The role of the social environment for health is evident in public health science [9,10] and for centuries it has been apparent in medicine to include social determinants to understand disease occurrence, health outcomes and prevention [11]. However, how the social environment affects health often appears to be an unspecified black box which remains after investigators have controlled for a range of individual characteristics [9,12]. In this perspective it is important to investigate "upstream" influences on more proximal determinants of health [13]. Stress is suggested to be one of these upstream influences by which health is affected [14].

The physiological mechanisms that mediates stress and their contribution to illness and manifest disease is not fully understood. One plausible and much studied pathway in the stress response is that of the neuroendocrin Hypothalamic-Pituitary-Adrenocortical (HPA) axis, which in turn controls other physiological systems like immune and inflammatory processes. Negative strain, stressors, cause negative affective states, e.g. anxiety, that alter the production of Corticotropin-Releasing Hormone (CRH) in the paraventricular nucleus of the hypothalamus. In response to levels of CRH, Adrenocorticotropic hormone (ACTH) is released from the pitiuitary gland out to the circulating blood ultimately reaching the cortex of the adrenal gland effecting synthesisation of cortisol, the primary effector of HPA-axis activation in humans. High cortisol level has potential deleterious effects. For example it cause prolonged wound healing [15] as well as a significant degree of cognitive impairment [16]. This stress-response through the HPA-axis is thougt to be a major physiologial mechanism through which stressors influence disease risk [17,18].

Currently, the best way to biologically measure the stress-response is to assess the activity of the HPA-axis through levels of cortisol. Measuring cortisol concentrations in saliva is the most extensively used method of assessing cortisol in stress research as it has some advantages: it is non-invasive and thus less stressful, cost effective, and saliva can be collected by non-medical personnel in many different environments. Unfortunately, current methods of measuring cortisol, including concentrations in saliva, blood or urine only cover a spot time interval and cannot detect stress longitudinally or retrospectively. This drawback makes the assessment of cortisol levels largely dependent on the time and situation of sample collection: Firstly, cortisol levels vary with a natural pulsatile variation during the day (circadian rhythm), which makes the time of the day important [19,20]. Secondly, the levels may also be influenced by situational factors like psychological [21] or physical [22] stressors since they have the ability to elevate cortisol levels up to one hour after onset before returning to baseline [21].

The shortcomings of the present cortisol measuring methods call for a better procedure in determining extended and long-term exposure to stress. Hair has been used as a substrate for measuring environmental agents, drugs or toxins for a number of years [23] and quite recently the ability to retrospectively measure steroid hormones [24,25] including cortisol levels in hair [23,26] was shown. Initial studies on rhesus monkeys showed that cortisol levels in their hair increased when exposed to the stress of relocation and that the cortisol levels also increased in serum and saliva, respectively [27,28]. Later, a few initial studies on humans indicated that the level of cortisol in hair may reflect the activity of the HPA-axis [29,30]. Furthermore, correlations between specific stressors and cortisol in hair in selected groups have been shown. For example, neonates have higher cortisol levels in hair depending on the period of stay at the neonatal intensive care [31]. For a group of patients with chronic pain, higher hair cortisol levels were found compared to a control group [32]. In a proof-of-concept study cortisol levels increased in scalp-near hair segment of pregnant women that corresponded to the well-known increase of blood cortisol levels in the third trimester of pregnancy [33]. An interesting study showed that hair cortisol levels representing three months prior to an acute myocardial infarction among 56 men were significantly higher than in hair from 56 controls who were hospitalised for other indications. This suggests that long term stress may be a contributing factor to myocardial infarction [34]. Recently, a study showed that long-term unemployed individuals had higher hair cortisol levels compared to controls [35].

Cortisol in hair might serve as a longitudinal biomarker of chronic stress and in public health research that would be of great value. However, it is warranted to validate the measurement of cortisol in hair in comparison to other indicators of extended stress. This present investigation serves as a pilot study to develop the methodology of cortisol in hair analysis and to test if cortisol in hair could be measured even in young healthy adults and to relate the cortisol levels to other indicators of stress and health for this group.

In a multiple regression model, indicators of perceived stress, perceived health and experience of serious life events were included. Only the indicator of experiences of serious life events the last three months exhibited an independent and significant (p = 0.042) explanation of increased levels of cortisol in hair.

