Home About Us Laboratory Services Forensic Science Communications Back Issues October 2001 Effect of Color and Curvature on the Concentration of...
This is archived material from the Federal Bureau of Investigation (FBI) website. It may contain outdated information and links may no longer function.

Effect of Color and Curvature on the Concentration of Morphine in Hair Analysis by Mieczkowski (Forensic Science Communications, October 2001)

Effect of Color and Curvature on the Concentration of Morphine in Hair Analysis by Mieczkowski (Forensic Science Communications, October 2001)

October 2001 - Volume 4 - Number 4

Research and Technology

Effect of Color and Curvature on the
Concentration of Morphine in Hair Analysis

Thomas M. Mieczkowski
Department of Criminology
University of South Florida
St. Petersburg, Florida

Introduction | Sample Frame | Purpose of the Study | Delimitation
Data Analysis Methods | Hypotheses |
Data Analysis: Morphine Concentration and Hair Color |
Discussion: Morphine Concentration and Hair Color |
Data Analysis: Morphine Concentration and Hair Curvature |
Descriptive Data | Discussion: Morphine Concentration and Hair Curvature
Discriminant Analysis
| Summary and Conclusion | References


Hair analysis for psychoactive drugs has become a progressively more common method for ascertaining toxicological evidence of drug use in forensic cases. Recently the Federal Food and Drug Administration gave approval for the use of hair analysis for determining heroin use. Currently a large number of commercial laboratories offer hair analysis services for psychoactive drugs, and a large number of corporations, some government agencies, and a number of criminal justice agencies use hair testing to identify potential drug abuse.

There has been concern that several factors associated with human hair may cause variation in the retention and recovery of specific analytes when using hair as a test matrix for toxicological analysis (Kidwell 1992; Skopp et al. 1997). Among the most important variables are different coloration and coloration patterns associated with hair and, to a lesser extent, the degree of coiling, curvature, or curling variations associated with human hair (Joseph et al. 1996; Knight et al. 1996). Coiling or curvature of hair has not been specifically discussed, but it is often used as the primary surrogate measure for race or ethnicity identification that has been noted as a potential confounder of hair analysis interpretation (Joseph et al. 1996). Researchers who have suggested an ethnic or race bias in hair analysis have de facto used kinkiness of hair as a race marker (Henderson et al. 1998). There has been an ongoing discussion in the literature about the importance of these factors. A number of published articles suggest that this putative systematic variation may not be important in hair analysis interpretation (Hoffman 1999; Kelly et al. 2000; Mieczkowski and Newel 1993, 2000b).

Two opiates widely used in medical practice as well as by abusers are codeine and morphine. Some studies have concluded that the incorporation of these opiates into hair is influenced by the melanin content of the hair. Research looking at opiates in human and animal populations, for example, has suggested that melanin binding accounts for higher concentrations of codeine and morphine in darkly pigmented hair (Rollins et al.1996; Wilkins et al. 2000). Kronstrand et al. (1999) published data on codeine recovery from hair as a function of hair melanin concentration and reported a strong correlation between codeine concentration and melanin concentration. However, there have been some interesting effects and variations noted in these reports. The strength of the contribution may be dependent on the specific opiate under scrutiny.

Pigmentation appears to play an important role in codeine retention in hair, for example, but a diminished role in morphine retention. Rollins (1995) observed, for instance, that given constant concentrations in the plasma of codeine and morphine, morphine appears to incorporate at approximately one half the rate as codeine. He suggested that a variety of factors including the variations in molecular properties of morphine in comparison to those of codeine might be responsible for this difference (Rollins 1995). Rothe et al. (1997) reported, consistent with Rollins, that morphine and 6-acetylmorphine were near-equivalence in concentration between pigmented and nonpigmented hair when comparing hair collected from the same individuals taking controlled doses of morphine. Wilkins et al. (2000) reported findings indicating the difficulty of positing pigmentation as a simple predictor of codeine concentration. Variations in codeine concentrations recovered from uniformly black hair (samples taken from several racial and ethnic groups) showed dramatic variations in concentration even though subjects were given equivalent controlled doses of codeine. The variations in concentration approached 300%, indicating the likelihood that factors other than pigmentation of hair play a major role in influencing concentration values. In their analysis, Mieczkowski and Newel (2000b) evaluated the outcome of a series of studies assessing hair assay values for opiates and hair color (Goldberger et al. 1991; Kintz et al. 1998). They reported that there was no statistically significant relationship between either parent drug or metabolites and hair color.

