Is Pneumonia a Risk Factor or a Risk Marker for Long-Term Mortality?

Pneumonia is a leading cause of morbidity and short-term mortality (usually measured in the first 30 days after diagnosis) (1–4). However, the potential long-term consequences of pneumonia remain an area of intense evaluation. Several studies have focused on the long-term risk for mortality among patients with pneumonia. In most studies, patients with more severe pneumonia had a higher risk for long-term mortality relative to patients with less severe pneumonia. As older patients with debilitating comorbidities, including diabetes and glucose homeostasis disturbances, are usually classified in the higher-risk categories, this could explain, at least in part, the higher risk observed among patients with more severe pneumonia (3–8).

Most previous studies have been relatively small in size or with short follow-up, restricted to patients hospitalized with pneumonia, or did not include a comparison group of patients without pneumonia. In this issue of the Journal, Eurich and colleagues (pp. 597–604) present a large, prospective cohort study that compares long-term mortality among adult patients enrolled during 2000–2002 with an emergency department visit or hospitalization for pneumonia and an age- and sex-matched comparison group selected among patients without pneumonia from the same settings and period (controls) (9). Investigators also used linked administrative databases to comprehensively monitor subsequent medical encounters for both pneumonia cases and patients without pneumonia. During a median follow-up of 9.8 years, long-term all-cause mortality was significantly higher among patients who had experienced pneumonia compared with controls, with an adjusted hazard ratio of 1.65 (95% confidence interval, 1.57–1.73). In particular, deaths involving the respiratory system were more common among patients who had experienced pneumonia than among controls (24% vs. 9%).

The increased risk for long-term mortality in patients with pneumonia relative to controls was consistently observed across all age groups and for both hospital and emergency department settings. The study design also allowed the exploration of subsequent hospitalizations and emergency department visits recorded during follow-up. The rates of all-cause and pneumonia hospitalizations and emergency department visits were also increased among patients who had experienced pneumonia relative to matched controls without pneumonia.

Efforts to improve prevention and management of pneumonia are clearly necessary and would reduce the related short-term mortality. However, whether the findings reported by Eurich and colleagues could be used directly to project long-term mortality reductions associated with improvements in pneumonia prevention and management remains unclear. By necessity, these studies are nonexperimental, and accounting for factors that are difficult to measure (e.g., patients’ frailty, smoking) is challenging (10). Establishing a causal association between pneumonia and long-term mortality is difficult. The potential effect of pneumonia prevention and management needs to consider whether pneumonia is a risk factor or a risk marker for long-term mortality. This distinction is important because if pneumonia is a causal risk factor, prevention of pneumonia (e.g., through effective vaccination) could indeed improve long-term survival. Alternatively, if pneumonia is only a risk marker of an underlying and possibly unrecognized process that increases the risk for death, pneumonia could be useful to identify patients at risk, but modifications of pneumonia occurrence would likely have no effect on long-term mortality.

It is interesting to note that more severe pneumonia has a stronger association with poor long-term survival among all patients with pneumonia and that the association between pneumonia and long-term mortality remains significant after statistical adjustment for age and measured comorbidities. Furthermore, many patients who experienced pneumonia suffered from recurrent events (3, 4). Whether some of these patterns represent manifestations of an underlying process is unknown. A provocative study conducted in young children with severe or recurrent invasive pneumococcal disease recently described the identification of clear, yet previously unrecognized immunodeficiencies in 26% of those children (11). Severe invasive disease in these instances served as a marker of those underlying processes. It is possible that some patients with more severe or recurrent pneumonia, especially those with no obvious risk factors, represent a distinct high-risk subgroup that could benefit from additional evaluation and preventative measures. Similarly, smoking increases the risk for pneumonia (10) and also increases the long-term risk for mortality. Thus, some smokers who experienced pneumonia will have a high long-term risk of mortality that could be attributed to smoking, rather than pneumonia. In this simplified scenario, targeting pneumonia would reduce the related short-term mortality risk, but would not necessarily affect the long-term risk. However, targeting the underlying process (i.e., smoking) would reduce pneumonia and the short- and long-term mortality.

Although several studies have consistently observed an association between pneumonia and long-term mortality, several questions remain unanswered. First, the biological plausibility of a causal association has not been clearly established. Increased preinfection levels of markers of inflammation are associated with an increased risk for pneumonia (12). Some studies have also noted an inflammatory and procoagulant state among patients with pneumonia that persists well beyond the clinical manifestations of the disease (13–15). There is a strong association between pneumonia and subsequent cardiovascular disease and decreases in renal function (2, 16). Similarly, existing or new-onset comorbidities, including diabetes and glucose disturbances, are also associated with a higher risk for mortality among patients with pneumonia (7). Although several mechanisms could be involved in these associations, a unifying explanation is yet to be formulated (3, 4). Second, although conditions that significantly increase all-cause mortality are of great interest, the use of the nonspecific all-cause long-term mortality as the study outcome is challenging, as pneumonia is unlikely to affect equally all causes of long-term mortality. The reported difference in deaths resulting from diseases of the respiratory system is valuable, but whether pneumonia increases the risk for long-term mortality as a result of a specific respiratory disease is unclear. That assessment may provide insights in potential underlying respiratory processes that either preceded or were triggered by the pneumonia event. Third, pneumonia encompasses a number of different causes, including several pathogens that are difficult to identify (1). In most previous studies, pneumonia has been used as a generic disease without regard for the specific etiology or the management received. Some studies have described an increased long-term mortality after pneumococcal pneumonia compared with the expected mortality from the general population (17). Nevertheless, those assessments and potential comparisons with pneumonia of different etiologies may be difficult because of the limitations of current bacterial and viral diagnostics (1). Whether the observed long-term mortality varies by the pneumonia etiology remains unclear.

Identification and recognition of patients with pneumonia as a group at high risk for long-term mortality seems warranted. Additional studies are needed to complement the thoughtful work by Eurich and colleagues and to clarify to what extent pneumonia is an independent cause of long-term mortality or whether pneumonia serves as a marker of another underlying process. Meanwhile, addressing known modifiable causes of pneumonia and long-term mortality, such as smoking, would maximize the public health effect on the short- and long-term consequences of pneumonia.