Gender role affects experimental pain responses: A systematic review with meta-analysis

Authors


  • Funding sources

    None declared.

  • Conflicts of interest

    None declared.

Correspondence

Oras Alabas

E-mail: o.a.alabas@LeedsMet.ac.uk

Abstract

Gender role refers to the culturally and socially constructed meanings that describe how women and men should behave in certain situations according to feminine and masculine roles learned throughout life. The aim of this meta-analysis was to evaluate the relationship between gender role and experimental pain responses in healthy human participants. We searched computerized databases for studies published between January 1950 and May 2011 that had measured gender role in healthy human adults and pain response to noxious stimuli. Studies were entered into a meta-analysis if they calculated a correlation coefficient (r) for gender role and experimental pain. Searches yielded 4465 ‘hits’ and 13 studies were eligible for review. Sample sizes were 67–235 participants and the proportion of female participants was 45–67%. Eight types of gender role instrument were used. Meta-analysis of six studies (406 men and 539 women) found a significant positive correlation between masculine and feminine personality traits and pain threshold and tolerance, with a small effect size (r = 0.17, p = 0.01). Meta-analysis of four studies (263 men and 297 women) found a significant negative correlation between gender stereotypes specific to pain and pain threshold and tolerance, with a moderate effect size (r = −0.41, p < 0.001). In conclusion, individuals who considered themselves more masculine and less sensitive to pain than the typical man showed higher pain thresholds and tolerances. Gender stereotypes specific to pain scales showed stronger associations with sex differences in pain sensitivity response than masculine and feminine personality trait scales.

1. Introduction

Several painful conditions are more prevalent in women than men including temporomandibular disorders, migraine, tension-type headache, fibromyalgia and irritable bowel syndrome (Hurley and Adams, 2008). Women report clinical pain to be more intense than men, with a greater frequency of painful episodes presenting at more body sites (Fillingim, 2009). Women also have greater sensitivity to experimentally induced pain with lower thresholds and tolerances to painful stimuli (Riley et al., 1998; Tashani et al., 2010; Neziri et al., 2011). Sex and gender differences in pain sensitivity response may be due to biological factors such as blood pressure, gonadal hormones and body size (Fillingim and Maixner, 1996; Riley et al., 1999; Tashani et al., 2010), and psychological factors such as fear of pain, catastrophizing, depression and anxiety (Osman et al., 2000; Riley et al., 2001; Tashani et al., 2010; Forsythe et al., 2011).

One psychosocial factor that may influence differences in pain sensitivity response is gender role (Bernardes et al., 2008). This refers to the culturally and socially constructed meanings that describe how women and men should behave in certain situations according to feminine and masculine roles learned throughout life (Myers et al., 2003). Traditionally, high levels of stoicism are related to men and high levels of sensitivity are related to women (Robinson et al., 2000).

What's already known about this topic?

  • Women have greater sensitivity to experimentally induced pain.
  • Different scales were developed to assess Gender role.
  • Gender role appears to influence response to experimental pain.

What does this study add?

  • Gender stereotypes specific to pain scales showed stronger associations with sex differences in pain sensitivity response than masculine and feminine personality trait scales.

Investigations into the influence of gender role on pain response have focused on the relationship between gender-related personality orientation and pain (Vierhaus et al., 2011). Otto and Dougher (1985) used the Bem Sex Role Inventory (BSRI, Bem, 1974) to measure gender role and examined its relation to experimentally induced pressure pain. They found a significant correlation between masculinity–femininity and pain threshold for male participants but not female participants. Men with higher masculinity scores showed higher pressure pain thresholds than men showing less masculinity. The Personal Attributes Questionnaire (PAQ, Spence et al., 1974) has been used to examine the relationship between gender-related personality types and pain sensitivity. For example, Dixon et al. (2004) found that masculinity–femininity scores correlated negatively with cold pain tolerance.

A Gender Role Expectations of Pain Questionnaire (GREP) was developed by Robinson et al. (2001) to measure sex-related stereotypic attributions of pain sensitivity, pain endurance and willingness to report pain. Wise et al. (2002) found that GREP provided a partial explanation for differences in response to painful heat stimuli between men and women. Defrin et al. (2009) found that GREP predicted heat pain tolerance but not pain threshold in an Israeli student population. Recently, we have reported that differences in response to experimental cold pain in healthy Libyans were mediated by GREP but not pain-related anxiety (Alabas et al., 2012).

