Public Perceptions and Influencing Factors of Non-National Immunization Program (Non-NIP) Vaccines in Shanghai: A Population-Based Study

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear All, 

 

Please, see the attached file

Comments for author File: Comments.pdf

Author Response

Comments 1: Abstract: Report only percentages and remove numbers

Response 1: Thank you for pointing this out. We agree with this comment. Accordingly, we have revised the Abstract by removing absolute numbers and retaining only percentages, in order to improve clarity and consistency with the journal's reporting style. The relevant revisions have been made in the Abstract (lines 30-34) of the revised manuscript: Results: Among the 753 respondents, with 15.5% of respondents reporting very high awareness, 18.7% reporting fairly high awareness, 32.1% reporting moderate awareness, 27.2% reporting somewhat low awareness, and 6.5% reporting complete unawareness. Acceptance levels were distributed as follows: 20.3% strongly in favour, 24.7% somewhat in favour, 45.9% neutral, 7.3% somewhat opposed, and 1.9% strongly opposed.

 

Comments 2: Abstract: Lines 39-40. On a 0-4 scale (0=completely unaware/unsupportive, 4=very aware/strongly supportive), …This sentence should be removed to methods.

Response 2: Thank you for this helpful suggestion. We agree with the reviewer's comment. Accordingly, we have relocated the description of how awareness and acceptance of non-NIP vaccines were measured from the Abstract to the Methods section, in order to improve structural clarity and align with standard reporting conventions. The relevant revisions have been made in the Methods section (lines 24-27) of the revised manuscript: Awareness and acceptance of non-NIP vaccines were measured using five-point Likert scales. On a 0-4 scale, where 0 = completely unaware/unsupportive and 4 = very aware/strongly supportive, respondents rated their level of understanding and endorsement of non-NIP vaccines.

 

Comments 3: Introduction: Since the audience of the Journal is international please, add more information regarding the Non-National Immunization Program and the National Immunization Program. For instance, which vaccines cover the NIP? What is the vaccination percentage for some important vaccines? Which vaccines are included in the non-NIP? Which of them Chinese prefer to do and in what percentages?

Response 3: Thank you for this helpful suggestion. We agree with the reviewer's comment. We agree that additional background information is necessary to improve the clarity and accessibility of the Introduction for an international readership. Accordingly, we have expanded the Introduction to provide a clearer and more detailed overview of China's National Immunization Program (NIP) and Non-National Immunization Program (non-NIP). Specifically, we have added descriptions of the vaccines covered by the NIP, typical vaccination coverage levels for key NIP vaccines, the main categories of non-NIP vaccines, and available evidence on uptake levels and commonly used non-NIP vaccines in China. These additions are intended to help international readers better understand the institutional context, coverage differences, and practical relevance of NIP and non-NIP vaccines in China. The relevant revisions have been made in the Introduction (lines 52-82) of the revised manuscript: In China, vaccines are broadly categorized into National Immunization Program (NIP) vaccines and Non-National Immunization Program (non-NIP) vaccines. NIP vaccines are funded and provided free of charge by the government as part of routine public health services and are administered according to the National Immunization schedule, such as hepatitis B vaccine, Bacillus Calmette-Guérin (BCG) vaccine, inactivated poliovirus vaccine (IPV), attenuated oral poliovirus vaccine (OPV), diphtheria-tetanus-pertussis (DTP) vaccine, measles-mumps-rubella (MMR) vaccine, live and inactivated Japanese encephalitis vaccines, group A and group A/C meningococcal polysaccharide vaccines, and hepatitis A vaccines (both live and inactivated). The routine vaccination coverage rate for NIP vaccines among eligible children in China has been consistently maintained above 90%, reflecting long-standing success in preventing major childhood infectious diseases.

