Invasive Pneumococcal Disease burden, clinical characteristics, serotypes’ distribution, immunization status and antimicrobial resistance: Evidence from 12-year hospital-based surveillance and cost analysis Authors: Riccardo Vecchio, Elizabeth Iskandar, Leandro Gentile, Paola Bertuccio, Laura Daprai, Patrizia Cambieri, Alba Muzzi, Silvia Deandrea, Simona Riboli, Lorella Cecconami, Danilo Cereda, Fausto Baldanti, Anna Odone Published: 2026 ABSTRACT ABSTRACTWe aimed to evaluate 12-y (2012–2024) burden, clinical and microbiological characteristics of Invasive Pneumococcal Disease (IPD) at one large research hospital – IRCCS Fondazione Policlinico San Matteo – in northern Italy, with focus on serotypes’ distribution, resistance trends, and vaccination impact on outcomes and costs. The study included 234 IPD cases. Data were obtained from medical records, microbiological reports, and vaccination registries. Statistical analyses included descriptive measures, multivariate regression models for risk factors (adjusted for sex, age group, comorbidities), and comparison of length of stay and costs between vaccinated and unvaccinated patients. Most cases occurred in males aged ≥65. Bacteremia with pneumonia was the most frequent presentation (55.6%). Obesity and splenectomy were associated with higher risk of severe outcome. The most common serotypes overall were 3 and 8. Among vaccinated patients serotypes 15A, 14, 19A, and 15C were more frequent. Macrolide resistance was detected in 26.9% of isolates and beta-lactam resistance in 14.9%. Collectively, 34.5% of cases were caused by serotypes preventable with Pneumococcal Conjugate Vaccine 13 (PCV13), an additional 5.1% by PCV15, 21% by PCV20 and 14.4% by V116 preventable serotypes. Vaccine-preventable serotypes accounted for 64% of cases, mostly (56%) in patients aged ≥65. Vaccinated patients reported shorter hospital stay (median 8 vs 16 d) and lower associated costs (€3313 vs €5101). IPD surveillance is critical to inform prevention strategies. Our findings quantify how much vaccination reduces disease severity and healthcare costs but highlight gaps in vaccine coverage against emerging serotypes due to replacement mechanisms.KEYWORDS: Invasive Pneumococcal Disease, serotype, antimicrobial resistance, clinical manifestations, epidemiology, hospital-based surveillance Background Streptococcus pneumoniae (S. pneumoniae) is a Gram-positive lancet-shaped bacterium. Ninety serotypes have been identified, distinguished by the polysaccharide composition of their capsule. It is carried asymptomatically by about 50% of individuals and is responsible for a wide range of infections, from mild conditions like otitis media, sinusitis, and bronchitis to more severe diseases, including bacteremia, pneumonia, and Invasive Pneumococcal Disease (IPD).1,2 IPD occurs when S. pneumoniae invades sterile sites, such as blood (bacteremia) or cerebrospinal fluid (meningitis), and is a leading cause of morbidity and mortality in both children and adults.3,4 It is estimated that Pneumococcal pneumonia causes more deaths worldwide than all other major causes of lower respiratory infections combined, including Haemophilus influenzae type B, Influenza virus, and Respiratory Syncytial Virus (RSV).5 Although the Coronavirus Disease (COVID-19) pandemic was associated to a general decrease in the incidence of other respiratory infections, including pneumococcal disease,6–10 S. pneumoniae currently remains the most common cause of bacterial respiratory infections in the community, particularly affecting children, elderly, and immunocompromised patients.6 In 2022, 17,700 confirmed cases of invasive pneumococcal disease (IPD) were reported in the European Union with a notification rate of 5.1 cases per 100,000, similar to pre-pandemic rates.11 In Italy, where national-level surveillance of Invasive Bacterial Diseases (IBD) was established in 2007, 1032 confirmed cases were reported in 2022, of which 308 in the Lombardy Region.11,12 At the national level, significant increase in IPD incidence was observed between 2021 and 2023, with 1783 cases of IPD confirmed in 2023.13 Clinical manifestations The clinical picture of individuals affected by IPD is characterized by pneumonia, bacteremia, or meningitis, often requiring intensive care support and potentially resulting in cognitive impairment, memory deficits, arrythmia, lung fibrosis and decline in lung function as most common sequelae.14 Other clinical manifestations may also occur in association with IPD, potentially affecting hospital outcomes, readmission rates, and long-term clinical consequences.15,16 In Italy, in 2023, the most common clinical presentation of invasive pneumococcal disease was pneumonia associated with bacteremia (50%), followed by isolated bacteremia (27%) and meningitis (20%).13 Age is a key factor in determining vulnerability to IPD.1,17 IPD rates were found to be higher in infants aged 1 y or less, in adults aged 65 or older and in males.11,13,18 Case fatality rate also increases with age.19 Furthermore, individuals with comorbidities are at higher risk of IPD. Children affected by asthma, hematologic malignancy, primary and acquired immunodeficiency, cardiovascular and metabolic diseases are more likely to develop invasive form of the diseases.20–25 In adults, prognostic factors for mortality include underlying chronic diseases, solid organ tumors, and immunosuppressed status; at-risk behaviors, such as alcohol abuse, are also associated with increased mortality risk.1,26,27 The incidence of pneumococcal infection in adults with hematologic malignancies or HIV is 20 times higher than in controls.28 Splenectomy or functional asplenia is an important risk factor29 and pneumococcal vaccination is generally recommended to individuals who undergo splenectomy.28,30 Serotype distribution Globally, serotypes 3, 8, and 22F are indicated among the most prevalent serotypes causing IPD.11,31,32 Researchers have attempted to link specific serotypes to disease severity and related outcomes.33–36 Serotype 3 is particularly concerning because of its high fatality risk, elevated incidence and mortality rates34 in addiction to its ability to resist antibody-mediated clearance. Moreover, despite its inclusion in various vaccine formulations, it has been linked to microbiological mechanisms of vaccine escape.37 Furthermore, a nonfunctional antibody response has been associated with low invasive serotypes (3, 19F, and 23F), leading to impaired long-term acquired immunity.38–40 Pneumococcal vaccines and immunological mechanisms Two types of pneumococcal vaccines have been progressively implemented in vaccination strategies worldwide targeting primarily children. First, the polysaccharide 23-valent vaccine (PPSV23) induces a short-term immune T cell-independent response. Its effectiveness