Natural hair color and questionnaire-reported pain among women in the United States

Dear editor,

Small-scale clinical studies have reported that individuals with naturally red hair have increased resistance to inhaled and subcutaneous local anesthetics and need more anesthetic to alleviate pain (Liem et al., 2005; Liem et al., 2004). Previous studies have also reported that red-haired individuals are more sensitive to thermal pain and dental pain (Binkley et al., 2009; Liem et al., 2005) and that individuals carrying MC1R variants associated with a red-hair phenotype have increased requirements for anesthetics (Liem et al., 2004) or increased fear of pain (Binkley et al., 2009). However, other studies based on an experimental setting using quantitative controlled measurements of graded pain stimuli reported entirely different findings. For example, Mogil et al showed greater tolerance of pain and an increased analgesic response among red-haired individuals with MC1R variants (Mogil et al., 2005; Mogil et al., 2003). Here we sought to revisit the problem with a different approach, the collection of information on self-reported pain from the Medical Outcomes Study Short-Form 36 Health Status Survey (SF-36) in the Nurses' Health Study (NHS) and NHS II. We recognize that self-reported assessments of pain and pain tolerance are subjective and cannot match experimental assessments of pain thresholds. Nevertheless, we believe that because nurses have high levels of education and are familiar with medical issues, the collected information is of high-quality and valid. An added advantage of using data in NHS and NHS II is that the information was collected prospectively and that it is based on a large number of participants, namely a total of 149,664 participants from the NHS (n=67,349) and NHS II (n=82,315). It has to be noted, however, that these cohorts only involved women, in fact mostly Caucasian women, and so extrapolations to men and other races/ethnicities require caution. Nevertheless, the baseline scores on SF-36 scales in NHS are quite comparable to a similarly aged group of working U.S. women (Coakley et al., 1998).

Detailed information on the Materials and Methods is shown in the supplementary information (Data S1). Briefly, self-reported natural hair color in early adulthood was categorized as red, blonde, light brown, dark brown, and black and assigned values of 5, 4, 3, 2, and 1, respectively. In 1992, 1996, and 2000 (NHS) and 1993, 1997, and 2001 (NHS II), participants completed the SF-36 to measure quality of life. We utilized the answers to two items concerning bodily pain, i.e., the extent of bodily pain and the extent of bodily pain interfering with normal work during the past four weeks. We assigned a score for each item (see Data S1) and evaluated the associations between natural hair color and pain in various ways (assessment of average age-adjusted as well as multivariate-adjusted combined pain scores; repeated-measures pain scores, referred to as “updated pain scores”; average pain scores of three surveys in each cohort; change in pain scores over four years; and pain scores assessed separately for the two questionnaire items). The regression coefficients (95% confidence intervals (CIs)) were calculated for each hair color using generalized linear regression models.

The mean age was 58.9 (SD=7.1) years in 1992 (range 46–72 years) in NHS and 38.1 (SD=4.6) years in 1993 (range 28–48 years) in NHS II. Overall, the pain scores tended to increase with age. Combining the two cohorts, each five-year increase in age was associated with a 1.18-point (95% CI: 0.89, 1.46) increase in pain score.

As shown in Table 1, the combined pain scores differed significantly according to hair color. In both cohorts, participants with lighter hair color reported higher pain scores. The updated pain score analysis showed that, compared with dark-haired individuals, the pain scores were 1.05 higher for dark brown-haired, 1.08 higher for light brown-haired, 1.25 higher for blonde-haired, and 1.54 higher for red-haired participants, after adjusting for potential confounders (P<0.0001 for all, last row of Table 1). The mean difference in pain score between red- and dark-haired participants was equivalent to the mean difference in pain score observed between participants (adjusted across all hair colors) who were 6~7 years apart in age. Each unit increment of lighter hair color was associated with a 0.19-point higher pain score (95% CI: 0.11–0.27) (P_trend<0.0001). Average pain score analysis yielded similar findings (Table 1). Furthermore, individuals with lighter hair color also showed a greater increase in pain score over a four-year time period. As shown in Table 2, when the results from NHS and NHS II were combined, each one-unit increment of lighter hair color was associated with a 0.17-point higher increase in pain score over 4 years (95% CI: 0.10–0.25 in pain score; P_trend<0.0001). Compared with black-haired women, red-haired women showed the largest increase in pain scores over time; the beta coefficient (95% CI) representing the relative change in pain score was 1.16 (0.74–1.57) in the combined cohorts (Table 2). The difference in pain score increase over four years between red-haired and dark-haired individuals was as large as the mean difference in pain score calculated regardless of hair color for participants who were five years apart in age. Similar results were obtained when the pain scores calculated from the two items on the questionnaire were analyzed separately. Compared with black-haired women, red-haired women had a score that was 1.88-point (P<0.0001) higher for the extent of bodily pain and 1.17 (P<0.0001) higher for the extent of bodily pain interfering with normal work. Red-haired women also had the largest increase in scores of each pain item over time; the beta coefficient (95% CI) was 1.27 (P<0.0001) for change in score for bodily pain and 1.11 (P<0.0001) for change in score for bodily pain interfering with normal work. In a secondary analysis, we examined each pain item as ordinal outcome and the results did not change significantly (data not shown).