Gender role appears to influence response to experimental pain (Greenspan et al., 2007) although the magnitude and the strength of the gender role on pain sensitivity are unclear (Chesterton et al., 2003; Garcia et al., 2007). The aim of this systematic review, with meta-analysis, was to evaluate the relationship between gender role and experimental pain responses in healthy human participants.

2. Methods

2.1 Data sources and search methods for identification of studies

This review used the guidelines for Preferred Reporting Items for Systematic Reviews and Meta-Analyses (Moher et al., 2009). The following computerized databases were searched for studies published between January 1950 and May 2011: MEDLINE, Academic Search complete, AMED, BIOSIS, CINAHL, SportDiscus, EMBASE, PsycARTICLES, PsycINFO, PubMed and ScienceDirect. Keywords used were: gender role OR femininity OR masculinity OR sex differences OR gender differences AND pain (Appendix 1).

2.2 Criteria for considering studies for this review

Studies were included if data were collected for:

  • healthy human participants at least 18 years of age;
  • gender role, and/or femininity and/or masculinity using a validated assessment instrument;
  • pain threshold, tolerance, intensity or unpleasantness to a painful stimulus;
  • relationships between pain response and gender role, and/or femininity, and/or masculinity.

All relevant studies were included irrespective of language and non-English reports were assessed and translated if necessary.

2.3 Study selection

Full articles of potentially relevant studies were screened for inclusion by two reviewers (O.A.A. and O.A.T.) with M.I.J. and G.T. acting as arbiters. Information extracted from eligible studies was experimental pain technique, gender role instrument and correlation coefficients between pain outcome and gender role measurements.

2.4 Statistical method for the meta-analysis

Studies were entered in the meta-analysis if they calculated a correlation coefficient for gender role and pain threshold or pain tolerance or pain ratings. Pain thresholds and pain tolerances were contained within the same analysis to improve the statistical power of the test (Thorn et al., 2004). Pain intensity and unpleasantness ratings were meta-analysed separately because of differences in the nature and direction of the measurements. If a study reported a regression rather than a correlation then it would be included in the meta-analysis if there were enough data to calculate a correlation-type effect size. If r 2 was given then r would be calculated, and if a study reported an effect size in the d form (e.g. odds ratios, relative risks, Glass's deltas) then it would be converted into r (Ellis, 2010). If a study reported more than one measurement from the same set of data an average effect size for each study was calculated (Ellis, 2010). The direction of correlation coefficients was standardized by ensuring that all scales assessing gender role and all measurements of pain were in the same direction in order to combine them in the meta-analysis. Studies were meta-analysed according to (1) masculine and feminine personality traits and (2) gender stereotypes specific to pain scales. The pooled mean effect size with 95% confidence intervals was calculated using Comprehensive Meta-analysis Software version 2.0 (Biostat, Englewood, NJ, USA) and Cohen's guidelines used to categorize the strength of the correlation coefficient with 0.10 taken as small, 0.30 as medium and 0.50 as large (Cohen, 1992; Sterne et al., 2001).

2.5 Exploration of heterogeneity

Heterogeneity in correlation coefficients between studies was tested by Cochran's Q statistic (Cochran's Q-test, Fleiss, 1981). If significant heterogeneity was found (i.e. Q statistic p < 0.05) pooled correlation coefficients were calculated using the random effects method (Sterne et al., 2001). The magnitude of heterogeneity was assessed by calculating I2 statistic, with 0% taken as no observed heterogeneity, 25% as low heterogeneity, 50% as moderate heterogeneity and 75% as high heterogeneity (Higgins and Green, 2006).

3. Results

There were 4465 references identified in the initial literature search of which 2176 were duplicates. Of the remaining 2289 references, 2230 were not relevant because they did not measure an experimentally induced pain outcome or investigate gender role. Fifty-nine potentially relevant references were identified (Fig. 1).

Figure 1.

PRISMA flow diagram of the process of identifying and including articles for the systematic review.

3.1 Excluded studies

Of the 59 potentially relevant articles, 46 were excluded because they did not quantify gender role, masculinity or femininity (n = 31), did not measure pain threshold or pain tolerance (n = 13), or used participants who were under 18 years old (n = 2) (Appendix 2).

