Introduction
Pneumococcal conjugate vaccines (PCVs) were mainly designed to prevent invasive pneumococcal disease (IPD) and have markedly reduced the overall pneumococcal disease burden in children throughout the world [Reference Feikin1, Reference Bennett2]. In Canada, PCV7 was authorized in 2001 and first introduced for universal childhood immunization with a 3 + 1 dose schedule in Alberta in 2002, in British Columbia in 2003 and in Ontario in 2005. A 2 + 1 dose off-label schedule was adopted upon its introduction in Quebec in 2004 and in 2007 in British Columbia (Table 1). PCV10 was used during a short period in Ontario and Quebec in 2009–2010. PCV13 was authorized in Canada according to a 2 + 1 or 3 + 1 schedule for children in 2010 and was introduced in all provinces and territories starting in 2010: 2 + 1 doses for healthy children and 3 + 1 doses for high-risk children [3]. Results from the 2021 (Canadian) Childhood National Immunization Survey indicate an overall 85% coverage (all PCV doses recommended) by age 2 years, 85% in Québec, 86% in Ontario, 81% in Alberta and 88% in British Columbia [4].
Study areas and populations, data source and pneumococcal conjugate programs
Table 1. Long description
The table contains four rows of data corresponding to different Canadian regions:
* Province of British Columbia: Population of 4,817,000 (all ages) and 221,000 (children under 5). P C V 7 introduced in 2003 (3 plus 1) and 2007 (2 plus 1); P C V 13 introduced in 2010 (2 plus 1). Data from Notifiable Disease Registry (Panorama) and Provincial Reference Laboratory (B C C D C).
* Calgary area, Alberta: Population of 1,641,000 (all ages) and 102,000 (children under 5). P C V 7 introduced in 2002 (3 plus 1); P C V 13 introduced in 2010 (2 plus 1). Data from C A S P E R Active, Population-Based, Laboratory-Identified I P D Surveillance Project.
* Toronto-Peel area, Ontario: Population of 4,113,000 (all ages) and 212,000 (children under 5). P C V 7 introduced in 2005 (3 plus 1); P C V 10 in 2009 (3 plus 1); P C V 13 in 2010 (2 plus 1). Data from T I B D N active, population-based, laboratory-identified I P D surveillance.
* Province of Quebec: Population of 8,300,000 (all ages) and 440,000 (children under 5). P C V 7 (2 plus 1) in 2004; P C V 10 (2 plus 1) in 2009 and 2018; P C V 13 (2 plus 1) in 2011. Data from Notifiable Disease Registry (M A D O File), Provincial Reference Laboratory (I N S P Q-L A B O file), and an on-going case-control study.
a Rounded.
PCV childhood programs in Canada led to a 48% reduction in all paediatric IPD cases and an 80% reduction in vaccine-type (VT) IPD cases from the pre-PCV era (2000–2003) to the PCV13 era (2011–2014) [Reference Bettinger5, Reference Wijayasri6]. The overall impact of a pneumococcal program results from the combination of direct protection provided to vaccinated children and the indirect (herd) protection generated in all age groups [Reference Feikin1, Reference Bennett2, Reference Lim7]. It is important to disentangle these two effects and to assess the direct vaccine protection that may vary according to the serotype (ST) [Reference Andrews8–Reference Savulescu10].
In Canada, there are surveillance systems on IPD collecting information on immunization status established in the province of British Columbia and Quebec, in the Calgary region, Alberta, and in the Toronto-Peel region, Ontario (covering about half of the Canadian population). The aim of this study was to estimate the effectiveness of PCV13 against IPD in children less than 5 years of age in these four regions according to the number of doses received, age at vaccine administration, and time since vaccine administration, focusing on ST-3 and ST-19A, which are the two PCV13 serotypes still causing a substantial proportion of IPD cases in children [Reference Griffith11].
Methods
Study population and design
The study analysed data collected from population-based surveillance systems in four geographically defined populations of Canada, and includes cases of IPD occurring between 1 January 2010, and 31 December 2019, in children aged 2 to 59 months (Table 1).
