Our study findings highlight five important findings in patients with IPF. First, the high incidence of pre-index multiorgan disease in IPF suggests that rather than a respiratory workup alone, at diagnosis, IPF patients should routinely be worked up for multiorgan disease. Second, the increased incidence of multiorgan disease in IPF cannot be attributed solely to the deleterious effects of smoking as the number of IPF patients with three or more comorbidities was almost double that seen in age, gender and smoking matched controls. Third, several comorbidities (cardiac disease, atrial fibrillation, heart failure, hypothyroidism, GERD, chronic renal failure) showed increased incidence in IPF compared to COPD, 7–10 years before the date of IPF/COPD diagnosis suggesting that mechanisms other than chronic systemic inflammation should be examined to explain increased IPF multimorbidity. Fourth, comorbidities in IPF are clinically important. Prevalent heart failure, chronic renal failure, cerebrovascular disease and abdominal and peripheral vascular disease, all independently associated with mortality, suggesting that IPF patients die from, rather than with these comorbidities. Lastly, incident comorbidities were comparable between IPF and COPD patients despite COPD subjects having a significantly longer follow up period. As disease mechanisms resulting in multiorgan damage appear to propagate through the disease course of IPF, regular routine patient assessment should therefore consider multiorgan evaluation.

The burden of comorbidity in patients with IPF has not been studied as extensively as it has in COPD patients. Estimates of comorbidity prevalence vary widely between studies, but the rates of multiple prevalent comorbidity identified in our study are in line with those documented in prior registry studies [9, 27]. Regarding individual comorbidities, previous studies have generated comparable estimates of the prevalence of hypertension [9, 28], cardiac disease [9, 27], thyroid disease [27], GERD [9], cerebrovascular disease [10], heart failure [10] in IPF populations (Table 1). The prevalence of diabetes was lower in our IPF cohort than prior studies (11% versus 23–45%) [9, 10, 29], but depression was more commonly seen in our study than in prior reports (23% versus 3–15%) [9, 10, 29]. Mendelian randomisation has demonstrated causal associations between hypothyroidism and GERD and the development of IPF in a prior study [30]. This observation aligns with our study where these comorbidities demonstrated increased incidence in subjects who later developed IPF. Conversely however, mendelian randomisation demonstrated that IPF has causal associations with a reduced risk of hypertension [30]. These findings are again supported by our study where hypertension was the only comorbidity more common in the matched control population than IPF subjects. Dementia was found to be less prevalent in the IPF cohort compared to the COPD and matched control populations. This may be an artefact of the small number of overall cases with dementia identified across all three study groups which could lead to spurious statistically significant differences. The limited extent of screening for dementia in the general population during the timeframe of data collection for this study also cautions us to interpret this finding with care.

Comorbidity has been more comprehensively studied in COPD. By matching pack-year smoking histories between IPF and COPD, our COPD population had a lower pack-year smoking exposure than might be seen in commonly reported COPD populations. Nevertheless, despite this reduced smoking burden in our COPD cohort, our estimates for comorbidity prevalence in COPD are similar to those previously published for general COPD cohorts. The prevalence of hypertension in COPD patients reported in the literature ranges from 30 to 52% [24, 25], in keeping with the 41% prevalence we report. Similarly, published prevalence estimates for cardiovascular disease [25, 31, 32], chronic kidney disease [24, 33], diabetes mellitus [24, 25] and heart failure [25] are all comparable to the results we report in the study (Table 1). This concordance between previously published comorbidity estimates in COPD and the estimates we report strengthens the likelihood that our estimates for comorbidity in IPF reflect real-world comorbidity prevalence.

The excess multiorgan comorbidities identified in COPD patients have been ascribed to the occurrence of chronic systemic inflammation, highlighted by the persistently raised levels of inflammatory mediators in peripheral blood [34]. This chronic inflammation can lead to widespread vascular endothelial dysfunction and in turn can result in cardiovascular, renal, and cerebrovascular end-organ damage [35]. IPF is not characteristically considered a disease of chronic inflammation, primarily due to the poor (at times, deleterious) effect of immunosuppressive therapies in the treatment of IPF [35, 36]. However, the impact of endothelial dysfunction in the aetiology of IPF [37,38,39] has received increased attention in recent years and has been speculated as a potential cause for multiorgan disease.

