Frequency of Perceived Poststress Symptoms in Inpatients, Outpatients and Healthy Controls: The Role of Perceived Exhaustion and Stress

Stress-related disorders are commonly considered to appear during periods of acute or chronic stress [1]. However, several symptoms occur most commonly after an episode of stress, during periods of rest [2]. Thus, it is difficult for patients and physicians to consider stress a trigger of the symptomatology and to determine adequate prevention strategies and treatment.

Stress is a threat to the normal, internal physiological equilibrium. To restore this equilibrium and adapt to external challenges, a complex behavioral and physiological stress response is initiated. The stress response is based on neuroendocrine and central processes that are crucial for the development of stress-related disorders [1]. While allostasis refers to the normal biological process of adaptation to stress, allostatic overload, as defined by Fava et al. [3], refers to the price the body pays for being forced to adapt to prolonged, heightened, or adverse psychosocial or physical situations [4, 5]. This can be physical but also purely psychological stress, as is known from learned helplessness or embitterment reactions [5, 6, 7, 8]. Allostatic overload may therefore not only cause symptoms under stress but also promote a poststress symptomatology: ‘A state of allostatic load entails some clinical manifestations that can be observed in daily practice. Examples may be provided by worsening of symptoms during weekends or vacation (inability to shut off response associated with lack of distraction entailed by work), or breakdowns which occur just when a stressor has terminated (caregivers of patients successfully recovering after a long struggle)' [3]. Allostatic overload may rather manifest clinical symptoms. Alterations in several biomarkers confirm a close association between allostatic overload, numerous neuroendocrine systems, and impaired immune and cardiovascular functioning [9].

Up to now, research has rarely focused on the pathophysiology of poststress symptoms. Some studies have described poststress headache as ‘weekend headache' or ‘weekend migraine' [10, 11], as these complaints occur more characteristically during rest at the end of the week [12, 13, 14]. Independently of this discussion, stress is known as a trigger for migraine [15, 16], peaking a few days before migraine attacks [17]. Recently, a ‘let-down headache' hypothesis was tested and a decline in perceived stress was shown to increase the chance of migraine onset 1.5-1.9 times, which was not the case if the stress load persisted [2].

Several other disorders seem to most commonly occur under poststress conditions, such as appendicitis [18], thrombosis [19], asthma [20, 21], infections [22, 23, 24], and myocardial infarction [25, 26]. Interestingly, it seems that the risk of myocardial infarction is not spread equally across the vacation time but peaks after the first 2 days of the vacation [27]. However, these observations have not always been confirmed [28, 29, 30]. Immunopathological responses such as asthma and inflammation contributing to cardiovascular disease have been described as a result of chronic stress [31].

In sum, poststress symptoms seem to include infections, asthma symptoms, cardiovascular diseases, and headache/migraine. In a representative healthy Dutch sample, additional poststress symptoms, such as muscular pain, fatigue, and nausea, were observed [32]. These authors reported a prevalence of up to 3.6% for a ‘weekend syndrome' and 3.2% for a ‘vacation syndrome'. The most frequently reported vacation symptoms were headache/migraine (54.4%), influenza-like symptoms (48.9%), fatigue (27.8%), muscular pain (26.7%), nausea (20.0%), and general pain (14.4%). The prevalence of some of these symptoms appeared higher on weekends than while on vacation, e.g. 67.7% for headache/migraine and 34.4% for fatigue. Most subjects reported recurrence of the same poststress symptoms. In addition, Vingerhoets et al. [32] observed that these subjects express an inability to relax and to switch off, as compared to people who rarely report such symptoms. This research group proposed some explanations for the development of poststress symptoms, ranging from environmental factors like exposure to certain chemicals, lifestyle factors such as alcohol or coffee consumption, adrenergic activity, and psychosocial factors [33].

Emotional exhaustion is probably the most common symptom of stress-related disorders, and it predicts direct, indirect, and intangible health costs [34]. Exhaustion is not well defined and is rather unspecific, sometimes described as a syndrome or a consequence of (work-related) stress. To describe exhaustion more precisely, we recently discriminated 6 endophenotypes of exhaustion. Poststress symptoms were mainly associated with an endophenotype of norepinephrine (NE) depletion [35].

Since little is known about the occurrence of poststress symptomatology in stress-related disorders, such as depression, anxiety, and somatoform disorders, the current work aimed to compare the frequency of poststress symptoms among healthy subjects and in- and outpatients. In addition, we were interested in detecting possible predictors of poststress symptoms.

To assess poststress symptoms, exhaustion, and psychiatric diagnoses, 3 questionnaires were filled out by inpatients, outpatients, and healthy controls.

