Main

Chronic pain is the most prevalent human health problem, affecting over one-quarter of the world's population, and is rising in incidence as the population ages1. Women are greatly overrepresented among patients with chronic pain2,3,4. A number of common chronic pain syndromes can only occur in women (including endometriosis, vulvodynia and menstrual pain). Furthermore, some highly prevalent chronic pain syndromes that are found in both sexes (including chronic fatigue syndrome, fibromyalgia, interstitial cystitis and temporomandibular disorder) occur overwhelmingly more often (in more than 80% of cases in which treatment is sought) in women. Last, the chronic pain syndromes with the highest prevalence overall — headache, migraine, low back pain, neck pain and knee pain (mostly osteoarthritis) — all have marked female predominance. (Fig. 1 shows estimates of excess female prevalence in large epidemiological studies of pain5). There are, of course, male-specific (chronic prostatitis), male-dominated (gout) and male-prevalent (cluster headache) pain states, but these tend to be less prevalent overall.

Figure 1: Sex differences in prevalence of chronic pain syndromes.
figure 1

The epidemiological data presented here are taken from Ref. 5 but were derived from large, general population-based (self-report) studies conducted via surveys or telephone interviews (see Supplementary information S1 for full citations). Data from clinical studies are not included because of bias associated with the fact that health care services are used more by women than by men. Each blue bar represents the excess prevalence of the pain condition in women reported in a single epidemiological study; the red bar to the right represents the median excess prevalence within the category. The definitions of pain prevalence (including current pain, 1–12-month pain duration or chronic pain) differed widely across the studies, but the definition in each was the same for males and females, and thus sex differences in prevalence can be compared directly. In some cases, the male–female difference scores plotted are averages of multiple prevalence estimates. The average of the category medians is 5.5% excess female prevalence. *Indicates the average of different age ranges. Indicates the average of different durations. §Indicates the average of different numbers of pain-related symptoms. ||Indicates the average of different pain locations).

Although the underlying reasons for the sex bias observed in pain are still hotly debated, the fact that clinical pain is more prevalent in women is well beyond doubt. This epidemiological reality, however, has been and continues to be largely ignored by the pain research community. Many jurisdictions now insist that clinical studies are performed on both sexes, but no such mandate exists for preclinical research. A recent literature search demonstrated male bias in experimental subject choice in eight out of ten biological disciplines6, and a huge male-orientated bias can also be observed in the preclinical pain literature7 (Fig. 2). Although much more attention has been paid to the topic of sex and/or gender differences in pain in the past few decades5 (and sex differences in neuroscience more generally8), there is little evidence to suggest that female mice and rats are becoming more popular as research subjects overall. The omission of female animals from preclinical experiments can have serious implications, as some sex differences are qualitative rather than quantitative, and failure to appreciate them can lead to either missing biological phenomena entirely or overgeneralization of findings (Box 1).

Figure 2: Subject choice and reporting practices in preclinical studies of pain.
figure 2

Data are from a survey of papers published between 1996 and 2005 reporting awake, behaving non-human animal pain experiments7. Seventy-nine percent of those experiments used male rodents only. Of studies using both sexes, most featured no discussion of whether sex differences were observed or not. In 3% of studies, the animals' sex was not even reported. No convincing trends in subject characteristics were observed within this 10-year period in any category (not shown).

The aims of this article are to re-evaluate the evidence indicating that women are more sensitive to pain than men, to examine factors that complicate strong conclusions as to the nature of this sex difference and to detail the various underlying mechanisms that have been proposed to explain it. The article will not attempt to comprehensively review the underlying mechanisms, but will attempt to categorize the types of explanations that have been put forward.

Are women more sensitive to pain?

