Fatigue is a normal experience, but for some persons its impact is profound, resulting in concerns regarding their health. In fact, fatigue is one of the commonest complaints reported in primary care, constituting up to 20 % of all presentations [1, 2] and the principal reason for consultation in 5-7 % of cases [1–3].

The psychosocial consequences of fatigue are becoming increasingly recognised. It is strongly associated with poor long-term quality of life [4] and work ability [5]. Indeed, excessively fatigued workers are estimated to cost employers $136.4 billion/year in the USA alone [6]. Despite this, patients’ commonly consider the symptom to be ignored by medical practitioners [7] and approximately 75 % of these patients are not followed up [8]. It may be that doctors are justifiably selective in their follow-up of these patients; however, this paucity of attention may also reflect the perception that fatigue does not affect traditional biomedical outcomes such as premature mortality.

A few studies have reported a possible association between fatigue and mortality, although these suffer from significant methodological limitations: (1) small sample sizes which prohibit the inspection of cause-specific mortality, (2) non-generalisable samples, (3) the use of non-validated single item measures of fatigue which preclude evaluation of dose response and (4) a limited consideration of potential confounders and mechanisms, clearly crucial in the context of such a multifactorial symptom [9, 10].

So why could fatigue result in premature mortality? Some have plausibly hypothesised that fatigue leads to harmful changes in behaviour [9]. Alternatively, the underlying pathophysiology of the symptom itself could lead to poor outcomes – fatigue is a commonly recognised feature of dysfunctional physiology, for example it is a cardinal feature of low rates of metabolism due to hypothyroidism, which in turn is an established risk factor of cardiovascular disease (CVD)-related mortality [11].

The aim of this study was to describe and then understand the reasons for an association between fatigue and mortality (both all-cause and cause-specific) in a large prospective population-based cohort.

Of the 25,639 participants who attended the baseline health assessment, complete SF36-VT data were available for 18,101. In this sample, there were a total of 4397 deaths (CVD: 1410, cancer: 1573, other causes: 1414) with over 300,322 person years of follow-up (mean 16.6 years).

Finally, the effect of excluding all deaths occurring within 2 years of follow-up (n = 4) upon the final models was negligible.

In this large population-based prospective study with long-term follow-up, fatigue was significantly associated with premature mortality, even after adjusting for multiple potential confounders. Compared with those in the lowest quartile of reported fatigue levels, participants in the highest quarter of reported fatigue levels had a 40 % higher risk of premature mortality. This association was observed for CVD-related, but not cancer-related, deaths. Thyroid status and CRP level attenuated these associations although these do not entirely explain the mechanisms which underpin this concerning observation.

As with all studies of this nature, the results must be considered in the context of some possible limitations.

First, although general practices in the UK are considered ideal population sample frames (because almost all people are registered to a general practitioner through the National Health Service), only 40–45 % of registrants provided baseline data, of which approximately 30 % did not complete the SF36-VT questionnaire. Nonetheless, previous descriptions of this cohort have failed to identify significant differences in population characteristics when compared to other national samples, except for a slightly lower prevalence of smokers [12], and no material differences in background characteristics were identified between SF36-VT responders and non-responders. Overall, these results appear generalisable when applied to middle- to old-aged members of the general population, although, owing to the study design, they are unlikely to fully apply to typical healthcare attendees principally presenting with significant fatigue. It is anticipated that such persons will complain of more severe levels of fatigue compared to this study’s modest definition of high fatigue (a self-reported score within the top 25 % of the general population over a period of 1 month). Chronic fatigue disorders, for example, are frequently characterised as disabling and lasting for >6 months [21]. Because our data evidences a consistent dose response between fatigue severity and mortality, these patients are likely to be at even greater risk of premature death in comparison to the population described in this study. These data may, however, not be used to infer the prognosis of patients with chronic fatigue syndrome because information on total fatigue duration was not measured and this may also represent a distinct disorder. Indeed, recent evidence suggests that such patients have a different risk factor profile compared to the general population in terms of mortality causes [22].

