Key Points
Question
What is the incidence of recurrent invasive pneumococcal disease (rIPD) and what characteristics are associated with increased risk?
Findings
This cohort study, which included 7006 adults in Calgary and Toronto who survived a primary episode of IPD, identified an incidence of rIPD of 1468 cases per 100 000 person-years in study year 1 and 146 cases per 100 000 person-years in study years 5 to 15, compared with 9.7 cases per 100 000 person-years in the overall adult population. Risk was higher in adults who were experiencing homelessness, had alcohol use disorder, or had immunocompromising conditions.
Meaning
These findings suggest that rIPD occurred at a persistently higher rate than primary IPD in the general adult population, with vulnerable populations being overrepresented in rIPD.
This cohort study estimates the incidence of recurrent invasive pneumococcal disease among adult patients in Canada and assesses risk factors for recurrence among patients with a primary episode of invasive pneumococcal disease.
Abstract
Importance
Recurrent invasive pneumococcal disease (rIPD) constitutes a clinically relevant proportion of all IPD cases. Improved understanding of these cases can inform priorities for prevention through vaccination.
Objective
To describe the incidence of rIPD and associated risk factors in adults with a primary episode of IPD.
Design, Setting, and Participants
This population-based, multicenter cohort study was performed during active surveillance for IPD in adults in Calgary and the Toronto-Peel regions of Canada from January 1, 2004, to December 31, 2022. rIPD was defined as IPD occurring 30 days or longer after a primary episode. Canadian reference laboratories performed serotyping; population data were obtained from Statistics Canada and the Alberta Interactive Health Data Application. Data were analyzed from September 3, 2024, to November 28, 2025.
Main Outcomes and Measures
Incidence of rIPD over time. Risk factors for rIPD were assessed using multivariable logistic regression.
Results
From 2004 to 2022, 7006 adult patients survived a primary episode of IPD, 274 (3.9%) of whom had rIPD. The median age at primary infection in patients with rIPD was 53.6 (IQR, 41.4-66.0) years; 168 patients (61.3%) were male. The incidence rate ratio (IRR) of rIPD compared with primary IPD rate in surveillance populations was highest in the first year after primary IPD (IRR, 152; 95% CI, 124-185). From 5 to 17 years after primary IPD, the IRR was 15 (95% CI, 11-20). Factors associated with recurrent disease included history of stem cell transplant or hematologic cancer (odds ratio [OR], 5.17; 95% CI, 3.35-7.98), HIV infection (OR, 4.47; 95% CI, 2.84-7.04), experiencing homelessness (OR, 1.87; 95% CI, 1.30-2.69), and alcohol use disorder (OR, 1.50; 95% CI, 1.07-2.11). Primary infection with serotype 3 (OR, 0.33; 95%, 0.16-0.67) or serotype 7F (OR, 0.42; 95% CI, 0.21-0.87), and being 65 years or older (OR, 0.55; 95% CI, 0.34-0.90) were associated with reduced odds of rIPD. At the primary episode, 4812 of 5470 patients (88.0%) were eligible for pneumococcal vaccine, and 1344 of 4812 (27.9%) had been vaccinated. Of patients with known vaccine status between the first and second episodes, 167 of 229 (72.9%) were eligible for vaccine and only 37 of 167 (22.2%) received one.
Conclusions and Relevance
In this cohort study, the risk of rIPD was higher than the risk of primary IPD throughout 17 years of follow-up. Most patients with primary IPD had indications for pneumococcal vaccine but were unvaccinated. These findings suggest that patients with IPD should be prioritized to receive recommended vaccination doses.
Introduction
Invasive pneumococcal disease (IPD) continues to be a major cause of morbidity and mortality at all ages despite programs for routine pneumococcal vaccination in children, older adults, and high-risk younger adults. IPD can also lead to complications and sequelae that include increased frailty, cardiac events, cognitive decline, and reduced life span.1,2,3 Although conjugate vaccines have reduced vaccine-serotype IPD incidence directly in recipients and indirectly in others,4,5 serotype replacement and infections by vaccine-evading serotypes (eg, serotype 3) result in continued challenges in IPD prevention.6,7,8,9
Recurrent IPD (rIPD) is defined as the occurrence of a new episode of IPD more than 30 days after an initial episode.10,11,12,13,14,15 rIPD has been reported in 2.2% to 5.3% of individuals with IPD, and with an estimated incidence 27 times that of primary IPD.10,11,12,13,14,15,16 Characteristics of rIPD have been described in case reports and case series.17,18,19,20,21,22 Immunocompromising conditions are known risk factors,12,14,23 and rIPD may indicate the presence of an underlying immune disorder.14,20 In this cohort study, we aimed to estimate the incidence of rIPD over time and to assess risk factors for rIPD at the primary episode.
Methods
Setting
This cohort study was approved by the Conjoint Health Research Ethics Board, University of Calgary, for the Calgary Area Streptococcus pneumoniae Epidemiology Research (CASPER) study, and by the Research Ethics Boards (REBs) of all Toronto Invasive Bacterial Disease Network (TIBDN) hospitals. All participants provided written or oral consent in the Toronto-Peel region, with an REB-approved waiver of consent for medical record review when participants could not be contacted. In Calgary, written consent was collected until 2018, when the REB approved a waiver of consent. This study followed Strengthening of the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.
