Invasive Pneumococcal Disease and Long-Term Mortality Rates in Adults, Alberta, Canada

Kristen A. Versluys; Dean T. Eurich; Thomas J. Marrie; Gregory J. Tyrrell


Emerging Infectious Diseases. 2022;28(8):1615-1623. 

In This Article


This study showed that an episode of IPD increases the risk for death not only in the short term, which is expected, but is also a prognostic marker in the intermediate- and long-term periods. The observed aHR for 30-day mortality rate was the highest estimate because acute infection is believed to be directly associated with death. As time after infection increases, risk for death is believed to be influenced by lasting sequalae after acute infection, which this study showed remains substantial.[20] Although the absolute difference in events per PY between cases and controls was nearly 4-fold higher in the initial 30-day period, the absolute event rate remained almost 50% higher throughout the entire follow-up period, irrespective of age or comorbidity level.

Our results were similar to those of previous studies (30-day mortality rate of 14% vs. 13%–21% published, >90-day mortality rate of 31% compared with 10%–42% previously published).[7,8,10,21] In terms of specific risk groups, like others, we observed a higher absolute rate difference in case-patients who had multimorbidities compared with those without multimorbidities irrespective of time frame.[7,8,22] However, the relative HRs compared with those for controls were highest in persons who did not have a comorbidity. A similar trend was seen with increasing age. Although persons <45 years of age had the lowest absolute rate difference in terms of events per PY, the relative increase in deaths compared with that of controls was the highest among persons <45 years of age, and the relative difference decreased with increasing age. Although published reports frequently describe male sex as being a risk factor for increased death from IPD,[7,8] our findings differ. We observed few differences in sex with respect to short- or long-term deaths. The reason for the discrepancy is unknown, but several previous studies were completed in specific populations and locations, whereas our analysis was a large population-based approach, which might partially or fully explain the reported differences. In addition, unknown confounding in previous studies or ours might also explain the differences.

Because our study spanned a wide period, it is useful to recognize advancements in medicine and preventive care for IPD. There have been decreases of aHRs over time, and the gap has decreased particularly in the past 5 years. Although the exact mechanisms of why this decrease is occurring is unknown, some possible explanations are increased use of vaccinations, herd immunity protection, and advances in use of antimicrobial drugs and supportive care.[1] In Canada, vaccine recommendations have been consistent with the 23-valent pneumococcal polysaccharide vaccine recommended for immunocompetent adults ≥65 years of age (the recommended target population) and immunocompromised adults 18–65 years of age. Estimated vaccine uptake in adults >65 years of age during 2014 was ≈37% (in Canada)[3] and increased to ≈53% during 2020–2021 (in Alberta).[23] One possible reason for the increase might be related to policy changes that enabled pharmacists in Alberta to provide routine 23-valent pneumococcal polysaccharide vaccine to eligible adults. In addition, changing serotype distribution and pathogenicity might have influenced differences in mortality rates observed between different periods.

Our study evaluated outcomes for up to 20 years in a cohort of IPD patients, covered a large sample size of persons who were identified from rural and urban areas, and had case ascertainment that is complete as a result of the provincial surveillance system and reportable requirements of IPD, but several limitations to our study should be recognized. First, because of the nature of the data, we were unable to account for some clinical differences (e.g., clinical markers such as blood pressure) that might have existed between patients who had IPD and controls. However, we adjusted for a well-known and validated Elixhauser comorbidity index, and controls were matched on site of care. Although it is not possible to adjust for every variable, our control matching on sex and age and adjustments for comorbidities provide a good understanding of IPD mortality rates. Second, the source of infection was not investigated for this study. It is hypothesized that persons who have nosocomial infections have worse outcomes than persons who have community-acquired infections,[8] and our sample was IPD based on community-acquired infections. Third, the statistical power was low in some stratum analyses in which there were fewer deaths, in particular persons who had limited comorbidities. Thus, CIs were wide and should be interpreted with caution. Moreover, all-cause death was used as the outcome as opposed to a more cause-specific death (i.e., infectious-related death), and cause of death data were not fully available for the cohort, particularly in the early years. Fourth, history of comorbidities was based on the well-validated Elixhauser comorbidity index by using a 5-year history before diagnosis. Thus, comorbidities that might have occurred before this period for which the patient never received any subsequent care or follow-up for the condition could potentially be misclassified. Moreover, it is possible that residual confounding might exist at the level of the individual person (e.g., adherence to treatments, severity of illness) or at the population level (e.g., access to clinical care), which we could not account for in our analyses. Thus, if potential differences exist in this regard between case-patients and controls, the estimates of mortality rates could be potentially confounded. Fifth, enrollment was limited to a single province in Canada, which might limit generalizability of our findings, However, Alberta has a population of >4 million persons, so we do not see this limitation as a major concern.

In conclusion, IPD confers increased short, intermediate, and long-term mortality rates, irrespective of age or comorbidity. In particular, short-term mortality rate outcomes are most noticeable compared with those for controls. However, persons who survive past 30 days are still at increased risk for death. In aging populations at risk, combined with increasing pneumococcal serotype switching and antimicrobial drug resistance,[12,13] IPD remains a major disease. Thus, focused efforts on prevention of IPD and how best to prevent downstream sequalae are required. We believe that our findings might help front-line clinicians in recognizing the high-risk nature of IPD patients, even after the acute event has been managed, and might assist in long-term postdischarge care plans and preventive strategies to mitigate the risk for longer-term adverse events in these patients.