The predominance of females among patients with chronic pain might be explained in one of three non-mutually exclusive ways. First, it is possible that women simply seek out health care services at higher rates than do men and/or are more willing to report pain on surveys than men, and thus will be tallied higher in epidemiological studies of various types. Second, it is possible that women have higher susceptibilities to common chronic pain syndromes than men and thus will be more likely to develop conditions that feature pain as a symptom. Last, it is possible that women have a greater sensitivity to and/or a lower tolerance of pain than men, leading to higher percentages of women crossing the threshold at which experienced pain rises to the level of a diagnosed 'pain syndrome'. In this case, pain levels in pain syndromes experienced by both sexes would be expected to be highest in women. Note that higher pain sensitivity in women might be due to biological sex differences in ascending pain transmission pathways, descending pain modulation pathways and/or any number of psychological phenomena that affect pain. There are also various possible explanations for apparent sex differences in analgesic responsivity (for example, to opioids)9; these could be due to differential drug pharmacokinetics or pharmacodynamics or simply to different starting pain levels.

It has been surprisingly difficult to determine which of the three scenarios outlined above provides the most convincing explanation for sex differences in chronic pain prevalence. Women do use health care services at rates exceeding those of men for painful and non-painful disorders alike10. There are multiple reports that suggest that pain levels within chronic pain syndromes are markedly higher in women than men, including a recently published review of 11,000 electronic medical records of men and women with the same diagnosis11. However, the real test of the hypothesis that pain sensitivity itself is higher in women requires controlled laboratory experimentation.

Laboratory studies of sex differences. Many studies of sex differences in pain sensitivity have been conducted (for recent examples, see Refs 12,13,14), and many reviews and meta-analyses of these studies exist5,9,15,18,19. As might be expected with a biological domain as heterogeneous as pain, the picture emerging from these studies is complex. Some studies show notable sex differences in pain sensitivity, whereas others do not. A persistent concern is that laboratory studies of pain sensitivity between the sexes are confounded by human subject–experimenter interactions involving gender role expectations, although contradictory data have been generated relating to this issue5,20. Overall, sex differences seem to be easier to evince in certain pain modalities than in others (such as in heat or pressure-induced pain compared with ischaemic pain), using certain dependent measures (such as tolerance as opposed to pain intensity or unpleasantness ratings) and at certain time points (such as early rather than late after introduction of the noxious stimulus), and exhibit small-to-moderate effect sizes (see Refs 5,15 for comprehensive recent reviews).

What has struck many researchers, however, is the fact that when differences are observed, they almost unanimously show that women have a higher sensitivity and lower tolerance to pain than men, report higher pain ratings and have a greater ability to discriminate among varying levels of pain. Nonetheless, a true consensus has been hard to reach; a consensus working group published a report in 2007 in which a direct statement that women were more sensitive to pain than men in the laboratory was conspicuously absent19. A recent review suggests that the informal consensus that women are more sensitive to pain is actually due to a bias related to participant selection criteria and an overemphasis on pain measures showing sex differences rather than ones that do not15. From a re-analysis of the relevant data (Box 2), I conclude that this critique is too conservative in its definition of what constitutes a sex difference; the evidence is actually overwhelmingly in support of the contention that women are more sensitive to pain, although the size and importance of this sex difference could be debated.

In addition to sex differences in pain, sex differences in response to opioid analgesics have also been intensely studied. A recent meta-analysis concluded that morphine is moderately more efficacious in women than in men in both clinical (largely patient-controlled analgesia) and experimental studies; however, the picture becomes far less clear for other μ-opioids and especially for mixed μ- and κ-opioid-acting compounds (such as butorphanol, pentazocine and nalbuphine)9. In contrast to the animal literature on sex differences in pain21, which generally supports the informal consensus of higher pain sensitivity in females, the animal literature on sex differences in opioid analgesia reaches a conclusion that dramatically opposes the human situation, with most studies showing increased μ-opioid analgesia in male rodents compared with female animals22,23. No explanation of this apparent species difference has been proposed. Of note, the single study examining this issue in non-human primates (Macaca mulatta) reported that male monkeys exhibited more analgesia from low-efficacy μ-opioid and κ-opioid agonists than did female animals24.

