Central Hair Loss in African American Women: Incidence and Potential Risk Factors

ORIGINAL ARTICLES

Central hair loss in African American women: Incidence and potential risk factors

EliseA.Olsen,MD,a Valerie Callender, MD,b Amy McMichael, MD,c Leonard Sperling, MD,d KevinJ.Anstrom,PhD,a Jerry Shapiro, MD,e Janet Roberts, MD,f Faith Durden, MD,g David Whiting, MD,h and Wilma Bergfeld, MDi Durham and Winston-Salem, North Carolina; Mitchellville and Bethesda, Maryland; Vancouver, British Columbia, Canada; Portland, Oregon; Dallas, Texas; and Cleveland, Ohio

Background: Although central scalp hair loss is a common problem in African American women, data on etiology or incidence are limited.

Objective: We sought to determine the frequency of various patterns and degree of central scalp hair loss in African American women and to correlate this with information on hair care practices, family history of hair loss, and medical history.

Methods: Five hundred twenty-nine subjects at six different workshops held at four different sites in the central and/or southeast United States participated in this study. The subjects’ patterns and degree of central scalp hair loss were independently assessed by both subject and investigator using a standardized photographic scale. Subjects also completed a detailed questionnaire and had standardized photographs taken. Statistical analysis was performed evaluating answers to the questionnaire relative to pattern of central hair loss.

Results: Extensive central scalp hair loss was seen in 5.6% of subjects. There was no obvious association of extensive hair loss with relaxer or hot comb use, history of seborrheic dermatitis or reaction to a hair care product, bacterial infection, or male pattern hair loss in fathers of subjects; however, there was an association with a history of tinea capitis.

Limitations: There was no scalp biopsy correlation with clinical pattern of hair loss and further information on specifics of hair care practices is needed.

Conclusions: This central scalp photographic scale and questionnaire provide a valid template by which to further explore potential etiologic factors and relationships to central scalp hair loss in African American women. ( J Am Acad Dermatol 2011;64:245-52.)

Key words: central centrifugal cicatricial alopecia; central scalp hair loss; hair loss in African American women.

high proportion of African American women seeking medical advice on their hair loss Abbreviations used: have central scarring hair loss, a condition CCCA: central centrifugal cicatricial alopecia A FPHL: female pattern hair loss that the North American Hair Research Society has MPHL: male pattern hair loss termed ‘‘central centrifugal cicatricial alopecia’’ (CCCA).1 Hair care products have long been impli- other than the temporal relationship with patchy hair cated as causes of hair loss in this population, but breakage, few data have been generated to prove

From the Duke University Medical Center, Durhama; Howard Conflicts of interest: None declared. University, Mitchellvilleb; Wake Forest University, Winston- Accepted for publication November 29, 2009. Salemc; Uniformed Services University of the Health Sciences, Reprints not available from the authors. Bethesdad; University of British Columbia, Vancouvere; North- Correspondence to: Elise A. Olsen, MD, Box 3294 Duke University west Dermatology and Research Center, Portlandf; Western Medical Center, Durham, NC 27710. E-mail: [email protected]. Reserve Dermatology, Clevelandg; Baylor Hair Research and Published online November 15, 2010. Treatment Center, Dallas,h and Cleveland Clinic, Cleveland.i 0190-9622/$36.00 Supported by the North American Hair Research Society and an ª 2010 by the American Academy of Dermatology, Inc. unrestricted educational grant from Procter & Gamble. doi:10.1016/j.jaad.2009.11.693

