Title: Reliability of the atd Angle in Dermatoglyphic Analysis

Authors and Institutions:

Emily K. Brunson1

Darryl J. Holman1

Laura L. Newell-Morris1

Christina M. Giovas1

Ann Striessguth2

1Department of Anthropology

2Department of Psychiatry and Behavioral Sciences

University of Washington

Seattle, WA 98195

Total Number of Pages: 16 Number of Figures: 1 Number of Tables: 2 Abbreviated title: Reliability of the atd angle Send Proof to:

Emily K. Brunson Department of Anthropology Box 353100 University of Washington Seattle, WA 98195-3100 USA Fax: (206) 543-3285 Email: ebrunson@u.washington.edu

Keywords: atd angle, dermatoglyphics, reliability

2 Brunson, et al.

ABSTRACT

The atd angle is a dermatoglyphic trait formed by drawing lines between the triradii below the

first and last digits and the most proximal triradius on the hypothenar region of the palm. This trait has

been widely used in dermatoglyphic studies, but several researchers have questioned its utility,

specifically whether or not it can be measured reliably. In this paper, we examine the measurement

reliability of this trait.

Finger and palm prints were taken using the carbon paper and tape method from 287

individuals with in utero exposure to alcohol and 175 unexposed controls. A sample of 100 matched

(right and left) prints was constructed from 75 randomly selected controls and 25 randomly selected

cases. Each atd angle was read twice, at different times, by Reader A, using a goniometer and a

magnifying glass, and three times by a Reader B, using Adobe Photoshop. Inter-class correlation

coefficients were estimated for the intra- and inter-reader measurements of the atd angles.

Reader A was able to quantify atd angles on 149 out of 200 prints (74.5%), and Reader B on

179 out of 200 prints (89.5%). Both readers agreed on whether an angle existed on a print 89.8% of

the time for the right hand and 78.0% for the left hand. Intra-reader correlations were 0.97 or greater

for both readers. Inter-reader correlations for atd angles measured by both readers ranged from 0.92

to 0.96. These results suggest that the atd angle can be measured reliably, and further imply that

measurement using a software program with an angle measurement tool to measure the angle on

digitized prints may improve this reliability.

3 Brunson, et al.

Dermatoglyphic traits are widely used markers in analyses of fetal development, developmental

disturbances, disease and genetics (e.g. Woolf and Gianas, 1976; David, 1981(b); Pursnani et al.,

1989; Arrieta et al., 1990; Wolman et al., 1990; Balgir, 1993; Balgir et al., 1993; Kohli and Bhalla,

2000; Paez et al., 2001; Neiswanger et al., 2002; Avila et al., 2003; Milicic et al., 2003). Most

dermatoglyphic traits develop in utero during weeks 17 through 24 and remain unchanged during an

individual’s lifetime (Babler, 1991). Therefore, analyses of dermatoglyphic traits can provide

information on intrauterine disturbances and possibly even genetic abnormalities (Babler, 1991;

Chakraborty, 1991; Meier, 1991). The most commonly analyzed dermatoglyphic traits include

descriptions of finger and hypothenar patterns, measurements of ridge counts between triradii (points

formed by the convergence of three patterns of ridges) and measurement of the atd angle, the angle that

exists between the a, d, and t triradii on the palm (Figure 1).

Although the atd angle has been widely used in dermatoglyphic studies, several researchers

have questioned its utility, specifically whether or not it can be measured reliably (Cummins and Mildo,

1961; Sharma, 1964; Mavalwala, 1979; Priest, 1979; David, 1981a). Measurement of the atd angle

involves locating three triradii and then measuring the angle between these points. Each step in this

process increases the possibility of reader error, including inconsistent identification of landmarks.

Measurement of atd angles is occasionally further complicated by the presence of additional a, d or t

triradii on some prints.

The purpose of our research was to determine the reliability of atd angle measurements within

and between readers. We also examined the literature in which results from the analyses of atd angles

4 Brunson, et al.

were reported in order to assess the consistency of associations between atd angles and developmental,

disease and genetic conditions of interest.

MATERIALS AND METHODS

Palm prints were taken using the carbon paper and tape method (Aase and Lyons 1971) from

287 individuals with in utero exposure to alcohol and 175 unexposed controls. A sample of 100

matched (right and left) prints was constructed from 75 randomly selected controls and 25 randomly

selected cases. We chose to include both affected and unaffected subjects in this sample because this is

a typical design for studies using the atd angle.