In Canada, since 1983, the National Advisory Committee on Immunization (NACI) recommended a dose of 23-valent pneumococcal polysaccharide vaccine (PPV23) for all adults older than 65 years and those aged 18 to 64 years at increased risk of IPD. Publicly funded programs for these risk groups were implemented in Ontario in 1996 and Alberta in 1997. In 2014, 13-valent pneumococcal conjugate vaccine followed by PPV23 was recommended for immunocompromised adults. In 2024, publicly funded programs were updated to a single dose of either 20-valent (PCV20) or 21-valent (PCV21) pneumococcal conjugate vaccine for all adults previously eligible for PPV23.24 Current guidelines do not consider prior IPD as an indication for vaccination.
The TIBDN and the CASPER programs conduct active population-based surveillance for IPD in residents of the Toronto-Peel region in Ontario (population in 2022, 4.54 million) and the Calgary Health Zone in Alberta (population in 2022, 1.78 million), respectively.25,26 Both define IPD as an acute systemic illness with S pneumoniae isolated from a normally sterile site (eg, blood, cerebrospinal fluid, pleural fluid).
Study Design and Data Collection
We pooled data from our population-based surveillance systems to create an inception cohort of adult patients (aged ≥18 years) who survived a primary episode of IPD between January 1, 2004, and December 31, 2022. Case patients were excluded if they were known to have had a primary episode of IPD prior to 2004 and had a single recurrence between 2004 and 2022. Patients with a primary episode prior to 2004 but multiple episodes after 2004 were included, but data prior to 2004 were excluded. rIPD was identified through ongoing surveillance, defined as IPD occurring 30 days or more after the start of a prior episode.10,11,12,13,14,15 Demographic and clinical data were collected by medical record review. Pneumococcal immunization history was obtained from the Alberta electronic immunization public health database in Calgary and from immunization records provided by consenting participants and their vaccine providers in the Toronto-Peel region.
Laboratory Methods
All Calgary isolates were received by the Calgary centralized microbiology laboratory, and all Toronto-Peel isolates were submitted by TIBDN laboratories to the central TIBDN laboratory at Mount Sinai Hospital. Isolates were confirmed as S pneumoniae by standard methods25 and serotyped by latex agglutination using commercial antisera (Statens Serum Institut) and/or Quellung reaction at the Alberta Public Health Laboratory (formerly the National Center for Streptococcus) in Edmonton, the central TIBDN laboratory, or Canada’s National Microbiology Laboratory in Winnipeg.
Statistical Analysis
Data analysis was conducted between September 3, 2024, and November 28, 2025, using StataNow/SE, version 19.5 (StataCorp LLC). The annual incidence of primary IPD (in cases per 100 000 adults per year) was calculated from pooling the incidence from the TIBDN and CASPER surveillance systems. Incidence of rIPD was calculated for the first recurrence as the number of rIPD episodes per 100 000 person-years of follow up in survivors. Because these data were collected by population-based surveillance, we did not have access to either all-cause mortality or data on out-migration from study regions.
The 30-day case-fatality rate for primary episodes and recurrences was compared using χ2 tests. Clinical and demographic characteristics of people with single episodes of IPD and those with rIPD were compared with univariable logistic regression. Adjusted multivariable logistic regression with backward elimination was used to examine factors associated with rIPD. Patient characteristics were assessed at the time of their primary episode. The model was adjusted for variables previously described as associated with rIPD or a higher incidence of IPD,12,14,26,27,28 as well as those identified as associated with rIPD in our univariable analyses. We categorized patients into 5 groups for comorbidities: no comorbidity, immunocompetent comorbidities, HIV, hematopoietic stem cell transplant or hematologic cancer, and other immunocompromising conditions. Immunocompetent comorbidities included any nonimmunocompromising conditions that are an indication for pneumococcal vaccination according to the NACI.25 HIV and hematologic cancer or stem cell transplant were assessed separately due to their higher association with recurrence in previous literature, association in univariable analysis, and high risk of primary IPD.12,14,26 Cox proportional hazards regression was performed as a sensitivity analysis to evaluate the association of IPD serotype, as well as clinical and demographic features over time (year of infection) with outcomes. Two-sided P < .05 indicated statistical significance.
Results
Cohort Derivation
In Calgary and Toronto, a total of 8676 episodes of IPD (6532 in Toronto and 2144 in Calgary) occurred in 8353 patients between January 1, 2004, and December 31, 2022. We excluded 28 patients with IPD episodes prior to 2004 and only 1 rIPD episode during study follow-up, as well as the 1319 patients who died within 30 days of onset of their primary IPD episode. Our inception cohort thus included 7006 patients who survived a primary episode of IPD, allowing them to be at risk for a recurrent episode (eFigure 1 and eTable 1 in Supplement 1).
Incidence of rIPD
Overall, 274 patients (3.9%) who survived their initial IPD episode had 334 rIPD episodes (274 had 1 recurrence, 39 had 2, 9 had 3, and 1 had 4). At the time of the primary IPD episode, 168 of 274 patients (61.3%) with rIPD were male and 106 (38.7%) were female. Compared with patients with primary IPD, those with rIPD were younger (median age, 53.6 [IQR, 41.4-66.0] vs 60.8 [IQR, 47.7-73.3] years; P = .001). Overall, 30-day mortality was 15.8% (1319 of 8325) after a primary episode, 12.4% (34 of 274) after first recurrence, 15.4% (6 of 39) after second recurrences, and 22.2% (2 of 9) after third recurrences (P = .45, χ2 test).