245 246 Olsen et al JAM ACAD DERMATOL FEBRUARY 2011 this. There are no published data on the incidence/- photographs of the subject’s scalp hair were taken prevalence of either scarring or nonscarring central and both subject and investigator rated this central hair loss in African American women or on poten- scalp hair loss using the aforementioned central tially related medical problems or hereditary factors scalp photographic scale. that may be causative or contributory factors. The first three workshops were held in the south- Moreover, until recently, there have been no stan- eastern United States and the last three workshops in dardized methods to measure the severity of the Cleveland, Ohio. The first three workshops were common central hair loss. designed to address the fre- Nine clinical investigators quency of the various de- with experience in hair dis- CAPSULE SUMMARY grees of central hair loss in orders from various parts of 233 African American d A validated photographic scale of central North America were identi- women in the general popu- hair loss in African American women has fied and first met at Duke lation and to correlate the proven useful to evaluate the frequency University Medical Center in degree of hair loss with an- of the various patterns of hair loss in the Durham, NC, in 2005. From swers to the questionnaire. general population and to assess the this meeting, a 6-point pho- To accomplish this, the work- relatedness of various hair care practices tographic scale using stan- shops were held incidental and associated medical conditions. dardized photographs of to meetings of African the scalp hair of African d Severe central hair loss occurs in 5.6% of American women without American women was re- African American women. previous information on the

fined with 0 = no hair loss d hair-related workshops. The 2 Chemical straightening is used by the and 5 = most severe hair loss majority of African American women first workshop was held in (Fig 1). Two subtypes of cen- without development of extensive Mitchellville, Maryland, as tral hair loss were identified central hair loss. part of an annual Health and and designated A (frontal ac- Beauty Symposium for centuation) and B (vertex ac- African American women. centuation). A questionnaire that addresses hair care Of the 825 women who attended the meeting, 150 practices and medical/hereditary factors was also participated in the hair workshop and were evalu- derived and later revised. ated independently by at least two of the four In this article, we report on the results of a investigators (V. C., A. M., E.A.O., and L.S.) in atten- multicenter trial in African American women de- dance. Two other smaller workshops held at general signed to evaluate the incidence and patterns of meetings of women members of two predominantly central hair loss using the aforementioned central African American churches were held in Wake-Forest hair loss photographic scale and the relationship of and Durham, North Carolina. At these latter two hair care practices, family history of hair loss, and meetings, more than 95% of all women (93 total) underlying medical conditions to the degree (if any) attending the meetings participated in the hair-re- of central scalp hair loss. lated workshops, and two investigators (A.M. and E.A.O.) independently evaluated each subject. METHODS Workshops four, five, and six were not part of Consent of subjects other meetings of African American women but were Institutional review board approval for the origi- stand-alone hair-related workshops with previous nal and all revised versions of the protocol and advertising in the African American community for consent form was obtained at the central site (Duke the workshops. One investigator (W.B.), in conjunc- University) and at the other 3 academic sites involved tion with one of her Cleveland Clinic colleagues, in the study. All coinvestigators were approved to evaluated all subjects for their degree of hair loss. partake in the study by one of the 4 institutional The information from the 296 women in attendance review boards. All subjects consented before enroll- at these last three meetings was added to that from ing in the study and before their data were collected. the 233 women who participated in workshops one, two, and three and used in the analysis of factors Overview of study design potentially related to the hair loss but was not used There were 6 study workshops involving a total of for determining the frequency of hair loss in the 529 African American women. At each workshop, general African American female population nor for participants filled out a detailed questionnaire on assessing the interrater variance in determination of their past and current hair care practices, medical the pattern of hair loss. In addition, the questionnaire history, and family history of hair loss. Standardized used in workshop six varied slightly from that used in JAM ACAD DERMATOL Olsen et al 247 VOLUME 64, NUMBER 2 the other workshops, so that answers to all questions in the questionnaire could not be consolidated from all six workshops.