Neuropattern Questionnaires. The Neuropattern Questionnaires (NPQ) consist of 4 different questionnaires, which were developed as a part of Neuropattern™, a new diagnostic tool to assess endophenotypes of the stress response [36, 37]. Here, only information about poststress symptoms from the NPQ - Patient Questionnaire (NQP-P), about perceived exhaustion from the NPQ - Symptom List (NQP-S), and the Patient Health Questionnaire (PHQ) were used.

NPQ-P. The NPQ-P is a self-administered questionnaire with a 4-point Likert scale and 195 items assessing 77 different characteristics of psychological and somatic stress symptoms (0 = never, 1 = sometimes, 2 = most times, and 3 = always; e.g. ‘In stressful situations, I experience a worsening of joint pain', ‘I sweat easily in stressful situations', and ‘I am unable to relax'). The different scales are based on literature reviews and our own research projects [36, 37]. A subscale measures somatic poststress symptoms, with 10 items addressing questions about shortage of breath, infections, illnesses, migraine/headache, digestive complaints, gastrointestinal pain, visual impairment each on weekends, holidays, or subsequent to stress (e.g. ‘I get sick not during but after a stress period' and ‘I get headache during recovery periods'). Groups with frequent or no poststress symptoms were compared according to their subscale mean (M) from 0 to 1.4 (never or sometimes) or above (most times or always). Information on reliability is reported in the Results.

NPQ-S. The NPQ-S is a self-administered questionnaire with a 5-point Likert scale that assesses the perceived stress response and strain from life events experienced within the last month (0 = never, 1 = rarely, 2 = sometimes, 3 = often, and 4 = very often). A subscale measures perceived exhaustion based on 15 items (e.g. tired-out, powerless, exhausted, and burned out). The NPQ-S was shown to be a reliable (Cronbach's α = 0.95, retest r_tt = 0.78) and valid scale to measure subjective exhaustion. It correlates with emotional exhaustion of the Maslach Burnout Inventory (r_s = 0.76; [38]) and shows an increased chance of drug consumption, visits to physicians and therapists, days of sick leave, and number of psychiatric diagnoses for exhausted patients compared to nonexhausted patients [34].

PHQ. The PHQ is a common, self-administered version of the Primary Care Evaluation of Mental Disorders [39]. Here, the German long version was applied, which additionally allows screening for psychosocial stress with 10 items [40](e.g. ‘Stress at work/school affects me' and ‘Worries about my health affect me'). The German version of the PHQ shows a high validity for several mental disorders [40, 41]. Cut-off points for depressive and somatic symptoms are suggested following the 5-10-15 rule [39]. As there are only limited data for cut-off points for the psychosocial stress scale [41], the same rule applied for the psychosocial stress scale.

The frequency of poststress symptoms was assessed in 3 samples: (1) healthy controls, (2) outpatients, and (3) inpatients with psychiatric diagnoses. These samples were chosen to compare the frequency of poststress symptoms in relation to different degrees of exhaustion. Due to their mental health conditions, the inpatient sample was unable to work and sought treatment in a rehabilitation hospital, while the outpatients were still able to meet daily life requirements. As described elsewhere, the inpatients perceived a significantly higher degree of exhaustion than the outpatients [34].

Healthy Controls. The first sample was comprised of healthy German patients matched for age, sex, and education. We recruited 984 respondents online via the market research institution Respondi AG (http://www.respondi.com/en/). Of these, 216 consented to answering the same questions again 25 months later to test the retest reliability. Respondi AG has a member community from which respondents and target groups can be recruited for online studies according to personal panel data. For the present study, a representative sample in terms of age, sex, and education was chosen. All participants were rewarded with Respondi AG credit points (a certain number of credit points can be converted into vouchers for online shops). Inclusion criteria were: physical and mental health, and age ranging from 20 to 70 years. Exclusion criteria were: psychiatric and somatic disorders, chronic complaints, and addiction to drugs or alcohol. All respondents answered the NPQ-P and demographic questions. Additionally, they had to indicate their health status (yes or no for: ‘I am physically and mentally healthy', ‘I don't suffer from chronic health complaints', and ‘I am not addicted to drugs or alcohol').