Influence of hormones. Determining whether women have different sensitivity to pain or analgesia compared with men is complicated by the hormonal cyclicity of women; the differential sensitivity might only be evinced in certain phases of this cycle. Much attention has been paid to this issue, although the relevant studies have been criticized for various methodological problems19,25. Some clinical pain conditions in women vary with the menstrual cycle25. A meta-analysis of experimental studies revealed that women have a higher pain threshold and tolerance during the follicular phase (with small-to-moderate effect sizes) in every stimulus modality except electrical pain, in which the highest pain thresholds were associated with the luteal phase26 (note that studies following the publication of this meta-analysis have produced conflicting results5,20). A more recent narrative review that used a different definition of menstrual phases compared with that used in the meta-analysis concluded that increased reactivity to pain occurs peri-menstrually and mid-cycle27.

Even one of the simpler relevant questions remains a matter of ongoing debate: are oestrogen and progesterone pronociceptive or antinociceptive? There are many extant reports of pain modulation in both directions by gonadectomy, oestradiol or hormone replacement therapy (with or without a progestin). Generally, if effects are seen, gonadectomy increases pain sensitivity, especially for acute pain28. By contrast, oestradiol and progesterone given to ovariectomized animals generally cause hypoalgesia28,29, if effects are observed. Human studies of clinical pain are even more complex, with a multitude of findings in both directions as well as null results. Craft29 speculates that this complexity may arise from the widespread distribution of oestrogen receptors in pain-relevant loci, possibly biphasic dose–response relationships, methodological inconsistencies and the ignored modulatory influence of other steroids such as testosterone, oestriol and oestrone.

Complicating interactions. The situation is complicated further by findings from animal studies that show robust interactions between sex and other factors in relation to pain sensitivity. The primary factor among these is genotype. Studies in mice21,30,31,32 and rats32,33,34,35,36 have demonstrated that sex differences in pain and analgesia can be demonstrated in certain strains but not others. The effects of gonadal hormones on pain-related traits are similarly strain-dependent21,37. Sex–strain interactions undermine the entire concept of sex differences in that they (at least partially) moot the question: which sex is more sensitive to pain? A more sophisticated perspective is that sex and genetic background (and their interaction) are both simply components of inter-individual variability that need to be explained. As might be expected given this interaction, genes (quantitative trait loci) with sex-dependent effects on pain trait variability have been uncovered38,39,40,41,42. A recent study uncovered a three-way interaction in both mice and humans between sex, genetics (AVPR1A genotype (AVPR1A encodes the vasopressin 1A receptor)) and acute stress43.

Recent mouse studies have revealed another surprising factor that interacts with sex to modulate pain: social interaction. In these studies, mice were placed in observational apparatuses in which some of the animals were in pain (which was induced by intraperitoneal injection of acetic acid) and some were not, and both mouse location and pain behaviour were measured. Unaffected female but not male mice approached cage mates (but not strangers) that were in pain and spent excess time in physical proximity to their hurting familiar44. This social approach appears to be an effective analgesic, as a negative correlation was obtained between contact time and pain behaviour44. Social interaction can also affect pain behaviour in male mice. When mice were tested in a dyad in which only one was injected with acetic acid, either stress-induced analgesia or stress-induced hyperalgesia was observed, depending on the threat level dictated by facets of the testing situation45; these effects are only seen in unfamiliar male mice.

Last, in rodents, sex has been shown to interact with prenatal or neonatal inflammation46,47 and/or prenatal or neonatal stress48,49,50,51 to affect pain sensitivity in adulthood.

Potential underlying mechanisms

Although, as outlined above, the debate over the existence of sex differences in pain is not yet over, some researchers have turned their full attention to the task of uncovering mechanisms underlying such differences. There are three operationally defined types of sex differences: sexual dimorphism, in which some end point exists in only one sex (nursing behaviour, for example), or in two contrasting forms in each sex (such as copulatory behaviours); sex differences in which an end point is found on a continuum on which the male and female average differs; and sex convergence and/or divergence, in which the end point is the same in both sexes but the underlying neural mechanisms are different52. The distinction between the latter two types can be considered the difference between 'quantitative' and 'qualitative' sex differences. Although attention has mostly been paid to documenting quantitative sex differences in pain, a growing number of examples of qualitative differences in pain have been reported31,39,43,53,54,55,56,57,58,59,60,61,62,63,64, and these promise to be far more important in the long run. As a practical matter, analgesics are routinely titrated according to their effect, which will effectively mitigate any sex differences along with other sources of inter-individual variability. Convergence or divergence in mechanisms underlying pain modulation in the sexes, by contrast, has direct and important consequences for analgesic drug development.