Data analysis Subjects. The characteristics of subjects from each workshop were analyzed individually and collectively. The results of the meetings are summarized where appropriate. Photographic scale. In our analysis, the severity of the central hair loss patterns as shown in the photographic scale was considered as follows: no hair loss = pattern 0; early hair loss = patterns 1 and 2; and probable CCCA = patterns 3 to 5. Although it is likely that patterns 3 to 5 on the photographic scale may be surrogate markers for CCCA, this study did not include histologic conﬁrmation of CCCA. A clinicohistologic correlation study is underway to address this issue. Investigator versus subject assessments. Investigator assessments of hair loss were analyzed both individually and collectively compared with the subject’s assessment of hair loss. The collective investigator assessment of the subject’s central hair loss (mode score where three or more investigators rated the subjects, lowest score where there was disagreement between two investigators) was used as the determinant of the degree of central hair loss and was correlated with the subject’s answers to the questionnaire. Statistical analysis. To simplify the results and to account for the fact that subjects cannot readily characterize hair loss in their vertex scalp (B sub- type) without the concomitant use of mirrors, subtypes A and B on the central photographic scale were merged into one category for analysis purposes. Categorical variables were summarized by percent- ages. Comparisons of categorical variables were conducted using x2 tests. Correlations between ordinal factors were computed using Spearman Rho. Comparisons of assessments by the different clinical investigators were conducted using Kappa statistics which correct for agreement caused by chance and offer a measure of agreement between two individuals: poor agreement: \0.20; fair agreement: 0.21-0.40; moderate agreement: 0.41-0.60; Fig 1. Central scalp alopecia photographic scale in Afri- substantial agreement: 0.61-0.80; almost perfect 2 agreement: 0.81-1.00. Two-sided P values were can American women. Reprinted from Olsen et al with permission from Wiley-Blackwell. considered statistically significant.

RESULTS are given in Table I. The mean age of the 529 subjects Five hundred twenty-nine African American was 48.3 years (range, 18-85 years) with 38.5% women participated in the study. The results for of subjects over 55 years of age and likely postmen- age and central hair loss pattern as determined by the opausal at the time of the survey. There were investigators at each meeting/site and collectively significant differences in the age of patients across 248 Olsen et al JAM ACAD DERMATOL FEBRUARY 2011

Table I. Characteristics of study subjects evaluated in each of the meeting workshops

Mitchellville, MD Wake Forest, NC Durham, NC Cleveland #1 Cleveland #2 Cleveland #3 Total No. of patients 150 60 23 33 103 160 529 Mean age, y (range) 45.9 (18-72) 52.1 (22-85) 47.4 (19-70) 42.2 (18-67) 46.4 (18-81) 55.7 (18-83) 48.3 (18-85) Central hair loss pattern as determined by investigator, n (%) 0 106 (70.7) 33 (55.0) 14 (60.9) 10 (30.3) 33 (32.0) 70 (43.8) 266 1 27 (18.0) 22 (36.7) 8 (34.8) 19 (57.6) 37 (35.9) 44 (27.5) 157 2 8 (5.3) 1 (1.7) 1 (4.4) 1 (3.0) 13 (12.6) 21 (13.1) 45 3 9 (6.0) 2 (3.3) 0 (0.0) 1 (3.0) 9 (8.7) 16 (10.0) 37 4 0 (0.0) 1 (1.7) 0 (0.0) 2 (6.1) 4 (3.9) 5 (3.1) 12 5 0 (0.0) 1 (1.7) 0 (0.0) 0 (0.0) 7 (6.8) 4 (2.5) 12

Table II. Comparison of investigator versus subject rating of central hair loss: All sites/subjects

Investigator rating of central hair loss pattern 0 1 2 3-5 Total Scores N (%) N (%) N (%) N (%) N Subject rating 0 180 (73) 78 (53) 10 (22) 4 (7) 272 1 49 (20) 41 (28) 13 (29) 4 (7) 107 2 5 (2) 7 (5) 8 (18) 14 (24) 34 3-5 12 (5) 22 (15) 14 (31) 36 (62) 84 Total 246 (49.2) 148 (29.8) 45 (9.1) 58 (11.7) 497

