Stress has long been recognised to be associated with asthma. Hippocrates stated that to prevent an asthma attack “the asthmatic should guard himself against his own anger” 1. Maimonides in his treatise of asthma suggested that “mental anguish, fear, mourning or distress” may cause asthma whereas “gaiety and joy” which “gladden the heart and stimulate the blood and mental activity” may have the opposite effect 2. In his treatise, H.H. Salter wrote that “asthma is essentially, and with perhaps the exception of a single class of cases, exclusively a nervous disease; the nervous system is the seat of the essential pathological condition” 3, and W. Osler referred to asthma as “a neurotic affection” 4. Thus, until the second half of the 20th century, asthma was predominantly viewed as a psychosomatic disorder in which emotional stress and imbalances in the nervous system were key factors in its aetiology. As a result, relief of anxiety was considered the main therapeutic intervention for “asthma nervosa”; as it has been referred to historically.
Although the increased awareness in the second half of the 20th century of other triggers of asthma (allergens and air pollution) has shifted the focus away from psychosocial factors, these old concepts are going through a renaissance, with an increasing number of publications reporting consistent positive associations between psychosocial stress and asthma. Associations with asthma have been shown for a wide range of stressors including: post-traumatic stress disorder (PTSD) 5, 6; psychiatric disorders 7, 8; community violence 9; stressful life events 10; partner violence and housing quality 11; war-related stress 12; neuroticism and relational problems 13; perceived safety 14; social depravity and high crime rates 15; anxiety and attention disorders 16; psychological distress 17; maternal anxiety 18; maternal depression and paternal PTSD 19; and work-related stress 20. Conversely, the prevalence of asthma has been shown to be elevated in those suffering from anxiety disorders 21.
The cross sectional nature of most of these studies has prevented firm conclusions from being drawn regarding causality and direction of the relationship. However, there is now a growing body of evidence within the last decade (table 1), and even prior to this 27, suggesting that emotional stress, anxiety and PTSD precedes the development of asthma both in children and adults. Similarly, there is evidence that asthma precedes panic disorders and that panic disorders may exacerbate pre-existing asthma 21. These associations remain even after adjusting for potential confounders such as smoking, socioeconomic status, body mass index, and familial and genetic factors, as assessed in twin studies 5. Thus, prospective studies appear to support a causal bidirectional association with psychosocial factors playing a role in both the development and prognosis of asthma 28. However, there are critical issues that have not been adequately addressed in the currently available studies. In particular, although many studies have used standardised questionnaires to assess asthma, or respiratory health more generally, few have included objective measures. This is generally not a major issue in most population-based asthma studies where standardised asthma questions show good agreement with a more objective clinical diagnosis of asthma 29; however, it may be problematic in studies of stress and asthma since psychosocial conditions are known to affect accurate symptom perception 30, 31. A difference in the incidence or prevalence of respiratory symptoms may, therefore, be at least partially accounted for by a difference in symptom perception between those with and without stress, rather than an objective difference in airway obstruction. Furthermore, psychological disorders have been associated with poor asthma self management 32, a factor that is not usually controlled for in studies of stress and asthma. Thus, any observed effects of stress on asthma (severity) may be indirectly mediated through poor asthma self management rather than being direct effects of stress itself. Finally, it is not clear whether the effects associated with emotional stress are specific to asthma, with some reports also showing associations with chronic obstructive pulmonary disease (COPD) and chronic bronchitis 24. These problems are exacerbated by uncertainty about the neuroimmunomodulatory mechanisms underlying the association between stress and asthma.
Studies involving objective measures of respiratory health effects, such as the one by Spitzer et al. 33 published in this issue of the European Respiratory Journal, may shed some light on these issues. The authors conducted cross-sectional analyses (nested within a cohort study) of the associations between lung function, as an objective measure of airway obstruction, and trauma exposure and PTSD in 1,772 adults from the general population. Trauma exposure, PTSD and respiratory symptoms were assessed using standardised questions, and lung function was assessed using spirometry. Similar to previous studies, those with PTSD (n = 28) had a significantly greater risk of having asthma symptoms, after adjustment for confounders. However, those with trauma (n = 887) had only slightly elevated risk and most associations were not statistically significant. Unadjusted analyses indicated that subjects with PTSD had lower forced expiratory volume in 1 s (FEV1), forced vital capacity (FVC) and FEV1/FVC. However, the PTSD group was substantially older, included more females, were shorter and had a greater proportion of former/current smokers compared to the reference group (no trauma and no PTSD). Lung function was also significantly lower in the group with exposure to trauma, but the deficits were smaller than those for the PTSD group. However, when the analyses were adjusted for potential confounders, the results looked very different; the largest lung function deficits were observed in the group with trauma exposure (significant for FEV1 and FEV1/FVC), and there was a negligible (nonstatistically significant) deficit in those with PTSD. Thus, after controlling for confounding, the lung function and symptom results were highly inconsistent, i.e. there were strong associations between PTSD and symptoms but no associations with lung function, whereas exposure to trauma showed a significant association with lung function but little association with symptoms.