Outpatients. The second sample consisted of outpatients with stress-related disorders. They participated in a randomized clinical trial with Neuropattern™ [36, 42]. The study protocol was approved by the Medical Association of Rhineland-Palatinate [Germany; reference No. 837.296.09 (6802)] and registered at the US National Institutes of Health (ClinicalTrials.gov identifier NCT01062880). Inclusion criteria were: diagnoses of depression, somatoform disorders or adjustment disorders, German as the mother tongue, written informed consent, and access to the Internet. Exclusion criteria were: simultaneous participation in other clinical trials, current psychotherapy, intake of psychotropic drugs, dexamethasone intolerance, pregnancy, and arrhythmia absoluta. Some of these exclusion criteria were chosen as part of the clinical trial. Here, electrocardiography and a dexamethosone suppression test were applied, too.All patients completed the German version of the PHQ [40], the NPQ-S, and the NPQ-P, among other evaluations. During routine visits, family physicians screened 558 outpatients, 420 of whom provided information about perceived poststress symptoms (NPQ-P), perceived exhaustion (NPQ-S), psychiatric diagnoses, and psychosocial stress (PHQ). These were the only data that were included in the current work.

Inpatients. The second sample consisted of 101 inpatients with stress-related disorders treated in a rehabilitation hospital of behavioral and psychosomatic medicine. Patients participated in a randomized clinical trial with Neuropattern™. This study was approved by the ethics committee of the German Federal Pension Fund. The study procedure is described elsewhere [42]. Here, we analyzed only data on the frequency of poststress symptoms (NPQ-P).

All data were first presented as percentages or M and standard deviations (SD) and checked for normality to apply appropriate statistical tests. To compare groups, χ² tests were used (with a significance level of 5% or less if adjusted because of multiple comparisons). Odds ratios (OR) were used to provide information about the proportion of risks for different groups, with 95% confidence intervals (CI), if appropriate. We calculated 3 nonhierarchical multiple logistic regression models and reported the unstandardized regression coefficients B, their SD, OR, degrees of freedom (d.f.) and the probability of the observed results (p). First, we tested whether poststress symptoms could be predicted by perceived exhaustion and psychosocial stress (model A). Second, their interaction term was added (model B), and, third, significant predictors from model A-B and demographic factors (age, gender, and education) were tested as predictors of poststress symptoms (model C). With respect to education, cases with ‘no/other education' were excluded to gain ordinal data. Noncategorical data were z-standardized to reduce multicolinearity and to gain standardized residuals. The linearity and multicolinearity of predictors were tested. Outliers above 2 SD were excluded case-wise and a 95% CI was applied. All analyses were performed using the IBM SPSS statistical software package version 22.

Descriptive data about the demographics of all 3 samples are provided in table 1. The healthy controls and outpatients showed an unexpected high education (58.7 and 52.2% with secondary education qualifying for university), whereas the inpatients showed a higher proportion of women (73.3%) and elderly people (42.2% were aged 50 years or older).

For healthy controls, the poststress symptoms scale of the NPQ-P showed an internal consistency of Cronbach's α = 0.83 and a retest reliability of r_tt = 0.682 (M_t1 = 0.40, SD_t1 = 0.44, M_t2 = 0.34, and SD_t1 = 0.39).

Demographic Factors. First, we analyzed whether the prevalence of poststress symptoms varied systematically across gender, age, and education (table 2). Poststress symptoms were more likely for women compared to men in all samples. Due to insufficient numbers of cases, our statistics did not consider the subjects with frequent poststress symptoms and ‘no or other education' for all samples (n_{healthy controls} = 0, n_outpatients = 3, and n_inpatients = 2), ‘basic secondary education' for healthy controls (n = 2) and inpatients (n = 3), and inpatients aged less than 30 years (n =2), respectively.

Prevalence. Next, the prevalence of poststress symptoms was compared across samples. A pronounced difference was observed. Poststress symptoms occurred ‘most times or always' in 2.9% of healthy controls, 20% of outpatients, and 34.7% of inpatients. The frequency of poststress symptoms differed significantly between these groups [χ² (2) = 172.5, p < 0.001]. The chance of frequent poststress symptoms was 8 times higher for outpatients compared to healthy controls (OR = 8.26; 95% CI 5.35-12.74), 2-fold higher for inpatients compared to outpatients (OR = 2.12; 95% CI 1.32-3.38), and 17-fold higher for healthy controls compared to inpatients (OR = 17.46; 95% CI 10.60-30.32). Depending on the sample, the most prevalent poststress symptoms were infections (16.9-51.9%) and gastrointestinal symptoms (13.8-54.7%), followed by headache (12.3-49.1%) and visual complaints (9.3-48.1%), while the rarest complaints were asthma symptoms (2.9-30.5%). All poststress symptoms were significantly more often reported by outpatients compared to healthy controls [all χ² (1) >84.0, p < 0.001, adjusted significance level (p_adjusted) = 0.005]. Additionally, headache and asthma symptoms also occurred more frequently in inpatients compared to outpatients [both χ² (1) >9.0, p = 0.002, p_adjusted = 0.005; table 3].