There are 32 patients without subject rating. the sites (P \ .0001 with five degrees of freedom). severity in meeting 1 versus meetings 2 and 3, However, the ages of participants in the first three respectively, revealing both a slight, but not statisti- meetings were not statistically significantly different cally significant, increase in agreement and the (KruskaleWallis P = .055). By investigator assess- reliability of the investigator grading over time. ment, the majority (65.7%) of subjects in workshops The ability of subjects to discern and to grade their one to three had no hair loss, 28.8% had mild hair own central scalp hair density/loss with the use of loss (patterns 1 and 2), and 5.6% had extensive hair the provided photographic scales was determined loss (patterns 3-5). In workshops four, five, and six, by the comparison of their scores to that of the where subjects with hair loss were overrepresented various investigators (Table II). Subjects had only fair compared to the general population, 38.1% of sub- agreement with investigators in discerning whether jects had no hair loss, 45.6% had mild hair loss, and hair loss was present or not (pattern 0-1 vs patterns 2- 16.2% had extensive hair loss. Across all sites, there 5) with kappa statistic of 0.55. Not surprisingly, was a strong positive correlation between age and subjects were poor at determining the degree of hair degrees of hair loss (Spearman correlation = 0.36, loss (patterns 0-5; kappa statistic of 0.27) and tended with P \.0001). to overestimate their degree of hair loss. We previously reported on the good agreement We queried women about possible infections that among 4 experienced investigators grading central could be related to hair loss: history of infections was hair loss at the ﬁrst of these 3 meetings with an exact not documented by culture or medical records. match in 72% to 86% of cases (kappa scores 0.46- There was no signiﬁcant difference in the reporting 0.67) and no investigator more than one grade off of bacterial or nonscalp fungal infections between from another on any assessment2: this held true for those with patterns 0-2 and those with patterns 3-5 all 3 meetings. Given that two of the investigators central hair (P = .98). There was, however, a statis- (A. M. and E. A. O.) independently rated subjects in tically significant association between reports of all of the first 3 meetings, we compared the agree- tinea capitis and patterns 3-5 versus patterns 0-2 ment between these two investigators at meeting (P = .009) although the total number of cases was 1 versus meetings 2 and 3 as representative of further extremely low (7 in 468 subjects [1.5%] with patterns experience with the photographic scale in a collab- 0-2 and 4 in 61 subjects [6.5%] with patterns 3-5). We orative trial. There was a kappa statistic of 0.46 (with also addressed whether there was an association of a 72% exact matches) versus kappa statistic of 0.58 reaction to a hair care product and the development (with 75% exact matches) for central hair loss of extensive hair loss: there was no association of a JAM ACAD DERMATOL Olsen et al 249 VOLUME 64, NUMBER 2