IPD was defined as illness with Streptococcus pneumoniae (Sp) identified by culture or PCR from a normally sterile site. Serotype identification was performed using the Quellung reaction with antisera from the Statens Serum Institut. Starting in 2018 in Quebec, PCR-positive specimens for Sp were serotyped using nucleic acid amplification test (NAAT) at the National Microbiology Laboratory, Public Health Agency of Canada, according to the CDC protocol. High-risk medical status was defined as any indication for a third primary PCV dose (including severe prematurity, immunodeficiency, hemoglobinopathy, asplenia, pulmonary, cardiac, hepatic, neurological or renal chronic disease, diabetes, cochlear implant, cerebrospinal fluid leak, malignancy/cancer, transplantation).
Information regarding risk factors and receipt of pneumococcal vaccine doses prior to the episode of IPD was obtained from medical records at hospitals and primary care clinics, parent and physician interviews, and vaccine registries with some variability according to surveillance systems. IPD cases with unknown or incomplete immunization data (dates of vaccine administration or PCV type), and children who received a PCV dose within 13 days prior to collection of the specimen defining their episode of IPD were excluded from analyses. Up-to-date immunization status was defined as having received 1 dose for a child aged <5 months, 2 doses for those aged 5–12 months, or 3 doses with the last dose received at 11 months of age or older if aged ≥13 months. Missing data on other variables were treated as a dummy variable in models.
Data analysis
The indirect cohort method was applied to assess vaccine effectiveness (VE) [Reference Broome, Facklam and Fraser12]. This is a variation of the case–control methodology in which cases are those with IPD caused by PCV13-serotypes (VT cases), and controls are those with IPD caused by non-PCV13 types or PCV13-related serotypes (NVT/NR controls). This method has the advantage of controlling for measured and unmeasured risk factors, as cases and controls are both IPD cases. PCV13-related serotypes are those belonging to serogroups included in PCV13, and because of possible cross-protection, IPD cases caused by these serotypes were excluded from the analyses.
Vaccine effectiveness (VE) was defined as [1 − odds ratio (OR)] × 100%, OR being the relative odds of a vaccine-type IPD case in individuals vaccinated with PCV13, compared to the odds of this outcome in the absence of vaccination. ORs were computed for specific PCV schedules compared to no immunization using logistic regression models including the Firth correction and adjusting for age group at disease (seven categories: 2–5, 6–11, 12–17, 18–23 months, 2, 3 and 4 years), calendar year at disease (10 single-year categories), and study area (four categories). In all models, the immunization status was classified into mutually exclusive categories (PCV13 schedules, PCV10 schedules, PCV7 schedules, mixed schedules, or unvaccinated, the reference for all VE estimates). PCV13 schedules were analysed as at least one PCV13 dose and according to the specified number of PCV13 doses, age, and time since the last vaccine dose (details on models are presented in the Supplementary Material). Main analyses were performed on all vaccine-type IPD. Subgroup analyses were performed separately and included only IPD cases caused by serotype 19A, serotype 3, other PCV13 serotypes (PCV10 + 6A), serotype 19F, and serotype 7F, compared to NVT/NR controls. VE against PCV10 serotypes and PCV-13 serotypes was also computed for two periods: 2010–2014 and 2015–2019.
VE was computed using fixed-effect logistic regression models with the province as a control variable. Also, as the adjustment for IPD risk factors made very little difference in VE estimates (mostly ≤1%, Supplement 2), and because risk factors were largely controlled by the indirect cohort design, final models did not include this variable. Overall adjusted VE (≥1 dose) was adjusted for year, province, age group at disease, and other PCV schedules, while all subgroup VE calculations were adjusted for year, province, and other PCV schedules. Details on each model are shown in the Supplementary Material.
Mixed models allowing a random effect for the province were also tested. However, results of these led to the conclusion that inter-system variation in the response was too low to attribute any variation to the random effect of this variable. Therefore, only fixed-effect models were retained for main analyses.
The small numbers of cases precluded in-depth analyses of VE against ST-7F and ST-19F. For these two serotypes, alternative models were used with fewer categories for calendar years (seven categories: single year from 2010 to 2013 and two-year categories from 2014 to 2019), for age (four categories: 0, 1, 2 and 3–4 years) and for other PCV schedules (any schedule containing PCV7 or PCV10 was combined). Also, VE was computed for ≥1 PCV13 dose only.