Non-National Immunization Program vaccines refer to vaccines not included in China's National Immunization Program and administered voluntarily at individuals' expense[2]. Common non-NIP vaccines in China include rabies vaccine, seasonal influenza vaccine, varicella (chickenpox) vaccine, rotavirus vaccine, enterovirus 71 (EV71) vaccine, and pneumococcal conjugate vaccines (PCV), among others. Among non-NIP vaccines, rabies, influenza, and varicella vaccines are typically the most frequently administered in China. Existing evidence from clinical trials and real-world studies has demonstrated that these vaccines are effective in reducing disease incidence and related complications across different age groups [3-6]. Compared with vaccines included in China's National Immunization Program (NIP), vaccination coverage for non-NIP vaccines was substantially lower. Relevant research indicated that the full primary series coverage for 13-valent pneumococcal conjugate vaccine (PCV13) was approximately 5.1% in 2019, rotavirus vaccine three-dose coverage was about 1.8%, and Haemophilus influenzae type b (Hib) vaccine three-dose coverage was approximately 25.0%; only first-dose varicella vaccine approached higher levels (around 67.1%) in selected settings, reflecting localized inclusion in some municipal programs rather than national free provision. These figures illustrate that non-NIP vaccine uptake remained far below the near-universal coverage of NIP vaccines and exhibited substantial regional and product-specific disparities [7].

 

Comments 4: Introduction: Lines 89-92 should be removed since this study included healthcare workers and your manuscript is about general population.

Response 4: Thank you for this comment. We agree with the reviewer's suggestion. Accordingly, we have removed the text in lines 89-92 to ensure consistency between the study population and the scope of the manuscript, which focuses on the general population.

 

Comments 5: Introduction: Please improve and update the introduction section by adding information regarding the factors that affect awareness and acceptance of vaccines worldwide and especially in China literature on this issue is very extensive.

Response 5: Thank you for this helpful and constructive suggestion. We agree that providing a broader overview of factors influencing vaccine awareness and acceptance is important for strengthening the Introduction. Accordingly, we have revised and expanded the Introduction to incorporate evidence from international and Chinese literature on determinants of vaccine awareness, attitudes, and acceptance. The relevant revisions have been made in the Introduction (lines 143–169) of the revised manuscript: Empirical studies across diverse settings have consistently demonstrated that population‑level vaccine‑related knowledge and attitudes exert a substantial influence on vaccination uptake [25, 26]. Multiple studies have demonstrated that higher vaccine knowledge and positive attitudes are strongly associated with greater vaccine uptake and acceptance, highlighting the importance of cognitive and belief factors in vaccination behavior [27, 28]. Conversely, widespread misconceptions and negative perceptions about vaccines can significantly reduce vaccination rates by increasing hesitancy and undermining public confidence [29–31].

In the Chinese context, empirical studies have identified several determinants of vaccine awareness and acceptance. For example, a survey of the general adult population in China found that knowledge and positive attitudes toward vaccine efficacy and safety were significantly associated with willingness to vaccinate, suggesting that higher levels of disease and vaccine knowledge can enhance acceptance in large-scale immunization campaigns [32]. Moreover, systematic reviews of HPV vaccination in mainland China have shown that awareness of the vaccine, understanding of disease risk, perceived vaccine safety, and cost considerations are among the key predictors of willingness to accept non-NIP vaccines, indicating that both informational and socioeconomic factors play important roles in shaping vaccine decisions [33]. In Shanghai, immunization services are delivered through an integrated public health network coordinated by the Centers for Disease Control and Prevention and community health centers, where National Immunization Program (NIP) vaccines are administered according to the national schedule and non-NIP vaccines are offered on a voluntary, self-paid basis. The municipal government has implemented supplementary programs to increase immunization coverage and uptake among residents, including provision of varicella and pneumococcal conjugate vaccine (PCV) vaccination services for targeted populations, as well as other supplementary immunization measures.

 

Comments 6: Introduction: At the end of the introduction, please mention the gap in the literature that your study fills in.

