The status of stressful life events as a risk factor for asthma is unclear and may be dependent on pre-existing allergic rhinitis. This study examined whether exposure to stressful life events predicted the onset of asthma in adults.
This is a prospective, population-based cohort study of 16,881 males and females, aged 20–54 yrs and free of diagnosed asthma at the beginning of the follow-up (January 1, 2004). Data about stressful life events were gathered with a postal survey. The onset of asthma was ascertained through national registers until December 31, 2005.
During the follow-up period, 192 incident cases of asthma were identified. High total exposure to stressful life events, as indicated by a cumulative severity score, predicted the onset of asthma (hazard ratio 1.96, 95% CI 1.22–3.13). This association was robust to adjustment for demographics, smoking and having a cat/dog at home and it was observed both among those with and without allergic rhinitis at baseline. Of the 10 most stressful life events, the illness of a family member, marital problems, divorce or separation and conflicts with a supervisor were associated with the onset of asthma.
Our study suggests that stressful life events may increase the onset of asthma.
Asthma is a disorder of the airways in which chronic inflammation is associated with airway hyperresponsiveness, leading to recurrent episodes of wheezing, breathlessness, chest tightness and coughing . It is one of the most common chronic diseases worldwide, especially in industrialised countries. Among the adult population, the reported prevalence figures vary from 1.9% in Greece to 18.4% in Scotland . In Finland, the prevalence of asthma is 4% among males and 7% among females  and the overall incidence rate during adulthood (i.e. 21–63 yrs) is 0.9 new cases per 1,000 person-yrs .
A large number of studies have examined the biological risk factors for asthma. There is strong support for an increased risk of asthma that is associated with genetic factors, allergens, respiratory infections, air pollutants, tobacco smoke  and allergic rhinitis (AR) . Recent studies suggest that AR and asthma, in fact, comprise a single syndrome known as the chronic allergic respiratory syndrome, which may affect one or two parts of the respiratory tract .
Recently, the role of psychosocial stress as a risk factor for asthma has gained increased attention. Stress is considered to affect the expression of asthma through multidimensional endocrine, neural, immune and behavioural processes , but clinical research to date has clarified only part of the causal mechanisms that may link stress to asthma . There is observational evidence that severe negative life events may increase the risk of children's asthma attacks over subsequent weeks  and may be associated with asthma symptoms among adolescents . The association between stressful life events and an increased risk of asthma has also been reported in four retrospective studies [12–15] and in at least one prospective study  among adults. In the latter study, which was based on 4,010 middle-aged participants in one town and its surroundings, breaking off a life partnership predicted incident asthma, but no other stressful events were found to have an effect . To our knowledge, however, no large-scale, population-based studies that explore the association between stressful life events and the onset of asthma are available.
In this study, we examine the association between stressful life events and the onset of asthma in a large population sample, taking into account pre-existing AR. We sought to determine whether: 1) the total exposure to stressful life events is associated with the onset of asthma among adults; 2) specific severe life events are particularly predictive of such an onset; and 3) the association between life events and the asthma onset is dependent on pre-existing AR.
MATERIALS AND METHODS
Study population and design
Data were drawn from the Health and Social Support (HeSSup) study, a longitudinal study on a population sample that is representative of the Finnish population in the following four age groups: 20–24, 30–34, 40–44 and 50–54 yrs at baseline (phase one) . Baseline characteristics were obtained from the phase one postal survey conducted in 1998. Exposure to life events during a 5-yr period between phase one and phase two was assessed with a follow-up questionnaire that was sent to all participants who were still living in Finland in 2003. Of the 25,901 respondents at baseline, 216 died before phase two, 234 had moved abroad and 969 could not be reached due to unknown addresses; therefore, a total of 1,419 baseline respondents did not participate in the follow-up survey. Of all 19,629 respondents at phase two, which revealed an 80% response rate, 18,900 consented to the use of their recorded health information in Finnish national registers. Out of that pool, we excluded any participants with prevalent asthma or asthma history at phase two (n = 2,016). Thus, the final sample consisted of 16,881 participants (6,663 males and 10,218 females) who had no indication of asthma at the beginning of the follow-up (fig. 1). They were followed for a 2-yr period from national health registers in order to detect the onset of new asthma cases (2004–2005) after the phase two survey.