A descriptive comparison of the prevalence of different poststress symptoms for psychiatric diagnoses (according to the PHQ-D) revealed a similar picture. For all PHQ-diagnoses, gastrointestinal symptoms were the most frequent (56.7-75.8%), followed by infections (50.0-59.5%), while asthma symptoms were again rare (9.8-33.3%; table 4). Given the fact that many patients had comorbid diagnoses (62.6% of outpatients) and different types of poststress symptoms (66.0% of outpatients), subgroups could not be statistically compared.

Exhaustion and Stress. In addition, the prevalence of poststress symptoms was compared between outpatients with or without perceived exhaustion (NPQ-S) and psychosocial stress (PHQ). The analysis revealed that the chance of poststress symptoms was about 4 times higher for exhausted and stressed patients compared to patients with no perceived exhaustion or psychosocial stress. However, the frequency of poststress symptoms was higher in exhausted patients (75.0%) than in patients with psychosocial stress (56.3%).

Finally, 3 multivariate logistic regression analyses were applied in the outpatient population. The first analysis revealed exhaustion and stress as highly significant predictors of poststress symptoms (table 5, model A). Furthermore, the main effects of exhaustion and stress, but not their interaction term, were significant predictors of poststress symptoms (table 5, model B). Thus, stress and exhaustion predicted poststress symptoms, likely independently of each other. Additionally, education turned out to be a significant predictor of poststress symptoms. However, education had OR and β values comparable to those of the other demographic variables (age and gender), which were not significant predictors of poststress symptoms (table 5, model C). Exhausted patients had a higher chance of poststress symptoms compared to nonexhausted patients (OR = 1.78-1.89, all p < 0.001). For psychosocial stress, the OR varied between 1.76 and 1.84 (all p < 0.001), and for education the OR was 1.45 (p = 0.006; table 5).

To the best of our knowledge, this is the first comprehensive report on poststress symptoms in patients with stress-related disorders. We found a high frequency of poststress symptoms in outpatients and even more in inpatients. Furthermore, our results inform about the influence of demographic factors, psychosocial stress, and perceived exhaustion as predictors of poststress symptoms. Our results are in line with other work on the frequency of leisure sickness or vacation syndrome in healthy subjects [32].

In this study, poststress symptoms appeared in 2.9% of healthy controls, 20.0% of outpatients, and 34.7% of inpatients across all diagnoses. The most prevalent poststress symptoms were infections and gastrointestinal symptoms in all samples and for all psychiatric diagnoses. In the outpatients, poststress symptoms were predicted by education, psychosocial stress, and perceived exhaustion. Amazingly, psychosocial stress was not the strongest predictor of poststress symptoms. Poststress symptoms were better predicted by perceived exhaustion. There was a higher prevalence of poststress symptoms for outpatients with a lower education and females across all samples. However, multivariate logistic regression revealed only education as a significant demographic predictor once psychosocial stress and perceived exhaustion were added. Further, the effects of psychosocial stress and perceived exhaustion on poststress symptoms were independent of each other, and the interaction between both was not a significant predictor.

Notably, exhaustion is a common, unspecific, and relevant symptom of many physical and mental disorders and an important prognostic indicator of direct, indirect, and intangible health costs [34]. Thus, it remains unclear whether exhaustion can be considered the cause or the consequence of poststress symptoms or whether both kinds of symptoms occur independently. Given the fact that 65.5% of the exhausted outpatients reported no poststress symptoms, one can assume that exhaustion is not an obligatory cause of poststress symptoms. However, poststress symptoms occurred more often in highly exhausted patients. Thus, it is likely that the intensity of exhaustion is closely associated with poststress symptoms. Notably, only 8.6% of patients with poststress symptoms showed no exhaustion.

Allostatic overload is a likely cause of poststress symptomatology. In a broader view on neuroendocrinology, the concept of allostatic load refers to the interaction between the brain and the body to maintain homeostasis. The adaptive function of hormones and autonomic activity to stressors may become cumulative and potentially damaging [43]. Further, the system may adapt to chronic demands by changing feedback loops and set points. This may explain why ‘entry into a relaxed condition may create an unpleasant state of withdrawal from one's physiological regulation. Such changes in hormones (…) provide a physiological basis for the individual to continue to seek a condition of high demand' [44]. Allostatic maladaptation is a function of the biological, biographical, behavioral, and current social environment - including epigenetic processes and socioeconomic status/lifestyle factors [43]. Fava et al. [3] identified a broad array of relevant criteria for allostatic overload, such as specific stressful situations, missing coping skills, psychopathological symptoms, impairment in social and occupational functioning, and a decrement of psychological well-being. Our data confirm the role of psychosocial stress as well as perceived exhaustion in the development of poststress symptoms. However, as both factors play an independent role, the complex interaction between experience, behavior, social environment, and biology still has to be examined in further studies.