Table III. Hair care history (ever used) along with current pattern of hair loss

Used (n = 529) Pattern 0 Pattern 1 Pattern 2 Pattern 3-5 Hair care history (ever used) N (%) (n = 266) N (%) (n = 157) N (%) (n = 45) N (%) (n = 61) N (%) Relaxer (P = .61) 476 (90) 237 (89) 143 (91) 39 (87) 57 (93) Hot comb (P = .35) 309 (58) 163 (61) 84 (54) 24 (53) 38 (62) Texturizer (P = .33) 61 (12) 29 (12) 21 (12) 2 (4) 9 (15) Braided with extensions 187 (35) 97 (38) 57 (34) 10 (22) 23 (38) (P = .29) Weaves or tracts (P = .22) 136 (26) 63 (25) 45 (27) 8 (17) 20 (33) scalp ‘‘rash’’ after shampoo or relaxer with extensive We then evaluated whether there was a relation- central hair loss (P = .47 for shampoo and P = .25 for ship of extensive central hair loss in mothers of relaxer). There was also no association of seborrheic subjects with pattern 3-5 loss, realizing that not only dermatitis/scaling scalp or eczema with any pattern hereditary factors but hair care practices might play a of central hair loss (P = .80 for seborrheic dermati- role in any similarity. There was a strong, statistically tis/scaling of scalp and P = .065 for eczema). signiﬁcant association of the severity of central hair We were interested in seeing if there was any loss in the subject and that of her mother (Spearman association of central hair loss with autoimmune or correlation P = .002). hormonal disorders. We did not ﬁnd any association Four hundred seventy-six (90%) of the 529 of thyroid disease and pattern of central hair loss women had ever used relaxers (Table III). There (P = .31). Type 1 or 2 diabetes mellitus was present in was no association of relaxer use and extensive 8.5% of women overall and was less common in central hair loss (P = .61). We asked subjects at what those with central hair loss patterns 1 and 2 than age relaxers were first used and compared the those with patterns 3-5 (P = .019). Surprisingly, a responses of the subjects with no central hair loss history of potential androgen-related disorders was (pattern 0), early central hair loss (pattern 1-2), and extremely common in subjects but was unrelated to extensive central hair loss (patterns 3-5) (Table IV). the degree of hair loss: adult acne was reported in Relaxer use began before 6 years of age in only two 34%, hirsutism (facial hair growth) in 48%, irregular (\1%) and between 6 and 15 years of age in 121 periods in 24%, and difficulty getting pregnant in 9% (37%) of the 328 subjects who had used relaxers in of the 529 women. Thirty-one percent (77/252) of studies one through five; the questionnaire was subjects with hirsutism also complained of irregular modified for study six, and data on the age of first periods, thus fulfilling the criteria for polycystic relaxer use were not available for this subset of ovarian syndrome.3 However, taking into account subjects. There was a statistically significant associ- the age of subjects and the presence of either facial ation with the age of first relaxer use and the current hair alone or facial hair with at least one other of the patterns of extensive loss (P = .01 for the Spearman above signs/symptoms, there was no relationship correlation) when comparing patterns 0 versus 3 to 5. between hyperandrogenism and central hair loss However, there was not a statistically significant pattern. association comparing patterns 1 and 2 versus 3 to We next looked at whether there was an associ- 5(P = .54 for the Spearman correlation). Of note, for ation of extensive hair loss with pattern hair loss in all users of relaxers, regardless of pattern of central male and female family members. There was no hair density, there was much more use of ‘‘no lye’’ relationship between central hair pattern 3-5 in versus ‘‘lye’’ relaxers. subjects and Hamilton Norwood patterns IV-VII Three hundred and nine of the 529 women (58%) (vertex to loss of all central scalp hair) male pattern had ever used hot combs, a much lower percentage hair loss (MPHL) in the subjects’ fathers (P = .18). than relaxers (Table III). There was no association of There was also no association of frontal MPHL in the use of hot combs ever and extensive hair loss (P = father and patterns 3-5 central scalp hair in the .35) (Table V). In contrast to the use of relaxers, in subject (P = .91). We evaluated the reported associ- those women who acknowledged the use of hot ation of women with polycystic ovarian syndrome combs and for whom we have age of first use, their (PCOS) and family members with MPHL3 and did not use began before 6 years of age in 32% to 33% of find a statistically significant association in those with subjects regardless of whether patterns 0, 1 to 2, or 3 both hirsutism and irregular periods (thus fulfilling to 5 (Table V) and by age of 15 in 87%. In contrast to the criteria for PCOS4) and vertex MPHL in the those who used relaxers, there was no association of subjects’ fathers (Spearman correlation P = .06). time of onset of use of hot comb use and current 250 Olsen et al JAM ACAD DERMATOL FEBRUARY 2011

Table IV. Onset of use of relaxers (1ye or not) in subjects with various patterns of central hair loss

Current hair loss pattern in women who have ever used relaxer (328/369*)

Age at time of Pattern 0 (n = 173) Pattern 1 (n = 104) Pattern 2 (n = 20) Pattern 3-5 (n = 31) first use, y n N (%) N (%) N (%) N (%) Total 0-5 2 2 (1) 0 (0) 0 (0) 0 (0) 2 6-15 121 78 (45) 30 (29) 5 (25) 8 (26) 121 16-24 155 76 (44) 53 (51) 11 (55) 15 (48) 155 25-33 30 10 (6) 13 (12) 3 (15) 4 (13) 30 34-43 14 6 (3) 5 (5) 0 (0) 3 (10) 14 44-53 4 1 (\1) 2 (2) 0 (0) 1 (3) 4 $ 54 2 0 (0) 1 (1) 1 (5) 0 (0) 2 Total 328 173 104 20 31 328

*Includes all subjects in workshops one to five; data on age of first relaxer use from workshop six could not be consolidated.