Response 6: Thank you for this important comment. We agree that explicitly stating the research gap strengthens the rationale and contribution of the study. Accordingly, we have added a clear statement at the end of the Introduction to highlight the specific gap in the existing literature that the present study aims to address. The relevant revisions have been made in the Introduction (lines 172-180) of the revised manuscript: Despite an increasing number of studies on vaccine awareness and acceptance, most research in China has focused on specific vaccines or limited populations, and few have systematically examined the cognitive levels and multifaceted determinants of awareness and acceptance across a broad range of non-NIP vaccines among general urban residents.. However, there is still a lack of comprehensive, population-based evidence on how sociodemographic, informational, and psychosocial factors jointly shape non-NIP vaccine perceptions in large urban settings such as Shanghai, which the present study aims to address.

 

Comments 7: Material and methods: Please, further explain the stratified random sampling. For instance, from how many districts did you select the five districts for your study.

Response 7: Thank you for this comment. We agree that additional clarification of the stratified random sampling procedure is helpful. Accordingly, we have revised the Methods section to explicitly explain how the five study districts were selected. The relevant revisions have been made in the Methods section (lines 200-230) of the revised manuscript: We employed a stratified multistage probability sampling design, consistent with standard survey sampling methodology to improve representativeness and precision. First, all 16 adminisative districts in Shanghai were stratified into categories (e.g., central urban vs suburban) to form strata, based on geographic and demographic characteristics. We chose five districts based on statistical efficiency and logistical feasibility given our target sample size, consistent with prior surveys, where a similar number of primary sampling units was sufficient to achieve both adequate representation and manageable field operations. The selected five districts included both highly urbanized central areas (such as districts with high population density and advanced health service infrastructure) and more socioeconomically diverse suburban areas, ensuring that the sample captured key variations in demographic, economic, and healthcare access characteristics across Shanghai's municipal population. Within each stratum, the five districts were selected using simple random sampling, in which each district had an equal probability of being chosen from the list of all eligible districts within that stratum.

Within each selected district, communities (administratively defined neighborhood units equivalent to residents' committees or residential communities, which serve as standard sampling units in Chinese household and health surveys) were listed, and simple random sampling was used to select four communities per district, ensuring that each eligible community had an equal probability of being chosen. Finally, and 35-40 residents per community were randomly recruited to complete a questionnaire. Within each selected community, 'residents' refers to adult individuals listed in the community's official household registry (sampling frame) who met the inclusion criteria, and recruitment was conducted by drawing from this registry using a simple random sampling approach to ensure that every eligible resident in the community had an equal probability of being selected from the sampling frame. The target number of 35-40 residents per community was derived from the overall sample size calculation to ensure sufficient precision for estimating key outcome proportions (e.g., awareness and acceptance levels) while accounting for design effect in a multistage sampling framework[34,35].

 

Comments 8: Material and methods: Please, further explain the inclusion criteria. For instance, can children participate in your study?

Response 8: Thank you for this comment. We agree that clarifying the inclusion criteria is important. Accordingly, we have explicitly stated in the Methods section that only adults aged 18 years or older were eligible to participate in the study, and that children were not included. The relevant revision has been made in the Methods section (line 234) of the revised manuscript: (5) aged 18 years or older

 

Comments 9: Material and methods: How did you collect the questionnaires? How did you approach the 760 potential participants? Please, check the response rate. It is impossible to be 99% in a cross

sectional study.

Response 9: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified how the questionnaires were collected, how potential participants were approached, and corrected the description of the response rate by distinguishing between the proportion of valid questionnaires and the overall participation rate among invited individuals. Specifically, questionnaires were administered through face-to-face interactions by trained survey staff, potential participants were approached in person at community health service centers or at their homes based on the sampling frame, and we clarified that the previously reported 99.08% referred to the proportion of valid questionnaires among those collected rather than the response rate for the entire invited sample. This revision can be found on Lines 243-246 of the revised manuscript: Questionnaires were administered primarily through face-to-face interactions by trained survey staff, the reported figure reflects the proportion of valid responses among the collected surveys (99.08%), and 760 of the 1000 (76.00%, individuals invited to participate) invited individuals completed the survey.