3.2 Characteristics of reviewed studies

Thirteen studies were included for review (Table 1). Sample sizes ranged from 67 to 235 participants with an age range of 18–35 years. The proportion of female participants in study samples ranged from 45% to 67%. Eight different instruments to measure gender role were used in the 13 studies. Only three studies examined the effect of sex of the experimenter on pain outcomes.

Table 1. Characteristics of the studies that were included in the systematic review (n = 13).
StudyCountrySample size (M/F)Instrument/dimensionPain stimuliPain measurementCorrelation coefficient (r)
  1. BSRI, Bem Sex Role Inventory; PAQ, Personal Attributes Questionnaire; EPAQ, Extended Personal Attributes Questionnaire; APBQ, Appropriate Pain Behaviour Questionnaire; GREP, Gender Role Expectations of Pain Questionnaire; HMI, Hyper-Masculinity Index; CPT, cold pressor test.
Otto and Dougher (1985)USA80 (40/40)BSRI/Masculinity–FemininityPressureThreshold

−0.31

0.08

Fillingim et al. (1999)USA209 (92/117)BSRI/Masculinity, FemininityHeatThreshold, toleranceNot calculated
Nayak et al. (2000)USA107 (53/54)APBQCPTTolerance0.57
India119 (60/59)
Myers et al. (2001)USA104 (54/50)BSRI/Masculinity, FemininityCPT

Threshold

Tolerance

0.13, −0.15

0.21, −0.22

Sanford et al. (2002)USA145 (66/78)EPAQ/FemininityCPTTolerance−0.17
Wise et al. (2002)USA148 (61/87)GREP/Stereotypical Willingness to Report PainHeat

Threshold

Tolerance

Unpleasantness

0.42

0.45

0.50

Thorn et al. (2004)USA235 (90/129)EPAQ/Masculinity–FemininityCPT

Tolerance

Intensity, unpleasantness

0.17

−0.29

Dixon et al. (2004)USA203 (91/112)PAQ/Masculinity–FemininityCPT

Tolerance

Intensity

Unpleasantness

−0.15

0.16

0.21

Robinson et al. (2004)USA67 (37/30)GREP/Stereotypical Willingness to Report PainHeatTemporal summation0.48
Pool et al. (2007)USA103 (49/28)

Gender Norms for Pain Tolerance Questionnaire

Gender Group Identification Questionnaire

Electrical shock

Threshold

Tolerance

Not calculated

Not calculated

Defrin et al. (2009)Israel72 (33/39)

GREP/Pain Sensitivity (self-same),

Pain Sensitivity (self-opposite),

Stereotypical Willingness to Report Pain (self-opposite)

Heat

Threshold,

tolerance

−0.28

−0.41, −0.27, −0.21

Reidy et al. (2009)USA195 (65/130)HMIElectrical shockTolerance0.40
Alabas et al. (2012)Libya114 (56/58)

GREP/Self-Sensitivity, Self-Endurance

Self-Endurance

CPT

Threshold

Tolerance

−0.32, 0.23

0.23

3.2.1 Masculine and feminine personality trait scales

Three studies used the BSRI (Otto and Dougher, 1985; Fillingim et al., 1999; Myers et al., 2001), one study used the PAQ (Dixon et al., 2004), two studies used the extended PAQ (EPAQ, Sanford et al., 2002; Thorn et al., 2004) and one study used the Hyper-Masculinity Index (HMI, Reidy et al., 2009). The BSRI is a 7-point scale measuring an individual's self-report of masculine or feminine characteristics based on sex-typed social desirability traits (Bem, 1974). The PAQ is three 8-item scales that measure socially desirable and stereotypically characteristic traits which was extended to capture the undesirable components of masculinity and femininity traits (EPAQ, Spence et al., 1974). The HMI is 30 forced choice items evaluating conformity to traditional male gender role (Mosher and Sirkin, 1984).

Most studies reported a relationship between gender role and pain sensitivity response. Otto and Dougher (1985) found that pressure pain threshold was significantly correlated with Masculinity–Femininity among men but not women. Reidy et al. (2009) found that pain tolerance of electrical shock correlated positively with Hyper-Masculinity (p = 0.001). Thorn et al. (2004) found that cold pain tolerance could be predicted from Masculinity–Femininity scores, although Sanford et al. (2002) found that only the Femininity scale showed a significant negative correlation with cold pain tolerance. Dixon et al. (2004) found that only Masculinity–Femininity correlated with cold pain tolerance, pain intensity and pain unpleasantness (p < 0.05). In contrast, Myers et al. (2001) reported no significant correlations between Masculinity or Femininity with cold pain responses.