Table 1.

Mean difference in pain score according to natural hair color

	Difference in pain score					Per one unit of hair color1	P for trend1
	Black	Dark brown	Light brown	Blonde	Red
Nurses' Health Study
Average score
Age-adjusted	0 (Ref)	0.92 (0.52, 1.33)	0.78 (0.37, 1.19)	0.95 (0.50, 1.40)	1.78 (1.24, 2.32)	0.17 (0.08, 0.26)	0.0002
Multivariate-adjusted2	0 (Ref)	1.14 (0.79, 1.48)	1.07 (0.72, 1.42)	1.28 (0.89, 1.66)	1.71 (1.25, 2.17)	0.19 (0.11, 0.26)	<0.0001
Updated score 3
Age-adjusted	0 (Ref)	1.05 (0.33, 1.77)	0.89 (0.17, 1.61)	0.99 (0.20, 1.79)	1.84 (0.88, 2.80)	0.15 (0.05, 0.25)	0.004
Multivariate-adjusted2	0 (Ref)	1.24 (0.64, 1.84)	1.16 (0.56, 1.76)	1.30 (0.64, 1.96)	1.70 (0.91, 2.49)	0.17 (0.04, 0.29)	0.009
Nurses' Health Study II
Average score
Age-adjusted	0 (Ref)	0.52 (0.17, 0.88)	0.59 (0.24, 0.95)	0.57 (0.19, 0.94)	1.16 (0.70, 1.62)	0.13 (0.06, 0.20)	0.0003
Multivariate-adjusted2	0 (Ref)	0.70 (0.35, 1.06)	0.78 (0.41, 1.14)	0.87 (0.49, 1.24)	1.19 (0.74, 1.63)	0.14 (0.08, 0.20)	<0.0001
Updated score 3
Age-adjusted	0 (Ref)	0.56 (−0.08, 1.20)	0.66 (0.02, 1.30)	0.71 (0.04, 1.38)	1.23 (0.41, 2.05)	0.16 (0.04, 0.28)	0.008
Multivariate-adjusted2	0 (Ref)	0.84 (0.20, 1.49)	0.97 (0.32, 1.63)	1.20 (0.52, 1.87)	1.38 (0.59, 2.16)	0.21 (0.10, 0.31)	0.0001
Nurses' Health Study and Nurses' Health Study II combined
Average score
Age-adjusted	0 (Ref)	0.71 (0.31, 1.11)	0.68 (0.41, 0.94)	0.74 (0.37, 1.10)	1.45 (0.85, 2.05)	0.14 (0.09, 0.20)	<0.0001
Multivariate-adjusted2	0 (Ref)	0.92 (0.49, 1.35)	0.93 (0.64, 1.22)	1.07 (0.67, 1.47)	1.45 (0.93, 1.96)	0.16 (0.11, 0.21)	<0.0001
Updated score 3
Age-adjusted	0 (Ref)	0.78 (0.30, 1.26)	0.76 (0.28, 1.24)	0.83 (0.31, 1.34)	1.49 (0.86, 2.11)	0.16 (0.08, 0.23)	<0.0001
Multivariate-adjusted2	0 (Ref)	1.05 (0.62, 1.49)	1.08 (0.63, 1.52)	1.25 (0.78, 1.72)	1.54 (0.98, 2.09)	0.19 (0.11, 0.27)	<0.0001

¹For the analysis of pain score associated with per-unit change of hair color, hair color was treated as continuous exposure, with black, dark brown, light brown, blonde, and red color assigned values of 1, 2, 3, 4, and 5 respectively.

²Calculated using linear regression considering repeated measures, adjusted for age, body mass index, physical activity, smoking status, menopausal status and postmenopausal hormone use, weekly sitting at home, personal history of hypertension, cardiovascular disease, type II diabetes, cancer, and autoimmune diseases, race, phobic anxiety, physical function scores, and antidepressant use.

³Repeated measures analysis of pain scores were used, which integrated the analysis of pain scores assessed in three surveys in each cohort (1992, 1996, and 2000 in NHS and 1993, 1997, and 2001 in NHS II) and permitted the examination of pain scores at all three available time points, taking into account the correlations between assessments.

Table 2.