 

Comments 10: 2.2. Questionnaire Development. Please further explain procedures that followed to create your study questionnaire. For instance, provide some references, explain the factor analysis, explain what is the KMO and why did you calculate this, for what items did you calculate Cronbach alpha, etc. Since you developed a questionnaire, you should present in detail the procedures that you followed to assess reliability and validity of it.

Response 10: Thank you for this detailed and constructive comment. We agree that further clarification of the questionnaire development and validation procedures is necessary. Accordingly, we have revised the Questionnaire Development section to provide a clearer description of the steps undertaken to establish content validity, reliability, and construct validity. Specifically, we have elaborated on the expert consultation process used to assess content validity, the pilot testing procedures conducted to evaluate item clarity and comprehensibility, and the subsequent item refinement. We also clarified the scope of the questionnaire, the constructs assessed, and the procedures used to evaluate internal consistency and suitability for factor analysis. The relevant revisions have been made in the Methods section (lines 266-295) of the revised manuscript: Through this expert review process, content validity of the questionnaire was established by ensuring that all items were relevant, clear, and aligned with the study objectives. In addition to expert consultation workshops, the complete questionnaire was pre-tested in a pilot survey with a small sample of community residents representative of the target population. The questionnaire consisted of 49 questions covering sociodemographic characteristics, awareness and acceptance constructs, and potential influencing factors. Completion of the questionnaire took approximately 15 minutes on average. The purpose of this pre-test was to evaluate item clarity, relevance, response burden, and respondent comprehension. Trained field staff conducted the pilot using the same structured interview mode planned for the main survey. Findings from the pre-test indicated that most items were comprehensible; however, several questions (e.g., those related to recent exposure to vaccine information and perceived affordability categories) were reworded to improve interpretability and reduce ambiguity. Minor modifications focused on simplifying phrasing and clarifying specific response options without changing the conceptual content of the items.

The final questionnaire achieved a Cronbach's α of 0.731, indicating acceptable internal consistency reliability for the set of items used to measure awareness and acceptance constructs. To evaluate construct validity, we conducted preliminary suitability tests for factor analysis: the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and Bartlett's test of sphericity were calculated. The KMO statistic assesses the degree to which variables share common variance, indicating whether the correlation matrix is suitable for identifying underlying factor structures. Bartlett's test of sphericity evaluates whether the observed correlation matrix significantly deviates from an identity matrix, with a significant P-value (P < 0.05) indicating that correlations among items are sufficiently large to justify factor analysis. These tests are widely used as preliminary steps in exploratory factor analysis to assess structural validity of newly developed instruments and provide evidence that the questionnaire items form coherent constructs. The observed KMO value of 0.716 and significant Bartlett's test (P < 0.001) indicate that the item correlation matrix was appropriate for further factor analysis and that the questionnaire possesses acceptable construct validity.

 

Comments 11: Sociodemographic information. Explain the way you measure these variables. For instance, what are the answers for distance to nearest vaccination site, mode of transportation, etc.

Response 11: Thank you for this helpful suggestion. We agree with this comment.  Accordingly, we have added Table 1 to explicitly present the measurement methods and response categories for all sociodemographic variables, including distance to the nearest vaccination site, mode of transportation, and other related variables. The relevant addition has been made in the Methods section (Table 1, line 319) of the revised manuscript.

 

Comments 12: Awareness and Acceptance. The same as above.

Response 12: Thank you for this helpful suggestion. We agree with this comment. Accordingly, we have added Table 1 (.line 319).

 

Comments 13: Additional Factors. The same as above.

Response 13: Thank you for this helpful suggestion. We agree with this comment. Accordingly, we have added Table 1 (line 319).

 

Comments 14: Ethical issues. Present the way you respect the ethical issues in your study. In particular, present the license of the Ethical Committee that allows you to conduct your study.

Response 14: Thank you for this important comment. We agree that ethical considerations should be clearly stated. Accordingly, we have revised the Methods section to explicitly describe the ethical approval for this study, including the approving ethics committee and the corresponding approval information. The relevant revisions have been made in the Methods section (lines 246-249) of the revised manuscript: The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Institutional Review Board (Ethics Committee) of the Shandong Center for Disease Prevention and Control (protocol code SDJK(K)2024-049-01; date of approval: 9 October 2024).