There are some reports during recent years where this novel method to measure cortisol in hair has been applied [29,30,32,34,35]. Although our laboratory methodology is in concordance with these, we have in this report managed to reduce the amount of human hair necessarily for reliable analysis to 5 mg instead of previously reported 25 to 45 mg [33]. This is of great value as we plan further studies in very young children.

The participants in this study were all young adults and university students, and could be considered as a relatively healthy cohort. Nevertheless, the measurement of cortisol concentrations in hair extracts retrospectively detected stress exposure i.e. experiences of serious life events. Cortisol levels thus seems to be related to serious life events even among young healthy adults, and not only in groups of ill adults who have some sort of known stressor compared to a control group. In one way this approach is a harder task than to be able to spot a difference between an exposed experimental group and a control group. The variable Serious Life Event is something that has happened or not, and it is usual to simply ask for such an event with a Yes/No question. However, one should be aware that the way in which a serious life event is perceived is a rather subjective factor. The influence of serious life events on physical and mental health has a long research tradition for over forty years from the early reports by Holmes and Rahe [40]. Their initial publication 1967 that tried to rate life events was criticized on methodological grounds concerning the usefulness of generalized magnitude ratings of life events. Later this quite simple measurement that is used in this report has been widely used in numerous publications [41]. When stressors like negative life events, and also chronic strains and traumas are measured comprehensively, their damaging impacts on health are substantial [42]. In a recent study, hair samples of PTSD patients had higher cortisol levels than those of traumatized controls [43], which support our findings that serious (traumatic) life events raises hair cortisol levels although traumatic load was assessed with a checklist of traumatic event types, not coded as yes or no. In this study there was also a positive correlation between number of different traumatic event types and hair cortisol levels.

We note that a surprisingly high number of students reported serious life events, 20 of 95 subjects. In this age group, possible explanation could be the illness/death of grandparents but also the divorcing parents.

In our report, the perceived stress measurement PSS [36] was weakly negatively correlated to increased cortisol levels but in the multivariate analyses the indicator of serious life events outperformed this indicator. Also, the PSS scale has been validated for a recall period of 4 weeks, but was used to assess 3 months of stress in the current study, which is a limitation. According to our results, the experience of serious life events seems to be the key factor and more important in raising cortisol levels in hair than perceived stress. There is always a risk for recall and reporting bias for retrospective self-reports [44]. Possibly, this is more conceivable for retrospective self-reports of perceived stress than for retrospective reports of actual life events. Nevertheless, a key point is that we do not really know the concordance between retrospective measured subjective and biological stress reactions.

A possible limitation in this study is that there is a risk of confounders like adrenal diseases not known to the participants, even though this is rather unlikely considering the rarity of the diseases. Almost one third of the participants (29%) reported some kind of longstanding medication but the use of pharmaceuticals were not correlated to levels of cortisol in hair in the present study. However, external contamination of the hair through e.g. use of cortisone-containing medications is unlikely but cannot fully be excluded. Coloured or bleached hair did not affect the concentration of cortisol in our analysis, although possible removal of some cortisol by repeated washing should also be considered, shown by Sauvé et al 2007 [29]. One could argue that the likelihood of removal of some cortisol from hair by repeated hair shampooing/washing would be low considering that studies have shown that washing with a solvent in the assay process seems to be an unnecessary process since it removes only small parts of the of the total hormone extracted [23,26]. But there are also studies showing that there is a "wash out" effect [31].

There are a hypothesis that stress- induced cortisol secretion may contribute to central fat and there by be a link between stress and risk for disease [45]. BMI was not asked for in this study, although it was not the aim of this study we can comment that the subjects were healthy young adults studying Health Sciences and no one (judged by eye sight) had class 1 obesity (BMI >30) or above. One student was very underweight; this was also the student who we later found out to have anorexia nervosa. This case also had extremely high levels of cortisol in hair, and was omitted from the analysis, as stated earlier.

Present biological cortisol markers like saliva, urine and blood suffer from the drawback of only covering spot time periods of up to 24 hours. Therefore, there is a need in stress research for biological markers reflecting long term and extended exposure to stress. Cortisol in hair has the potential of becoming a new indicator for stress exposure of major life stressors over periods of months and could even be used in clinical settings. An obvious advantage of analysing cortisol in human hair is the ease of sample collection and transportation as well as being non-invasive. It is also easy to store samples since hair does not decompose like other body fluids or tissues.