter was evaluated under a coolwhite light (mean = 46.589, SD = 1.167), neutral-white light (mean = 47.368, SD = 1.211) or warm-white light (mean = 47.145, SD = 1.193). No statistically significant interaction effect of examiner × temperature and examiner × illuminance was observed (p > 0.05). Finally, the interaction effect of examiner × temperature × illuminance was statistically significant (F[4,116] = 2.931, p < 0.05), indicating that there was a combined effect of examiner, colour temperature and illuminance level on the estimated age-at-death, yielding a 9.2% of the variance explained by these combined factors. Table6 shows the multiple comparisons based on the effects of the interaction examiner × temperature × illuminance on the estimated age-at-death. These comparisons indicated statistically significant differences between the experienced and intermediate examiner on the estimated age-at-death in all combinations of colour temperature and illuminance (p < 0.01), except the following combinations of colour temperature and illuminance (p > 0.05): coolwhite light and medium level of illuminance, neutral-white light and minimum level of illuminance and warm-white light and medium level of illuminance. Table 5 ANOVA for the estimated age-at-death after the application of the Prince and Ubelaker formulae [3], depending on examiners and light settings (colour temperature and illuminance) during the collection of the different dental measurements of the sample Abbreviations: SS, sum of squares; df, degrees of freedom; MS, mean squares, F, F-statistic; p, p-value; ηp2, partial eta square Factor SS df MS F p ηp2 Examiner 335.136 1 335.136 8.763 0.006 0.232 Temperature 7.277 2 3.638 2.672 0.078 0.084 Illuminance 52.340 2 26.170 12.424 0.000 0.300 Examiner × temperature 1.452 2 0.726 0.430 0.653 0.015 Examiner × illuminance 0.812 2 0.406 0.207 0.814 0.007 Temperature × illuminance 23.483 4 5.871 6.574 0.000 0.185 Examiner × temperature × illuminance 11.639 4 2.910 2.931 0.024 0.092 Table 6 Multiple comparisons based on the interaction examiner × temperature × illuminance on the estimated age-at-death Abbreviations: SE, standard error; p, p-value; 95% CI, 95% confidence interval for difference 95% CI Examiner comparison Temperature Illuminance Mean differences SE pLower Upper Experienced vs. intermediate Cool white Maximum 1.568 0.540 0.007 0.464 2.672 Medium 1.130 0.671 0.103 − 0.242 2.502 Minimum 1.699 0.616 0.010 0.440 2.958 Neutral white Maximum 1.303 0.435 0.006 0.413 2.194 Medium 2.002 0.618 0.003 0.738 3.266 Minimum 1.333 0.682 0.060 − 0.061 2.727 Warm white Maximum 2.107 0.567 0.001 0.948 3.266 Medium 1.285 0.697 0.075 − 0.140 2.710 Minimum 1.752 0.582 0.005 0.561 2.943