 

Comments 15: Statistical Analysis. Further explain. For instance, why did you use Kruskal-Wallis and ordinal logistic regression? How did you deal with the requirements to apply ordinal logistic regression in your data? Which descriptive statistics did you use?

Response 15: Thank you for this detailed comment. We agree that additional clarification of the statistical analysis is important. Accordingly, we have revised the Statistical Analysis section to more clearly explain the rationale for the statistical methods used and the assumptions assessed. Specifically, we clarified that descriptive statistics were presented as frequencies and percentages for categorical and ordinal variables. We also explained that the Kruskal-Wallis H test was used because awareness and acceptance were measured on ordinal scales and did not meet normality assumptions, making non-parametric comparisons across multiple independent groups appropriate. In addition, we elaborated on the use of multivariable ordinal logistic regression to identify factors associated with different levels of awareness and acceptance, and clarified how key model assumptions were assessed, including the proportional odds assumption (test of parallel lines), overall model fit (likelihood ratio test), and multicollinearity (variance inflation factors). The relevant revisions have been made in the Methods section (lines 362-385) of the revised manuscript: Descriptive statistics were first used to summarize participants' sociodemographic characteristics as well as the distribution of awareness and acceptance levels of non-NIP vaccines, including frequencies and percentages for categorical variables and ordinal outcomes. Because the dependent variables (awareness and acceptance of non-NIP vaccines) were measured on ordinal scales and did not meet the assumptions of normal distribution, non-parametric methods were applied. Descriptive statistics were also presented using frequencies and percentages to summarize the distribution of ordinal and categorical variables. Accordingly, the Kruskal-Wallis H test was used to compare awareness and acceptance scores across different subgroups defined by sociodemographic characteristics and potential influencing factors. This approach is appropriate for comparing ordinal outcomes among multiple independent groups. Variables that showed statistical significance in the Kruskal-Wallis tests were subsequently entered into the multivariable ordinal logistic regression models for further analysis.

Ordinal logistic regression models were applied to identify factors associated with awareness and acceptance levels of non-NIP vaccines. The proportional odds assumption was evaluated using the test of parallel lines, while overall model fit was assessed using the likelihood ratio test. A non-significant result for the parallel lines test supports the proportional odds assumption, and a significant likelihood ratio test indicates that the model with predictors fits the data better than the null model. Multicollinearity among explanatory variables was assessed using variance inflation factors (VIFs), and variables with VIF > 10 were considered to have concerning multicollinearity; no explanatory variables exceeded this threshold. All statistical tests were two-sided, and a P-value < 0.05 was considered statistically significant.

Comments 16: Results: Participant Characteristics. Present these characteristics in a Table to understand your study population.

Response 16: Thank you for this suggestion. We agree that presenting participant characteristics in a table improves clarity and readability. Accordingly, we have added a table summarizing the sociodemographic characteristics of the study participants. Details are provided in Table 2 (line 395).

 

Comments 17: Table 1. Present percentages in the column "Total respondents".

Response 17: Thank you for this comment. We agree with the reviewer's suggestion. Accordingly, we have added percentages to the "Total respondents" column in Table 3 to improve clarity and consistency of data presentation. The relevant revision has been made in Table 3 (line 419) of the revised manuscript.

 

Comments 18: Table 1. Please, consider again the statistical tests you performed. For instance, to investigate the association between two ordinal variables it would be better to use Spearman's correlation coefficient. Moreover, in case you find statistical associations with Kruskal-Wallis, you should then perform Mann-Whitney by applying Bonferroni correction to further examine differences between two groups.

Response 18: Thank you for this detailed methodological comment. We appreciate the reviewer's suggestions regarding alternative non-parametric statistical approaches. We would like to clarify the rationale underlying our analytical strategy and the specific role of the Kruskal-Wallis tests in this study.