Atd angles from these 200 prints were read by two readers (A and B) using the criteria set forth

for atd angle measurement by Elbualy and Schindeler (1971). Each angle was read twice by Reader A,

through the process of placing enlarged photocopies of the prints into a transparent plastic sleeve,

locating and marking the a, d and t triradii on the sleeve, and drawing straight lines from a to t and from

d to t. The atd angle was then read to the nearest degree using a basic goniometer. Reader B read

each angle three times from images created by digitally scanning each print at a resolution of 400 dots

per in (DPI). The prints were digitally enlarged so that the a, d and t triradii could be marked, and then

the Photoshop (version 8.0, Abobe Systems, Inc., San Jose, CA) angle measurement tool was used to

measure the resulting angle. All angle measurements were rounded to the nearest degree.

To increase independence between successive measurements of the prints, each set of readings

was made at different times, weeks or months apart. Markings from previous readings were not saved

5 Brunson, et al.

by either reader to ensure that subsequent readings were completely independent. Prints were read in a

random order; and right and left prints from the same individual were not read at the same time.

When two a or two d triradii were encountered, the more radial and more ulnar triradius,

respectively, was used to determine the angle. When more than one t triradius was encountered in a

single print, only the more proximal triradius was used (t instead of t’ and t’ instead of t”) in accordance

with the methods proposed by David (1981a).

To determine if the two readers measured the same atd angle, we used a type of randomization

test. For each print, we randomly selected one of the two readings from Reader A and one of three

readings from Reader B. A two-tailed t-test (α=0.05) was used to test whether the mean of the

difference between pairs of readings (Reader A - Reader B) was significantly different from zero. We

repeated this procedure one million times, and counted the number of times the sum of the difference

was significantly different from zero. Intraclass correlation coefficients (ICC) were then estimated to

determine the degree of association within and between readers.

RESULTS

Triradii were not readable on all prints; therefore, atd angles could not be determined for all

200 prints. Reader A was able to read atd angles on 149 out of 200 prints (74.5%), and Reader B

was able to read the angles on 179 out of 200 prints (89.5%). Both readers agreed on whether an

angle existed on a print 89.8% of the time on the right hand and 78.0% on the left hand.

For angles that both readers were able to measure, Reader A’s readings tended to be slightly

higher (mean 45.6°) than Reader B’s (mean 44.8°), but Reader A’s readings were less variable (SD

6 Brunson, et al.

6.3°, compared to SD 7.4° for Reader B). These trends remained true even when measurements from

the right and left hands were examined separately (Table 1). The randomization test yielded no

significant differences (0 of 1,000,000 times) strongly suggesting that Reader A’s and Reader B’s

readings are interchangeable—that is, the slight upward bias in Reader A’s readings relative to Reader

B’s readings is not significant.

Reliability within readers, as assessed by the ICC analysis, showed that both readers were

consistent in their measurements. Reader A had an ICC of 0.97 and Reader B had an ICC of 0.99 for

all three pairwise comparisons (reading 1 vs. reading 2, reading 1 vs. reading 3, and reading 2 vs.

reading 3). Considering the reliability of atd angle measurements between readers, the ICC analysis

suggested that there is more variability between readers than within the same reader, but that the

correlations were still well above 0.9, ranging from 0.92 to 0.96 (Table 2).

DISCUSSION

The purpose of this study was to assess the reliability of measuring the atd angle. The high

ICCs obtained in our analysis suggest that atd angles can be measured reliably both within and between

readers. The results of the randomization test further suggest that two different readers, using different

measurement tools (a goniometer versus a digital angle measurement tool), can measure atd angles

without bias. These findings lend support for a trait that has been widely used in the literature on birth

defects, neurological disorders and diseases. A reader who is well-trained in recognizing the landmarks

and measuring the angle should provide results consistent with those of other readers equally well-

trained.

7 Brunson, et al.

In addition, our findings lend support for the method of using software programs, such as

Adobe Photoshop, to measure dermatoglyphic traits on digitized images. Using such a method, Reader

B was able to measure atd angles on 30 more prints (15% of the total sample) compared to Reader A,

who used a magnifying glass and goniometer to measure the angles on photocopies of the prints.

Additionally, while the intra-observer ICCs were high (>0.97) for both readers, correlations were

consistently higher for Reader B. Because both readers were trained using the same methods and

because they cross-checked their methods during training to ensure they were reading the prints with the

same methods and a comparable level of accuracy, the differences in the number of prints each reader

was able to read and the differences in the intra-observer ICCs between the readers likely reflect the

difference in the tools each reader used to measure atd angles on the prints.

It seems from our analysis that reading the prints in a digital format provided the best results.

Not only did using Photoshop allow for the prints to be magnified and the lines of the palms to be

enhanced when necessary, but the angle measurement tool in this program also reduced the error that

could occur from a reader misreading an angle on a goniometer. We recommend that in the future

researchers consider digitally scanning the prints and then using some type of software program to

digitally measure the atd angle and even other dermatoglyphic traits. We used an ordinary flat-bed

scanner to scan the images at 400 DPI using an eight-bit grayscale.