Stressful life events
Life events that occurred between phases one and two were measured by a list of 21 negative events, as in earlier studies [18–20]. For each event, the questionnaire included the following categories: never, within the previous 6 months, within the previous 5 yrs and >5 yrs ago. The subjects classified the severity of each event as: 1, not so burdensome; 2, burdensome; and 3, extremely burdensome. The events that had occurred during the previous 5 yrs (i.e. in the period between the questionnaires) were considered in this study. We excluded ‘illness causing work disability of >21 days’ and ‘retirement’ because such events might be related to asthma. To assess the total exposure to life events within the 5 yrs of follow-up for each individual, we calculated, from the 2003 survey responses, a cumulative sum score of the severity of events that had occurred within the previous 6 months or 5 yrs. As weights, we used means of squared severity ratings for each event as in our earlier studies [20, 21]. Those with a zero score were defined as having no exposure. The exposed were divided into tertiles: light (cumulative sum score 2.74–5.53), moderate (6.15–12.07) and heavy (12.09–55.41). In addition to the total exposure to stressful life events, we measured the nine most burdensome categories of specific events: death of a family member (death of own child, weight (w) 7.42; death of a spouse, w 7.01); emotional, physical or sexual violence (w 6.16), severe illness in a family member (w 5.53), death of the mother (w 5.04), a major increase in marital problems (w 4.92), divorce or separation (w 4.75), severe conflicts with a supervisor (w 4.63), severe financial difficulties (w 4.37) and death of the father (w 4.25) .
Case definition for asthma
Using the unified personal identification code system, which covers all Finnish citizens, we linked the survey responses to records from three independent and comprehensive Finnish national health registers in order to identify incident cases of asthma. The identification of the cases was based on the clinical diagnosis from two registers (i.e. the Drug Reimbursement Register and the Hospital Discharge Register) of the treating physician (for special reimbursement for medicine costs or hospitalisation) or detailed information about the prescribed medication (the Drug Prescription Register). A participant was classified as having incident asthma when the caseness was verified for the first time by at least one of the three data sources between January 1, 2004 and December 31, 2005.
First, we used the Drug Reimbursement Register of the Social Insurance Institution of Finland, which contains information on persons entitled to special reimbursement for certain chronic diseases, such as asthma. Patients who apply for special reimbursement must attach a detailed medical certificate that was prepared by the treating physician, who also provides data to confirm the diagnosis. The application is then reviewed by a physician in the Social Insurance Institution to determine whether the uniformly defined requirements for the disease are met. From this register, participants were defined as incident asthma cases if they were recorded in the Central Drug Register as being eligible for asthma treatment for the first time during the follow-up.
Secondly, we used prescription data to assess the beginnings of medical treatment for asthma. In Finland, the National Social Insurance Scheme at the Social Insurance Institution provides basic reimbursement (currently 42%) for all filled outpatient prescriptions that are recorded in the Drug Prescription Register according to the World Health Organization's Anatomical Therapeutic Chemical (ATC) Classification  and by the date of purchase. We identified all participants with two or more prescriptions for drugs for obstructive airway diseases (ATC code R03) in any year during the follow-up by using the day of the first purchase as an indicator of the onset of asthma.
Thirdly, we obtained data from the Hospital Discharge Register of the National Institute for Health and Welfare, which includes records on all in-patient hospital admissions. This register is comprised of countrywide information on virtually all hospitalisations. We obtained the discharge dates and the corresponding main diagnoses for hospitalisation due to asthma (International Classification of Disease-10 J45).
All of the individuals who were identified as having an onset of asthma in any of these registers before January 1, 2004 were excluded from the analysis, as were those individuals who reported a life-time diagnosis of asthma in either survey. Of all 192 incident cases of asthma that were detected on any of the three registers, 190 (99%) cases had purchases of prescribed medication for asthma and 41 (21%) cases had a clinical diagnosis of asthma noted in hospitalisation or special reimbursement records. Of the latter, 39 (95%) cases also had purchases of prescribed medication for asthma.
All baseline characteristics were measured at phase one. The possible risk factors for the development of asthma that were considered in this study included AR, exposure to dogs or cats  and smoking . The participants were asked whether a doctor had previously told them that they have or have had AR (yes/no). The subjects were classified as having AR versus not having AR (baseline). AR is a risk factor for asthma, particularly for the allergic phenotype [6, 24]. Information about a pet dog/cat was obtained by asking whether the participant has a pet. The variable was dichotomised as having a pet dog/cat at home versus not having one. The participants reported their current and past smoking habits, smoking rate and number of cigarettes . They were classified as nonsmokers, past smokers or current smokers. Sociodemographic background variables included sex, age group, level of education (basic/vocational/college/university) and marital status (single, divorced or widowed versus married or cohabiting).