3.2.2 Gender stereotypes specific to pain scales

Four studies used the GREP (Wise et al., 2002; Robinson et al., 2004; Defrin et al., 2009; Alabas et al., 2012). The GREP consists of twelve 100-mm visual analogue scales assigned to measure three dimensions of sex-related stereotypic attributions: Pain Sensitivity, Pain Endurance and Willingness to Report Pain (Robinson et al., 2001). Pain Sensitivity describes a respondent's expectation of how sensitive they are to a noxious stimulus when compared with a typical man and a typical woman. Pain Endurance describes the respondent's expectation of how much pain they are prepared to endure compared with a typical man and a typical woman. Willingness to Report Pain describes the respondent's expectation of how willing they are to report pain compared with a typical man and a typical woman. In addition, for each dimension the respondent reported:

  • their perception of a typical man's response to pain compared with a typical woman's response to pain;
  • their perception of a typical woman's response to pain compared with a typical man's response to pain.

Nayak et al. (2000) used the 14-item Appropriate Pain Behaviour Questionnaire to assess perception of appropriate behaviour in men and woman. Pool et al. (2007) used a Gender Norms for Pain Tolerance Questionnaire consisting of a 9-point Likert scale to rate how an ideal man and ideal woman would respond to pain tolerance, and a Gender Group Identification Questionnaire consisting of two 9-point scales to measure the importance an individual ascribes to being like an ideal man or woman (Pool et al., 2007).

All of these studies reported at least one correlation between GREP and pain sensitivity response. Wise et al. (2002) found that Stereotypical Willingness to Report Pain predicted heat pain threshold (p < 0.01), pain tolerance (p < 0.01) and pain unpleasantness (p < 0.01), suggesting that participants who considered themselves as more willing to report pain than the typical man reported higher pain unpleasantness ratings and had lower pain thresholds and pain tolerances. Robinson et al. (2004) found that Stereotypical Willingness to Report Pain predicted the magnitude of temporal summation to thermal stimulation (p = 0.18). Defrin et al. (2009) found that heat pain threshold was negatively correlated with the expectation of sensitivity to painful stimuli of oneself compared with the opposite sex. Interestingly, heat pain tolerance predicted the expectation of sensitivity of oneself compared with others of the same and opposite sex. Heat pain tolerance also predicted the willingness to report pain of oneself when compared with the opposite sex. A study conducted by our team found that pain threshold was correlated with Self-Sensitivity (p < 0.001) and with Self-Endurance (p < 0.001) in a population of Libyan participants (Alabas et al., 2012). Self-Endurance was also significantly correlated with pain tolerance (p < 0.001). Nayak et al. (2000) found a significant correlation between cold pain tolerance and appropriate pain behaviour (p < 0.01). Pool et al. (2007) found that both men and women believed that the ideal man should tolerate more pain than the ideal woman. However, only men who strongly identified themselves with the male gender norm showed higher electrical pain tolerances.

3.3 Meta-analysis

Two studies did not report a correlation coefficient or regression equation (Fillingim et al., 1999; Pool et al., 2007) and one study measured temporal summation to thermal pain rather than pain threshold, tolerance or pain ratings (Robinson et al., 2004) and were excluded from the meta-analysis. Ten studies (1505 participants; 669 men and 836 women) were included in the meta-analysis and reported correlations between gender role and response to noxious cold (Nayak et al., 2000; Myers et al., 2001; Sanford et al., 2002; Dixon et al., 2004; Thorn et al., 2004; Alabas et al., 2012), heat (Wise et al., 2002; Defrin et al., 2009), pressure (Otto and Dougher, 1985) and electrical stimuli (Reidy et al., 2009).