Explanations of sex differences can be grouped into various categories (Box 3). For example, one can ask about the reason that sex differences exist (the ultimate cause) or the mechanisms underlying them (the proximate cause). Two hypotheses have been put forward regarding the ultimate causes of sex differences in pain and analgesia. One suggests that male and female mammals are under divergent adaptive pressure with respect to the evolution of pain modulatory circuitry owing to the presumably more common exposure to traumatic pain in males and visceral pain in females. However, there seems to be no extant data directly supporting this possibility. A second idea is based on the observation that the neural systems underlying lordosis behaviour and analgesia in rats have extensive anatomical and neurochemical overlap (for example, both feature important roles for opioid receptors in midbrain loci such as the periaqueductal grey). The theory postulates that pain inhibitory circuitry may thus have 'piggybacked' on top of previously existing reproductive circuitry in the midbrain and brain stem65. According to this view, the reason for sex differences in analgesia is simply that there are already sex differences in reproductive behaviour.

Of perennial interest is the extent to which sex differences are due to sex-specific physiologies that are determined developmentally or to shared physiologies that are reversibly modulated in adulthood by gonadal steroids, the levels of which vary dramatically between the sexes. Evidence for organizational versus activational effects of gonadal hormones on pain traits is mixed, with some studies showing robust effects of neonatal male castration and female testosterone-induced masculinization on pain behaviour66,67,68,69,70 and others being strongly supportive of the primacy of circulating hormone levels in the adult71,72,73,74,75,76,77. The classic organizational versus activational dichotomy has recently been supplemented by the realization that direct genetic effects are possible as well: traits such as pain might be affected by genes on the Y chromosome, X-linked genes escaping inactivation, parentally imprinted genes and/or allelic mosaicism78. Using the 'four core genotypes' model79, such genetic effects on acute thermal and tonic inflammatory pain have been observed80,81, although the precise mechanism underlying the increased pain sensitivity of XX-containing (but not necessarily gonadally female) mice is not yet known.

A number of more specific explanations for sex differences in pain have been put forward, spanning the range of the sociological, psychological and biological sciences (Box 3). Some of these explanations purport that sex differences in pain are essentially an artefact. For example, it has been asserted that the on pain scales, the pain label “worst pain imaginable” is probably affected by childbirth experience82 such that the overall size of the pain scale is larger in most women than in men. If so, a woman's five out of ten numerical rating would actually represent higher pain perception than the same rating given by a man. Another example concerns gender role expectations, which are due to sex-specific socialization. Males may be discouraged from expressing pain behaviours, whereas females are 'permitted' to do so, which might lead to biased reporting in one sex and not the other83. Other explanations purport that sex differences in pain are secondary to known sex differences in some other experience (such as abuse), psychological state (such as anxiety) or strategy (such as coping) that is known to affect pain. For example, if depression is associated with worsened pain and women are more likely to experience depression, women will have more pain for a reason lying outside the core of pain physiology. In a number of studies, controlling for one of these variables completely abolished the observed sex differences84,85,86,87.

Explanations rooted explicitly in pain biology, either at the systems levelor neurochemical level, purport that aspects of neural processing of pain feature either quantitative or qualitative sex differences. Three such explanations — involving qualitative sex differences in the midbrain, spinal cord and the primary afferent — are currently more comprehensively documented than the others.

Multiple laboratories have observed that the midbrain–brain stem neural circuit subserving stress-induced analgesia, κ-opioid (and possibly μ-opioid) analgesia, morphine tolerance and morphine hyperalgesia in mice contains NMDA-type glutamate receptors (NMDARs) in males49,56,57,88,89,90 but not females; in many cases, females seem to use melanocortin 1 receptors (MC1Rs) instead39,54,91. In these studies, pharmacological antagonism with non-competitive NMDAR antagonists (for example, MK-801) blocks the phenomena in male but not female mice at any dose. In female subjects, genetic dysfunction or pharmacological antagonism of MC1R does not actually block the phenomena but instead renders them newly sensitive to blockade by MK-801. Similarly, ovariectomy leads to MK-801 sensitivity, whereas chronic oestrogen or acute progesterone treatment reinstates resistance to MK-801. The parsimonious interpretation of this body of research would suggest that two alternative pathways exist for opioid analgesia and hyperalgesia; one (the NMDAR pathway) normally used by males and one (the MC1R pathway) normally used by females. Depending on hormonal levels, females can access the 'male' system when their own system is compromised. Although most of the research has been carried out in mice, for κ-opioid analgesia this sex difference is seen in humans as well, with genetic dysfunction of MC1R leading to increased pentazocine analgesia in women but not men39.