In the present analysis, the Kruskal-Wallis H test was applied as a preliminary screening tool to explore potential associations between awareness/acceptance levels and a wide range of explanatory variables. Importantly, the explanatory variables considered in Table 3 comprised a heterogeneous set of variable types, including nominal variables (e.g., gender, occupation, residence type), ordinal variables (e.g., self-reported living standard). As a result, the variables were primarily treated as unordered categorical variables for the purpose of initial screening. Under these conditions, Spearman's rank correlation coefficient—which is designed for monotonic associations between two ordinal or continuous variables—was not uniformly applicable across all variable combinations examined.

Furthermore, the Kruskal-Wallis tests were not intended to serve as the main inferential analysis, nor to support detailed pairwise group comparisons. Instead, their primary purpose was to identify candidate variables potentially associated with awareness and acceptance outcomes. Variables identified at this stage were subsequently entered into multivariable ordinal logistic regression models, which constituted the main analytical framework of the study. These multivariable models allowed for adjusted estimation of associations while accounting for confounding and provided more interpretable effect estimates for ordered outcome variables.

Accordingly, post-hoc pairwise Mann-Whitney U tests with Bonferroni correction were not performed following the Kruskal-Wallis tests. Given that the study's primary conclusions are based on multivariable ordinal logistic regression rather than univariate group comparisons, additional post-hoc testing was not aligned with the analytical objectives and could have led to unnecessarily conservative inference due to multiple-testing adjustments.

In summary, Kruskal-Wallis tests were used solely for exploratory screening, while multivariable ordinal logistic regression served as the principal method for evaluating associations. Thank you again for your comment.

 

Comments 19: Table 2. Present percentages in the column "Total respondents"

Response 19: Thank you for this comment. We agree with the reviewer's suggestion. Accordingly, we have added percentages to the "Total respondents" column in Table 4 to improve clarity and consistency of data presentation. The relevant revision has been made in Table 4 (line 431) of the revised manuscript.

 

Comments 20: Table 2. Please, consider again the statistical tests you performed. For instance, to investigate the association between two ordinal variables it would be better to use Spearman's correlation coefficient. Moreover, in case you find statistical associations with Kruskal-Wallis, you should then perform Mann-Whitney by applying Bonferroni correction to further examine differences between two groups.

Response 20: Thank you for this detailed methodological comment. We appreciate the reviewer's suggestions regarding alternative non-parametric statistical approaches. We would like to clarify the rationale underlying our analytical strategy and the specific role of the Kruskal-Wallis tests in this study.

In the present analysis, the Kruskal-Wallis H test was applied as a preliminary screening tool to explore potential associations between awareness/acceptance levels and a wide range of explanatory variables. Importantly, the explanatory variables considered in Table 3 comprised a heterogeneous set of variable types, including nominal variables (e.g., gender, occupation, residence type), ordinal variables (e.g., self-reported living standard). As a result, the variables were primarily treated as unordered categorical variables for the purpose of initial screening. Under these conditions, Spearman's rank correlation coefficient—which is designed for monotonic associations between two ordinal or continuous variables—was not uniformly applicable across all variable combinations examined.

Furthermore, the Kruskal-Wallis tests were not intended to serve as the main inferential analysis, nor to support detailed pairwise group comparisons. Instead, their primary purpose was to identify candidate variables potentially associated with awareness and acceptance outcomes. Variables identified at this stage were subsequently entered into multivariable ordinal logistic regression models, which constituted the main analytical framework of the study. These multivariable models allowed for adjusted estimation of associations while accounting for confounding and provided more interpretable effect estimates for ordered outcome variables.

Accordingly, post-hoc pairwise Mann-Whitney U tests with Bonferroni correction were not performed following the Kruskal-Wallis tests. Given that the study's primary conclusions are based on multivariable ordinal logistic regression rather than univariate group comparisons, additional post-hoc testing was not aligned with the analytical objectives and could have led to unnecessarily conservative inference due to multiple-testing adjustments.