3.3.1 Masculine and feminine personality traits

Six studies (945 participants; 406 men and 539 women) reported correlations between masculine and feminine personality traits and response to noxious cold (Myers et al., 2001; Sanford et al., 2002; Dixon et al., 2004; Thorn et al., 2004), pressure (Otto and Dougher, 1985) and electrical stimuli (Reidy et al., 2009) (Fig. 2). There was a significant amount of heterogeneity across studies (Cochran's Q = 20.66; p = 0.001; I2 = 76%) so a random effects model was used. There was a significant positive overall correlation between masculine and feminine personality traits and pain threshold and tolerance (r = 0.17, z-test = 2.56, p = 0.01). This suggested that those participants who considered themselves more masculine and less feminine exhibited higher pain thresholds and tolerances. The effect size of the overall correlation was small.

Figure 2.

Forest plot of the correlation coefficients (with 95% confidence intervals) between masculine and feminine personality trait scores and pain sensitivity. Main effect data are shown in the bottom row. CI, confidence interval.

3.3.2 Gender stereotypes specific to pain

Four studies (560 participants; 263 men and 297 women) reported correlations between gender stereotypes specific to pain and response to noxious heat (Wise et al., 2002; Defrin et al., 2009) and cold (Nayak et al., 2000; Alabas et al., 2012) (Fig. 3). A significant amount of heterogeneity across studies was observed (Cochran's Q = 13.46; p = 0.004; I2 = 77%) so a random effects model was used. There was a significant negative overall correlation between gender stereotypes specific to pain and pain threshold and tolerance (r = −0.41, z-test = −4.63, p < 0.001). This suggested that individuals who considered themselves less sensitive than the typical woman showed higher pain thresholds and tolerances. The effect size of the overall correlation was moderate.

Figure 3.

Forest plot of the correlation coefficients (with 95% confidence intervals) between gender stereotypes specific to pain scores and pain sensitivity. Main effect data are shown in the bottom row. CI, confidence interval.

3.3.3 Pain ratings and gender role scales

Only three studies (570 participants; 242 men and 328 women) reported correlations between gender role and response to noxious heat (Wise et al., 2002) and cold (Dixon et al., 2004; Thorn et al., 2004). A significant amount of heterogeneity across studies was observed (Cochran's Q = 13.46; p = 0.004; I2 = 77%) so a random effects model was used. There was a significant positive correlation between gender role and pain ratings (r = 0.33, z = 3.41, p = 0.001). This suggested that study participants who were more willing to report pain than the typical man reported higher pain intensity and unpleasantness. Study participants who had higher values in emotional vulnerability scales also reported higher pain intensity and unpleasantness. The effect size of the overall correlation was moderate.

3.4 The effect of the sex of the experimenter on pain responses

Only three studies evaluated the effect of sex of the experimenter on pain measurements. Both Otto and Dougher (1985) and Thorn et al. (2004) found no effect of the sex of the experimenter on pain outcomes. Fillingim et al. (1999) found that participants reported lower pain thresholds when tested by a female investigator, although this just failed to reach statistical significance (p = 0.055). We decided to search the literature for additional studies using the original search strategy with the addition of the following keywords: Sex AND experimenter, OR Gender AND experimenter, AND Pain. Studies were screened for eligibility in a similar manner to that previously described.

Fourteen studies were eligible for analysis, including the three studies from our review of gender role (Table 2). The effect size (d) of the difference in pain response according to the sex of experimenter was calculated for each study. If t or F statistics were reported these were converted to d. We did not calculate the pooled effect size (d) of the effect of the sex of the experimenter on pain response because of variations in pain outcomes, statistical analyses and presentation of data between the studies. Only one study provided means and standard deviations (Tashani et al., 2010). Effect size values of 0.20 were taken as small, 0.50 as medium and 0.80 as large (Cohen, 1992).

Table 2. Characteristics of the studies that examined the effect of sex of the experimenter on pain outcomes (n = 14).
StudyCountrySample size (M/F)Pain stimulusPain measurementOutcomeEffect size (d)
  1. CPT, cold pressor test.
  2. –: No available data.
Otto and Dougher (1985)USA80 (40/40)Pressure painPain sensitivityNo significant effect
Fillingim et al. (1999)USA209 (92/117)Heat painPain sensitivity

No significant effect

Individuals showed lower pain threshold when tested by female investigator

0.22
Levine and De Simone (1991)USA68 (35/33)CPTPain ratings

Men reported lower pain ratings when tested by a female investigator

Women had higher pain ratings when tested by a male investigator

0.61

0.24

Feine et al. (1991)Canada40 (20/20)Heat painPain ratingsNo significant effect0.12
Bush et al. (1993)USA47 (23/24)Heat painPain ratings