has been extensively assessed in adults aged 65 y or older, estimating an effectiveness against all-type IPD of 32% and against PPSV23-serotype IPD of 41%, and also suggesting reduced effectiveness against the highly prevalent serotype 3.41 Second, the Pneumococcal Conjugate Vaccines (PCVs) induce a T-dependent response with consequent production of memory B cells, leading to the development of immunological memory. Effectiveness of PCV 7-valent and PCV 13-valent conjugate vaccines has been largely documented both in children and older adults.42–46 National and regional policies and recommendations In Italy, PPSV23 was first introduced in 2000 for children aged 2 y or older. Later, in 2007, PCV7 was introduced and then replaced in 2010 by PCV13. The Italian National Immunization Plan 2017–2019 recommended the sequential administration of PCV13 and PPSV23 to individuals aged 65 or older.47 In October 2021, the Advisory Committee on Immunization Practices (ACIP) updated its recommendations for pneumococcal vaccination in adults due to the introduction of two new vaccines.48 The updated recommendations focus on using either the 20-valent pneumococcal conjugate vaccine (PCV20) alone or the 15-valent pneumococcal conjugate vaccine (PCV15) in combination with PPSV23. PCV15 has demonstrated a safety profile comparable to PCV13 in infants and children with higher immune responses for serotypes 3, 22F and 33F.49 Remarkably, the predicted effectiveness of PCV20, additionally covering serotype 8, was comparable to PCV7 and PCV13 for 7 of the 13 shared serotypes (5, 6A, 7F, 9 V, 18C, 19F, and 23F), with decreased effectiveness against the remaining serotypes (1, 3, 4, 6B, 14, and 19A).50 Moreover, a new 21-valent vaccine formulation for individuals aged 18 and older was approved by the European Medicines Agency (EMA) in 2025.51 Currently, four vaccines are available in Italy: PPSV23, PCV13, PCV15 (since 2022) and PCV20 (since 2022). Italian regions have a certain degree of autonomy in programming their own vaccination strategies. As of January 2025, the Lombardy Region’s vaccine coverage among infants up to 2 y old is approximately 95%, with slight variations between provinces. However, among individuals aged 65 and older, coverage remains low, ranging from 7.4% to 4.4% in different areas. Recently, the vaccination schedule in the Lombardy Region has been updated and includes PCV15 for infants and PCV20 for adults aged 65 and older, as well as for individuals at risk. Vaccination campaigns’ impact It is widely recognized that the serotype prevalence is strongly influenced by vaccination campaigns and uptake. An increase in IPD rates among highly vaccinated populations was associated to the mechanism known as “serotype replacement.” After the introduction of PCV7, an increasing amount of cases due to serotypes not covered by PCV7 vaccine was documented.52 The same was registered after the introduction of PCV13.53 A study conducted in England and Wales, published on The Lancet, reported that the incidence of IPD caused by non-PCV13 serotypes doubled following the introduction of PCV7 and accelerated after 2013. Serotypes 8, 12F, and 9N accounted for more than 40% of IPD cases by 2017.54 This evidence was recently confirmed by others, which observed an increase in disease caused by serotypes 8 and 12F, not included in PCV13.7 A study conducted in the United Kingdom identified an increase in serotypes 8, 9N, 10A, 12F, 15B/C, 23B, and 24F after the PCV13 adoption, with 8 and 12F (now targeted by PCV20) causing the greatest increase.55 In 2018, ECDC reported that 75% of IPD cases in children were caused by serotypes not included in any PCV formulation. In contrast, among cases aged 65 y and older, 73% were caused by serotypes included in the 23-valent polysaccharide vaccine.11 Routine pediatric immunization with PCV13 has significantly reduced the incidence of vaccine-serotype IPD also in older adults in most countries through indirect protection of unvaccinated individuals.56 However, this effect has plateaued in high-income countries, with residual vaccine-serotype disease persisting and an increasing burden of IPD caused by non-vaccine serotypes.57 Notably, tendency toward higher mortality rates and more severe outcomes was observed for vaccine-non-preventable serotypes.10,58 Also in Italy, cases related to PCV13 serotypes decreased after the adoption of PCV13 vaccine – excluding serotypes 3, 8 and 19A. Specifically, PCV13 serotypes accounted for 62% of cases in the pre-PCV13 era, compared to 45% in the post-PCV13 era.59 Serotype and antimicrobial resistance The introduction of vaccination tools has had a significant impact also on pneumococcal antimicrobial resistance patterns. After the introduction of the vaccine, β-lactam resistance dropped in PCV13 serotypes, emerging in non-vaccine serotypes.60,61 PCV13 serotypes 14, 19A, and the non-PCV13 serotype 11A have been recently identified as the most common penicillin-resistant serotypes in Spain.62 A similar shift has been observed for macrolides susceptibility.63 In this regard, erythromycin resistance has been recently associated with certain serotypes, such as serotypes 19A and 24F in Spain64 and serotypes 19F, 19A, 6C, 15A in Portugal.65 In Italy, from 2015 to 2023, the percentage of S. pneumoniae isolates resistant to penicillin remained relatively stable, with values reaching 12.9% in 2023. For erythromycin resistance, a steady increase has been observed since 2018, rising from 20.3% to 26.2% in 2023. High resistance percentages were also noted for tetracycline (23.9%) and clindamycin (22.6%).6 Healthcare costs In addition to considerable clinical burden, associated IPD-induced healthcare costs are significant.66–68 However, although data are available about the cost-effectiveness of pneumococcal vaccination for prevention of IPDs,69,70 scant evidence is available about the impact of IPDs on costs at a hospital-level. Aim of the study General objectives of this study are to map the clinical burden of IPDs over a 12-y study period in the research hospital IRCCS Fondazione Policlinico San Matteo in northern Italy, focusing on clinical manifestations and outcomes. Additionally, our research aims to describe the microbiological aspects of the disease, including serological patterns, antimicrobial resistance, and temporal trends. Impacts on organizational aspects, such as hospital stay length and costs, were analyzed. Moreover, we provide a focus on the impact of vaccination campaigns on both microbiological and hospital organizational aspects. Methods Study design and setting A retrospective observational study was conducted by researchers from the University of Pavia in collaborations with the Microbiology and Virology department of IRCCS Fondazione Policlinico San Matteo. In addition to descriptive analyses, an