Studies have documented that the neural circuitry subserving analgesia from sex steroids and morphine in the rat spinal cord show profound sex divergence60,61,92, and recent work suggests that the core of the sex difference stems from a bias in μ- and/or κ-opioid heterodimer expression59,93. These heterodimers — the formation of which is under regulation by spinal oestrogen synthesis93 — are vastly more prevalent in the female rat spinal cord than in the male spinal cord and are activated by endogenous dynorphin 1–17, which itself can be released by intrathecal injection of morphine, producing effects that are not seen in monomeric κ-opioid receptors59. These results may explain the known dependency of morphine analgesia in female (but not male) mice on the κ-opioid receptor gene41.

Robust regulation of pain-relevant signal transduction pathways, especially those involving protein kinases Cɛ, Cδ and A, by sex hormones has been observed63,94,95,96. Investigators have shown, for example, that hyperalgesia induced by the activation of β2-adrenergic receptors is dependent on protein kinase Cɛ in male but not female rats94; this effect is mediated by direct oestrogen inhibition of the peripheral nociceptor95.

Perhaps the biggest conceptual difficulty in explaining sex differences in pain and analgesia is that many of the proposed explanations seem fully able to account for observed differences. Can they all be simultaneously true? Of course, it is entirely conceivable that these explanations interact with each other. For example, if the feminine gender role dictates that pain will be more intense, that would be associated with expectancies for increased pain, which could engage pain facilitatory systems (much like a nocebo effect) and neurochemical elements associated with them (such as cholecystokinin and cannabinoid 1 receptors97,98) in a sex-dependent manner.

Future directions

The subfield of sex differences in pain is at an interesting juncture: some still doubt that there is anything to be studied or anything worth studying, whereas others are well into their search for underlying mechanisms. As mentioned above, most preclinical researchers simply ignore the issue entirely. Why are female rats and mice avoided? I believe that the simple combination of inertia and fear of oestrous cycle-related variability are sufficient explanations and that this fear is, in fact, unfounded. The sex-specific coefficients of variation (mean-corrected standard deviations) have been examined in large archival data sets using acute thermal and tonic inflammatory pain tests in mice7, and in fact it is male mice that generally display (nonsignificantly) higher variability. Whatever oestrous cycle-related variability there might be, dominance hierarchies (and associated fighting) in cages of male mice provide a male-specific source of variability that may be equal or greater in magnitude.

Although I contend that the evidence for both clinical and experimental sex differences is overwhelming, it is true that the clinical impact of this research is very limited at this point in time20. I believe this is because most investigators continue to study quantitative sex differences in pain instead of focusing on the far more important qualitative differences. Regardless of the overall sensitivity of men and women to pain and pain inhibition, ignoring sex-specificity in neural circuits subserving these phenomenon will continue to complicate drug development efforts. There are already too many examples of qualitative sex differences in pain biology to simply assume that similar biological mechanisms exist in male and female subjects. Wilful blindness to sex differences risks both over-generalizing findings made in male subjects and missing the opportunity to discover female-specific mechanisms. The Toll-like receptor 4 (TLR4) sex difference described in Box 1 is a perfect example. Those findings predict that spinal TLR4 antagonism, even if it is successful in blocking pain in men, would be ineffective in women. However, findings from my group's research necessitate the existence of an as-yet-uncharacterized alternative (non-TLR4) mechanism. Given that women with chronic pain greatly outnumber men, to ignore female-specific pain biology is to do a great ethical disservice to the majority of people with this condition.