Since the introduction of pneumococcal polysaccharide vaccines in the twentieth century, a number of studies have been conducted to assess its disease burden after the licensing of PCVs in various regions around the world. However, among the limited studies reporting on the pneumococcal burden in China, most were based on WHO or Maternal and Child Epidemiology Estimation Collaboration country-specific estimates or focused on residents in local areas. To the best of our knowledge, this is the first systematic review and meta-analysis to estimate the burden of bacterial and PM among under-fives in China in the post-PCV13 period. The present study serves as an important complement to the limited data on PM and indicates that PM remains a significant health problem for children in China. We found that the annual incidence rates for PBM, CBM, and PM in China were 12.12, 11.17, and 2.10 cases per 100,000 children younger than 5 years, respectively. Generally, SBM patients were recognized from the typical clinical syndromes they manifested, and bacterial meningitis cases should be confirmed by lumbar punctures and pathogen detection [20]. The ratio of PM incidence rate to that of confirmed bacterial meningitis reflected the contribution of S. pneumoniae. In our study, over 20% of the CBM cases were caused by S. pneumoniae, indicating that S. pneumoniae may be a major contributor to the disease burden in under-fives. This distribution pattern could potentially provide significant guidance for expediting diagnosis and implementing precise therapeutic interventions.

Compared with the estimated incidence rate of PM of 2.10 cases per 100,000 children in the present study, Chen et al. reported that 1.33 cases occurred among 100,000 children younger than 5 years in the year 2010 [21]. Though incidence rates for PM were calculated based on the all-cause incidence rate of meningitis and the etiological proportion of S. pneumoniae in the aforementioned study, and we obtained pneumococcal incidence rates directly from the included studies, the consistency between our results strengthens the overall validity of the present study. Apart from that, we also report a pooled mortality rate for all-cause meningitis of 20.69 deaths per 100,000 children younger than 5 years, which was in accordance with the trend shown by the GBD study [22]. Despite using different methods and data sources to estimate the mortality rate, the GBD study produced similar results, suggesting the robustness of our findings. Nevertheless, a lack of abundant records hampered the analysis of the mortality rate in CBM and PM cases. It was estimated that there were 1617.16 (95% CI: 454.35–5744.78) PM cases and 548.86 (95% CI: 474.80–627.62) deaths among children younger than 5 years in China in 2020, which may seem relatively small compared with the number of cases (8686, 95% CI: 5213–11,980) and deaths (1114, 95% CI: 669–1537) in 2015 obtained by Wahl et al. [23]. These differences may be attributed to the designed models and the sources of data used. Wahl et al. used mortality data from GBD-sourced modeled estimates combined with the pathogen-specific CFR to generate pneumococcal cases, while we utilized the pooled incidence rate to estimate the number of pneumococcal deaths. Similarly, both of the studies considered PCV use and hypothesized a higher CFR for patients without access to care.

There were certain discrepancies between our estimated CFR for PM and GBD-sourced results (24.59% versus 4.84%). The estimates from the GBD databases were generated using more complex models with global surveillance data, only a few of which were from China. In the present study, most of the CFR records from PM cases were based on hospital-based case reports; only one out of the six (16.67%) studies were based on community-based surveillance data. Usually, hospital-based studies tend to report cases manifesting more typical clinical symptoms, and physicians are likely to pay attention to severe cases [24]. It is possible that the hospital-based case reports were those with higher CFRs, which may have led to our overestimated CFR for PM. In addition, Lai et al. applied a literature-based CFR of PM when modeling the disease burden caused by S. pneumoniae, which was also extracted from mostly hospital-based studies, and the CFR of 12.85% further reinforced the validity of our findings [10].

Similar age distribution patterns of the CFR were observed for bacterial meningitis and PM. A steep decrease in the bacterial meningitis CFR from the period 1980–2005 to 2006–2015 indicates that improved accessibility to health care and the introduction of vaccines over time may help to prevent avoidable deaths. Regarding disease severity across geographic regions, a lower CFR of PM was found in eastern China, which may be attributed to a better healthcare capacity and higher PCV coverage. PCVs were one of the most inequitably distributed vaccines among non-NIP vaccines in China, due primarily to wealth disparity [25]. The eastern China region is more developed and the higher income levels there lead to better access of the residents to healthcare and possible immunization. However, it was observed that the pooled CFR of bacterial meningitis cases in eastern China was relatively high. This may be attributed to the inclusion of mostly rural cases in selected PM studies.

Our analysis of positively detected pathogens demonstrated that S. pneumoniae is one of the most significant pathogens of bacterial meningitis. Its dominant role in the etiologic distribution remained stable across different age groups, which was in accordance with previous studies [26]. Compared with the 1980–2006 period, the drastic decrease in both the pooled detection rate of and the proportion of cases with S. pneumoniae implies that PCVs helped prevent bacterial meningitis cases associated with S. pneumoniae. However, it is difficult to estimate the contribution of S. pneumoniae to bacterial meningitis accurately since antimicrobial treatment can distort pathogen detection and pre-diagnostic antibiotic usage was not available in most studies.

Timely antimicrobial therapies are crucial for improving survival and preventing adverse sequelae [27], but it is worth noting that inappropriate usage of antibiotics has led to worldwide resistance to penicillin and other antibiotics, threatening the traditional treatment of PM [28]. In the present study, S. pneumoniae displayed pooled resistance rates of 71.50% and 29.56% towards penicillin and cefotaxime, respectively, whereas all of the S. pneumoniae strains were reported to be sensitive to vancomycin, which is recommended by van de Beek et al. as an advanced treatment of PM [29]. Given that culture remains the gold standard for detecting S. pneumoniae, we only included studies that used cultured specimens for analysis, which eliminated the potential bias introduced by the use of different specimen collection methods.

Data extracted from two studies identified 6B, 14, 19A, 19F, and 23F as the serotypes of S. pneumoniae most commonly detected in bacterial meningitis, which is consistent with previous studies [21]. The majority of the reported serotypes were covered by PCV7 (78.95%) and PCV13 (91.10%), while PPSV23 showed full coverage. The high serotype coverage of PCVs provides useful information for the further management and promotion of PCVs in China, encouraging the potential inclusion of PCVs in the NIP.

There are certain limitations of our study. First, due to inadequate data, subgroup analysis with covariates including age, study year, and gender was not feasible. The insufficiency of available information underscores the need for further investigation. Second, estimates of incidence rates and case numbers may suffer from an inherent problem of underestimation, which is mainly attributable to the overuse of antibiotics in the early stage of the disease and the incomplete ascertainment of cases when children do not reach a health facility for diagnosis [30]. Thirdly, since China has not integrated PCVs into the NIP, caution should be exercised when attempting to generalize the estimates to other countries or regions with different PCV strategies. While our study provides valuable insights into the burden of PM in China, further studies using surveillance data are necessary to help develop targeted policies, particularly for disease mortality. Authorized local surveillance systems should be established to monitor the prevalence, antimicrobial resistance, and serotypes of pneumococcal strains that cause meningitis, which could provide valuable insights into the prevention of PM. In addition, comprehensive immunization programs targeting children at risk will help reduce the incidence of PM with great efficiency.