All analyses were performed by using PASW Statistics for Windows software, Release 18.0.0 (SPSS Inc., Chicago, IL, USA). The associations between the baseline characteristics and the onset of asthma, as well as those between life events that occurred between phase one and phase two and the onset of asthma, were analysed via the Cox proportional hazard model. Follow-up began on January 1, 2004 (i.e. the year after the phase two survey) and ended upon the first occurrence of the outcome measure, death, or December 31, 2005, whichever came first. We calculated hazard ratios and their 95% confidence intervals.
We studied the associations of baseline characteristics and risk factors with the onset of asthma during the risk time. Then we studied the association between the total exposure to life events and onset of asthma, first adjusting the analysis for all demographics and finally adding the risk factors to the model. Adjusting in the same way, we analysed the association between the 10 most stressful life events and asthma onset in the risk time.
The sex difference in the association of the cumulative severity score to the onset of asthma was assessed with the interaction term gender* cumulative severity score. Because no significant interaction was found (p = 0.43) we analysed males and females in combination adjusting for sex.
We performed two sensitivity analyses. First, we tested whether the effect of life events on asthma were dependent on AR. This analysis was carried out among two subgroups: participants with (n = 4,029) and without AR (n = 12,726). Secondly, we tested whether the association between life events and the subsequent onset of asthma was dependent on case definition. This analysis was carried out in subgroups of participants whose case definition was based on the clinical diagnosis of the treating physician (i.e. special reimbursement for medicine costs or hospitalisation, n = 41) and those whose caseness was determined according to prescribed medication (n = 190).
During a mean±sd follow-up of 2.0±0.12 yrs, 192 participants were diagnosed with asthma. Past smoking habits, older age and the presence of AR were associated with an increased risk (table 1).
Stressful life events and asthma
Altogether, 13,050 (79%) participants in our study reported that they encountered a stressful event during the 5-yr exposure window between phase one and phase two. Of these participants, 4,659 (36%) experienced one event, 7,846 (47%) experienced two to five events and 545 (3%) endured six to 12 events. There were 3,366 participants who reported no events.
The cumulative severity rating of the events ranged from 0 to 55.41 (median 6.97). Table 2 shows that those in the highest tertile of exposure to life events had a two times (95% CI 1.26–3.20) greater risk of asthma during the follow-up, compared with participants who did not experience any stressful life events. This relative risk remained unchanged after taking into account adjustments for demographics, smoking habits and exposure to dogs or cats.
Analyses of specific life events showed that seven of the nine categories of the most severe events were associated with a 1.5–1.7-fold risk of asthma during the follow-up period (table 3). Statistically, a significantly increased risk of asthma onset was observed after a divorce or separation, major increase in marital problems, severe illness in a family member, severe conflicts with a supervisor and severe financial difficulties. These associations were robust to all adjustments with one exception; the category ‘severe financial difficulties’ was not associated with asthma risk in the fully adjusted model. The only categories that showed no association with asthma onset were ‘death of a family member’ and ‘death of own father.’
Figure 2 illustrates the cumulative hazard curves of asthma onset by exposure to stressful life events. The expected dispersion of hazard curves between categories of cumulative severity scores was observed across the entire follow-up period (p for log rank test = 0.006).
To test the robustness of our findings we excluded the participants with AR. This analysis replicated the findings for all participants. The hazard of asthma onset for those who were most severely exposed to life events was 1.9 times higher (95% CI 1.04–3.43) than that for the non-exposed (table 3). In an analysis that included the participants with AR only, the risk estimate was practically the same. However, it failed to reach statistical significance due to lower numbers. The nonsignificant interaction (p = 0.46) between AR and life events is consistent with these findings.
A further sensitivity analysis testing whether the association between life events and a subsequent onset of asthma was dependent on case definition also replicated findings from the main analysis. Determining incident asthma on the basis of drug prescriptions (in analysis n = 187), the hazard ratio for high exposure was 2.09 (95% CI 1.33–3.33). If incident asthma was determined solely on the basis of clinical diagnosis (i.e. special reimbursement for asthma or hospitalisation, n = 41), the hazard ratio after a high level of exposure to stressful life events remained practically the same but did not reach statistical significance (HR 2.41, 95% CI 0.88–6.63). These associations were adjusted for age and sex; light and moderate exposure scores were combined due to the small number of cases (data not shown).