Second, there is the potential for pre-existing baseline morbidities, especially CVD, to exaggerate the true association between fatigue and premature mortality. We attempted to address such reverse causality within the analysis by excluding participants who reported baseline CVD and cancer. To further exclude the influence of possible subclinical disease, deaths within 2 years were excluded and baseline aspirin use, a surrogate marker of high CVD risk, was submitted as a covariate. Neither resulted in a meaningful modification of the final models.

Third, there are potential measurement errors in the assessment of exposures. We only used a single measure at one time point to assess each confounder and so did not take into account possible changes over the follow-up period. That said, such random measurement error would probably attenuate any associations observed, so the estimated differences in risk are likely to be larger than those observed.

In relation to other prognostic factors of mortality, which are already considered key targets for intervention in modern-day health services, the observed unadjusted mortality risk of high fatigue (HR 1.63, 95 % CI 1.40–1.89) is similar to being a current smoker (HR 1.35, 95 % CI 1.24–1.48, see Additional file 1: Table S2) in this same sample (although direct comparisons of HRs are not possible owing to differences in the measurement scale). These results are even more meaningful from a public health perspective given the highly prevalent nature of fatigue. Hardy et al. [10] and Moreh et al. [9] also reported similarly sized effects of fatigue on mortality (adjusted HR 1.40, 95 % CI 1.08–1.93 and 1.31, 0.92–1.86, respectively) although their study populations were restricted to the elderly and were small (n = 492 and n = 605, respectively), so did not determine cause-specific mortality [9, 10]. Furthermore, both utilised an unvalidated single item question to measure fatigue, failed to investigate key putative mechanisms such as thyroid status and inflammation, and did not account for the possibility of reverse causality as we have.

Knowledge regarding the mechanisms which cause fatigue is limited and the precise pathophysiology of fatigue has yet to be elucidated. Possible insights may be derived from research examining the concept of ‘burnout’ – a multidimensional construct consisting of emotional exhaustion, cognitive weariness and physical fatigue. Intriguingly, this is also associated with all-cause mortality, in particular CVD [23]. This state has been characterised by reduced hypothalamic pituitary adrenal axis responses [24], low grade inflammation [25] and impaired fibrinolytic capacities [26], all of which have also been independently associated with CVD [27–29]. It is interesting that our study also identified inflammation, as measured by CRP, to be a potentially relevant mechanism. We have also identified the importance of TSH. Of the multiple functions of this hormone, its positive influence on the basal metabolic rate and on catecholamine sensitivity best explain why low levels commonly manifest as fatigue [30]. Furthermore, the hypothyroid state is associated with accelerated atherosclerosis, systemic vascular resistance and subsequent CVD [31]. Though this study does not directly inform whether thyroxine replacement (in those with high abnormal TSH level) would attenuate the mortality risk of fatigue, the data in combination with existing biological knowledge does strongly allude to this.

The persisting significant risk despite adjustment of almost all routinely considered associations highlights the need for greater therapeutic options in this field. This cannot be achieved without further research in this area to unravel the likely complex mechanisms. Based on our findings, such biological mechanisms may be shared with CVD, providing a starting point for further investigation, but other behavioural and psychosocial factors, for example work-shift patterns, should also be considered.

Being in the highest quartile for self-reported fatigue is significantly predictive of death and on a par with recognised risk factors such as active smoking. The risk is likely to be even greater for those presenting with more severe symptoms and is especially associated with CVD-related death. This study therefore indicates the potential seriousness of this frequently dismissed symptom [7]. The attenuation of risk by both behavioural and clinical factors in our models suggests a potential benefit of a comprehensive bio-psychosocial approach. In the future, fatigued persons may benefit from both non-pharmacological interventions, for example physical activity, and pharmacological interventions, for example anti-inflammatory agents.

This study presents the first reliable evidence linking the common symptom of fatigue with excess mortality. Patients have long recognised the importance of fatigue [7]. The biomedical community should now align with this patient priority and more optimally research this problem because further elucidation and targeting of the mechanisms may not only improve crucial patient reported outcomes but conceivably improve life longevity as well.

CI, confidence interval; CRP, C-reactive protein; CVD, cardiovascular disease; EPIC, European Prospective Investigation into Cancer and Nutrition; Hb, haemoglobin; HR, hazard ratio; ICD, International Classification of Disease; ONS, Office for National Statistics; SF36-VT, Short form 36 vitality, TSH, thyroid-stimulating hormone