No significant effect

Individuals reported less pain when tested by an investigator of a different sex

Carter et al. (2002)USA80 (40/40)Heat painPain sensitivityParticipants demonstrated longer pain tolerance when tested by female experimenter0.47
Kallai et al. (2004)Germany160 (80/80)CPT

Pain threshold

Pain tolerance

Pain intensity

No significant effect

Individuals tolerate more pain when tested by investigator of opposite sex

Individuals expressed stronger pain when tested by a female investigator

0.32

0.36

Essick et al. (2004)USA34 (17/17)ThermalPain sensitivityNo significant effect
Gijsbers and Nicholson (2005)UK64 (32/32)Pressure painPain threshold

Men had higher pain threshold when tested by an opposite sex investigator

No such interaction was found for women

0.61

0.025

Thorn et al. (2004)USA235 (90/129)CPT

Tolerance

Expressed pain

No significant effect

No significant effect

Jackson et al. (2005)Australia

91 (34/57)

126 (50/76)

CPT

Pain tolerance

Pain ratings

No significant effect

No significant effect

0.20

0.17

Weisse et al. (2005)USA343 (156/187)CPT

Pain tolerance

Pain ratings

Not calculated

Not calculated

Aslaksen et al. (2007)Norway64 (32/32)Heat painPain ratingsMen reported lower pain ratings when tested by a female investigator0.56
Tashani et al. (2010)Libya58 (29/29)CPT

Pain sensitivity

Pain ratings

No significant effect

Individuals reported higher pain intensity when tested by an investigator of opposite sex

0.24

0.77

Seven studies found no effect or interaction for sex of the experimenter on the response to noxious stimuli of male and female participants (Otto and Dougher 1985; Feine et al., 1991; Bush et al., 1993; Fillingim et al., 1999; Essick et al., 2004; Thorn et al., 2004; Jackson et al., 2005). In two studies that manipulated the experimental environment by asking experimenters to dress in a manner that emphasized their gender roles, male participants who were tested by a female investigator reported significantly lower pain ratings (Levine and De Simone, 1991) and higher pain thresholds (Gijsbers and Nicholson, 2005) compared with men who were tested by an investigator of the same sex. No such interaction was found for women in these studies. Kállai et al. (2004) found that both men and women showed higher cold pain tolerances when tested by an experimenter of the opposite sex and expressed stronger pain when tested by a female investigator. Carter et al. (2002) found that participants tolerated more pressure pain only when tested by a female investigator. Aslaksen et al. (2007) found that men reported lower pain ratings when tested by female investigators compared with testing conducted by male experimenters. Tashani et al. (2010) reported that pain intensity rating was higher in both men and women in the presence of an investigator of the opposite sex.

4. Discussion

This meta-analysis of 13 studies found that gender role was related to response to experimentally induced pain in healthy human participants. Masculinity was positively associated with pain threshold and pain tolerance and this finding is consistent with previously published narrative reviews (Fillingim et al., 2009; Mogil and Bailey, 2010). Our findings support claims from gender schema theory that learned masculinity encourages stoicism (Sanford et al., 2002) and encourages displays of withstanding pain (Myers et al., 2003). We found that femininity was associated with greater sensitivity to painful stimuli and this may be one of the factors contributing to a greater proportion of women presenting with clinical pain (Mogil and Bailey, 2010).

Our review provides evidence that gender role is related to pain sensitivity response in healthy human participants. The relationship between gender role and pain sensitivity response has also been explored in populations of pain patients using masculine and feminine personality trait scales. Ali et al. (1998) found a positive correlation between the feminine scale of the PAQ and illness behaviour in irritable bowel syndrome patients. It has been shown that femininity was associated with poorer health outcomes among rheumatoid pain patients (Trudeau et al., 2003) and that weak masculine identity measured using the BSRI was associated with an early stage of chronic prostatitis-like symptoms (Ku et al., 2002). Higher masculinity measured using PAQ was shown to predict severe post-myocardial infraction chest pain (Helgeson, 1991).

