Applicability of the mandibular canine index for sex estimation: a systematic review

Abstract

Background

The Mandibular Canine Index (MCI) comprises a method of sex estimation by teeth that presents controversial results in the literature.

Main body

This systematic review aims to expose whether MCI can be used as a method of reliable sex estimation. A literature search was performed using the keywords “canine,” “sex,” “gender,” “determination,” “estimation,” “dimorphism,” “assessment,” “forensic” in the databases Pubmed, Scopus, Lilacs, Scielo, and Web of Science. In addition, manual searches were carried out on the reference lists of the selected articles to cover the largest number of articles of interest as possible. Studies that performed the measurements only on maxillary canines, scientific conferences abstract books, case reports and literature reviews were excluded. The assessment of methodological quality and risk of bias was carried out based on a checklist for cross-sectional studies and another for accuracy studies. Thus, 53 articles were selected, 13 of which were accurate and 40 were cross-sectional. All accuracy articles were assessed as low risk. Among cross-sectional articles, seven were considered to be of low risk, 31 of moderate risk, and two of high risk. The accuracy of the sex estimate by MCI was verified and, despite varying among studies, the minimum and maximum values found were, respectively, 20% and 87.5% for women and 40.6% and 94% for men.

Conclusion

The accuracy of the MCI was variable and should be used with caution and as an auxiliary method of sex estimation.

Background

The construction of the victim’s biological profile comprises an important stage of the human identification process in which knowledge of forensic anthropology is applied in order to reduce the number of potential suspects (Francisco et al. 2013). This initial screening is performed when primary identification methods need to be applied to bodies with an advanced state of decomposition, carbonized, fragmented or skeletonized, and it prevents time and resources from being wasted (Francisco et al. 2013; Silva et al. 2015).

In extreme situations in which it is not possible to easily distinguish human remnants from non-humans (Dias et al. 2012; Oliveira et al. 2008; Silva et al. 2013), the construction of the biological profile must be initiated by determining the species so that, afterwards, parameters such as sex, age, stature and ancestry can be estimated (Silva et al. 2013). Estimating sex, in turn, is particularly important because it is an information that will guide the methods to be applied in estimating the other parameters that will form the victim's biological profile (Krishan et al. 2016; Rösing et al. 2007).

To estimate sex in adults, cranial elements can be used (Spradley and Jantz 2011) and, in forensic practice, the bones of the pelvis and skull are widely used because they have morphological and morphometric characteristics evidenced by pubertal hormones (Koelzer et al. 2019; Krishan et al. 2016; Rogers 2005; Rösing et al. 2007; Sinhorini et al. 2019), although the presence of youthful traits and androgenic characteristics may be present (Krishan et al. 2016; Silva et al. 2015).

In addition to the bone elements, research was carried out to ascertain or even quantify the presence of sexual dimorphism in the teeth, especially of the canine tooth in the male sex as it is evolutionarily associated with hunting activities and primate survival (Rao et al. 1989; Vijayan et al. 2019). Thus, the use of teeth in sex estimation is motivated by the resistance that these organs present, which would be particularly useful in the identification of extremely fragmented bodies or incomplete bones (Acharya et al. 2011; Azevedo et al. 2019; Vijayan et al. 2019).

Rao et al. (Rao et al. 1989) proposed a simple and practical method for sex estimation that was based on the ratio between the mesiodistal measurement of the mandibular canine tooth and the lower intercanine distance, which is why this method became known as the Mandibular Canine Index (MCI). Thus, this systematic review aims to answer the following question: the MCI can be applied as a reliable method for sex estimation of an unknown person?

Main text

Material and methods

Search strategy

The searches were conducted on September 20th, 2019. In order to find relevant studies, there were no language restrictions, a range of publication year was not used, and the search was performed in the electronic databases Pubmed, Scopus, Lilacs, Scielo, and Web of Science. The search strategy was adequate for each database and the following keywords were used: canine, sex, gender, determination, estimation, dimorphism, assessment, forensic, as shown in Table 1. Manual searches were carried out on the reference lists of the selected articles to verify whether previous searches in the databases failed to identify any studies of interest.

Table 1 Quantity of articles found in each database according to the search strategy

Article eligibility

The PICO strategy (P—Population; I—Intervention; C—Comparison; O—Outcome) is fundamental for the construction of the appropriate research question, which determines a comprehensive bibliographic search, which incorporates the best scientific data available on the studied topic (Santos et al. 2007). Thus, this systematic review included articles that submitted patients (P), who had the sex previously known by those responsible for the research (C), to certain dental measures to calculate the Mandibular Canine Index (I) in order to estimate the sex of these individuals (O). Studies that performed the measurements only on maxillary canines, abstract books of scientific congresses, case reports, and literature reviews were excluded.