analytical approach was applied to assess the association between selected risk factors and severe outcomes (death or ICU admission) within the study population. As an IRCCS (Research Hospital), the hospital routinely integrates clinical care and research activities. At hospital admission, patients provide informed consent for the use of their clinical data and biological samples for research purposes, in accordance with institutional policies and current regulations. The research protocol was approved by the Ethics Committee of IRCCS Fondazione Policlinico San Matteo in April 2024. The study included all patients admitted between January 1, 2012, and December 31, 2024, diagnosed with IPD through microbiological testing (S. pneumoniae isolated from sterile sites such as blood or cerebrospinal fluid). No specific exclusion criteria were applied. For each patient, we collected data on socio-demographics (sex, age, deprivation index and at-risk behaviors), clinical manifestations and comorbidities, microbiological features (serotypes and resistance profiles), organizational aspects (length of hospital stay and associated costs) and vaccination status. Data were collected from multiple sources. Socio-demographic data, including sex and date of birth, were extracted from hospital records, while deprivation index (as deciles) was traced using the patient’s residential address via Italian National Institute of Statistics (ISTAT) public registry.71 At-risk behaviors, clinical data and comorbidities were retrieved from discharge letters, hospital discharge forms (SDO), and other medical forms. The hospital’s Microbiology and Virology Unit provided data on serotypes and antimicrobial resistance profiles. Serotypes marked as “Not vital” were unsuccessfully characterized. Hospital stay duration and associated costs were provided by the hospital’s Medical Direction. Pneumococcal vaccination data (PCV7, PCV13, PCV15, PCV20, PPSV23) were retrieved from the vaccination registry of the Health Protection Agency of Pavia (ATS Pavia) or the regional vaccination registry (ARVAX) in the case of patients residing outside the Pavia province. Only the last vaccination received before hospital admission was considered. All the information were collected through a specific electronic Case Report Form (eCRF). To guarantee data anonymization, two separate projects were created within Research Electronic Data Capture (REDCap). The first project contained identifying information, including the subject’s name, surname, date of birth, and tax code. During the creation of a subject’s profile in this project, the software automatically generated a unique code. This code was then associated with the corresponding case in the second project. The second project contained all the variables necessary for the clinical research, excluding any identifying information. By extracting data solely from the second project, researchers could access anonymized data for analysis, ensuring compliance with data protection regulations. This dual-project structure allowed the principal investigator (PI) to maintain traceability of each case through the unique code, while ensuring that data used for the study remained anonymized for the research team. Microbiological procedures and criteria S. pneumoniae isolates described in this study were obtained from blood cultures and/or cerebrospinal fluid (CSF) cultures and were analyzed as part of the routine diagnostic pipeline at the Microbiology and Virology department of IRCCS Fondazione Policlinico San Matteo during the study period (January 2012–December 2024). If multiple samples were available from the same patient, only the earliest one in chronological order was included. Blood cultures were grown in BD BACTEC™ blood culture media and incubated in the BD BACTEC FX automated system (Becton Dickinson and Company, Franklin Lakes, New Jersey, United States) according to the manufacturer’s instructions. CSF samples and positive blood cultures were subjected to Gram staining, culturing, species identification, and Antimicrobial Susceptibility Testing (AST) according to laboratory procedures. Culturing involved streaking samples onto Columbia agar supplemented with 5% sheep blood, incubated at 37°C under aerobic conditions, and onto Chocolate agar with PolyViteX, incubated at 37°C in a 5% CO2 atmosphere. Due to the low specificity of MALDI-TOF mass spectrometry (Bruker Daltonics GmbH) in distinguishing S. pneumoniae from S. mitis and other viridans group streptococci, colony identification was performed using a 6 mm optochin disc containing 5 µg of ethylhydrocupreine hydrochloride, placed on a bacterial lawn in an area of dense growth. The plate was then re-incubated under the same conditions for an additional 18–24 hours. After incubation, results were assessed by measuring the zone of inhibition around the optochin disc. If the inhibition zone was ≥14 mm, the organism was identified as Streptococcus pneumoniae (optochin-sensitive). In high-risk patients requiring a rapid response, identification was performed using the PCR-based BioFire® FilmArray® Blood Culture Identification 2 (BCID2) or BioFire® FilmArray® Meningitis/Encephalitis (ME) syndromic panels (bioMérieux SA, Marcy-l’Étoile, France). AST was performed using the Kirby-Bauer disk diffusion method and the concentration gradient diffusion assay (Liofilchem) on Mueller–Hinton Fastidious agar (Liofilchem). Results were interpreted according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) breakpoints, from version 2.0 (2012) to version 14.0 (2024), based on the year of isolation (www.eucast.org). Isolates obtained from blood cultures were stored at −80°C and then sent to the Regional Reference Laboratory at the Policlinico di Milano for serotyping using the ImmuLex™ Pneumotest-Latex kit. This test employs latex agglutination with antisera pools to identify serotypes, particularly those covered by the 23-valent vaccine. The procedure involves preparing an overnight culture of pneumococcal isolates, mixing them with latex reagents on microscope slides, and observing agglutination within 5–10 seconds. To confirm serotyping, the latex test is complemented by the Quellung (Neufeld) reaction. This method detects capsular polysaccharides using specific antibodies, leading to a visible capsule swelling under a 40× optical microscope. The test is performed by mixing pneumococcal colonies with factor sera on a slide and observing for capsule enlargement, indicating a positive reaction. However, in some cases, strains are non-viable during analysis, preventing serotype determination. These are classified as “Non Vital” in the study. Analysis plan Descriptive statistics were calculated, with means and standard deviations reported for continuous variables and frequencies and percentages for categorical