These conflicting findings are not easily explained but could, in part, be due to the very small sample size of the PTSD group (n = 28). This is particularly inadequate for comparisons involving lung function (as a marker of asthma) since asthma is characterised by variable (or episodic) airway obstruction, and a sizable proportion of asthmatics therefore have normal, or close to normal, lung function most of the time. Also, the authors adjusted the analyses for some unusual “confounders” including “asthma/bronchitis” and “medication” (we assume medication refers to asthma medication) which, given their functional relationship with lung function, may have been inappropriate and may have affected the precision and validity of the analyses. In particular, asthma may be an intermediate variable in the causal pathway between stress and lung function in which case it is inappropriate to adjust for it. Also, it is feasible that “panic attacks” and “negative emotional states” (two other variables that the analyses were adjusted for) were highly associated with PTSD and/or trauma which may have led to multicollinearity, once again potentially resulting in invalid risk estimates. Finally, several studies have shown differential effects of emotional stress on respiratory health between males and females 24, 34. Therefore, rather than treating sex as a confounder, it might have been better to conduct analyses stratified by sex, although given the low number of subjects in the PTSD group (and low proportion of males) this would not have been possible for all subgroups. The authors provide no indication as to whether these issues were considered when the multiple regression model was developed, and it is therefore not clear whether or not they play a role in explaining the surprising and inconsistent results. If we assume that population size, statistical problems and effect modification by sex in the multivariate analyses were not significant issues then what else could have explained these mixed results? As indicated previously, stress may alter symptom perception, but this would only be a plausible explanation for the findings if symptom perception was different for those with PTSD and those with trauma, with the PTSD group over-reporting symptoms. Differential effects of PTSD and trauma on self asthma management may have also played a role, with better medication adherence for the PTSD group resulting in improved lung function. To fully explain the conflicting results, both phenomena would have had to take place. However, in the absence of data on these specific issues, these potential explanations remain speculative.
To our knowledge there are only three other population-based studies assessing the relationship between psychosocial factors and lung function. In a cross-sectional study of 5,486 patients, Goodwin et al. 35 found a consistent inverse relationship between various mental health problems and obstructive or restrictive lung function assessed by spirometry. A prospective study of 3,885 middle-aged adults by Pembroke et al. 24 showed that psychological distress (PD; assessed by questionnaire) was associated with respiratory symptoms such as breathlessness and bronchitis, as well as reduced FEV1, but these effects were only seen in females. In males, PD was associated with an increased risk of bacterial infection but was not associated with other respiratory symptoms or lung function. A recent cross-sectional study (nested in a longitudinal study) in 2,443 young adults by Hayatbakhsh et al. 34 found an increased risk of asthma symptoms and asthma medication in individuals suffering from neurological disorders (including PTSD), but no association with lung function and anxiety, panic or PTSD was found. However, a significant inverse association was shown between (12 months and lifetime) major depression and FEV1 and forced expiratory flow at 25–75% of FVC, but only in males. These studies, in addition to the study by Spitzer et al. 33, suggest some evidence of associations with lung function but the results are inconsistent. Therefore, it remains uncertain whether the respiratory symptoms associated with psychosocial factors are due to airway obstruction or other factors not directly related to asthma (or other respiratory conditions such as COPD).
As mentioned previously, the uncertainty about the association between stress and asthma is further complicated by the relative lack of data regarding the underlying mechanisms. Spitzer et al. 33 suggest that the lung function changes observed in the traumatised group may be due to inflammation. This assumption (few studies on stress and asthma have collected data on airway inflammation to date) is largely driven by the general consensus that asthma is an inflammatory disease. In addition, there is some, but limited, evidence suggesting that stress (through altered regulation of the hypothalamic-pituitary-adrenal axis and sympathetic-adreno-medullary nervous system) may have immunomodulatory effects involving an atopy-, or Th2-, biased response favouring allergic outcomes 36, 37. If airway inflammation is indeed the main mechanism through which stress causes asthma, as has been suggested, then it would be expected that stress would also be associated with the exhaled nitric oxide fraction (FeNO) which has been increasingly recognised as an objective measure of airway inflammation in (atopic) asthmatics. However, the few studies that have been conducted on stress and FeNO in small groups of asthmatics and nonasthmatics experimentally exposed to stress also showed mixed results. In particular, some studies showed an (inconsistent) stress-associated increase in FeNO in asthmatics 38–40 and nonasthmatics 40, whereas another study found a stress-associated decrease in FeNO in nonasthmatics and no change of FeNO in asthmatics 41. In addition to neuroimmunomodulatory effects, it is plausible that direct neurogenic mechanisms may underlie at least part of the association between stress and asthma 42, but these mechanisms have rarely been studied.
Where to from here? Further prospective and cross sectional studies are needed to clarify the relationship between stress, respiratory disease and objective parameters of respiratory function. In addition to collecting valid data on psychosocial factors and potential confounders including general aspects of lifestyle 43, these studies would ideally involve repeated spirometry, atopy testing and objective measures of both airway inflammation and neurogenic mechanisms, and could provide the scientific underpinning of future intervention studies. Of course stress is only one of a range of potential causal exposures for asthma and intervention of stress, therefore, will not prevent all cases of asthma 44. However, it is a feasible candidate for intervention, which has the added benefit of potentially reducing other stress-related conditions including cardiovascular disease, diabetes, hypertension and obesity, as well as being of importance to public health in itself, a view shared with the classic authoritative scholars such as Hippocrates and Maimonides.
Footnotes
Support Statement
The Centre for Public Health Research is supported by a Programme Grant from the Health Research Council of New Zealand.
Statement of Interest
None declared.
- ©ERS 2011