We found that the overall effect size of the correlation coefficients between masculine and feminine trait scores and pain sensitivity was small whereas the overall effect size of the correlation coefficients between gender stereotypes specific to pain scores and pain sensitivity was moderate. This suggests that when assessing the role of gender on pain response, larger effect sizes are detected using instruments that capture gender stereotypes specific to pain. Robinson et al. (2001) and Defrin et al. (2009) suggest that instruments assessing gender stereotypes specific to pain reflect different constructs than instruments assessing gender role in general. For example, gender stereotypes specific to pain do not reflect masculinity and femininity in a broader sense. There is a dearth of research on the relationship between masculine and feminine personality trait scales and pain-specific gender scales and we found no studies in our sample that used both types of scales. This would be a useful avenue for future research.

We found that pain report was positively correlated with scales measuring Emotional Vulnerability (Dixon et al., 2004; Thorn et al., 2004) or Willingness to Report Pain (Wise et al., 2002). In other words, individuals who were more emotionally vulnerable or more willing to report pain expressed higher pain intensity and pain unpleasantness ratings. As this finding was based on three heterogeneous studies, we remain cautious about our interpretation. A study by Forsythe et al. (2011) using cold pressor pain found that there were race and sex differences for pain appraisals and catastrophizing and that the sex of participant affected affective-motivational aspects of pain more than sensory-discriminative aspects of pain. We combined sensory (threshold) and affective (tolerance) outcomes for pain for meta-analysis and could not verify these findings. Nevertheless, it has been suggested that social learning influencing affective aspects of pain is more likely to contribute to sex differences in pain response than physiological differences (Robinson et al., 2001).

Greenspan et al. (2007) have recommended that the sex of the investigator should be reported in all laboratory studies assessing pain sensitivity response, yet only three of the 13 studies in our review examined the effect of the sex of the experimenter. We found additional 11 studies that had focused on sex differences in pain sensitivity and the sex of the experimenter but had not measured gender role response. The effect sizes for the influence of the sex of the experimenter on pain sensitivity response ranged from small (d = 0.12, Feine et al., 1991) to moderate (d = 0.77, Tashani et al., 2010), suggesting that the sex of the experimenter influenced sex differences in pain response. A comprehensive systematic review is needed to confirm or otherwise our preliminary post hoc analysis of the interaction between the sex of the experimenter and pain outcome.

The ethnicity and cultural background of participants and experimenters also influences pain response in laboratory (Campbell et al., 2005; Watson et al., 2005; Weisse et al., 2005; Rahim-Williams et al., 2007; Forsythe et al., 2011) and clinical settings (Edwards et al., 2001; Green et al., 2003). Recently, we have shown that Libyan women had higher pain sensitivity to cold noxious stimuli than Libyan men, and that both Libyan men and women had heightened pain sensitivity compared with their counterparts in Europe (Tashani et al., 2010; Alabas et al., 2012). Hsieh et al. (2011) found that Chinese participants exposed to Chinese experimenters and Chinese language reported higher affective pain and more non-verbal pain behaviour than Chinese participants exposed to Euro-Canadian experimenters and English language. Greater attention needs to be given to the reporting of ethnic concordance between investigators and participants in experimental studies and clinical trials.

4.1 Limitations

Heterogeneity was present between studies. This may have been due in part to combining threshold and tolerance data to improve statistical power. Other sources of heterogeneity were variations in the demographic characteristics of participants and gender role measurement methods. We used a random effects model to account for this heterogeneity and remain cautious about our inferences because of the small number of studies in our analysis. Publication bias may also have been present although it was not possible to test because of too few studies (Borenstein, 2005).The impact of variations in the sex of the experimenter and other ethno-sociocultural factors influencing pain outcomes in the laboratory setting could not be accounted for between studies, and there clearly needs to be better documentation of such factors in study reports.

4.2 Summary and conclusion

In conclusion, individuals who considered themselves more masculine and less sensitive to pain than the typical man showed higher pain thresholds and tolerances. Gender stereotypes specific to pain scales showed stronger associations with sex differences in pain sensitivity response than masculine and feminine personality trait scales.

Author contributions

All authors contributed equally to the study protocol, search strategy and screening of articles. OAA and OAT extracted data with MIJ and GT acting as arbiters for any disagreement. OAA and OAT conducted the meta-analysis. First draft was written by OAA, OAT and MIJ. All authors contributed to discussion section.