Sample Frame

To further assess the possible influence of pigmentation on the concentration of morphine in hair, data was presented on the basis of a random selection of 95 morphine-positive cases retrieved from a database produced by a major hair analysis laboratory (Psychemedics Corporation, Culver City, California). These 95 cases were selected from approximately 6,000 cases that had a positive assay for some drug. These 6,000 cases were selected from a sampling frame of 80,000 cases submitted to the laboratory for analysis of the presence of five illicit psychoactive drugs (cocaine, cannabinoids, opiates, amphetamines, and PCP). The samples were sent as part of a preemployment or employment drug screening. The only inclusion criteria used in constituting the sample frame were that cases included a confirmed positive hair assay for morphine with a concentration value of 0.2 ng/mg of sample mass. It is important to emphasize that it is very rare for persons in preemployment and employment categories to test and confirm positive for morphine. Furthermore, when stratifying the data further by hair color, the number of cases available for analysis can become extremely small. This is further exacerbated by the confounding effect that hair coloration is distributed unevenly, with natural red and natural light blond hair being relatively scarce.

The testing methods used by the Psychemedics Corporation are based on the protocols developed by Baumgartner (Baumgartner 1995; Baumgartner et al. 1995; Baumgartner and Hill 1996). This protocol includes a complex wash-and-wash analysis, enzymatic digestion of the washed hair, radioimmunoassay screening of the digest, and GC/MS or GC/MS/MS confirmation of any radioimmunoassay-positive result. For details of the chemical analysis protocol, please consult the cited publications.

Purpose of the Study

The study consists of the presentation and analysis of data for the 95 selected cases. The analysis is directed at two purposes. The first is to determine whether there is a statistically significant relationship between the hair color categorization and the morphine concentration reported for each sample. The second is to determine whether there is a relationship between the hair curvature or curliness and the morphine concentration reported for each sample. Technical personnel of the testing laboratory, using internal categorization standards, performed the assessments and assigned the characterizations of the hair by color and curvature. Curvature was assessed by visual examination of the sample laid on a glass surface. Hair was classified as curved if it exceeded values of 02 on the Bailey and Schleibe (1985) curvature measurement scale, with most hair so classified being above 05 on the scale. An internally developed color schema derived from hair cosmetic industry color comparison charts was used to assess sample color on a categorical basis. Color classification was a five-fold schema: black, brown, blond, red, and gray/gray mixed.


This study has several notable delimitations. First, the number of samples is relatively small, especially for light-colored hair. Red hair is entirely absent. Although samples were taken from a very large initial database, the rate of opiate-positive samples is generally very low in preemployment populations. The representation here is a reflection of the real distribution of these cases in the general employment-seeking population. This becomes further exacerbated by the subdivision of these cases into hair color categories. Samples that could be determined to be dyed, bleached, or tinted or colored cosmetically were excluded. The resultant number of persons with blond and red hair that occurs naturally is low in the general population. The appearance of these colors to the casual observer in everyday experience gives the impression that these color categories should be more frequent, but this is because a substantial number of individuals modify their natural hair color through cosmetic processes. Consequently, the data offered do not permit any statistical inference because of the convenience sampling method.