The study protocol was approved by research ethics boards (REB) in each of the four surveillance systems: the Comité d’éthique de la recherche du CHU de Québec – Université Laval, University of British Columbia REB, University of Calgary Conjoint Health REB and Sinai Health REB. De-identified IPD lists were shared with the coordination centre according to data transfer agreements in compliance with applicable laws in each province.
Results
Study population
From 1 January 2010 to 31 December 2019, a total of 1,280 IPD cases were reported in the study populations of which 1116 (87%) had a serotype identified (Figure 1). Of these, 232 were caused by ST-19A (21%), 81 by ST-3 (7%), 76 by other PCV13-serotypes (7%), 118 by PCV13-related serotypes (11%), and 609 were NVT/NR (55%). A total of 349 cases were excluded: those with unknown serotype (n = 164), those with PCV13-related serotypes (n = 118), those with unknown or imprecise immunization status (n = 44), and those who received a PCV13 dose within 13 days prior to their IPD-defining specimen collection (n = 23).
Selection of cases of invasive pneumococcal disease in study areas.
Figure 1. Long description
A flowchart starting at the top with a box containing 1280 eligible I P D cases aged 2-59 months notified during 2010-2019 period.
Below this, the flow splits into two paths. A dotted arrow to the left points to a dashed box of 164 unknown serotype. A solid arrow points down to a box of 1 116 with serotype identified.
From the 1 116 box, three arrows branch out.
1. To the left, a solid arrow leads to 389 P C V 13 included serotypes, consisting of 232 serotype 19 A, 81 S T 3, and 76 other P C V 13 serotypes.
2. In the center, a dotted arrow leads to a dashed box of 118 P C V 13-related serotypes, listing specific counts for serotypes 23 B, 23 A, 9 N, 6 C, 7 C, 7 B, 19 B, and 19 C.
3. To the right, a solid arrow leads to 609 N V T / N R serotypes.
Below the 389 P C V 13 box, a solid arrow leads to 374 complete documentation of P C V immunizations. A dotted arrow with the number 15 indicates cases with incomplete documentation.
Below the 609 N V T / N R box, a solid arrow leads to 580 complete documentation of P C V immunizations. A dotted arrow with the number 29 indicates cases with incomplete documentation.
Both the 374 and 580 boxes have dotted arrows (labeled 7 and 16 respectively) pointing to a central dashed box of 23 cases with a P C V dose received within 0-13 days before specimen collection date.
The final boxes at the bottom are:
- Bottom-left: 367 P C V 13 included serotypes (217 serotype 19 A, 78 serotype 3, 72 other P C V 13 serotypes).
- Bottom-right: 564 N V T / N R I P D controls included in the analyses.
The final analysis included 367 IPD cases caused by PCV13 serotypes and 564 NVT/NR IPD controls. Of these, 15 IPD cases (8 ST-3 cases, 1 ST-19A case, 1 ST-1 case, and 5 NVT/NR controls) were identified by PCR only, and the others were identified by culture. Of the PCV13-serotypes, 217 were ST-19A (59.1%), 78 ST-3 (21.2%), 28 ST-7F (7.6%), 22 ST-19F (6.0%), 5 ST-1 (1.4%), 5 ST-14 (1.4%), 4 ST-6A (1.1%), 3 ST-23F (0.8%), 1 ST-5 (0.3%), 1 ST-6B (0.3%) and 1 ST-4 (0.3%).
The distribution of cases and controls according to age groups, calendar years, immunization schedule, and provinces is presented in Table 2. There were some differences between VT cases and NVT/NT controls by age group, with a higher proportion of VT cases aged 2 years or older. There were also more VT cases occurring during the 2010–2013 period, and more NVT-NR controls immunized with PCV13 only.
Characteristics of selected invasive pneumococcal disease (IPD) cases and controls
Table 2. Long description
The table is structured with a primary header row for Cases (P C V 13-serotypes) and Controls (N V T-N R serotypes). Cases are further subdivided into All P C V 13-serotypes, Serotype 19 A, Serotype 3, and Other P C V 13-serotypes. Each category includes N (count) and % (percentage).
* Age group: The largest control group is 1 year (40%), while cases are more evenly distributed across 2-11 months (21%), 1 year (25%), 2 years (20%), 3 years (16%), and 4 years (18%).
* Calendar year: Cases peaked in 2010-2011 (49%) and declined to 11% by 2018-2019. Conversely, controls increased from 13% in 2010-2011 to 20% in 2018-2019.