The annual incidence of recurrent disease was highest in the first year after the primary IPD episode: 1468 cases per 100 000 person-years (95% CI, 1206-1788 cases per 100 000 person-years), declining to 456 cases per 100 000 person-years (95% CI, 313-665 cases per 100 000 person-years) in study year 4 and remaining stable from study year 5 to study year 17 at an overall incidence of 149 cases per 100 000 person-years (95% CI, 114-194 cases per 100 000 person-years) (Figure 1). During the study period (excluding 2020-2021 during the COVID-19 pandemic), the overall mean rate of IPD in the general population under CASPER and TIBDN surveillance was 9.7 cases per 100 000 person-years (range, 7.7-12.4 cases per 100 000 person-years) (eFigure 2 in Supplement 1). The incidence rate ratio (IRR) comparing rIPD incidence to incidence of IPD in the general population was 152 (95% CI, 124-185) in the first year after IPD and 15 (95% CI, 11-20) from study year 5 to study year 17 (Figure 1).
Figure 1. Dot Plot Showing Incidence of Recurrent Invasive Pneumococcal Disease Over Time After a Primary Episode.
Error bars and blue shading represent 95% CIs.
Time to Occurrence of rIPD
The median time from initial episode to first recurrence was 22.0 months (IQR, 7.9-47.8 months; range, 1.1-193.6 months). Of the 323 episodes of recurrent disease, 295 had serotypes available for multiple episodes (eg, both the primary episode and first recurrence or isolates from both the first and second recurrence). The serotype was the same for 70 of 295 paired episodes (23.7%). The proportion of recurrences caused by an isolate of the same serotype declined from 24 of 35 (68.6%) when the time to recurrence was 31 to 89 days to 8 of 51 (15.7%) when the time to recurrence was 1 to 2 years and 8 of 127 (6.3%) when the time to recurrence was more than 2 years (Table 1).
Table 1. Association of Time Between Episodes of Invasive Pneumococcal Disease and Recurrences Due to the Same vs a Different Serotype.
| Time between episodes, d | No. of subsequent episodes with same serotype as the prior episode/total No. of recurrent episodes (%)a |
|---|---|
| 31-89 | 24/35 (68.6) |
| 90-179 | 16/34 (47.1) |
| 180-269 | 10/36 (27.8) |
| 270-365 | 4/22 (18.2) |
| 366-730 | 8/51 (15.7) |
| >730 | 8/127 (6.3) |
There were 295 episodes of recurrent invasive pneumococcal disease in which both primary and recurrent episodes had isolates available for serotyping. Patients with multiple recurrences are included with each recurrence compared for the most recent prior episode, as long as serotypes were available for both infecting isolates.
Serotype Distribution in IPD and rIPD
Of 7329 episodes, 6890 (94.0%) had serotypes available. Among 6323 patients with a single episode of IPD, the most common serotypes were 3 (653 [10.3%]), 19A (550 [8.7%]), and 22F (548 [8.7%]) (Figure 2). For the 262 patients with rIPD who had a serotyped primary episode, the most common primary episode serotypes were 19A (27 [10.3%]), 22F (23 [8.8%]), and 4 (18 [6.9%]) (Figure 2). The most common serotypes in the 260 serotyped first recurrences were 19A (23 [8.8%]), 4 (23 [8.8%]), and 22F (20 [7.7%]). Serotypes 3 and 7F were less common in rIPD compared with primary IPD (Figure 2) and were significant in univariable analysis (unadjusted odds ratio [OR] for ST3: 0.26 [95% CI, 0.13-0.53; P < .001]; unadjusted OR for ST7F: 0.36 [95% CI, 0.18-0.73]; P = .005]) compared with other serotypes. For all 6890 IPD episodes, 3882 (56.3%) were caused by serotypes included in 15-valent pneumococcal conjugate vaccine, 4897 (71.1%) by serotypes included in PCV20, and 4912 (71.3%) by serotypes included in PCV21. For the 260 first recurrences with serotyped isolates available, 164 (63.1%) were of serotypes included in PCV20, 184 (70.8%) of serotypes included in PCV21, and 233 (89.6%) of serotypes included in one or both vaccines.
Figure 2. Bar Graph Showing Serotype Distribution of Infecting Isolates in a Single Episode and in Recurrent Invasive Pneumococcal Disease.
Among serotypes not included in any conjugate vaccine (NVT), only serotype 6C (at 172 [2.5%]) of isolates was associated with more than 103 (1.5%) of cases. Serotypes 15B and 15C are reported at 15B/C because a common single step mutation distinguishes them, and their infections cannot reliably be distinguished.
Vaccine Eligibility and Prior Vaccination
Among the 5470 patients with known vaccination history, 4812 (88.0%) were pneumococcal vaccine eligible prior to their primary episode, but only 1344 (27.9%) had received a dose. Six hundred fifty-eight people (12.0%) were younger than 65 years and had no known NACI-defined comorbidity or social or behavioral risk factor making them eligible for pneumococcal vaccine (Table 2). Of these 9 patients (1.4%) had recurrent disease, for an estimated annual rate of 80.5/100 000 (95% CI, 3.6/100 000 to 528/100 000). Data on vaccination between primary and first recurrent episodes of disease were available for 229 patients, of whom 167 (72.9%) were eligible to receive a vaccine dose after their primary episode, and 37 of these (22.2%) received a dose before their recurrent episode.