Table V. Onset of hot comb use compared with current hair loss

Current hair loss pattern in women who have ever used hot combs (269/369*)

Age at time of Pattern 0 (n = 133) Pattern 1 (n = 88) Pattern 2 (n = 18) Pattern 3-5 (n = 30) first use, y n N (%) N (%) N (%) N (%) Total 0-5 87 42 (32) 28 (32) 7 (39) 10 (33) 87 6-15 147 73 (56) 49 (56) 10 (56) 15 (50) 147 16-24 23 12 (9) 7 (8) 0 (0) 4 (14) 23 25-33 6 4 (3) 2 (2) 0 (0) 0 (0) 6 34-43 3 2 (1) 1 (1) 0 (0) 0 (0) 3 44-53 2 0 (0) 1 (1) 0 (0) 1 (3) 2 $ 54 1 0 (0) 0 (0) 1 (5) 0 (0) 1 Total 269 133 88 18 30 269

*This includes all subjects in workshops 1-5; the question was not asked in workshop 6. extensive hair loss (P = .96 for Spearman correlation has been called ‘‘hot comb alopecia’’ secondary to its comparing central hair loss patterns 0 vs 3 to 5 and frequent occurrence in those using hot combs to P = .87 for comparing 1 and 2 vs 3 to 5). straighten their hair7 or ‘‘follicular degeneration Use of braids with extensions, weaves or tracts, syndrome’’ based on the characteristic histopatho- and texturizers were not associated with extensive logic findings on a representative scalp biopsy.8 That hair loss (Table III; P = .29 for braids with extensions, this central scarring hair loss occurs much more P = .22 for weaves or tracts, and P = .33 for commonly in African American than Caucasian texturizers). women has led to a focus on the hair care practices We attempted to address whether the current of African American women as the sole cause of this central scalp hair loss pattern appeared to select for problem. Our study was initiated to gather data on any particular hair care practice (Table VI). Women the type, degree, and frequency of central hair loss in with extensive hair loss were less likely to now use African American women and its relationship to age, relaxers than those without extensive loss (52% for family history of hair loss, concurrent medical prob- patterns 3-5 vs 67% for those with patterns 1-2), but lems, and hair care practices. clearly extensive hair loss did not curtail the use of The photographic scale developed by the group relaxers. Hot combs continued to be used much less was found to be a reliable tool for rating the general often than relaxerse17%, 19%, and 30% in those with severity of central hair loss when used by physician pattern 0, 1, or 2, respectively, and 25% in those with investigators and should be able to be used to patterns 3 to 5. standardize the assessment of the severity of hair loss in clinical trials. However, subjects that had no DISCUSSION previous instructions or training on its use were not Central centrifugal cicatricial alopecia (CCCA) is reliably able to self-identify their degree of hair loss: the primary reason that African American women whether subjects could more closely mirror investi- seek medical consultation for their hair,5,6 but the gators with a simpliﬁed scale or if the use of the incidence and causality are not known. This hair loss current scale were preceded by a training session JAM ACAD DERMATOL Olsen et al 251 VOLUME 64, NUMBER 2