Initially, the selection was made by reading the titles and abstracts of the articles found. If the abstract was not available or if any doubts persisted regarding inclusion or not, the studies were downloaded and read in full. Some full texts were not found by manual search or using the Unpaywall software. The authors were asked to provide them via ResearchGate (http://www.researchgate.net), however, to no avail. Scientific papers found in more than one database were considered only once. The search and selection of the articles were performed by two different researchers independently. Then, the selected studies were checked by both researchers. When there was doubt about include or exclude any article, a third examiner was consulted.

Assessment of quality and risk of bias

The methodological quality assessment used two checklists proposed by The Joanna Briggs Institute: (The Joanna Briggs Institute 2017, 2020) one for cross-sectional studies, named “Checklist for Analytical Cross-Sectional studies” (The Joanna Briggs Institute 2020), which has an observational design and participants are selected only by the inclusion and exclusion criteria (Setia 2016), and another directed to the accuracy studies, named “Checklist for Diagnostic Test Accuracy Studies” (The Joanna Briggs Institute 2017). In the analysis of cross-sectional studies (Table 2), topics were discussed on the details of the chosen samples, such as the inclusion and exclusion criteria, and demographic data; the measurements taken and the examiners (validity, reliability, training); possible confounding factors and on statistical analysis. When the information requested in each item was answered with “Yes,” the score of 11.11% was applied; the value zero was assigned when the question received the answer “U” (uncertain) and when “N/R” (not/reported) was used, the value 5.5% was determined. The percentage column relates to the percentage of “Yes” that each study presents. Thus, when the percentage of “Yes” remains below 49%, the risk is considered “High”; if the percentage is between 50 and 69%, it is classified as “Moderate”; and “Low” is applied if the percentage is equal to or greater than 70%.

Table 2 Criteria for quality assessment of cross-sectional studies (The Joanna Briggs Institute 2020)

For the accuracy studies (Table 3), items related to the sample and the method of carrying out the measurements were verified. Regarding the samples, the surveys were evaluated according to the sampling method used and the exclusion criteria applied. For the evaluation of the measurements made, the following characteristics were observed: the blinding of the examiners, the data of the reference standard and if the time of analysis of the reference was compatible with the time of analysis of the variable of interest. In this case, when the information requested in each item was answered with “Yes,” a score of 12.5% was used; the value of 6.25% was applied if the question received the answer “Unclear” and when “No” was used, the value zero was determined. The percentage column relates to the percentage of “Yes” that each study presents. Thus, when the percentage of “Yes” remains below 49%, the risk is considered “High”; if the percentage is between 50 and 69%, it is classified as “Moderate”; and “Low” is applied if the percentage is equal to or greater than 70%.

Table 3 Criteria for quality assessment of accuracy studies (The Joanna Briggs Institute 2017)

Results

From the search in the electronic databases, 1806 articles were found, of which 1590 remained eligible after the removal of duplicates. Then, reading the titles and abstracts allowed to select 61 studies and, after applying the inclusion and exclusion criteria, 53 articles were included in this systematic review, as illustrated in Fig. 1. A description of the studies involved in this work containing the year of publication, title, online journal in which they were published, and the respective Quartile is shown in Table 4.

Fig. 1

Flowchart of the search results in the databases

Table 4 Title, journal and quartile of the included articles

From the analysis of the methodological quality of the accuracy studies, Table 5 explains that only Rajarathnam et al. commented on the sampling method applied (Rajarathnam et al. 2016), while all articles scored positively on the items on the precaution of using case-control and inappropriate exclusions, the performance of measures in a standardized way and the inclusion of all patients in the analysis. For questions regarding the blindness of the survey examiners, the affirmative answer was received only by Azevedo et al. (Azevedo et al. 2019) and Silva et al. (Silva et al. 2016). Finally, no article scored positively in relation to the information on the reference data for comparison with those found by the studies.

Table 5 Outcomes of the assessment of the risk of bias within eligible accuracy studies

It is clear that everyone reached a score above 70%, and classified as low risk. Azevedo et al. (Azevedo et al. 2019) and Silva et al. (Silva et al. 2016) received the maximum score in seven of the nine questions applied, adding a percentage of 88.77%. Other values received were 83.16% and 77.55% (Table 4), directed to the studies that gained the highest value in six and five items of the methodological evaluation, respectively.