While we have shown that the atd angle can be measured reliably, the question still remains as

to whether or not the atd angle is useful in applied studies. We reviewed the current literature in which

atd angles have been used to investigate a variety of conditions, including autism (Arrieta et al., 1990;

Wolman et al., 1990; Milicic et al., 2003), schizophrenia (Balgir et al., 1993; Paez et al., 2001; Avila et

8 Brunson, et al.

al., 2003), cleft lip/palate (Woolf and Gianas, 1976; Balgir, 1993; Kohli and Bhalla, 2000; Neiswanger

et al., 2002) and genetic heart conditions (David, 1981(b); Pursnani et al., 1989). The results of these

studies were largely contradictory, with more than half finding no differences in the atd angles of

subjects with a given condition versus unaffected controls. When associations were found, the direction

of the relationship often differed between studies: for example Wolman and colleagues (1990) found no

difference between atd angles in children with and without autism; Arietta and colleagues (1990) found

atd angles were wider in autistic girls, but that there was no difference between normal controls and

autistic boys; while Milicic and colleagues (2003) found atd angles were wider among autistic boys, but

not autistic girls, when compared to normal controls.

While it is possible that measurement error resulting from multiple or untrained readers is

responsible for these contradictory results, we suggest that factors other than reader reliability should

also be considered as possible causes for the discrepancies. First, it is important to consider that atd

angles change with age (the angle is larger in children than adults) and with sex (the angle is larger in men

than in women; Cummins and Mildo, 1961). If these two variables are not taken into consideration,

biases may result if the age or sex-composition of the samples differs among studies. Mavalwala (1979)

has also shown that if a print is made with the fingers together, a different atd angle measurement will be

obtained than if the angle is measured from the same person with their fingers spread apart. Thus,

researchers must also be careful in obtaining consistent hand adduction during collecting the handprints.

Finally, David (1971, 1981a) has shown that great care must be taken in determining which angle to

measure (atd, at’d or at’’d). If researchers are not measuring the angle from the same t triradius, it is

impossible to determine whether the atd angle is smaller or larger because the t triradius is actually

9 Brunson, et al.

displaced or because the researchers were simply measuring the prints differently. Like David (1981a),

we suggest that researchers use the most proximal t triradius available, as this most likely represents the

“true” t triradius and not just an overabundance of hypothenar patterns. We strongly recommend that

researchers use the procedures of Mavalwala and David so that results among studies will be

comparable.

CONCLUSION

Based on the results of this research, we suggest that the atd angle can be measured reliably

whether the readings are made by individual or multiple readers and that the reliability of this

measurement may be improved by reading digitized prints with the help of a software program such as

Adobe Photoshop. Our review of the current literature showed that the atd angle provides inconsistent

relationships with traits like autism and schizophrenia. While several possibilities exist for these

discrepancies, based on the results of our study, we suggest that the usefulness of the atd angle in

dermatoglyphic studies may be increased if researchers agree on a set of methods for print collection

and measurement.

ACKNOWLEDGMENTS

We thank Carrie Kuehn for her work in collecting some of the prints used in this study. This

research was supported by a grants from NIH NIAAA (R21-AA013704 and R37-AA01455) and a

grant from the Alcohol and Drug Abuse Institute, University of Washington.

10 Brunson, et al.

LITERATURE CITED

Aase JM, Lyons RB. 1971. Technique for recording dermatoglyphics. Lancet 1:432-433.

Arrieta MI, Martinez B, Criado B, Simon A, Salazar L, Lostao CM. 1990. Dermatoglyphic analysis of

autistic Basque children. American Journal of Medical Genetics 35:1-9.

Avila MT, Sherr J, Valentine LE, Blaxton TA, Thaker GK. 2003. Neurodevelopmental interactions

conferring risk for schizophrenia: A study of dermatoglyphic markers in patients and relatives.

Schizophrenia Bulletin 29:595-605.

Babler WJ. 1991. Embryologic development of epidermal ridges and their configurations. In Plato CC,

Garruto RM, Schaumann BA, Paul MW, editors. Dermatoglyphics: Science in Transition.

New York: Wiley-Liss. p. 95-112.

Balgir RS. 1993. Dermatoglyphics in cleft lip and cleft palate anomalies. Indian Pediatrics 30:341-

346.

Balgir RS, Murthy RS, Wig NN. 1993. Genetic loadings in schizophrenia: a dermatoglyphic study.

Israel Journal of Medical Sciences 29:265-268.