From a psychobiological perspective, stress-induced exhaustion has been related to different dysregulations of the stress response network in the brain, from which only one condition has been related to poststress symptoms [35, 45]. The authors assume that during periods of prolonged stress the dorsal noradrenergic bundle, arising from the locus caeruleus, is permanently activated. Under such conditions the stores of NE cannot be adequately refilled by a new synthesis of NE. This situation may not become relevant as long as the system is stimulated by stress. However, after termination of the stress (at the end of the day) or during rest periods (weekends or vacation) the unstimulated release of NE is expected to be proportional to the availability of NE. This may result in an imbalance of the central and peripheral adrenergic and serotonergic plus vagal systems, which promotes the poststress symptomatology [35]. From such a viewpoint, it is not surprising that exhaustion is only partially associated with poststress symptoms.

Vingerhoets et al. [32] were the first to report gender differences in poststress symptoms. They reported a prevalence of 3.2-3.6% for men and 2.7-3.2% for women. In this study, we found a higher prevalence of poststress symptoms in women (4.4, 24.3, and 41.9%) compared to men (1.7, 15.0, and 14.8%) in healthy controls, outpatients, and inpatients, respectively. These data suggest that poststress symptoms occur most frequently in women, particularly in those who seek hospital treatment for mental illness. Several other findings support comparable gender differences for perceived stress [46] and exhaustion [47]. Recently, Kocalevent et al. [48] addressed this issue from the perspective of a resources-demands model in the general population. First, the authors pointed to small effect sizes for gender effects on chronic stress. Secondly, gender effects seem to be moderated by other variables; in their study, resources (defined by self-efficacy and optimism) were a stronger predictor of exhaustion in women compared to men.

Age was not a significant predictor of poststress symptoms. However, it appeared to be marginally significant. We assume this is due to a rather small effect size and a lack of power in our study. Educational level was the only significant demographic predictor. Most likely, people with a higher education are better able to control their working conditions and to actively cope with ongoing demands. This is probably due to resources like coping mechanisms, a higher self-esteem, control beliefs, social support, and optimism [49, 50]. These findings have great therapeutic value as they underline the importance of adequate coping strategies, individual beliefs, and the importance of a positive attitude in (post-)stress management. Perceived exhaustion is rather an independent and essential predictor of these symptoms, whereas the role of demographic factors seems to be more complex given the smaller effect size.

In sum, the prevalence of poststress symptoms is remarkably frequent and about 17 times higher in inpatients compared to healthy controls. Triggers of poststress symptoms arise not only from the workplace but also from family life and leisure activities [10]. This work gives a comprehensive perspective on the prevalence and early signs of such conditions and common poststress symptoms.

Our measures of poststress symptomatology still need to be validated in future studies. Exploratory evidence, however, shows significant correlations between the number of days of inability to meet everyday requirements and the subjective extent of work ability in outpatients. In outpatients, poststress symptoms are associated with impaired work ability and the number of psychiatric diagnoses according to the PHQ. Thus, the validity of poststress impairments is not unlikely under suitable assessment conditions, preferably in longitudinal prospective studies. In such a context, one may even expect such associations to be mostly reported by patients, who tend to report multiple complaints.

Aside from animal experiments, the neurobiological mechanisms of poststress symptoms in humans are poorly understood. If depletion of NE stores is relevant, nutritional or pharmacological treatments may be promising for the prevention of poststress symptoms (L-tyrosine, protein ingestion; see the study of Deijen [51] for more information). In humans, tyrosine reduced central stress-related symptoms, such as decreased tension, fatigue, confusion, headache, dizziness, and cognitive impairment [52]. Clinical studies are needed to test the relevance of both NE depletion in poststress symptomatology and the function of stress management, relaxation, sufficient work breaks, and periods of regeneration in the prevention of both NE depletion and poststress symptoms. Lastly, an integrative medical approach should focus on biological and biographical factors such as life events and experiences and aim for a healthy interaction between the brain and the body [53]. Clinical practice and relevance and specific biological phenomena and treatments, as well as psychosocial factors, should be the focus of further research [54].

We gratefully acknowledge the invaluable support of Christel Neu and her coworkers of the study office.

The clinical trial in outpatients was funded by the Stiftung Innovation of the State of Rhineland-Palatinate, Germany.

There are no conflicts of interest.