variables. Group comparisons were performed using χ2 or Fisher’s exact test for categorical variables. The risks of severe outcomes, namely death or ICU admission, according to demographic characteristics, clinical characteristics, medical history, and vaccination status, were estimated in terms of relative risks (RR) and their 95% confidence intervals (CI) by carrying out adjusted log-binomial models. The adjusted variables included sex, age group, and comorbidities (i.e., number of diseases). The non-parametric Wilcoxon rank-sum test was used to compare differences in the distribution of hospital stay and costs between PNC vaccination groups (vaccinated vs not vaccinated). The correlation between hospital stay and costs were evaluated using scatter plots, correlation coefficients, and linear regression model. Analyses were conducted using Python and SAS software version 9.2. The correlation between hospital stay and costs was assessed using scatter plots, Pearson correlation coefficient, and a linear regression model. Analyses were performed using Python and SAS software, version 9.2. Results In total, 234 cases were reported at IRCCS Fondazione Policlinico San Matteo between 2012 and 2024. The number of IPD cases per year is reported in Figure 1. The lowest number of cases was reported in 2020, with less than 10 cases, while the highest number was reported in 2018, with 30 cases. A steady increase was reported after 2021, with 24 cases reached in 2024. Number of IPD cases, by year reported at IRCCS San Matteo, 2012–2024. Clinical burden, manifestations and outcomes Table 1 reports the study population sociodemographic and clinical characteristics. Fifty-two per cent (n = 122) of IPD cases were males. The age distribution shows that IPD cases occurred more frequently in individuals aged 65 and older (54.7%, n = 128), followed by those aged 18–59 (25.6%, n = 60), 60–64 (10.7%, n = 25), and 0–9 (8.1%, n = 19), with the lowest proportion in the 10–17 age group (0.9%, n = 2). The deprivation index, divided into quintiles, indicates that most cases occurred in the least deprived group (quintile 1, 52.1%, n = 122), with progressively smaller proportions in quintiles 2 through 5 (25.2%, 12.8%, 3.8%, and 1.7%, respectively). In terms of at-risk behaviors, 4.3% (n = 10) had a history of alcohol abuse, whereas smokers accounted for almost one-fifth of cases (23.9%, n = 56 cases). Bacteremia with pneumonia was the most frequent clinical presentation (55.6%, n = 130), followed by isolated bacteremia (24.4%, n = 57), and meningitis (20.1%, n = 47). With reference to comorbidities, 17.1% of IPD patients (n = 40) were affected by one disease at the time of admission, 23.5% (n = 55) by two, 20.1% (n = 47) by three diseases, 27.8% (n = 65) by four or more comorbidities, whereas 11.5% (n = 27) had no comorbidities. In particular, hypertension was the most frequently observed comorbidity (46.2%, n = 108), followed by a diagnosis of oncological disease (38.9%, n = 91), including solid tumors (22.6%, n = 53) and hematologic malignancies (18.4%, n = 43). Cardiovascular disease affected 31.6% of IPD patients (n = 74), while 27.4% were affected by diabetes mellitus (n = 64), and 17.5% by pulmonary disease (n = 41). Hepatic disorder was reported in 16.7% of cases (n = 39), chronic renal failure in 13.2% (n = 31), and neurological disease in 10.3% (n = 24). Patients underwent immuno-suppressive treatment at the time of admission in 10.3% of cases (n = 24). Less frequent conditions included past transplants (both solid organ and stem cell transplants) and obesity (9.0% each, n = 21), anemia (8.5%, n = 20), neck and head surgery (5.6%, n = 13), thyroid disorder and splenectomy or functional asplenia (4.7% each, n = 11), and immunodeficiency (4.3%, n = 10). When the analysis was restricted to patients aged 18 y or older, clinical burden, manifestations, and outcomes were largely consistent with those observed in the overall study population (Supplementary Tables 1.1, 1.2, and 1.3). Hospital readmissions within 40 d related to IPD and clinical sequelae up to 6 months after discharge are summarized in Supplementary Tables 2 and 3, respectively. Table 1.Distribution of 234 IPD cases according to sociodemographic characteristics, clinical characteristics, comorbidities, clinical outcome, ICU admission and vaccination status. Pavia, IRCCS San Matteo, 2012–2024. n%Sex12252.1MalesFemales11247.9Age group198.10–910–1720.918–596025.660–642510.765+12854.7Deprivation index (quintiles) 112252.125925.233012.8493.8541.7Missing10–At-risk behaviorsAlcoholism104.3Smoking5623.9Clinical Manifestation Meningitis4720.1Bacteremia5724.4Bacteremia and pneumonia13055.6ComorbiditiesNumber of diseases 02711.5 14017.1 25523.5 34720.1 4+6527.8Diseases Immunodeficiency104.3 Neck and head surgery135.6 Anemia208.5 Oncological disease9138.9 Hematologic malignancy4318.4 Solid tumor5322.6 Transplant219.0 Immuno-suppressive treatment2410.3 Chronic renal failure3113.2 Diabetes mellitus6427.4 Hepatic disorder3916.7 Cardiovascular disease7431.6 Neurological disease2410.3 Hypertension10846.2 Pulmonary disease4117.5 Thyroid disorder114.7 Obesity219.0 Splenectomy or functional asplenia114.7Patient in-hospital outcome Discharge13959.4Discharge to other hospital setting (e.g. from acute to subacute setting)3213.7Discharge/Transfer to other healthcare facilities166.8Death4720.1ICU admission Yes4820.5PNC vaccination Yes2711.5 Distribution of 234 IPD cases according to sociodemographic characteristics, clinical characteristics, comorbidities, clinical outcome, ICU admission and vaccination status. Pavia, IRCCS San Matteo, 2012–2024. Table 2 reports the associations between selected sociodemographic and clinical characteristics, vaccination status, and the risk of severe outcome. Compared to meningitis, both bacteremia with pneumonia (RR: 0.61; 95% CI: 0.41–0.89) and bacteremia without pneumonia (RR: 0.61; 95% CI: 0.41–0.89) were associated with a significantly lower risk of severe outcome. Among comorbidities, obesity was significantly associated with an increased risk of ICU admission or death (RR: 1.86; 95% CI: 1.22–2.83). An increased risk of severe outcome was also observed in patients with four or more comorbidities (RR: 1.14; 95% CI: 0.72–1.82]) compared to those with none or one disease. Increased risk was also observed in patients who underwent transplantation (RR: 1.47; 95% CI: 0.82–2.61), and in those with anemia (RR: 1.10; 95% CI: 0.60–2.04), oncological diseases, in particular hematologic malignancies (RR: 1.36; 95% CI: 0.88–2.10), chronic renal failure (RR:1.03: 95% CI: 0.59–1.80), diabetes (RR: 1.29; 95% CI: 0.87–1.93), and hypertension (RR: 1.18; 95% CI: 0.77–1.79). When considering only the subgroup of IPD cases caused by vaccine-preventable serotypes (n = 126), the risk of severe outcome was lower among vaccinated