* Risk factor: Most subjects had no risk factors (80% of cases, 73% of controls). ‘Yes’ for risk factors accounted for 17% of cases and 21% of controls.
* P C V status: 55% of cases had P C V 7 or P C V 10 containing schedules, while 73% of controls had P C V 13 only. 13% of cases and 6% of controls had No P C V.
* Province: Quebec had the highest representation for both cases (55%) and controls (55%), followed by Ontario (20% cases, 19% controls), British Columbia (15% cases, 17% controls), and Alberta (10% cases, 10% controls).
* Totals: All cases N = 367; All controls N = 564.
a NVT/NR: Non-vaccine type/nor related.
Vaccine effectiveness
VE results are presented in Table 3. VE of ≥1 PCV13 dose against all VT IPD was 70% [95% confidence interval (CI) 49%, 83%] but was lower against ST-19A IPD at 58% [14%, 80%], and even lower against ST-3 IPD at 46% [−36%, 79%], but higher against other PCV13 serotypes with 90% [78%, 96%]. The same pattern was observed for VE with three PCV13 doses.
Adjusted effectiveness of 13-valent pneumococcal conjugate vaccine to prevent invasive pneumococcal disease in four Canadian regions, 2010–2019, according to different immunization schedules
Table 3. Long description
The table presents vaccine effectiveness V E and 95 percent confidence intervals C I for P C V 13.
Model 1: Greater than or equal to 1 dose at any age. Controls V plus 414. All serotypes V E 70 percent. Serotype 19 A V E 58 percent. Serotype 3 V E 46 percent. Other serotypes V E 90 percent.
Model 2: 3 doses at any age. Controls V plus 235. All serotypes V E 77 percent. Serotype 19 A V E 76 percent. Serotype 3 V E 50 percent. Other serotypes V E 92 percent.
Model 3: Up-to-date U T D status.
- 1 dose U T D: All serotypes V E 17 percent. Serotype 3 V E minus 72 percent.
- 2 doses U T D: All serotypes V E 75 percent. Serotype 3 V E 73 percent.
- 3 doses U T D: All serotypes V E 78 percent. Serotype 3 V E 50 percent.
Model 4: By age at disease onset.
- 2 to 11 months: All serotypes V E 72 percent.
- 12 to 23 months: All serotypes V E 89 percent.
- Greater than or equal to 24 months: All serotypes V E 51 percent.
Model 5: By time since last dose (3 doses).
- 0 to 5 months: All serotypes V E 94 percent.
- 6 to 17 months: All serotypes V E 72 percent.
- Greater than or equal to 18 months: All serotypes V E 48 percent.
Footnote: U T D is defined as 1 dose at less than 5 months, 2 doses at 5 to 12 months, or 3 doses at 13 months or older.
V+: number of vaccinated cases or controls VE: estimated vaccine effectiveness.
UDT: Up-to-date immunization status was defined as having received 1 dose at age < 5 months, 2 doses at age 5–12 months, or 3 doses at age ≥ 13 months, with the last dose received at 11 months of age or older.
For all serotype categories, the highest protection with ≥2 PCV13 doses was seen from 12 to 23 months of age (82% against ST-19A, 83% against ST-3, 97% against other PCV13 serotypes) and decreased thereafter in children aged 24 to 59 months, with lowest VE for ST-3 than for the other serotype categories (43% against ST-19A, 13% against ST-3, 77% against other PCV13 serotypes) (Table 3, Model 4).
Similarly, the protection provided by three PCV13 doses was the highest during the first 6 months following the administration of the 3rd dose (94% against all PCV13 serotypes) and waned thereafter (48% 18 to 48 months after last dose). The waning was especially marked for ST-3 with high protection during the first 6 months (89% in the first 6 months) and no substantial protection thereafter (33%, 6–18 months after last dose and −1%, 18 to 48 months after last dose) (Table 3, Model 5).
VE for 3 PCV-13 doses against all PCV10 serotypes +6A with an observation period of less than 1 year following the last vaccine dose was 99% [85%, 100%] for the period 2010–2014 (52 cases and 247 controls), and was 99% [75%, 100%] for the period 2015–2019 (20 cases and 317 controls). Against all PCV-13 serotypes, VE was 97% [90%, 99%] for the period 2010–2014 (277 cases and 247 controls), and was 88% [69%, 96%] for the period 2015–2019 (90 cases and 317 controls).