The checklist for cross-sectional studies showed that only two studies did not clearly provide the inclusion and exclusion criteria. Regarding the second question, five studies did not describe the sample in detail. About the standardization of measures, all articles received a positive score; however, for the requirements on confounding factors, no article scored positively. When verifying the statistical analysis, seven studies did not clearly inform the statistical method used (Table 6).

Table 6 Outcomes of the assessment of the risk of bias within eligible cross-sectional studies

It appears that only seven studies were considered low risk for obtaining a 75% percentage and only Muhamedagic and Sarajlic (Muhamedagić and Sarajlić 2013) and Yadav et al. (Yadav et al. 2002) reached lower percentages, being classified as high risk. The remaining studies received percentages between 50 and 68.75% and, therefore, were identified as of moderate risk.

The samples of the selected studies involved the population of some countries in Asia, Africa, Europe, and America, with the Indians being the most investigated (n = 37), and the sample size varied between 50 (Krishnan et al. 2016; Muthukumar and Thenmozhi 2018) and 1000 (Gargano et al. 2014; Jacob et al. 2018) participants. The youngest individuals analyzed were 14 years old (Dhakar et al. 2012; Duraiswamy et al. 2009; Sherfudhin et al. 1996); the most advanced age examined was 60 years old (Gargano et al. 2014; Sassi et al. 2012) and some studies (Atreya et al. 2019; Muhamedagić and Sarajlić 2013; Muller et al. 2001) did not inform the studied age group.

Most of the selected studies analyzed only the right and left mandibular canines. However, some authors have also included the evaluation of the right and left maxillary canines. The measurements were performed on plaster models, digital models, radiographs and in situ evaluation. With regard to the examiners who performed the measurements, participation ranged from zero to ten in the surveys that provided this information (Table 7).

Table 7 Description of included studies

The data obtained by the studies included in this review are shown in Table 8. The lowest values found for the average of the mandibular canine index were 0.1896 ± 0.0175 in women and 0.1921 ± 0.0185 (Kakkar et al. 2013) in men, and the largest, discovered by Vijayan, Jayarajan, and Jaleel (Vijayan et al. 2019), were 0.521 ± 0.012 in women and 0.444 ± 0.010 in men for the right mandibular canine index and 0.526 ± 0.014 in women and 0.447 ± 0.012 in men for the left mandibular canine index.

Table 8 Results found by the included studies

Canine sexual dimorphism was analyzed, and the highest rate found was 15.24% for the right canine, and 16.74% for the left canine (Ibeachu et al. 2012). The accuracy of the sex estimation by MCI was verified and, despite varying among studies, the minimum and maximum values found were, respectively, 20% (Anu et al. 2018) and 87.5% (Rao et al. 1989) for women and 40.6% (Jacob et al. 2018) and 94% (Silva et al. 2016) for men. The results found led some studies (Acharya et al. 2011; Acharya and Mainali 2009; Atreya et al. 2019; Gargano et al. 2014; Hosmani et al. 2013) to the conclusion that the canine mandibular index is not reliable for sex estimation, while others (Bakkannavar et al. 2015; Dhakar et al. 2012; Divyadharsini and Kumar 2019; Kaushal et al. 2004; Kumawat et al. 2017) guaranteed the applicability of this index in forensic practice.

Discussion

MCI is a widely researched method of sex estimation by teeth, although its practical application remains controversial. When applying MCI in different populations, divergent levels of accuracy were found (Table 8), and information related to monomorphism and reverse dimorphism of canine teeth as a result of human evolution was also reported (Boaz and Gupta 2009; Prabhu and Acharya 2009), which makes the use of MCI in forensic practice as an even more dubious method of sex estimation.

The divergence related to the use of the canine mandibular index in sex estimation is present even in studies of the same population. The Indians were the most evaluated by the studies included in this systematic review and, while some authors stated that MCI has a statistically significant sexual dimorphism, presenting rates starting at 80% (Dhakar et al. 2012; Gandhi et al. 2017; Kaushal et al. 2004; Kumawat et al. 2017; Latif et al. 2016; Paramkusam et al. 2014; Patel et al. 2017; Rao et al. 1989; Singh et al. 2015; Sreedhar et al. 2015), others showed accuracy below 50% (Acharya et al. 2011; Anu et al. 2018; Hosmani et al. 2013; Jacob et al. 2018; Krishnan et al. 2016; Srivastava 2010) supporting the idea that MCI is not a significant sex estimation method.