individuals than in unvaccinated ones (RR: 0.94; 95% CI: 0.41–2.14), although the association was not statistically significant. Table 2.Absolute and relative risk (RR) and corresponding 95% confidence intervals (CI) of death or ICU admission according to demographic characteristics, clinical characteristics, medical history, and vaccination status. Pavia, IRCCS San Matteo, 2012–2024. Ref.Death or ICU admissionUnivariateRR (95% CI)aAdjustedRR (95% CI)bn%n%Sex8368.03932.0 Males1 Females7062.54237.51.17 (0.82–1.67)–Age Group1578.9421.1 0–91 10–17150.0150.0 18–593761.72338.31.61 (0.70–3.69)–60–641456.01144.01.85 (0.76–4.47)–65+8667.24232.81.38 (0.62–3.08)–Clinical manifestation2348.92451.1 Meningitis11Bacteremia4070.21729.80.58 (0.36–0.95)0.57 (0.35–0.94)Bacteremia and pneumonia9069.24030.80.60 (0.41–0.88)0.61 (0.41–0.89)Number of diseaes4364.22435.8 0–1112–37169.63130.40.85 (0.55–1.31)0.87 (0.56–1.36)4+3960.02640.01.12 (0.72–1.73)1.14 (0.72–1.82)Immunodeficency14564.77935.3 0111880.0220.00.57 (0.16–1.98)0.57 (0.16–2.01)Neck and head surgery14364.77835.3 No11Yes1076.9323.10.65 (0.24–1.79)0.64 (0.23–1.76)Anemia14165.97334.1 No11Yes1260.0840.01.17 (0.66–2.07)1.10 (0.60–2.04)Oncological disease9667.14732.9 No11Yes5762.63437.41.14 (0.80–1.62)1.14 (0.78–1.66)Hematologic malignancy12967.56232.5 No11Yes2455.81944.21.36 (0.92–2.02)1.36 (0.88–2.10)Solid tumor11764.66435.4 No11Yes3667.91732.10.91 (0.59–1.41)0.91 (0.58–1.42)Transplant14266.77133.3 No11Yes1152.41047.61.43 (0.88–2.33)1.47 (0.82–2.61)Immuno-suppressive treatment13865.77234.3 No11Yes1562.5937.51.09 (0.63–1.89)1.00 (0.55–1.84)Chronic renal failure/dialysis13466.06934.0 No11Yes1961.31238.71.14 (0.70–1.85)1.03 (0.59–1.80)Diabetes mellitus11668.25431.8 No11Yes3757.82742.21.33 (0.93–1.91)1.29 (0.87–1.93)Hepatic disorder12564.17035.9 No11Yes2871.81128.20.79 (0.46–1.34)0.74 (0.42–1.28)Cardiovascular disease10565.65534.4 No11Yes4864.92635.11.02 (0.70–1.49)0.95 (0.61–1.49)Neurological disease13765.27334.8 No11Yes1666.7833.30.96 (0.53–1.74)0.81 (0.42–1.57)Hypertension8668.34031.7 No11Yes6762.04138.01.20 (0.84–1.70)1.18 (0.77–1.79)Pulmonary diseases12263.27136.8 01113175.61024.40.66 (0.38–1.17)0.62 (0.34–1.10)Thyroid disorder14565.07835.0 No11Yes872.7327.30.78 (0.29–2.08)0.67 (0.24–1.84)Obesity14568.16831.9 No11Yes838.11361.91.94 (1.31–2.86)1.86 (1.22–2.83)Splenectomy or functional asplenia11.9. Yes––No15268.86931.2 PNC vaccination1763.01037.0 Yes1.08 (0.64–1.83)1.12 (0.61–2.04)cNo13665.77134.311aEstimates from unadjusted log-binomial regression models.bEstimates from log-binomial regression models adjusted by sex, age group, and number of diseases.cEstimate considering only the subgroup of 126 IPD cases with serotypes included in the PPSV23 vaccine and serotype 6A was: 0.94 (0.41–2.14). Absolute and relative risk (RR) and corresponding 95% confidence intervals (CI) of death or ICU admission according to demographic characteristics, clinical characteristics, medical history, and vaccination status. Pavia, IRCCS San Matteo, 2012–2024. aEstimates from unadjusted log-binomial regression models. bEstimates from log-binomial regression models adjusted by sex, age group, and number of diseases. cEstimate considering only the subgroup of 126 IPD cases with serotypes included in the PPSV23 vaccine and serotype 6A was: 0.94 (0.41–2.14). Microbiological features and impact of vaccination on serotype distribution Figure 2 reports the distribution of pneumococcal vital serotypes over the study period. The most frequent serotype was serotype 3, which accounted for 15.2% (n = 30) IPD cases, followed by serotype 8 (11.2%, n = 22). Other common serotypes included serotypes 19A, 19F, 14, and 12B (n = 9). Serotypes 6C and 15A equally accounted for 4.1% cases (n = 8) followed by serotype 23B (3%, n = 6). Serotypes 11A, 33F, 22F, and 23A where isolated in 2.5% cases (n = 5). All the other serotypes accounted for less than 2.5% IPD cases. Serotypes 3 and 8 temporal distribution over the study period showed a fluctuating pattern with a sharp decrease in the number of cases between 2020 and 2021 and a significant increase in 2022 for both serotype 3 and 8. The proportion of IPD cases associated with serotype 3 increased steadily after 2021, reaching the highest share of cases in 2024 (3.5%, n = 7). In contrast, a significant decrease in serotype 8–associated cases was registered in 2023. Distribution of serotypes (only vital serotypes) and temporal trends for serotypes 3 and 8. Pavia, IRCCS San Matteo, 2012–2024. Penicillin susceptibility (Figure 3) varied across different pneumococcal serotypes. No significant resistance patterns were detected across the most frequent serotypes (3 and 8); however, a notable difference was reported in their macrolide susceptibility profiles. While all serotype 8 strains were susceptible to macrolides, 38.3% of serotype 3 strains exhibited macrolide resistance. Among all isolated strains, macrolide resistance was detected in 26.9% (63/234), while beta-lactam resistance was detected in 14.9% (35/234) with MIC values for penicillin, when evaluated, ranging between 0.125 and 2 µg/mL (EUCAST Clinical Breakpoint Tables v. 14.0). Resistance profiles for macrolides were observed frequently and involved various serotypes. Most macrolide resistances were reported in serotypes 3 (n = 11), 19F (n = 7), 6C (n = 6), 19A (n = 4) and 33F (n = 5). Serotype 12B was not associated to any resistance profile. Serotype 19F showed both beta-lactam and macrolide resistance in 66.7% of cases (n = 6). Regarding levofloxacin, no resistance profiles were identified. Ceftriaxone was the most utilized antimicrobial (more than 64% of cases, Supplementary Figure 1); only one strain isolated in the year 2012 (serotype 14), showed a slightly elevated MIC value of 2 µg/mL but no resistant patterns were found (Figure 3). No resistance patterns were found for linezolid and vancomycin (Supplementary Figure 2), while Supplementary Figure 3 reports temporal trends for penicillin MIC values. Antimicrobial susceptibility by antimicrobial class and pneumococcal serotypes. Penicillin, macrolides, ceftriaxone and levofloxacin. Pavia, IRCCS San Matteo, 2012–2024. Figure 4 reports IPD vaccine preventable cases by pneumococcal vaccine type. Among IPD cases where serotyping of the causative strain was possible (n = 195), 34.9% of cases (n = 68) were associated with serotypes preventable included in PCV13 and shared with PCV15, PPSV23 and PCV20. Of these, serotype 3 was the most common serotype, accounting for 44.1% (n = 30) of cases, followed by serotypes 19A, 19F, and 14. Moreover, 5.1% (n = 10) were associated with PCV15-exclusive serotypes (not included in PCV13 but shared with PCV20 or PPSV23), notably 33F and 22F. An additional 21.0% of cases (n = 41) were associated with PCV20-additional serotypes also included in PPSV23 (i.e., 8, 10A, 