VE of ≥1 PCV13 dose against ST-7F (28 cases) and ST-19F (22 cases) were 87% [49%, 97%] and 92% [75%, 97%], respectively.
Discussion
Results of this study confirm differences in PCV13 effectiveness against disease due to different serotypes, with effectiveness being lower against serotype 3 and serotype 19A IPD than against all other vaccine serotypes. These results are in line with correlates of protection derived from effectiveness and immunogenicity studies in the UK, showing an average ELISA IgG correlate of protection of 0.98 μg/mL for all PCV13 serotypes, 1.00 μg/mL for ST-19A, 2.83 μg/mL for ST-3, and values below 1.00 μg/mL for other PCV13 serotypes except for ST-19F [Reference Andrews8]. Differences in PCV13 effectiveness according to serotype have also been observed in other studies in which lower values were observed for ST-3 and ST-19A [Reference Andrews9, Reference Savulescu10, Reference Andrejko13]. In our study, the decreasing VE against all PCV-13 included serotypes observed during the 2024–2019 period may be explained by an increased proportion of ST-3 cases in the PCV-13 serotypes category over time. As expected, the protection provided by three PCV13 doses waned over time after the last dose for all serotypes, and this phenomenon was especially marked for ST-3. This was also observed in two other studies using the same design [Reference Andrews9, Reference Savulescu10].
In countries such as Canada, where PCV13 has been used for many years with a high uptake in children, ST-3 and ST-19A are still among the most frequent IPD serotypes identified in all age groups [Reference Griffith11, Reference Garcia Quesada14]. This may be explained by high antibody concentrations that are required to prevent IPD, as discussed above, and also carriage acquisition. In a recent study in the United States, the average ELISA IgG correlate of protection derived from PCV13-induced antibody levels for prevention of Streptococcus pneumoniae colonization in children was 5 μg/mL, five times the value to prevent IPD [Reference Pham and Pichichero15]. The limited impact of PCV13 childhood programs on the overall circulation of these two serotypes is thus not surprising.
In our study, the indirect cohort method was applied to measure VE. The main assumption underlying the method is that immunization with PCV13 does not affect the probability of infection and disease with non-vaccine serotypes [Reference Broome, Facklam and Fraser12]. If vaccinated children are at increasing risk of infection and disease with non-vaccine serotypes, VE values would be overestimated. The hypothesis of bias was tested for PCV in different former studies. In the pivotal randomized trial on PCV7 among children in the US, there was no evidence of an increased risk of disease among pneumococcal vaccine recipients caused by serotypes not contained in the vaccine [Reference Black16]. In an analysis of US Centers for Disease Control and Prevention’s Active Bacterial Core surveillance data, VE results in children based on the indirect cohort design were almost identical to those based on the classic case-(healthy) control design [Reference De Serres, De Serres, Pilishvili and Link-Gelles17]. In two studies on PCV13, sensitivity analyses were performed, and results showed that VE overestimation would be small unless the probability of IPD caused by non-PCV13 serotypes is markedly different between vaccinated and unvaccinated individuals [Reference Savulescu10, Reference Andrews18]. All these observations are reassuring as to the validity of the indirect cohort method to estimate VE in observational studies. The advantage of the indirect cohort analysis is that controls are also IPD cases (of NVT/NR serotype) and share most of the same risk factors, allowing control for several potential confounders; indeed, comorbidities were not confounding (Supplement 2). On the contrary, the year of disease onset was the most important factor to control for in all analyses (Supplement 2). This is related to the fact that PCV13 was implemented without catch-up, with PCV13 immunization coverage among the <5 year old population rising progressively, while the risk of VT disease changed over the years, because of the gradual establishment of herd immunity. As a result, year is associated with VT IPD risk and immunization status, thus an important potential confounding factor. Adjustment for single calendar year in categories, as performed in our analysis, is therefore essential to control for changing epidemiology.
A precise comparison of results from observational effectiveness studies using the same design (i.e., indirect cohort method) cannot be readily performed as there are differences in the schedules considered, in the age at which children are observed and in the duration of follow-up. These characteristics should always be specified when presenting results to allow meaningful comparisons. Also, presenting a VE value combining all PCV13 serotypes is difficult to interpret as it is influenced by the serotype distribution among the cases. In publications, serotype-specific VE should be reported even if confidence intervals are large.