Additionally, the study is delimited by not knowing the actual dosages consumed, and it cannot be excluded from possibility that the results here are a measurement of dose preference differences, which correspond to hair color. However, that presumption is no more or less defensible than the assumption that in a large preemployment population, the dosage ranges throughout many cases is likely to resemble a Gaussian distribution, which is the assumption made here. Ideally, of course, one would like to conduct a controlled-dose experiment. This, however, presents some problems as well. Perhaps most notably, the ability to conduct such an experiment on a very large scale and to conduct it throughout a wide-dose range.

The data and analysis offered here are designed to suggest what may be possible and reasonable in considering the hypotheses.

Data Analysis Methods

The independent variables in this analysis are the color category of the hair sample and the curly/straight dichotomous category for the hair sample. The dependent variable is the concentration value of morphine in the hair sample. The assessment of the case data is done by univariate ANOVA for color categorizations and by independent sample t-test for hair curvature. Each independent–dependent variable relationship is also assessed by discriminant analysis. Multiple statistical methods are used as indicated to provide a more comprehensive assessment of the possible relationship. The criteria used to determine statistical significance is p = .05 for all methods. The analysis of this data set was done with SPSS v 10.0. The particulars of the data file follow in Table 1.

Table 1. Structure of the Data File



Value labels
ID: Case identifier Nominal None
HRCOLOR: Hair color Nominal 1 = Black
2 = Brown
3 = Dark brown
4 = Blond
5 = Gray
Straight/curly hair
Nominal 1 = Straight
2 = Curly
Morphine concentration
(ng/10 mg)

The hypotheses for the analysis are specified as follows:

Hypothesis 1
The null hypothesis: There is no significant relationship between the concentration of morphine recovered from each hair sample and the color categorization. The alternative hypothesis: Hair samples that are darker in color will have a higher concentration value than those that are lighter in color.

Hypothesis 2
The null hypothesis: There is no significant relationship between the concentration of morphine recovered from each hair sample and the curvature categorization. The alternative hypothesis: Hair samples that have discernible physical curvature will have statistically significant differences in morphine concentration when compared to those that lack curvature (straight).

Data Analysis: Morphine Concentration and Hair Color

Table 2 reports the mean value (M) and standard deviation (SD) for the recovered morphine for each hair color category.

Table 2. Mean Concentration of Morphine by Hair Color

Hair color

Mean morphine concentration
(ng/10 mg)

Standard deviation

Number of samples
(N = 95)
Black 10.69 8.32 42
Brown 15.02 10.17 13
Dark brown 13.40 9.10 32
Blond 7.50 3.80 5
Gray 10.10 5.31 3
Overall mean 12.01 8.71  

Figure 1. Box-and-whisker (chart) plot shows morphine concentration in hair according to hair color.
Figure 1. Box-and-whisker plot shows morphine concentration in hair according to hair color. Click for enlarged image.

Figure 1 presents the box-and-whisker plot of morphine concentration by hair color. The horizontal bars in the box represent the group mean, the box boundaries represents the 25th and 75th percentile values, and the whiskers delineate the cases within 1.5 “hspreads” (box lengths) of the mean.

Table 3 contains the outcome of a one-way (univariate) analysis of variance (ANOVA) for all color combinations contrasting the groups by morphine concentration. ANOVA seeks to compare the degree of within-category variance to between-group variance to assess the possible significance of difference by the independent variable category set. Table 3 indicates that there is no significant effect for hair color for morphine concentration.

Table 3. Analysis of Variance, Hair Color (IV) and Morphine Concentration (DV), All Hair Colors

  df SS MS F
Morphine in hair by hair color Between groups Combined 4 364.479 91.120 1.213***
Linearity 1 11.657 11.657 .155***
Deviation from linearity 3 352.822 117.607 1.566***
Within groups   90 6761.154 75.124  
Total   94 7125.633    

*p = .311. **p = .695. ***p = .203.

Further analysis by Tukey’s honestly significant difference (HSD) method (also known as the Tukey a procedure), which permits complete pair-by-pair comparisons for all possible color combinations and their contrasts, does not show significance for any color combination.

Discussion: Morphine Concentration and Hair Color

Statistical analysis by univariate ANOVA of these 95 cases supports the following observations regarding color and morphine concentrations.