11A, 12F, and 15B): among these, serotypes 8 and 15B accounted for the majority of cases (n = 22 and n = 7, respectively). A further 3.6% of cases (n = 7) were attributable to serotypes 9N and 20, both shared by PPSV23 and V116. Cases caused by serotypes exclusively included in V116 - and not present in any other licensed vaccine – accounted for 14.4% of the total (n = 28). Furthermore, 21.0% of IPD cases (n = 41) were linked to serotypes not covered by any currently available pneumococcal vaccine. IPD vaccine preventable cases by vaccine type. Pavia, IRCCS San Matteo, 2012–2024. Figure 5 describes the IPD cases serotype distribution among vaccinated patients. IPD cases were mostly caused by serotypes 15A (14.8%, n = 4) and 14 (14.8%, n = 4). Serotype 19A accounted for 11.1% of cases (n = 3), whereas 15C for 7.4% (n = 2). Serotypes 23A, 33F, 1, 18A, 10B, 7C, 8, 10A, 15B, 11A and 3 accounted for 1 case. Vaccine-preventable cases occurred in patients vaccinated with vaccine formulations containing the IPD causing serotype except for the cases caused by serotypes 33F, 8, 15B and 11A. Overall, 37.5% of cases (n = 9) were attributable to serotypes included in PCV13 or shared with its formulation across PCV15, PCV20, or PPSV23. A further 16.7% (n = 4) of cases were associated with PCV20-additional serotypes, also included in PPSV23, while 4.2% (n = 1) were due to serotypes covered by PCV15, PCV20, and PPSV23. Serotypes exclusively included in V116 (i.e., not found in other currently licensed vaccines) accounted for a relevant 29.2% of cases (n = 7). Finally, 12.5% of cases (n = 3) were caused by serotypes not included in any of the evaluated vaccine formulations, representing non-preventable cases under current immunization strategies. Serotype distribution among patients vaccinated with any pneumococcal vaccine formulation: overall distribution and vaccine-preventable serotypes. Pavia, IRCCS San Matteo, 2012–2024. Age-stratified analysis is reported in Figure 6. Across all age groups, the majority of IPD cases were caused by serotypes included in vaccines (n = 126). In the 65–85 y age group, vaccine-included serotypes were responsible for most cases. Age distribution of IPD cases: vaccine-preventable serotypes vs. serotypes not included in commercially available vaccines. Hospital stay length, costs and organizational impact of vaccination Overall, hospital stay length ranged between 0 and 112 d with an interquartile range of 7–24 d, and a median of 15 d. Hospital costs ranged between 336 euros and 106,420 euros, with an interquartile range of 3,791–7,967, and a median of 5,101 euros (Supplementary Figure 4), with positive correlation between hospital stay and costs. The linear regression model yielded an R2 coefficient of 0.32, corresponding to a Pearson correlation coefficient of 0.56. The intercept of the model was 3,328.23, and the slope was 124.65 (Supplementary Figure 5). Median value was higher in unvaccinated patients (16 d) than in vaccinated individuals (8 d) (Supplementary Table 4). Longest hospital stay lasted 47 d among vaccinated patients in comparison with 112 d among the unvaccinated ones (Figure 7). Difference between groups was statistically significant (p < .05). Hospital stay (days) among vaccinated and unvaccinated patients. Pavia, IRCCS San Matteo, 2012–2024. With reference to costs, minimum value was 465 euros for vaccinated group and 336 euros among unvaccinated ones (Figure 8). The highest amount was registered among unvaccinated (19,855 euros) whereas the highest cost value within vaccinated group was 13,573 euros. Median cost value was higher in unvaccinated (5,101 euros) in comparison with vaccinated patients (3,313 euros) (Supplementary Table 5). Hospital costs (euros) among vaccinated and unvaccinated patients. Pavia, IRCCS San Matteo, 2012–2024. Discussion We present a comprehensive analysis of hospital-based surveillance of IPD cases over a 12-year period, highlighting changes in serotype distribution, antimicrobial resistance patterns, and the impact of vaccination on clinical and organizational outcomes. Occurrence of IPD cases at the IRCCS Fondazione Policlinico San Matteo fluctuated over time. Notably, the COVID-19 pandemic had a significant impact, leading to a marked reduction in IPD cases. As suggested in scientific literature, the decline in IPD cases between 2020 and 2021 can be attributed to the public health measures and restrictions implemented during the pandemic, such as lockdowns, social distancing, and mask mandates, which may have also curtailed the transmission of respiratory pathogens, including S. pneumoniae.6–10 By 2022, IPD cases began to steadily rise again, gradually approaching pre-pandemic levels. This trend was confirmed in 2023 and 2024. As expected, the majority of IPD cases occurred in males aged 65 y or older, in line with available evidence collected in scientific literature.1,3,11,17,18 Older adults are known to be at higher risk of IPD due to age-associated immune system decline and presence of comorbidities.17,26 Bacteremia associated with pneumonia was the most frequent clinical manifestation, followed by isolated bacteremia and meningitis. This distribution was consistent with the national data.13 At the time of admission, patients were affected by comorbidities in the majority of cases. One fourth of the study population was affected by 4 or more conditions, highlighting the role of chronic conditions as important risk factors for IPDs. Females, subjects aged 60–64 y, subjects with comorbidities, obese or who underwent splenectomy were at higher risk of severe outcomes, including death or ICU admission. Vaccinated IPD cases had lower risk of severe outcomes compared to unvaccinated patients. Consistently with international, European and National surveillance data, serotypes 3 and 8 were the most prevalent serotypes accounting for 15.2% and 11.2% of cases, respectively.11–13 Temporal trend for serotype 3 showed a steady increase in related IPD cases after the COVID-19 pandemic, with peaks reached in 2024. As reported in the literature, serotype 3–associated IPD cases are at higher risk of severe outcomes, likely due to its ability to escape the immune response, even in vaccinated patients.37,38 Moreover, in our study, serotype 3 strains exhibited high macrolide resistance (38.3%). With regard to antimicrobial resistance patterns, since 2019, EUCAST has introduced significant changes in the way AST is reported for Streptococcus pneumoniae, particularly with the redefinition of the intermediate (I) category and its impact on clinical decision-making. Previously, AST results classified isolates into three categories: Susceptible (S), Intermediate (I), and Resistant (R).72 However, in 2019, EUCAST redefined the intermediate category as “Susceptible, Increased Exposure” (I), shifting the