Table 2. History of Pneumococcal Vaccine Receipt Prior to Primary Episode of Invasive Pneumococcal Disease by Vaccine Eligibility Categorya.
| Vaccination status at primary episode | Vaccine eligibility category, No. (%)b | |||
|---|---|---|---|---|
| Not eligible (n = 658) | Healthy and aged ≥65 y (n = 339) | Aged <65 y with risk condition (n = 2727) | Aged ≥65 y with risk condition (n = 1746) | |
| Not vaccinated | 647 (98.3) | 230 (67.8) | 2304 (84.5) | 934 (53.5) |
| Received 1 vaccine dosec | 10 (1.5) | 88 (26.0) | 345 (12.7) | 677 (38.8) |
| Received ≥2 vaccine doses | 1 (0.2) | 21 (6.2) | 78 (2.9) | 135 (7.7) |
Overall, 5470 of 7006 individuals (78.1%) with a primary episode of disease during the study period had complete information for both vaccination status and underlying comorbidities and behavioral risk factors.
Defined per National Advisory Committee on Immunization guidelines at the time of infection and includes all adults 65 years or older and those aged 18 to 64 years with a medical comorbidity or a behavioral characteristic placing them at risk for invasive pneumococcal disease.24
The single vaccine dose given was 23-valent pneumococcal polysaccharide vaccine in 1077 of 1120 cases (96.2%), 13-valent pneumococcal conjugate vaccine in 36 of 1120 (3.2%), and undocumented or other in 7 of 1120 (0.7%).
Patient Characteristics Associated With rIPD
Variables included in the multivariable logistic regression model for risk factors associated with rIPD are shown in Table 3. In the final model, variables associated with increased odds of rIPD were the presence of alcohol use disorder (AUD) (OR, 1.50; 95% CI, 1.07-2.11), experiencing homelessness (OR, 1.87; 95% CI, 1.30-2.69), and being immunocompromised, particularly due to HIV infection (OR, 4.47; 95% CI, 2.84-7.04) and hematologic cancer or history of stem cell transplant (OR, 5.17; 95% CI, 3.35-7.98). The odds of rIPD were reduced in patients with primary episodes caused by serotypes 3 (OR, 0.33; 95%, 0.16-0.67) or 7F (OR, 0.42; 95% CI, 0.21-0.87) and in those 65 years or older (OR, 0.55; 95% CI, 0.34-0.90) (Table 3). To assess whether the association of serotypes 3 and 7F with reduced rIPD risk was due to changing rates of serotype-specific disease over time, Cox proportional hazards regression incorporating the year of infection was performed (eTable 2 in Supplement 1) and time did not alter the significance of any variables, including serotypes 3 and 7F.
Table 3. Characteristics of Primary Episodes of IPD and Their Association With at Least 1 Episode of Recurrent Infectiona.
| Characteristic | Risk of recurrence, No./total No. (%) | Unadjusted | Adjusted | ||
|---|---|---|---|---|---|
| OR (95% CI) | P value | OR (95% CI) | P value | ||
| Site (n = 7006) | |||||
| Toronto-Peel | 172/5222 (3.3) | 1 [Reference] | NA | 1 [Reference] | NA |
| Calgary | 102/1784 (5.7) | 1.78 (1.38-2.29) | <.001 | 1.35 (1.01-1.80) | .04 |
| Sex (n = 7005) | |||||
| Male | 168/4075 (4.1) | 1 [Reference] | NA | NAb | NA |
| Female | 106/2930 (3.6) | 1.15 (0.89-1.47) | .28 | ||
| Age group, y (n = 7006) | |||||
| 18-34 | 30/627 (4.8) | 1 [Reference] | NA | 1 [Reference] | NA |
| 35-49 | 87/1423 (6.1) | 1.30 (0.85-1.98) | .23 | 1.14 (0.69-1.77) | .56 |
| 50-64 | 82/2136 (3.8) | 0.79 (0.52-1.22) | .29 | 0.73 (0.46-1.15) | .17 |
| ≥65 | 75/2820 (2.7) | 0.54 (0.35-0.84) | .006 | 0.55 (0.34-0.90) | .02 |
| Housing status (n = 6674) | |||||
| Housed | 207/5975 (3.5) | 1 [Reference] | NA | 1 [Reference] | NA |
| Experiencing homelessness | 61/699 (8.7) | 2.66 (1.98-3.59) | .001 | 1.87 (1.30-2.69) | .001 |
| AUD (n = 6580) | |||||
| No | 191/5490 (3.5) | 1 [Reference] | NA | 1 [Reference] | NA |
| Yes | 76/1090 (7.0) | 2.08 (1.58-2.74) | <.001 | 1.50 (1.07-2.11) | .02 |
| Smoking status (n = 6150) | |||||
| Does not smoke | 135/3800 (3.6) | 1 [Reference] | NA | NAb | NA |
| Smokes currently | 113/2350 (4.8) | 1.37 (1.06-1.77) | .02 | ||
| Injection drug or crack cocaine use (n = 6583) | |||||
| No | 241/6240 (3.9) | 1 [Reference] | NA | NAb | NA |
| Yes | 26/343 (7.6) | 2.04 (1.34-3.11) | .001 | ||
| Comorbidities present (n = 6586)c | |||||
| None | 66/2298 (2.9) | 1 [Reference] | NA | 1 [Reference] | NA |
| ≥1, No immunocompromise | 80/2448 (3.3) | 1.14 (0.82-1.59) | .43 | 1.40 (0.98-2.00) | .06 |
| HIV | 38/248 (15.3) | 6.12 (4.00-9.35) | <.001 | 4.47 (2.84-7.04) | <.001 |
| SCT or hematologic cancer | 47/540 (8.7) | 3.22 (2.19-4.75) | <.001 | 5.17 (3.35-7.98) | <.001 |
| Other immunosuppression | 36/1052 (3.4) | 1.20 (0.79-1.81) | .39 | 1.74 (1.12-2.72) | .01 |
| Vaccination prior to first IPD episode (n = 5508) | |||||
| None | 181/4144 (4.4) | 1 [Reference] | NA | NAb | NA |
| 1 Dose, any pneumococcal vaccine | 52/1129 (4.6) | 1.05 (0.77-1.45) | .73 | ||
| ≥2 Doses any pneumococcal vaccine | 14/235 (6.0) | 1.39 (0.79-2.43) | .25 | ||
| Site of infection (n = 6654) | |||||
| Bacteremia without focus | 36/722 (5.0) | 1 [Reference] | NA | NAb | NA |
| Pneumonia or empyema | 200/5092 (3.9) | 0.78 (0.54-1.12) | .18 | ||
| Meningitis | 14/411 (3.4) | 0.67 (0.36-1.26) | .22 | ||
| Other | 18/429 (4.2) | 0.83 (0.47-1.49) | .54 | ||
| Serotype (n = 6585) | |||||
| All other serotypes | 246/5446 (4.5) | 1 [Reference] | NA | 1 [Reference] | NA |
| Serotype 3 | 8/661 (1.2) | 0.26 (0.13-0.53) | <.001 | 0.33 (0.16-0.67) | .002 |
| Serotype 7F | 8/478 (1.7) | 0.36 (0.18-0.73) | .005 | 0.42 (0.21-0.87) | .02 |
Abbreviations: AUD, alcohol use disorder; IPD, invasive pneumococcal disease; NA, not applicable; OR, odds ratio; SCT, stem cell transplant.
Cases missing data for relevant variables were removed from multivariable models. The final model retained 6188 cases (88.3%). The number for univariable analysis varies based on missing data in the variable; totals are given in parentheses in the header for each variable.