Research carried out in Nepal (Acharya and Mainali 2009; Atreya et al. 2019), Portugal (Azevedo et al. 2019; Silva et al. 2016) and Uruguay (Gargano et al. 2014; Sassi et al. 2012) showed that the mandibular canine index showed insufficient capacity to estimate sex and should be used with caution. However, MuhamedagIic and Sarajlic tried to find out if mandibular canines could provide elements that estimate sex in the population of Bosnia and Herzegovina and reported that the right MCI indicated greater accuracy in relation to other measures (Muhamedagić and Sarajlić 2013).

This contrast can be explained by the fact that both genetic and structural formation are different between individuals, even though they are part of the same population (Jacob et al. 2018). In addition, the results for the MCI can be diverse depending on the geographic area in which each individual lives, being necessary to carry out a random sampling in each region to obtain results with greater accuracy (Singh et al. 2015).

On the other hand, the mesiodistal distance of both mandibular canine teeth, one of the components of the MCI formula, was able to estimate sex with an 85.8% level of general accuracy (78% for males and 91.4% for the female sex), while the intercanine distance presented a level of 71.7% (74% for the male sex and 70% for the female sex) of accuracy in the research by Azevedo et al. (Azevedo et al. 2019). When in the form of MCI, Azevedo et al. found levels of general accuracy of 63.3 and 64.2% for left and right canine teeth, respectively (Azevedo et al. 2019). Other studies (Acharya et al. 2011; Divyadharsini and Kumar 2019; Ibeachu et al. 2012; Paramkusam et al. 2014; Patil et al. 2015) also confirm the results found by Azevedo et al. when evidencing that the mesiodistal length of the canines revealed a greater degree of accuracy for sex estimation than the MCI (Azevedo et al. 2019). Questions about the influence of the intercanine distance on the accuracy of the MCI can be raised, especially if factors such as the shape of the lower dental arch, the presence of dental crowding or diastemas, or even dental movement resulting from dental absences are considered.

When thinking about the sexual dimorphism of canine, there is a disagreement about which tooth would be able to more accurately predict sex. The left canine showed a higher degree of dimorphism for some authors (Gandhi et al. 2017; Kaushal et al. 2004; Kaushal et al. 2003; Paramkusam et al. 2014; Rajarathnam et al. 2016; Reddy et al. 2008; Sharma and Gorea 2010; Singh et al. 2015) and the value of 16.74% was found by Ibeachu, Didia, and Orish for that parameter (Ibeachu, Didia and Orish et al. 2012). However, there is also evidence in the literature that the right canine provides higher rates of sexual dimorphism. Vishwakarma and Guha found that the right canine was more dimorphic than the left, with levels of 12.51 and 10.15%, respectively (Vishwakarma and Guha 2011).

The applicability of the canine to differ men from women is related to the action of the Y chromosome, which regulates the thickness of the dentin, influencing the size of the tooth (Garn et al. 1967). However, in addition to genetic factors, the performance of ethnic, environmental, nutritional, and cultural factors can attribute to teeth’s different rates of sexual dimorphism in different studies and populations (Nagalaxmi et al. 2014).

In addition, the different applied methodologies may have played an important role in the variability of the results obtained by the studies. Patel et al. performed measurements of dental dimensions in plaster models and in the individual’s oral cavity in an Indian population and, when comparing them, found that plaster models contributed to a better analysis, providing a high degree of accuracy (Patel et al. 2017). On the other hand, other studies also carried out in India identified that the measurements made intraorally and in the plaster models did not differ from each other and showed to be similar and accurate (Kaushal et al. 2003; Rajarathnam et al. 2016).

The use of plaster models to perform dental measurements has advantages over intraoral analysis, since conditions such as the existence of spacing, inclined teeth, rotations, interproximal contacts, and anatomical variations can intervene in the accuracy and repeatability of measurements of teeth in the cavity oral (Zilberman et al. 2003). Thus, in addition to facilitating the analysis of measurements when the elements mentioned above are present, plaster models allow them to be examined at another time to reduce errors during measurements due to fatigue (Patel et al. 2017).

Thus, from a methodological point of view, the use of plaster models allows studies to approach the forensic routine in which the expertise teams have skeletonized or fragmented human remains. However, since the plaster model production process may suffer environmental and operator interference, it is possible that research using skeletonized human remnants of known sex may elucidate the role of MCI in estimating sex between different populations.

Conclusions

The accuracy of the MCI was shown to vary among different populations and even within the same population. Because of this, MCI must be considered an auxiliary method in estimating sex, but its application must be viewed with great caution. On the other hand, the mesiodistal length of the canine showed a high degree of sexual dimorphism.