In this population-based prospective cohort study of almost 17,000 participants who had no indication of asthma at the beginning of the follow-up period, a high exposure to stressful life events was associated with a two-fold increased risk of the subsequent onset of asthma as measured by reliable national health registers. The elevated risk was not accounted for by baseline characteristics, such as demographics, smoking habits or exposure to dogs or cats and it was of the same magnitude both among those with and without AR. We also found that specific severe events (i.e. a divorce or separation, a major increase in marital problems, severe illness in a family member and severe conflicts with a supervisor) were predictive of the onset of asthma.
Our findings support the biopsychosocial model of stress , which suggests that stressful life events may alter the psychological, immunological and endocrine systems and contribute to the onset of asthma. To the best of our knowledge, the relationship between total exposure to negative life events and the onset of asthma has not been previously reported. The findings of this study are in agreement with prior studies that suggest an association of stressful life events with various health problems ranging from sleep disturbances and sickness absence to depression and breast cancer [19, 20, 26, 27].
To date, only a few studies have prospectively investigated the association between stressful life events and adult asthma. The only longitudinal cohort study of which we are aware followed a population-based sample of 4,010 middle-aged adults from Germany via questionnaire surveys at baseline (1992–1995) and follow-up (2002 and 2003) . Exposure to three stressful life events was measured (i.e. unemployment, having broken off a life partnership and death of someone close). Having broken off a life partnership was associated with a risk of incident asthma that is 2.2 times higher, while no excess risk was found for the other two life events. However, this study was based on a self-report of asthma. In addition, some earlier clinical and retrospective studies noted that adults with acute asthma report that they have experienced a divorce or marital problems before the onset of the disease more often than the non-asthmatic controls [12, 14]. In our study, based on a follow-up of a population sample with no indication of asthma at baseline, individuals who had interpersonal conflicts indicated by divorce or separation, a major increase in marital problems and severe conflicts with a supervisor experienced a risk of incident asthma that was 1.6–1.7 times higher in the future. In agreement with previous studies, the death of someone close, such as a family member or one's own mother or father, was not associated with increased risk of asthma [15, 16]. The fact that the case definition was based on objective indicators of asthma increased the credibility of our findings.
We also noted a robust association between the severe disease of a family member and incident asthma. In line with this result, Salminen  observed that a severe disease of a family member was more common among acute asthma patients than among non-asthmatic controls. A smaller-scale analysis found, in contrast, no association between family death or illness and diagnosed asthma . It is noteworthy that there are some differences between the studies in life event variables in the analyses and in the study populations. We are not aware of any other studies on violence or financial difficulties as predictors of the onset of asthma and, in this study, no robust association between these events and asthma onset was found.
The strengths of this study are that data were included from a large sample of working-age adults, a study design that allowed the determination of temporary order between exposure to stressful life events and the onset of the disease and the measurement of asthma from national health registers rather than relying on self-reporting. In Finland, the validity of the national registers has been found to be high , reasonably accurate and highly reliable for epidemiological study purposes . We were also able to control a number of sociodemographic elements and asthma risk factors. At phase one, the response rate was relatively low (40%), and there may have been differences between respondents and non-respondents regarding the frequency of asthma and stressful life events. However, >80% of these individuals participated in the phase two survey and, of them, 96% consented to the linking of data from national health registers. Thus, it is unlikely that the longitudinal associations between life events and incident asthma would be spurious due to the low participation at phase one.
This is apparently the first epidemiological study of a working-age population that investigates the association between stressful life events and the onset of asthma from reliable health registers. We found that the severity of all stressful life events was associated with the onset of asthma and observed associations with a divorce or separation, a major increase in marital problems, a severe illness in a family member and severe conflicts with a supervisor. Taking into account the nature of asthma as a multifactorial disease, the finding of this study that some life events may trigger the onset of asthma in adults deepens the understanding of the disease's psychosocial risk factors.
This study was supported by the Academy of Finland (grant numbers 110451, 117604, 124271, 128089 and 129262). M. Kivima¨ki and J. Vahtera were supported by the Academy of Finland (projects 124271, 124322, 128089 and 129262). M. Kivima¨ki is also supported by the BUPA Foundation, UK, the National Institute on Aging (R01AG034454), the National Heart, Lung, and Blood Institute (R01HL036310), NIH, USA and the New OSH ERA Programme grant, EU.
Statement of Interest
- Received October 16, 2009.
- Accepted September 26, 2010.
- ©ERS 2011