focus from an ambiguous middle category to one that emphasizes the possibility of successful treatment under optimized dosing conditions.73 This change aimed to highlight that isolates classified as “I” were not necessarily less responsive to treatment but instead required higher doses or altered administration strategies to achieve therapeutic efficacy.74 Our microbiology laboratory implemented this reporting approach by the end of 2021, ensuring that susceptibility results were aligned with these updated guidelines. One notable example of this change in practice is the reporting of levofloxacin susceptibility. The actual susceptibility profile of S. pneumoniae strains remained unchanged throughout the study period, as reflected by MIC values and disk diffusion inhibition zone diameters. Under the current EUCAST guidelines, to classify an isolate as fully susceptible (S) to levofloxacin, it must have a MIC below an arbitrary threshold of 0.001 µg/mL. Since standard susceptibility testing methods can not accurately differentiate MICs at such low concentrations, nearly all isolates in our study population fall into the “Susceptible, Increased Exposure” (I) category by default, despite no actual change in their susceptibility profile (Figure 3). Since 2020, EUCAST has refined its criteria for detecting β-lactam resistance mechanisms in S. pneumoniae, using either the oxacillin disk diffusion test or benzylpenicillin MIC values.75 A benzylpenicillin MIC > 0.06 mg/L or an oxacillin zone diameter <20 mm indicates the presence of a β-lactam resistance mechanism. Based on the exact MIC value, clinical context, and site of infection, strains are classified as susceptible (S), susceptible with increased exposure (I), or resistant (R). According to these criteria, 14.9% of isolates (n = 35) exhibited β-lactam resistance. Resistant strains were most commonly associated with serotypes 19F (n = 7), 19A (n = 4), and 23B (n = 4). Importantly, no significant antimicrobial resistance patterns were detected for serotype 3 when considering first line antimicrobial classes, such as penicillin and ceftriaxone, whereas resistance profiles were reported for macrolides (almost half of the serotype 3 cases). Serotype 8 was not found to be associated with resistance profiles for these antimicrobials. Significant reduction in macrolides susceptibility was also found for serotypes 19A, 14, 6 and 33F, confirming available evidences.64,65 Serotypes 3, 19F and 19A have been also reported in literature among those related to higher risk for severe outcomes.35 Linezolid and vancomycin were not found to be associated with any resistance patterns. These results may be interesting when compared with the cumulative antimicrobial consumption at the IRCCS Fondazione Policlinico San Matteo during the study period with ceftriaxone, levofloxacin and vancomycin being the most utilized antimicrobial classes. Overall, 34.9% of cases were associated to serotypes preventable with PCV13, PCV15, PPSV23 or PCV20 (with serotype 3 accounting for the majority) and additional 21% cases with serotypes covered by PCV20 and PPSV23 (with serotype 8 accounting for 22 cases). Among the six most prevalent serotypes, 5 were due to serotype preventable with PCV20 vaccine. Interestingly, a shift in serotypes prevalence was noted when analyzing IPD cases occurred in vaccinated patients. Among the four most frequent, two non-vaccine preventable serotypes (notably 15A and 15C) as well as serotypes 14 and 19A stood out. The replacement mechanism as result of massive vaccination campaigns in children is widely described in literature and it was clearly observable also in our study.7,42–46,52,54,55,76 Shifts in serotype prevalence make it essential to maintain constant surveillance of microbiological dynamics and to update immunization strategies. Since serotype replacement poses a concern for public health programmes, vaccine formulations will need to be updated progressively as new evidence emerges. The rise in non-PCV13 serotypes highlights the need to consider these serotypes in future vaccine formulations and to maintain robust surveillance of emerging ones. Indeed, consistently with previous evidence, serotype prevalence is strongly influenced by vaccination campaigns and uptake. After the introduction of PCV7, an increasing amount of cases due to serotypes not covered by PCV7 vaccine was documented.52 The same was registered after the introduction of PCV13.53 A study conducted in England and Wales, published on The Lancet, reported that the incidence of IPD caused by non-PCV13 serotypes doubled following the introduction of PCV7 and accelerated after 2013. Serotypes 8, 12F, and 9N accounted for more than 40% of IPD cases by 2017.54 This evidence was recently confirmed by others, which observed an increase in disease caused by serotypes 8 and 12F, not included in PCV13.7 A study conducted in the United Kingdom identified an increase in serotypes 8, 9N, 10A, 12F, 15B/C, 23B, and 24F after the PCV13 adoption, with 8 and 12F (now targeted by PCV20) causing the greatest increase.55 In 2018, ECDC reported that 75% of IPD cases in children were caused by serotypes not included in any PCV formulation. In contrast, among cases aged 65 y and older, 73% were caused by serotypes included in the 23-valent polysaccharide vaccine.11 Routine pediatric immunization with PCV13 has significantly reduced the incidence of vaccine-serotype IPD also in older adults in most countries through indirect protection of unvaccinated individuals.56 However, this effect has plateaued in high-income countries, with residual vaccine-serotype disease persisting and an increasing burden of IPD caused by non-vaccine serotypes.57 Notably, tendency toward higher mortality rates and more severe outcomes was observed for vaccine-non-preventable serotypes.10,58 Also in Italy, cases related to PCV13 serotypes decreased after the adoption of PCV13 vaccine – excluding serotypes 3, 8 and 19A. Specifically, PCV13 serotypes accounted for 62% of cases in the pre-PCV13 era, compared to 45% in the post-PCV13 era.59 Notably, our study found that almost 35% of cases were not caused by vaccine preventable serotypes. In this regard, in January 2025 the European Medicines Agency (EMA) approved a new 21-valent vaccine formulation (V116) for individuals aged 18 and older and a marketing authorization was issued in March 2025.51 The recommendation was based on results from a Phase 3 trial, which found V116 to be non-inferior to PCV20 for 10 common serotypes and superior to PCV20 for all serotypes unique to V116.77 In Italy, V116 was authorized by the Italian Medicines Agency (AIFA) in May 2025.78 In our study, almost 30% of cases in vaccinated patients were caused by serotypes exclusively targeted by V116. However, this estimate refers to potential coverage and should