For the multivariable model, sex was included but was not significant. Vaccination before primary episode, smoking status, infection drug use or crack cocaine use, and site of infection were removed by backward elimination due to nonsignificance in multivariable model.
Illness classified as immunocompetent or immunocompromised, and underlying illnesses whose presence makes adults younger than 65 years eligible for pneumococcal vaccine per National Advisory Committee on Immunization recommendations.24
Discussion
The proportion of IPD that was rIPD in Calgary and Toronto from 2004 to 2022 was 3.9%, similar to previously reported proportions ranging from 2.1% to 5.3%.11,12,13,14,15,16 The median age at the time of the primary episode was younger among patients with rIPD than among those with single episodes. The Active Bacterial Core Surveillance study in the US also described people aged 18 to 64 years at increased risk for recurrence compared with adults 65 years or older.12 This may be in part due to younger adults having more years of life in which to have a future episode of IPD, but is an important contrast to IPD in general where older adults are at highest risk for disease and poor outcomes.
In Canada, the NACI statement on high-risk conditions for IPD include individuals with AUD and those experiencing homelessness.24 In Calgary and Toronto, rIPD was associated with these characteristics, and other studies have also observed higher levels of AUD in individuals with rIPD compared with those who have had single episodes.11 AUD can have adverse effects on the immune system and predispose individuals to both pneumococcal infections29,30 and mortality from IPD.31 One study showed an 11-fold higher incidence in adults with AUD vs healthy adults.32 Furthermore, AUD is a risk factor for hepatic cirrhosis, which is also a risk for IPD.30 Some independent risk factors for rIPD may be clustered together. For example, in Calgary, prior reports have highlighted the importance of experiencing homelessness for younger individuals having IPD, as well as AUD, HIV infection, and serotype-specific community outbreaks.27,28
The NACI currently recommends vaccination with PCV20 or PCV21 for any adult at risk for IPD.24 Immunocompromising comorbidities such as a history of stem cell transplant, hematologic cancers, and HIV infection were associated with increased odds of rIPD in Toronto, Calgary, and elsewhere.12,13 Patients with these conditions are at high risk for IPD in general,26 and vaccines may offer only limited protection. With the availability of both PCV20 and PCV21, individuals at highest risk may benefit from receipt of both vaccines.24 The PCV20 vaccine contains important serotypes not included in PCV21, including serotypes 4 and 9V. Serotype 4 in particular was a reemerging cause of IPD in younger adults in our regions during the COVID-19 pandemic.9
Overall, 88.0% of patients were vaccine eligible prior to their primary episode of IPD, and many had multiple risk factors. However, only 27.9% had been vaccinated prior to their primary episode, and only 22.2% of eligible patients received a dose of vaccine after the primary episode. In Canada in 2023, a National Survey estimated that 38.5% of adults were vaccinated against pneumococcal disease.33
To our knowledge, our finding that primary infection with isolates of serotypes 3 and 7F was associated with lower odds of rIPD is novel and unexplained. The association with serotype 7F is clinically meaningful but not statistically significant, and the variation in the incidence of serotype 7F over time may have resulted in confounding. However, the incidence of disease due to serotype 3 has been stable over time, and adjustment for time did not alter the association (eTable 1 in Supplement 1).8,34 Serotype 3 has been associated with higher morbidity and mortality,35,36 suggesting that immune responses to an infection might result in increased protection against other serotypes. Conversely, a recent meta-analysis reported that serotype 7F IPD infections have a lower risk of mortality than other serotypes.36
Limitations
This study has some limitations. We analyzed an inception cohort of patients who survived 1 episode of IPD and could develop recurrent IPD; therefore, the results can only be generalized to individuals who have had a primary episode of IPD. Data for prior vaccination were missing for 1498 of 7006 patients (12.4%) at both sites, and smoking history was missing for 856 of 7006 patients (12.2%) at both sites due to incomplete information in medical record reviews. Medical record review data may also be inaccurate in the assessment of injection drug use, AUD, homelessness, and duration and dose of corticosteroid use. We would also have missed recurrent disease if a patient moved in or out of our population areas between a primary and a recurrent episode. We did not have data on mortality due to other causes or loss to follow-up by out-migration. We did not have data on follow-up of the cohort apart from rIPD cases. Any of these limitations would bias our incidence estimates toward the null, as denominators will be inflated. However, the absence of these data means that we could not perform a time-to-event analysis with appropriate censoring and competing risk. Although Cox proportional hazards regression revealed that time to rIPD event after primary IPD was not a significant risk factor for recurrence, it could not include censoring or competing risk. Thus, the results of our analysis of risk factors for recurrence should be interpreted with caution.