Rank order of mean concentrations of morphine by hair color from high to low is as follows:

1. Brown

2. Dark Brown

3. Black

4. Gray

5. Blond

Although there appears to be a concentration gradient, which decreases from darker to lighter hair, the one-way ANOVA is not significant for morphine concentration and hair color category, F(90, 4) = 1.213, p = 0.311. The R2 value for this relationship is 0.002 and the h2 value is 0.051. If the nominal values for hair color are transformed into a dichotomous contrast (black, dark brown in contrast to blond/gray), there is not a significant difference in morphine concentration, F(93, 1) = 1.446, p = 0.232, R2 = .015. These values indicate that color contributes negligibly to variations in concentration values and is not a function of a dilution effect by the number of dark categories.

For the color contrasts, the Levene Test for homogeneity of variance is not significant, F(90, 4) = 1.941, p = 0.110, indicating that the groups have homogeneous variances. Subsequent examination by Tukey’s HSD analysis shows no significant morphine concentration effect for any color combination, nor are there any homogeneous concentration subsets by color.

Data Analysis: Morphine Concentration and Hair Curvature

The same analytic approach can be applied to assessing the potential that gross hair morphology as straight or curled hair may have a relationship to morphine concentration in hair. The only required modification is that the curly/straight variable is dichotomous, and thus instead of a one-way ANOVA (which is used for three or more group comparisons), a t-test to assess the significance of the hypothesized relationship can be applied.

Table 4. Mean Concentration Values of
Morphine for Straight and Curled Hair

Curvature morphology Mean morphine concentration
(ng/10 mg)
Number of samples
(N = 95)
Straight 12.26 8.79 76
Curled 11.02 8.54 19
Total 12.01 8.71  

Descriptive Data

Figure 2. Box-and-whisker plot shows a comparison of curly and straight hair according to morphine concentration in hair.
Figure 2. Box-and-whisker plot shows a comparison of curly and straight hair according to morphine concentration in hair. Click for enlarged image.

Of the 95 cases under analysis, 76 are classified as straight, and 19 are classified as curled. All cases have a confirmed positive hair assay for morphine. Table 4 presents the mean values and associated standard deviations for the two categories.

Figure 2 is a box-and-whisker plot of this same contrast.

Table 5 presents the results of a t-test contrasting the straight and curled hair samples by morphine concentration. The Levene Test indicates that the criterion of equality of group variance is met and that the t value does not attain significance. Assessment of this relationship by a general linear model indicates that the R2 value is 0.003.

Table 5. Results of t-test for Mean Difference in Morphine Concentration by Hair Curvature

t-test for equality
of means
t df Mean difference
Morphine in hair (ng/10 mg) .554 93 1.2408

p = .581, two-tailed.

Discussion: Morphine Concentration and Hair Curvature

Statistical analysis by t-test and univariate ANOVA of these 95 cases supports the following observations regarding hair curvature and morphine concentration. Neither one-way ANOVA nor an independent sample t-test is significant for morphine concentration and hair curvature, t = 0.554, df = 93; F(93, 1) = .306. The R2 value for this relationship is 0.003. These values indicate that hair curvature or curliness does not contribute in a meaningful way to variations in morphine concentration values.

Discriminant Analysis

Last, a discriminant analysis on this data was performed. Discriminant analysis is a technique using aspects of multivariate ANOVA and regression. It permits the analysis of a quantitative predictor variable (or a series of quantitative predictor variables) and the ability of these variables to successfully classify outcomes or states. In this case the predictor variable is morphine concentration. The approach of discriminant analysis is such that it tests the following concept: If the concentration of morphine for any particular sample is known, that sample is assigned to a predicted group membership. The predicted memberships are compared to the empirically known group membership, and the utility of the relationship between predictor and category is assessed on the basis of agreement between predictive outcome and known outcome.