Table VI. Current hair care practice versus subjects’ could lead to hair loss in this country and Africa: 67% current pattern of central hair loss of Kenyan women in one study reported relaxer- related problems, the most common being hair loss Current pattern of hair loss (type unspecified) and burns (scalp).11 In our study, Pattern Pattern Pattern Pattern 0 1 2 3-5 Total there was no association of extensive hair loss and Current hair (n = 266) (n = 157) (n = 45) (n = 61) (n = 529) reaction to hair care products, although the type of care practices N (%) N (%) N (%) N (%) N (%) reactions was not specified. In future studies in the Relaxer 176 (66) 104 (66) 32 (70) 32 (53) 344 (65) US, we need to go beyond the query of age of onset Hot comb 44 (17) 32 (19) 14 (30) 15 (25) 105 (20) or current usage of relaxers or other hair care Braids with 35 (14) 20 (12) 6 (13) 11 (18) 72 (14) practices to determine if there are differences in the extensions way relaxers or permanent waves are neutralized, Braids without 34 (13) 18 (11) 3 (7) 11 (18) 66 (13) the concentrations of active chemicals, the frequency extensions of application, and the severity and number of Weaves 23 (9) 18 (11) 7 (15) 11 (18) 59 (11) reactions to the products subjects have had. We Curly perm 9 (4) 3 (2) 2 (4) 2 (3) 16 (3) also need to consider whether there are other hair Jheri curl 15 (6) 14 (8) 5 (11) 5 (8) 39 (7) Twist 17 (7) 9 (5) 2 (4) 3 (5) 31 (6) care practices, such as the use of heat, hair color, Texturized 3 (1) 7 (4) 2 (4) 2 (3) 14 (3) greases, and pomades, being used either concomi- tantly or sequentially with relaxers or permanent waves that could, in combination, lead to a permanent compromise of normal follicular growth. needs further evaluation. Taylor et al9 previously Extensive hair loss was not found to be related to reported a similar but less profound disconnect any history of scalp scaling, seborrheic dermatitis, or between experienced observers and male subjects eczema. The incidence of tinea capitis (6.5%), how- in rating their Hamilton-Norwood pattern of MPHL. ever, was high in African American women with Because we collapsed the data on the A and B central hair loss patterns 3 to 5 and speaks to the subtypes of central loss in this study, we need further need to address early all potentially reversible scalp information on both the relative frequency with conditions that could lead to prolonged periods of which the frontal versus vertex predominant patterns scalp inflammation. That African American women are present in African American women with central may be particularly susceptible to tinea capitis was hair loss and whether there is any particular medical shown in a retrospective study carried out in an or genetic association with the frontal versus vertex academic dermatology practice in which only 60% of pattern of central hair loss. the patients were of African descent but all of the Overall, 65.7% of the 233 women in workshops culture-documented tinea capitis occurred in African one to three, which we felt was indicative of the Americans including seven of the nine cases in general population, did not have any central hair African American women.12 loss, a finding similar at all three screenings. In this The incidence of extensive (patterns 3-5) central group, 24.5% of women had pattern 1, 4.3% had hair loss at our three meeting sites in the southeast- pattern 2, and 5.6% had pattern 3 to 5 central scalp ern United States was 5.6%. If central hair loss hair loss. The use of relaxers per se did not appear to patterns 3 to 5 can be used as a surrogate marker be related to severity of hair loss because it was for CCCA, then although the incidence of this con- equally high in women with no or minimal hair loss, dition is relatively low in African American women, (87%-91%) versus those with extensive hair loss the incidence is far greater than other types of (93%). There was obvious difficulty in giving up cicatricial alopecia and the total number of affected relaxer use even with extensive hair loss because African American women is high. The overall inci- 53% of women with patterns 3 to 5 continued to use dence of CCCA in a group of African women who relaxers. Despite the purported potential for hair loss commonly use relaxers (2.7%)10 is not too dissimilar from hot comb use, there was no association with to our findings. The location of hair loss in African severity of hair loss and use of hot combs. American women with central patterns 1 to 5 mimics The use of chemical processing of scalp hair by that seen in Caucasian women with female pattern women of African descent is not unique to the US but hair loss (FPHL) Ludwig patterns 1 to 3, the differ- is also commonly seen in women in African coun- ence being the less frequent occurrence of extensive tries: Khumalo et al10 reported that almost 50% of hair loss in Caucasian women (;1% incidence of black South African women relax their hair and that Ludwig pattern 3)13 and the paucity of reports of CCCA exists in this population