Our sample size was too small to assess vaccine effectiveness in preventing rIPD. The higher risk of rIPD due to the same serotype in the first year after initial infection raises the question of whether infection has provided any protective immunity in these patients. While the point estimate of the incidence of rIPD in patients without recognized risk factors at the time of their primary episode was high, the 95% CIs are wide, and it is not certain to what extent their risk exceeds that of the general population. Our multivariable model controlled for demographic and clinical factors that may be associated with specific serotypes; however, we did not control for other exposures such as antecedent viral infections nor for variation in innate or adaptive immune responses that may influence the likelihood of recurrent infections.
Conclusions
In this cohort study, we aimed to inform recommendations for pneumococcal conjugate vaccine use in adults by estimating the incidence of rIPD over time, describing its clinical and microbiological characteristics, and assessing risk factors for occurrence. We found that immunocompromised adults, individuals experiencing homelessness, and individuals with AUD had increased risk of rIPD, and prior vaccination rates were low in all groups. The risk of rIPD was highest early after the primary episode, but substantial increased risk persisted for at least 17 years. More than three-quarters of recurrences were caused by a different serotype than the primary episode, and serotypes in PCV20 covered 71.1% of isolates causing rIPD while those in PCV21 71.3%, with 89.6% of isolates being of serotypes included in one or the other vaccine. The risk of rIPD in patients without indications for pneumococcal vaccine prior to their primary episode was substantially higher than the general population risk, although 95% CIs were wide.
These observations suggest that eligible unvaccinated individuals should be vaccinated promptly after an episode of IPD and, because many adults at risk of rIPD are likely to have challenges accessing vaccines, consideration should be given to providing this vaccine during the index hospitalization.10,37 However, in this specific population, vaccine recommendations are based on expert advice, without direct evidence that vaccination will reduce the risk of rIPD. Further research is needed to assess the effectiveness of conjugate vaccines in preventing rIPD to better understand the immunologic factors that are associated with the risk of same-serotype recurrence during the year after the primary episode, and to further investigate our finding that infection with serotypes 3 and 7F are associated with a reduced risk of rIPD.
eFigure 1. Calgary and Toronto Multicenter Surveillance Populations for Final Inclusion in Inception Cohort to Examine Recurrent IPD
eFigure 2. Incidence of IPD per 100 000 Adults per Year in Toronto-Peel and Calgary Regions Combined
eTable 1. Clinical and Demographic Features of People at the Time of Their Primary IPD Episode Stratified by Those With a Single IPD Episode and Those Who Had Recurrent IPD
eTable 2. Time-to-Event Analysis Using Cox Proportional Hazards Regression to Examine Association of Time With rIPD and Risk Factors
Nonauthor Collaborators
Data Sharing Statement
References
- 1.Brooks LRK, Mias GI. Streptococcus pneumoniae’s virulence and host immunity: aging, diagnostics, and prevention. Front Immunol. 2018;9:1366. doi: 10.3389/fimmu.2018.01366 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Jit M. The risk of sequelae due to pneumococcal meningitis in high-income countries: a systematic review and meta-analysis. J Infect. 2010;61(2):114-124. doi: 10.1016/j.jinf.2010.04.008 [DOI] [PubMed] [Google Scholar]
- 3.Lucas MJ, Brouwer MC, van de Beek D. Neurological sequelae of bacterial meningitis. J Infect. 2016;73(1):18-27. doi: 10.1016/j.jinf.2016.04.009 [DOI] [PubMed] [Google Scholar]
- 4.Feikin DR, Kagucia EW, Loo JD, et al. ; Serotype Replacement Study Group . Serotype-specific changes in invasive pneumococcal disease after pneumococcal conjugate vaccine introduction: a pooled analysis of multiple surveillance sites. PLoS Med. 2013;10(9):e1001517. doi: 10.1371/journal.pmed.1001517 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Bennett JC, Deloria Knoll M, Kagucia EW, et al. ; PSERENADE Team . Global impact of ten-valent and 13-valent pneumococcal conjugate vaccines on invasive pneumococcal disease in all ages (the PSERENADE project): a global surveillance analysis. Lancet Infect Dis. 2025;25(4):457-470. doi: 10.1016/S1473-3099(24)00665-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wijayasri