Mean difference in change of pain score over four years associated with natural hair color

	Difference in change of pain score					Per one unit of hair color1	P for trend1
	Black	Dark brown	Light brown	Blonde	Red
Nurses' Health Study
Age and baseline score-adjusted2	0 (Ref)	0.45 (0.01, 0.90)	0.40 (−0.05, 0.85)	0.60 (0.11, 1.10)	0.81 (0.22, 1.40)	0.10 (0.01, 0.20)	0.03
Multivariate-adjusted3	0 (Ref)	0.64 (0.20, 1.09)	0.60 (0.16, 1.05)	0.86 (0.37, 1.36)	0.96 (0.38, 1.55)	0.13 (0.04, 0.23)	0.006
Nurses' Health Study II
Age and baseline score-adjusted2	0 (Ref)	0.77 (0.35, 1.18)	0.96 (0.54, 1.37)	0.99 (0.55, 1.43)	1.31 (0.77, 1.85)	0.19 (0.11, 0.27)	<0.0001
Multivariate-adjusted3	0 (Ref)	0.83 (0.36, 1.30)	1.04 (0.56, 1.51)	1.12 (0.63, 1.62)	1.34 (0.77, 1.92)	0.21 (0.13, 0.29)	<0.0001
Nurses' Health Study and Nurses' Health Study II combined
Age and baseline score-adjusted2	0 (Ref)	0.62 (0.31, 0.93)	0.69 (0.14, 1.23)	0.82 (0.44, 1.19)	1.07 (0.59, 1.56)	0.15 (0.07, 0.23)	0.0003
Multivariate-adjusted3	0 (Ref)	0.73 (0.41, 1.05)	0.81 (0.39, 1.23)	0.99 (0.64, 1.34)	1.16 (0.74, 1.57)	0.17 (0.10, 0.25)	<0.0001

¹Hair color values were as indicated in Table 1.

²Calculated using linear regression considering repeated measures, adjusted for age and the pain score at the baseline year of each analysis (scores in 1992 for the analysis of change in 1996, and scores in 1996 for the analysis of change in 2000 in NHS; scores in 1993 for the analysis of change in 1997, and scores in 1997 for the analysis of change in 2001 in NHS II).

³Additionally adjusted for body mass index, physical activity, smoking status, menopausal status and postmenopausal hormone use, weekly sitting at home, personal history of hypertension, cardiovascular disease, type II diabetes, cancer, and common autoimmune diseases, race, phobic anxiety, physical function scores, and antidepressant use.

It was conceivable that the above differences in hair color-associated pain scores were due to differences in overall health. Neither the detailed health status information presented in Table S1 nor additional sensitivity analyses performed by excluding specific groups of participants supported this assumption, however. For instance, selective exclusion of participants with major chronic diseases, major autoimmune diseases, or in addition the lowest quartile of the physical function scores did not change the above findings significantly, nor did, for that matter, restriction to Caucasians. Lastly, adjusting for weekly dosage of major over-the-counter pain killers or excluding current users of these medications did also not change the results appreciably (data not shown). However, as information on the use of prescribed pain medications was not available, we cannot assess the association between hair color and the overall use of pain medication.

The above results show that lighter hair color, in particular red hair, is associated with an increase in self-reported pain. They hence suggest, though they do not prove, that increased pain perception in individuals with lighter hair color might be due to a decrease in pain thresholds and not, at least not exclusively, an increased resistance to pain medications. The underlying mechanisms, however, are not clear. An obvious candidate potentially responsible for the observed differences is the endogenous opioid β-endorphin (Barfield et al., 2013). β-endorphin is known to reduce nociception and is derived from the same precursor, pro-opiomelanocortin (POMC), that also gives rise to the MC1R ligand α-melanocyte-stimulating hormone (α-MSH) (Cui et al., 2007; Dores et al., 2014). In fact, experimental data have suggested a link between β-endorphin, susceptibility to pain, and pigmentation. In mice, for instance, UV exposure not only leads to a tanning response but also a systemic elevation of β-endorphin levels and an increase in pain thresholds (Cui et al., 2007; Fell et al., 2014). Increased levels of plasma β-endorphin and β-lipotropin (β-LPH, another POMC derivative) after UV exposure have also been associated with enhanced pigmentation (Belon, 1985; Kauser et al., 2004; Levins et al., 1983). It is conceivable, though not yet demonstrated, that individuals with lighter hair color may have generally lower levels of β-endorphin, but other mechanisms such as genetic variations in CACNG2 (a voltage-dependent calcium channel) that have been associated with pain responses (Nissenbaum et al., 2010; Sorge et al., 2012) may have to be considered as well.

Our study also revealed an interesting association between aging and pain. This association has also been studied previously, particularly for individuals at advanced age. For example, in one study, older individuals reported more head and chest pain, but less genital pain (Shega et al., 2014), but another study did not find age-related patterns to pain reporting among older adults (Riley et al., 2014). We here found not only that aging in general is significantly associated with increased pain perception but also that the age-related increase in pain is greater among red-haired compared to black-haired women.

In conclusion, our analysis of two large, prospective, well-established cohorts suggest an association between lighter hair color and increased questionnaire-reported pain, particularly for those with red hair. Further studies are warranted to confirm our findings in other populations and to investigate the common genetic predisposition underlying the associations as well as the biological mechanisms.