Table 6. Classification Resultsa of Discriminant Analysis of Morphine Concentration and Hair Color

Hair color Predicted group membership
Black Brown Dark
Blond Gray
n % n % n % n % n %
(N = 42)
1 2.4 11 26.2 3 7.1 23 54.8 4 9.5
(N = 13)
1 7.7 6 46.2 0 .0 5 38.5 1 7.7
Dark brown
(N = 32)
2 6.3 14 43.8 0 .0 11 34.4 5 15.6
(N = 5)
0 .0 0 .0 1 20.0 3 60.0 1 20.0
(N = 3)
1 33.3 1 33.3 0 .0 1 33.3 0 .0

a10.5% of original grouped cases correctly classified.

A discriminant analysis was performed for hair color categorizations and for curvature characterizations relying on morphine concentration as the predictor variable. Analysis assumptions used were the following: Equiprobability was assigned for all groups for prior probabilities for the analysis, and the covariance matrix used was a within-groups design. The outcomes are presented here as classification tables, which indicate the degree of success of the classification on the basis of a predicted group membership versus actual group membership. Tables 6 and 7 show by count and percentage the coincidence of predicted and original group membership, using morphine concentration as the predictor variable.

Table 7. Classification Resultsa for Discriminant Analysis of Morphine Concentration and Hair Curvature

Hair curvature Predicted group membership
Straight Curly
n % n %
(N = 76)
30 39.5 46 60.5
(N = 19)
7 36.8 12 63.2

a44.2% of original group cases correctly classified.

In cases of color-versus-curvature categorization, the selected variable was less effective as a predictor relative to random chance assignments. For hair color prediction, the rate of correct classification was approximately 10%, about half of the value of correct predictions that would result from random chance. For curvature, the value is approximately 44%, slightly less than the 50% value that would be true of random assignment. These outcomes are consistent with the t-test and ANOVA outcomes of the previous analysis and reinforce the findings that neither hair color or hair curvature appears to be significantly related to morphine concentration.

Summary and Conclusion

Analysis of the samples does not reveal a statistically significant effect at p = .05 between hair color or hair curvature and the concentration of morphine recovered by chemical analysis. For multigroup comparisons (color) or dichotomous comparisons (curvature) there is a very low value for R2, indicating that for this particular drug these variables do not have significant explanatory power associated with variations in the value of the drug’s concentration. A second assessment using discriminant analysis is consistent with the outcomes for the first set of ANOVA and t-test results. Classification results show that for color and curvature the assignment performance is degraded over random chance when using either independent variable as a predictor. The hypotheses identified are thus not supported by these findings. These results may or may not hold true when evaluating other important opiates and their concentration in hair.

The data here suggest that the relationship between color, curvature, and morphine concentration may not be a meaningful effect. But the sample size and the assumption of approximate dose equivalencies, as noted, seriously delimit the study. A controlled-dose, large-scale clinical experiment would be the most fruitful approach to definitively evaluating the presence and degree of these possible effects and better assessing the hypotheses specified in this study. The use of controlled-dose experiments is an unlikely approach because of human subjects and ethical constraints. Compiling and analyzing a larger database, which would result in a larger representation in the scarcer color categories, is a more pragmatic approach. Furthermore, the dosage equivalency assumption becomes more plausible with an increase in analytic N. This effort is currently underway and may produce a clearer picture of the results of the study presented here.


Bailey, J. and Schleibe, S. The precision of the average curvature measurement in human head hair. In: Proceedings of the International Symposium on Forensic Hair Comparisons. Federal Bureau of Investigation, U.S. Government Printing Office, Washington, DC, 1985.

Baumgartner, W. A. Hair analysis method. U.S. Patent 5,466,579 (1995).

Baumgartner, W. A., Cheng, C., Donahue, T., Hayes, G., Hill, V., and Scholtz, H. Forensic drug testing by mass spectrometric analysis of hair. In: Forensic Application of Mass Spectrometry. I. Yinon, ed. CRC Press, Boca Raton, Florida, 1995, pp. 61–94.