The main limitation of this study and of most others on the same topic is small numbers of cases in many analyses. Logistic regression models often failed to converge using datasets with small numbers of cases, and VE estimates and their confidence intervals may be somewhat biased [Reference Greenland, Mansournia and Altman19, Reference Gosho20]. In our study, regrouping to decrease the number of categories was performed when needed, and the Firth correction was systematically applied in modelling to reduce biases [Reference Stolte, Herbrandt and Ligges21]. In fine, all models presented converged. Exact methods may provide less biased estimators and confidence intervals than those produced by logistic regression models but require extensive computing facilities and are labour-intensive. Also, no adjustment for multiple testing was made in our analyses, as the goal was to provide VE estimates that could be compared using a common metric for confidence intervals and not to test the absolute effectiveness of a particular vaccine or schedule.
In our study, only 6% of IPD cases (22/389) and 7% of IPD controls (45/609) were excluded from analyses because of incomplete documentation of immunization status or because a PCV-13 dose was administered less than 14 days before disease onset (Figure 1). Even if the immunization status of excluded cases was markedly different from those included in analyses, a bias in VE estimates would be small.
Excluded cases with unknown serotype could be different from those included with known serotype in terms of clinical presentation and serotype distribution. Among 164 IPD cases with unknown serotype in our study, 45% had empyema or a pneumonia with pleurisy. Therefore, VE observed in this study represents protection against IPD mostly identified by culture, as it is the case for most estimated VE already published in the literature. In the future, more attention should be paid to VE against IPD cases identified by PCR in pleural fluids.
Conclusion
In this study, PCV13 was effective against all vaccine serotypes. However, the direct protection was lower against ST-19A and ST-3 than against other vaccine serotypes. As ST-3 and ST-19 are still among the most frequent causes of IPD in children, more effective vaccines against these serotypes would be of value.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0950268826102271.
Data availability statement
Data came from confidential, anonymized medical information of patients and were made available to the research team in charge of analyses for exclusive use to the study objective. Data are therefore unavailable because of these restrictions.
Author contribution
Conceptualization: G.D., N.B., A.M., M.N., J.D.K., P.D.W.; Data curation: G.D., A.M., M.N., J.D.K., H.S.; Formal analysis: G.D., N.B.; Investigation: G.D., A.M., M.N., J.D.K., H.S.; Methodology: G.D., N.B., A.M., M.N., J.D.K., P.D.W.; Writing - original draft: G.D., N.B., P.D.W.; Writing - review & editing: G.D., N.B., A.M., M.N., J.D.K., M.S., H.S., P.D.W.; Project administration: A.M., M.N., J.D.K., M.S., P.D.W.; Funding acquisition: M.S., P.D.W.
Funding statement
The study was supported by the Canadian Immunization Research Network. The sponsor had no role in the design and conduct of the study, collection, management, analysis, and interpretation of the data, and preparation, review, and approval of the manuscript. This research received no other external financial or non-financial support. M.S. is supported via a salary award from the BC Children’s Hospital Foundation.
Competing interests
The study was supported by the Canadian Immunization Research Network. A.MG. has investigator-initiated research grants (paid to her institution), from Pfizer, Merck, and Sanofi, and has received honoraria from AstraZeneca, Cidara, GlaxoSmithKline, Leyden Laboratories, Merck, Moderna, Pfizer, and Seqirus for participation in advisory boards, DSMBs, and outcome adjudication committees. J.D.K. has been an investigator on projects funded by GSK Canada, Moderna Canada, and Pfizer Canada, all outside the submitted work. M.S. has been an investigator on projects funded by GlaxoSmithKline, Merck, Moderna, Pfizer, and Sanofi-Pasteur. All funds have been paid to his institute, and he has not received any personal payments. All funds have been paid to the University of Calgary, and he has not received any personal payments. M.S. is supported via a salary award from the BC Children’s Hospital Foundation and has been an investigator on projects funded by GlaxoSmithKline, Merck, Moderna, Pfizer, and Sanofi-Pasteur. All funds have been paid to his institute, and he has not received any personal payments. All other co-authors have no conflicts of interest to disclose.
Open access