be interpreted in light of current age-specific indications. According to the European prevalence study on healthcare-associated infections and antibiotic use in long-term care and other extra-hospital healthcare facilities (HALT Pilot Study 2022), pneumococcal vaccine coverage in long-term care facilities and other extra-hospital healthcare settings remains alarmingly low, with only 2% coverage.79 This is particularly concerning given that these facilities primarily host older and vulnerable patients, who are at the highest risk for IPD and its severe complications. Low vaccination rates in this high-risk population underscores a critical gap in preventive healthcare, especially considering that among the 126 cases of IPD caused by vaccine-preventable serotypes (Figure 6), only 11.9% had been vaccinated. Of these, 66.6% of cases were caused by a serotype included in the administered vaccine, suggesting possible vaccine failure or waning immunity. The remaining 33.3% of vaccinated patients were infected by a serotype not covered by the vaccine they received. This distribution suggests that while vaccination has likely contributed to reducing IPD cases caused by vaccine-included serotypes, a considerable burden remains, especially in older adults. With reference to organizational aspects related to IPD cases admission, great variability was observed in hospital stay length and hospital cost, with a positive correlation between the two variables and an average estimated increase in costs of 124.65 euros for each additional day of stay. Interestingly, median values for both hospital stay length and hospital costs were lower in vaccinated group in comparison with unvaccinated. In this regard, our results can provide insights not only into the direct effects of vaccination campaigns on serotype distribution and clinical manifestations but also into their indirect effects on hospital length of stay, clinical management, and associated healthcare costs. This study has both strengths and limitations. Study strengths include our multidisciplinary approach: microbiological, clinical and administrative data were collected, linked and interpreted to provide an integrated view of the IPD burden, clinical and microbiological characteristics at the local level. Moreover, a wide time window allowed us to investigate temporal transitions and the impact of vaccination campaign on serotype distribution. Furthermore, we investigated the impact of IPD cases on hospital’s organizational aspects and costs. Among limitations it is noteworthy that serotyping could not identify the causative serotype in 15.8% of cases, reducing the accuracy of the serotype distribution assessment and hindering the finding of potential trends. Secondly, we were not able to identify the vaccination status in individuals residing outside Lombardy Region. Furthermore, vaccination status in Lombardy Region is collected since 2000 and data are accessible through the regional immunization registry. However, data collection was not systematized before 2021, and gaps in vaccination status information may have led us to underestimate vaccination coverage among IPD patients in the present study. For the same reason, it was not possible to assess the timing of vaccine administration. This ultimately underscores the importance of strengthening monitoring systems and integrating data from community services, hospital settings and General Practitioners’ activity. Finally, the limited sample size did not allow for the detection of statistically significant associations between the investigated risk factors and clinical outcomes. Notably, only a limited number of cases occurred in individuals under 18 y of age, which may affect the interpretation of age-specific findings. However, analyses restricted to adult patients (≥18 y) yielded results consistent with those observed in the overall study population, suggesting that the inclusion of pediatric cases did not materially influence the main findings. Moreover, the low number of IPD cases observed over the study period may limit the ability to robustly assess temporal trends. Our retrospective study is a first attempt to describe IPD clinical manifestations, microbiological characteristics, and organizational implications in the context of a tertiary level University hospital. Furthermore, the implementation of a dedicated eCRF enables the development of a prospective registry, allowing researchers and clinicians to monitor the burden, trends, and patterns of IPD cases and eventually functioning as hospital surveillance system in the next future. Data on the clinical presentation and antimicrobial resistance of IPD at the regional level remain limited, highlighting the need for further investigation. Indeed, scaling our approach at a regional level and exploring serotype distributions, their associations with clinical manifestations, as well as the role of contributing factors in the disease onset, could provide researchers and healthcare workers with deeper insights into the microbiological and clinical dynamics, eventually informing the planning, implementation, evaluation and monitoring of targeted preventive interventions. Stakeholders, such as policymakers and scientific associations, could also benefit from these findings by better targeting preventive measures, such as vaccination, for individuals at higher risk. Supplementary Material Supp_rev_ref.docx Biography Anna Odone (MD, MSc, MPH, DrPH, PhD) is Full Professor and Director of the School of Public Health at the University of Pavia. Medical doctor by training, Prof. Odone received an M.Sc. in Epidemiology from the London School of Hygiene and Tropical Medicine (London, UK) and an M.P.H. in Health Policy and Management as a Fulbright Scholar from the Harvard School of Public Health (Boston, US). She holds a PhD in Medical Sciences from the University of Parma. She is Coordinator of the National PhD programme in “One Health Approaches to Infectious Diseases and Life Science Research”. Her research is in the field of infectious diseases epidemiology, prevention and control, with a focus on vaccines, immunization and healthcare-associated infections. Funding Statement MSD, Italy, provided financial support [grant n. 100714]. The opinions expressed in this paper are those of the authors and do not necessarily represent those of Merck Sharp & Dohme LLC. Disclosure statement This research was partly funded by MSD Italy (grant n. 100714). The funder had no role in the study design, data collection, analysis, interpretation of data, or writing of the manuscript. The authors declare no other conflicts of interest. Supplemental material Supplemental data for this article can be accessed online at https://doi.org/10.1080/21645515.2026.2670827 References Associated Data Supplementary Materials Supp_rev_ref.docx