Baumgartner, W. A. and Hill, V. Hair analysis for organic analytes: Methodology, reliability issues and field studies. In: Drug Testing in Hair. P. Kintz, ed. CRC Press, Boca Raton, Florida, 1996, pp. 223–265.

Goldberger, B., Caplan, Y., Maguire, T., and Cone, E. Testing human hair for drugs of abuse: Identification of heroin and 6-acetylmorphine as indicators of heroin use, Journal of Analytical Toxicology (1991) 155:226–232.

Henderson, G., Harkey, M., Zhou, C., Jones, R., and Jacob, P. Incorporation of isotopically labeled cocaine into human hair: Race as a factor, Journal of Analytical Toxicology (1998) 22:156–164.

Hoffman, B. H. Analysis of race effects on drug test results, Journal of Occupational and Environmental Medicine (1999) 41:612–614.

Joseph, R., Tsung-Ping, S., and Cone, E. In vitro binding studies of drugs to hair: Influence of melanin and lipids on cocaine binding to Caucasoid and Africoid hair, Journal of Analytical Toxicology (1996) 20:338–344.

Kelly, R. C., Mieczkowski, T., Sweeney, S., and Bourland, J. Hair analysis for drugs of abuse: Hair color and race differentials or systematic differences in drug preferences, Forensic Science International (2000) 107:63–86.

Kidwell, D. A. Hair analysis: Techniques and potential problems. In: Recent Developments in Therapeutic Monitoring and Clinical Toxicology. I. Sunshine, ed. Marcel Dekker, New York, 1992, pp. 555–563.

Kintz, P., Bundeli, P., Brenneisen, R., and Ludes, B. Dose-concentration relationships in hair from subjects in a controlled heroin-maintenance program, Journal of Analytical Toxicology (1998) 22:231–235.

Knight, J. M., Eliopoulos, C., Klein, J., Greenwald, M., and Koren, G. Passive smoking in children: Racial differences in systematic exposure to cocaine by hair and urine analysis, Chest (1996) 109:446–450.

Kronstrand, R., Forstberg-Peterson, S., Kagedal, B., Ahlner, J., and Larson, G. Codeine concentration in hair after oral administration is dependent on melanin content, Clinical Chemistry (1999) 45:1485–1494.

Mieczkowski, T. and Newel, R. An analysis of the racial bias controversy in the use of hair assays. In: Drug Testing Technology. T. Mieczkowski, ed. CRC Press, Boca Raton, Florida, 2000a, pp. 313–348.

Mieczkowski, T. and Newel, R. An evaluation of patterns of racial bias in hair assays for cocaine: Black and white arrestees compared, Forensic Science International (1993) 63:85–98.

Mieczkowski, T. and Newel, R. A statistical examination of hair color as a potential biasing factor in hair analysis, Forensic Science International (2000b) 107:13–38.

Rollins, D. E. Disposition of codeine into hair. In: Proceedings of the International Conference and Workshop for Hair Analysis in Forensic Toxicology. R. deZeeuw, I. Hosani, and S. Al Munthiri, eds. Ministry of the Interior, Abu Dhabi, United Arab Emirates, 1995, pp. 118–135.

Rollins, D. E., Wilkins, D. G., and Krueger, G. G. Codeine disposition in human hair after single and multiple doses, European Journal of Clinical Pharmacology (1996) 50:391–397.

Rothe, R., Pragst, F., Hunger, J., and Thor, S. Effect of pigmentation on the drug deposition in hair of grey-haired subjects, Forensic Science International (1997) 84:53–60.

Skopp, G., Potsch, L., and Moellet, M. On cosmetically treated hair: Aspects and pitfalls of interpretation, Forensic Science International (1997) 84:43–52.

Wilkins, D., Rollins, D., Slawson, M., and Augsburger, M. Pharmacokinetic Studies of the Disposition of Drugs of Abuse Into Human Hair. Presented at the California Association of Toxicologists, Napa, California, 2000.