Appendix 1: Detailed overview of the search queries

Search query 1: MEDLINE

  • #1 gender role [ti and ab]
  • #2 femininity [ti and ab]
  • #3 Masculinity [ti and ab]
  • #4 Sex differences [ti and ab]
  • #5 Gender differences [ti and ab]
  • #6 pain [ti and ab]
  • #7 OR/ 1-5
  • #8 6 AND 7

Search query 2: EBSCO

Including the following database: Academic Search complete, CINAHL, SportDiscus, and PsycoArticles and PsychoInformation

  • #S1 gender role [ti]
  • #S2 gender role [ab]
  • # S3 femininity [ti]
  • #S4 femininity [ab]
  • #S5 masculinity [ti]
  • #S6 masculinity [ab]
  • #S7 gender differences [ti]
  • #S8 gender differences [ab]
  • #S9 sex differences [ti]
  • #S10 sex differences [ab]
  • #S11 pain [ti]
  • #S12 pain [ab]
  • #S13 S1 AND S2
  • #S14 S3 AND S4
  • #S15 S5 AND S6
  • #S16 S7 AND S8
  • #S17 S9 AND S10
  • #S18 S11 AND S12
  • #S19 OR/ S13-S17
  • #S20 S18 AND S19

Search query 3: Medical Subject Headings (MeSH)

(gender identity [MeSH] OR femininity [MeSH] OR masculinity [MeSH] OR sex characteristics [MeSH]) AND (pain [MeSH] OR pain measurement [MeSH] OR pain threshold [MeSH])

Appendix 2: Excluded studies (ordered by study year)

IDStudyReason of exclusion
 1Lash et al. (1990)No experimental pain measurements were taken.
 2Feine et al. (1991)No gender role instrument was used.
 3Thomas and Rose (1991)No gender role instrument was used.
 4Gillespie and Eisler (1992)No experimental pain stimulus was used.
 5Osman et al. (1993)No experimental pain stimulus was used.
 6Lautenbacher and Rollman (1993)No gender role instrument was used.
 7Carlson et al. (1998)No gender role instrument was used.
 8Fillingim et al. (1999)No gender role instrument was used.
 9Sheffield et al. (2000)No gender role instrument was used.
10Sullivan et al. (2000)No gender role instrument was used.
11Osman et al. (2000)No experimental pain stimulus was used.
12Robinson et al. (2001)No experimental pain stimulus was used.
13Keogh and Herdenfeldt (2002)No gender role instrument was used.
14Jones and Zachariae (2002)No gender role instrument was used.
15Fillingim et al. (2002)No gender role instrument was used.
16Chesterton et al. (2003)No gender role instrument was used.
17Myers et al. (2003)No experimental pain stimulus was used.
18Robinson et al. (2003a)No experimental pain stimulus was used.
19Robinson et al. (2003b)No gender role instrument was used.
20Robinson and Wise (2003)No experimental pain stimulus was used.
21Jones et al. (2003)No gender role instrument was used.
22Lowery et al. (2003)No gender role instrument was used.
23Frot et al. (2004)No gender role instrument was used.
24Robinson et al. (2004a)No experimental pain stimulus was used.
25Robinson and Wise (2004)No gender role instrument was used.
26Logan and Gedney (2004)No gender role instrument was used.
27Edwards et al. (2004)No gender role instrument was used.
28Zacny and Beckman (2004)No gender role instrument was used.
29Sarlani et al. (2004)No gender role instrument was used.
30Jackson et al. (2005)No gender role instrument was used.
31Hobara (2005)No experimental pain stimulus was used.
32Fillingim et al. (2005)No gender role instrument was used.
33Keogh et al. (2005)No gender role instrument was used.
34Myers et al. (2006)Participants under 18 years old.
35Keogh et al. (2006)No gender role instrument was used.
36Garofalo et al.(2006)No gender role instrument was used.
37Pud et al. (2006)No gender role instrument was used.
38Aslaksen et al. (2007)No gender role instrument was used.
39Garcia et al. (2007)No gender role instrument was used.
40Well et al. (2008)No experimental pain measurements were taken.
41Keogh and Denford (2009)No experimental pain stimulus was used.
42Goodin et al. (2009a)No gender role instrument was used.
43Goodin et al. (2009b)No gender role instrument was used.
44Vierhaus et al. (2011)Participants under 18 years old.
45Fowler et al. (2011)No gender role instrument was used.
46Defrin et al. (2011)No experimental pain stimulus was used.

Ancillary