S, Hillier K, Lim GH, Harris TM, Wilson SE, Deeks SL. The shifting epidemiology and serotype distribution of invasive pneumococcal disease in Ontario, Canada, 2007-2017. PLoS One. 2019;14(12):e0226353. doi: 10.1371/journal.pone.0226353 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Grewal R, Hillier K, Deeks SL, et al. Invasive pneumococcal disease epidemiology and serotype replacement after the introduction of the 13-valent pneumococcal conjugate vaccine in Ontario, Canada, 2007-2022. Open Forum Infect Dis. 2024;11(6):ofae275. doi: 10.1093/ofid/ofae275 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Garcia Quesada M, Peterson ME, Bennett JC, et al. ; PSERENADE Team . Serotype distribution of remaining invasive pneumococcal disease after extensive use of ten-valent and 13-valent pneumococcal conjugate vaccines (the PSERENADE project): a global surveillance analysis. Lancet Infect Dis. 2025;25(4):445-456. doi: 10.1016/S1473-3099(24)00588-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Ricketson LJ, Kellner JD. Changes in the incidence of invasive pneumococcal disease in Calgary, Canada, during the SARS-CoV-2 pandemic 2020-2022. Microorganisms. 2023;11(5):1333. doi: 10.3390/microorganisms11051333 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Malo JA, Ware RS, Lambert SB. Estimating the risk of recurrent invasive pneumococcal disease in Australia, 1991-2016. Vaccine. 2021;39(40):5748-5756. doi: 10.1016/j.vaccine.2021.08.095 [DOI] [PubMed] [Google Scholar]
- 11.Mufson MA, Hao JB, Stanek RJ, Norton NB. Clinical features of patients with recurrent invasive Streptococcus pneumoniae disease. Am J Med Sci. 2012;343(4):303-309. doi: 10.1097/MAJ.0b013e31822d9860 [DOI] [PubMed] [Google Scholar]
- 12.King MD, Whitney CG, Parekh F, Farley MM; Active Bacterial Core Surveillance Team/Emerging Infections Program Network . Recurrent invasive pneumococcal disease: a population-based assessment. Clin Infect Dis. 2003;37(8):1029-1036. doi: 10.1086/377736 [DOI] [PubMed] [Google Scholar]
- 13.Turett GS, Blum S, Telzak EE. Recurrent pneumococcal bacteremia: risk factors and outcomes. Arch Intern Med. 2001;161(17):2141-2144. doi: 10.1001/archinte.161.17.2141 [DOI] [PubMed] [Google Scholar]
- 14.Einarsdóttir HM, Erlendsdóttir H, Kristinsson KG, Gottfredsson M. Nationwide study of recurrent invasive pneumococcal infections in a population with a low prevalence of human immunodeficiency virus infection. Clin Microbiol Infect. 2005;11(9):744-749. doi: 10.1111/j.1469-0691.2005.01221.x [DOI] [PubMed] [Google Scholar]
- 15.Sanz JC, Rodríguez-Avial I, Ríos E, Ramos B, Cercenado E. Recurrent pneumococcal invasive disease in the region of Madrid during a five-year period. Infection. 2014;42(3):475-483. doi: 10.1007/s15010-013-0571-3 [DOI] [PubMed] [Google Scholar]
- 16.Font B, Llimiñana C, Fontanals D, Pineda V, Segura F. Eleven-year study of recurrent pneumococcal bacteremia. Eur J Clin Microbiol Infect Dis. 2001;20(9):636-638. doi: 10.1007/s100960100580 [DOI] [PubMed] [Google Scholar]
- 17.Murad Y, Hung TY, Sadarangani M, et al. ; Members of IMPACT . Clinical presentations and outcomes of children in Canada with recurrent invasive pneumococcal disease from the IMPACT Surveillance Network. Pediatr Infect Dis J. 2022;41(4):e166-e171. doi: 10.1097/INF.0000000000003454 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bingen E, Doit C, Bouillie C, et al. Multiple recurrences and relapse of Streptococcus pneumoniae meningitis. Lancet. 1995;346(8970):311. doi: 10.1016/S0140-6736(95)92197-4 [DOI] [PubMed] [Google Scholar]
- 19.Whitecar JP Jr, Reddin JL, Spink WW. Recurrent pneumococcal meningitis: a review of the literature and studies on a patient who recovered from eleven attacks caused by five serotypes of Diplococcus pneumoniae. N Engl J Med. 1966;274(23):1285-1289. doi: 10.1056/NEJM196606092742302 [DOI] [PubMed] [Google Scholar]
- 20.Rodríguez-Créixems M, Muñoz P, Miranda E, Peláez T, Alonso R, Bouza E. Recurrent pneumococcal bacteremia: a warning of immunodeficiency. Arch Intern Med. 1996;156(13):1429-1434. doi: 10.1001/archinte.1996.00440120087008 [DOI] [PubMed] [Google Scholar]
- 21.Coccia MR, Facklam RR, Saravolatz LD, Manzor O. Recurrent pneumococcal bacteremia: 34 episodes in 15 patients. Clin Infect Dis. 1998;26(4):982-985. doi: 10.1086/513924 [DOI] [PubMed] [Google Scholar]
- 22.Gelfand EW, Rao CP, Minta JO, Ham T, Purkall DB, Ruddy S. Inherited deficiency of properdin and C2 in a patient with recurrent bacteremia. Am J Med. 1987;82(3 Spec No):671-675. doi: 10.1016/0002-9343(87)90122-7 [DOI] [PubMed] [Google Scholar]