It is possible that a genetic predisposition to vascular endothelial dysfunction may explain the high incidence of cardiac, cerebrovascular, renal, and abdominal and peripheral vascular disease seen in a subset of our IPF population. Microvascular abnormalities are common in IPF, with fibrotic areas being far less vascularised than non-fibrotic areas. This abnormal vascularization in IPF may well inhibit normal repair mechanisms or conversely, be a compensatory mechanism that limits fibrogenesis [40] and abets the theory of IPF as a disease of abnormal wound healing in response to repeated injury of alveolar epithelial cells [41]. The pathogenesis of IPF also involves aberrant local immune responses [42], which may influence the systemic immune landscape and thus contribute to multimorbidity.

The identification of excess telomere shortening and cellular senescence [43] in pulmonary fibrosis has also raised the possibility that IPF represents a pulmonary manifestation of premature ageing. Large volumes of senescent cells have been found to accumulate in IPF lungs [44], but have only shown a limited presence in COPD lungs [45]. Type II alveolar epithelial cells (AECs) have been shown to have short telomeres [46] and to uniquely express senescence markers in IPF [47]. These findings have in turn been shown to lead to age-dependent lung remodelling and fibrosis in mouse models [48]. Lastly, telomere length in peripheral blood leukocytes has been shown to predict survival in IPF patients [49]. Multisystem ageing in a subset of IPF patients could therefore explain the increased levels of comorbidity observed in our study when compared to COPD or matched control subjects.

There were limitations to this study. The diagnosis of IPF relied on electronic health record codes which could have been misclassified by general practitioners. However, we believe miscoding to be limited at best, as survival in our IPF cohort corresponds well with expected life expectancy over the period in which data was collected. Similarly, identification of comorbidities relied on individuals seeking health care services and GPs undertaking relevant tests. It is possible that patients who avoided health services or testing could have been assigned as having no comorbidities. Our prevalent comorbidity rate may therefore have been an underestimate. We were not able to assess data on IPF disease severity (lung function, imaging), hospital or intensive care admission data, or antifibrotic treatment in our population as the study period covered the spectrum of pre-IPF diagnostic guidelines and changes to diagnostic guidelines. We were also unable to access data on the severity of individual comorbidities or their mode of diagnosis but we do not believe these factors would have influenced our estimates of prevalent comorbidities. It is possible that IPF patients with large numbers of comorbidities might not have been as aggressively managed in a hospital setting (regarding intubation and ventilation) when compared to patients with COPD and similar numbers of comorbidities. This in turn could account for some of the impact of multiple comorbidities on survival identified in our IPF patients. The lower pack-year smoking history in our COPD population is notable when compared to more general COPD populations described in the literature. Nevertheless, the fact that comorbidity prevalence rates identified in our study are in keeping with prior reports of comorbidity prevalence in general COPD cohorts suggests that the lower pack-year history in our data may not have influenced our results unduly. Notably, we did not exclude patients with connective tissue disease (CTD) from the COPD or matched control cohorts. CTDs are relatively common, and only a minority have associated fibrosing lung disease. Excluding CTD patients, who often have multi-system disease, might have strongly biased our findings against the presence of comorbidities in the control population. Therefore, when considering that patients with CTD were present in the negative control cohort, our findings of increased comorbidity in the IPF cohort would appear to be strengthened. A final limitation of the study design and analysis is the potential measurement error in variables such as smoking pack-year history that may introduce a degree of residual confounding, as well as the lack of recording of BMI in almost 50% of cases across the three cohorts.

In conclusion our study shows the excess burden of prevalent comorbidities seen in IPF which is not explained by smoking-related damage and is greater than that seen in COPD where systemic inflammation is thought to drive multiorgan disease. The mortality impact of these comorbidities emphasizes the need to consider multiorgan mechanisms of injury in patients that develop IPF.