as well. Relaxers are a scarring alopecia in the central scalp in Caucasian frequent cause of both hair loss and problems that compared to African American women. Olsen14 has 252 Olsen et al JAM ACAD DERMATOL FEBRUARY 2011 recently reiterated that there can be follicular drop- us a wealth of new data on hair loss in African out (confirmed histologically) with extensive FPHL American women, it is only the starting place in and has hypothesized that CCCA may be a scarring which to evaluate the causality and treatment options form of female pattern hair loss, perhaps induced by for this common condition. chronic inflammation or infection induced or aggra- The authors thank Melissa H. Piliang, MD, and Angelia vated by hair care practices. This concept, especially Kyie, MD, of the Cleveland Clinic for their role in helping to in lieu of the frequency of patterns 1 and 2 in African assess subjects’ hair loss at the Cleveland Clinic workshops. American women which resembles female pattern They also thank Andrea Bazakas, BS, and Eric Lai, BS, of hair loss in Caucasian women, needs further explo- Duke University Medical Center for their role in organizing ration including biopsy confirmation. the data for the study analysis. Any hereditary nature of central hair loss in African American women remains unclear: although REFERENCES 27% of women with pattern 3 to 5 hair loss had 1. Olsen E, Bergfeld W, Cotsarelis G, Price V, Shapiro J, Sinclair R, et al. Summary of NAHRS-sponsored workshop on cicatricial mothers with pattern 3 to 5 central hair loss, the alopecia, Duke University Medical Center, February 10 and 11, association could either be genetic or secondary to 2001. J Am Acad Dermatol 2003;48:103-10. common hair care practices. 2. Olsen EA, Callender V, Sperling L, McMichael A, Anstrom KJ, The relatively high incidence of clinical signs of Bergfeld WD, et al. Central scalp alopecia photographic scale hyperandrogenism in African American women with in African American women. Dermatol Ther 2008;21:264-7. 3. Carey AH, Waterworth D, Patel K, White D, Little J, Novelli P, and without hair loss is important to note and should et al. Polycystic ovaries and premature male pattern baldness broaden the history taking and evaluation of African are associated with one allele of the steroid metabolism gene American women with hair loss to include a search CYP17. Hum Mol Genet 1994;3:1873-6. for signs or laboratory results indicative of hyper- 4. Rotterdam ESHRE/ASRM-Sponsored PCOS Consensus Work- androgenism and its associated conditions. shop Group. Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syn- This study has limitations. One limitation is that we drome. Fertil Steril 2004;81:19-25. used central scalp hair patterns 3 to 5 as a surrogate 5. Sperling LC. Scarring alopecia and the dermatopathologist. phenotypic marker for CCCA but do not have biopsy J Cutan Pathol 2001;28:333-42. confirmation of the follicular loss and typical inflam- 6. Whiting DA, Olsen EA. Central centrifugal cicatricial alopecia. matory findings seen histologically with early CCCA. Dermatol Ther 2008;21:268-78. 7. LoPresti P, Papa CM, Kligman AM. Hotcomb alopecia. Arch Second, data collected based on memory of distant Dermatol 1968;98:234-8. events are always subject to bias, particularly ques- 8. Sperling LC, Sau P. The follicular degeneration syndrome in tions related to when an event occurred. Third, black patients: ‘‘hot comb alopecia’’ revisited and revised. Arch despite asking questions on frequency of hair care Dermatol 1992;128:68-74. treatments, we were not able to determine whether a 9. Taylor R, Matassa J, Leavy JE, Fritschi L. Validity of self reported male balding patterns in epidemiological studies. BMC Public particular hair care practice, such as relaxers, might Health 2004;4:60-4. be problematic or safe if used in a particular way. 10. Khumalo NP, Jessop S, Gumedze F, Ehrlich R. Hairdressing and Fourth, we did not address whether the type, inten- the prevalence of scalp disease in African adults. Br J Dermatol sity, and/or frequency of heat or the use of emollients 2007;157:981-8. and greases used on the scalp contribute to infection, 11. Etemesi GA. Impact of hair relaxers in women in Nakuru, Kenya. Int J Dermatol 2007;46(suppl):23-5. inflammation, and follicular destruction. Fifth, we do 12. Silverberg NB, Weinberg JM, DeLeo VA. Tinea capitis: focus on not have enough information to address whether African American women. J Am Acad Dermatol 2002; tight braids, weaves, or extensions, if anchored in a 46(Suppl):S120-4. particular fashion, could be related to extensive 13. Dinh QQ, Sinclair R. Female pattern hair loss: current treatment central hair loss. Finally, this was not a population- concepts. Clin Intervent Aging 2007;2:189-99. 14. Olsen EA. Female pattern hair loss and its relationship to based study so no prevalence information can be permanent/cicatricial alopecia: a new perspective. J Investig gleaned from it. Therefore, although this study gives Dermatol Symp Proc 2005;10:217-21.