- 23.McEllistrem MC, Mendelsohn AB, Pass MA, et al. Recurrent invasive pneumococcal disease in individuals with human immunodeficiency virus infection. J Infect Dis. 2002;185(9):1364-1368. doi: 10.1086/339882 [DOI] [PubMed] [Google Scholar]
- 24.National Advisory Committee on Immunization . Pneumococcal vaccines: Canadian Immunization Guide. 2025. Accessed October 25, 2025. https://www.canada.ca/en/public-health/services/publications/healthy-living/canadian-immunization-guide-part-4-active-vaccines/page-16-pneumococcal-vaccine.html
- 25.Ricketson LJ, Nettel-Aguirre A, Vanderkooi OG, Laupland KB, Kellner JD. Factors influencing early and late mortality in adults with invasive pneumococcal disease in Calgary, Canada: a prospective surveillance study. PLoS One. 2013;8(10):e71924. doi: 10.1371/journal.pone.0071924 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Shigayeva A, Rudnick W, Green K, et al. ; Toronto Invasive Bacterial Diseases Network . Invasive pneumococcal disease among immunocompromised persons: implications for vaccination programs. Clin Infect Dis. 2016;62(2):139-147. doi: 10.1093/cid/civ803 [DOI] [PubMed] [Google Scholar]
- 27.Vanderkooi OG, Church DL, MacDonald J, Zucol F, Kellner JD. Community-based outbreaks in vulnerable populations of invasive infections caused by Streptococcus pneumoniae serotypes 5 and 8 in Calgary, Canada. PLoS One. 2011;6(12):e28547. doi: 10.1371/journal.pone.0028547 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Lemay JA, Ricketson LJ, Zwicker L, Kellner JD. Homelessness in adults with invasive pneumococcal disease (IPD) in Calgary, Canada. Open Forum Infect Dis. 2019;6(10):ofz362. doi: 10.1093/ofid/ofz362 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Grau I, Ardanuy C, Calatayud L, Schulze MH, Liñares J, Pallares R. Smoking and alcohol abuse are the most preventable risk factors for invasive pneumonia and other pneumococcal infections. Int J Infect Dis. 2014;25:59-64. doi: 10.1016/j.ijid.2013.12.013 [DOI] [PubMed] [Google Scholar]
- 30.Bhatty M, Pruett SB, Swiatlo E, Nanduri B. Alcohol abuse and Streptococcus pneumoniae infections: consideration of virulence factors and impaired immune responses. Alcohol. 2011;45(6):523-539. doi: 10.1016/j.alcohol.2011.02.305 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Harboe ZB, Thomsen RW, Riis A, et al. Pneumococcal serotypes and mortality following invasive pneumococcal disease: a population-based cohort study. PLoS Med. 2009;6(5):e1000081. doi: 10.1371/journal.pmed.1000081 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Kyaw MH, Rose CE Jr, Fry AM, et al. ; Active Bacterial Core Surveillance Program of the Emerging Infections Program Network . The influence of chronic illnesses on the incidence of invasive pneumococcal disease in adults. J Infect Dis. 2005;192(3):377-386. doi: 10.1086/431521 [DOI] [PubMed] [Google Scholar]
- 33.Adult National Immunization Coverage Survey. Government of Canada . 2023. Accessed November 29, 2025. https://www.canada.ca/en/public-health/services/immunization-vaccines/vaccination-coverage/adult-national-immunization-coverage-survey-2023-results.html
- 34.Griffith A, Golden AR, Lefebvre B, et al. Invasive pneumococcal disease surveillance in Canada, 2021-2022. Can Commun Dis Rep. 2024;50(5):121-134. doi: 10.14745/ccdr.v50i05a02 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Luck JN, Tettelin H, Orihuela CJ. Sugar-coated killer: serotype 3 pneumococcal disease. Front Cell Infect Microbiol. 2020;10:613287. doi: 10.3389/fcimb.2020.613287 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Darkwah S, Somda NS, Mahazu S, Donkor ES. Pneumococcal serotypes and their association with death risk in invasive pneumococcal disease: a systematic review and meta-analysis. Front Med (Lausanne). 2025;12:1566502. doi: 10.3389/fmed.2025.1566502 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.McCosker LK, El-Heneidy A, Seale H, Ware RS, Downes MJ. Strategies to improve vaccination rates in people who are homeless: a systematic review. Vaccine. 2022;40(23):3109-3126. doi: 10.1016/j.vaccine.2022.04.022 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eFigure 1. Calgary and Toronto Multicenter Surveillance Populations for Final Inclusion in Inception Cohort to Examine Recurrent IPD
eFigure 2. Incidence of IPD per 100 000 Adults per Year in Toronto-Peel and Calgary Regions Combined
eTable 1. Clinical and Demographic Features of People at the Time of Their Primary IPD Episode Stratified by Those With a Single IPD Episode and Those Who Had Recurrent IPD
eTable 2. Time-to-Event Analysis Using Cox Proportional Hazards Regression to Examine Association of Time With rIPD and Risk Factors
Nonauthor Collaborators
Data Sharing Statement