Chakraborty R. 1991. The role of heredity and environment on dermatoglyphic traits. In Plato CC,

Garruto RM, Schaumann BA, Paul MW, editors. Dermatoglyphics: Science in Transition.

New York: Wiley-Liss. p. 151-191.

11 Brunson, et al.

Cummins H, Mildo C. 1961. Finger Prints, Palms and Soles: An Introduction to Dermatoglyphics.

New York: Dover Publications Inc.

David TJ. 1971. The palmar axial triradius t: A new method of location. Human Heredity 21:624-627.

David TJ. 1981(a). The corrected atd angle. Human Heredity 31:283-285.

David TJ. 1981(b). Dermatoglyphs in congenital heart disease. Journal of Medical Ethics 18:344-

349.

Elbualy MS, Schindeler JD. 1971. Handbook of Clinical Dermatoglyphics. Coral Gables, FL:

University of Miami Press.

Kohli S, Bhalla V. 2000. Bilateral asymmetry of hand dermatoglyphics in deaf-mute and hare-lip and

cleft palate patients. Man in India 80: 185-193.

Mavalwala J. 1979. The atd angle: A critique. Birth Defects: Original Article Series 15:169-172.

Meier RJ. 1991. Applications of dermatoglyphics to anthropologic populations. In Plato CC, Garruto

RM, Schaumann BA, Paul MW, editors. Dermatoglyphics: Science in Transition. New

York: Wiley-Liss. p. 253-265.

Milicic J, Petkovic ZB, Bozikov J. Dermatoglyphics of digito-palmar complex in autistic disorder: family

analysis. Croatian Medical Journal 44:469-476.

12 Brunson, et al.

Neiswanger K, Cooper ME, Weinberg SM, Flodman P, Keglovits AB, Liu Y, Hu DN, Melnick M,

Spence MA, Marazita ML. 2002. Cleft lip with or without cleft palate and dermatoglyphic

asymmetry: evaluation of a Chinese population. Orthodontics and Craniofacial Research

5:140-146.

Paez F, Apiquian R, Fresan A, Puig A, Orozco B, De la Fuente JR, Sidenberg D, Nicolini H. 2001.

Dermatoglyphic study of positive and negative symptoms in schizophrenia. Salud Mental

24:28-32.

Priest JH, Brantley KE, Tishler PV. 1979. Disagreements in dermatoglyphic pattern assignment and

their effect on quantitative dermatoglyphic analysis. Birth Defects: Original Article Series

15:173-176.

Pursnani ML, Elhence GP, Tibrewala L. 1989. Palmar, dermatoglyphics in essential hypertension.

Indian Heart Journal 41:119-122.

Sharma A. 1964. Morphilogically-evidenced extent of fluctuation of BHI values at t, t’ and t” for the

corresponding different single angular values of angle atd. Eastern Anthropologist 17:23-30.

Wolman SR, Campbell M, Marchi ML, Deutsch SI, Gershon TD. 1990. Dermatoglyphic study in

autistic children and controls. Journal of the American Academy of Child and Adolescent

Psychiatry 29:878-884.

13 Brunson, et al.

Woolf CM, Gianas AD. 1976. Congenital cleft lip and fluctuating dermatoglyphic asymmetry.

American Journal of Human Genetics 28:400-403.

14 Brunson, et al.

Table 1. Summary statistics for atd angle measurements by reader

Reader* A1 A2 B1 B2 B3 right hand data

mean 45.6 46.8 45.4 45.7 45.6 SD 6.7 6.8 8.0 8.0 7.9 CV 14.7 14.5 17.6 17.5 17.3 minimum 36 37 33 33 33 maximum 73 74 80 80 78 number 77 77 77 77 77

left hand data

mean 45.2 46.1 44.4 44.7 44.5 SD 5.8 5.8 6.0 6.1 6.1 CV 12.8 12.6 13.5 13.6 13.7 minimum 35 36 35 35 36 maximum 65 66 65 65 65 number 71 71 71 71 71 * A1= Reader A’s first reading, B1= Reader B’s first reading

15 Brunson, et al.

Table 2. Reliability coefficients for atd angle measurements.

intra-observer reliability

Reader* A1 / A2 B1 / B2 B1/B3 B2/B3 right hand 0.97 0.99 0.99 0.99 left hand 0.97 0.99 0.99 0.99 inter-observer reliability

A1 / B1 A1 / B2 A1/B3 A2 / B1 A2 / B2 A2/B3 right hand 0.96 0.96 0.96 0.94 0.95 0.95 left hand 0.96 0.96 0.96 0.92 0.92 0.92 * A1= Reader A’s first reading, B1= Reader B’s first reading

16 Brunson, et al.

FIGURE CAPTIONS

Figure 1. Landmarks used to measure atd angle.

17 Brunson, et al.