In summary, Kruskal-Wallis tests were used solely for exploratory screening, while multivariable ordinal logistic regression served as the principal method for evaluating associations. Thank you again for your comment.

 

Comments 21: Please, present also results from the univariate analysis. Also, present results from all influencing factors in the table 3. Also, in the table 3 you should present the categories from the dependent variables to better understand the results.

Response 21: Thank you for pointing this out. We agree with this comment. The results of the univariate analysis are now explicitly presented in Table 3 (line 419), to further enhance interpretability, the categories of the dependent variables are explicitly presented in Table 5 (formerly Table 3; line 454), allowing readers to better understand how outcome levels are distributed and modeled in the multivariable ordinal logistic regression analyzes.

 

Comments 22: Please, present also results from the univariate analysis. Also, present results from all influencing factors in the table 4. Also, in table 4 you should present the categories of the dependent variables to better understand the results.

Response 22: Thank you for pointing this out. We agree with this comment. The results of the univariate analysis are now explicitly presented in Table 4 (line 431), to further enhance interpretability, the categories of the dependent variables are explicitly presented in Table 6 (formerly Table 4; line 479), allowing readers to better understand how outcome levels are distributed and modeled in the multivariable ordinal logistic regression analyzes.

 

Comments 23: Please, further explain how your findings can help to improve public health especially in China.

Response 23: Thank you for pointing this out. We agree with this comment. we have expanded the discussion to explicitly articulate how our results can inform public health practice, health communication strategies, and policy development in China (lines 695-733) :

4.5. Public Health Relevance and Study Strengths

This study's multimethod analytical approach—starting with Kruskal–Wallis tests to identify potential predictors and followed by multivariable ordinal logistic regression—enabled robust identification of factors independently associated with both awareness and acceptance of non-NIP vaccines. The use of ordinal measures for awareness and acceptance and appropriate statistical models strengthened the validity of effect estimates, while consideration of a broad set of explanatory variables including sociodemographic characteristics, provider recommendation, education participation, and affordability provided a comprehensive view of determinants shaping vaccine perceptions. Although the cross-sectional design limits causal inference and self-reported data may be subject to recall bias, the findings illuminate key behavioral and informational barriers that can be targets for public health action.

The identified associations between knowledge, attitudes, and acceptance suggest clear directions for public health policy. Tailored communication strategies that emphasize evidence-based information on non-NIP vaccine effectiveness and safety, combined with enhanced provider recommendation and structured educational campaigns, could mitigate misconceptions and improve vaccine confidence. Furthermore, aligning financing mechanisms to reduce perceived financial barriers may help translate positive attitudes into actual uptake. These insights can inform efforts to refine China's immunization framework, particularly in considering how non-NIP vaccines might be better integrated into population health planning to reduce disease burden and promote equitable access.

5.Conclusions

This study systematically assessed the levels of knowledge and acceptance of non-NIP vaccines among residents of Shanghai. The findings indicate that overall awareness and acceptance remain below optimal levels, characterized by a pronounced phenomenon of relatively high acceptance but insufficient understanding. These results underscore the need to reinforce the professional recommendation role of healthcare providers and to implement comprehensive educational interventions aimed at improving access to accurate vaccine information, particularly among older adults and low-income populations. Concurrently, efforts to enhance vaccine acceptance should focus on strengthening the dissemination of evidence-based vaccine information, expanding public education initiatives, optimizing vaccine payment mechanisms, and improving affordability. Together, these findings highlight key demand-side barriers relevant to public health practice in China and suggest that exploring diversified financing strategies such as targeted subsidies or expanded insurance support, may promote more equitable access to high-value non-NIP vaccines. Collectively, these measures could contribute to improving non-NIP vaccine uptake and enhancing the effectiveness and equity of China's immunization system.

Reviewer 2 Report

Comments and Suggestions for Authors

Overall comments: This study assesses awareness and knowledge levels to determine the elements that influence attitudes towards non-National Immunisation Program vaccinations among Shanghai residents. However, a thorough statistical analysis is lacking in this paper, and there is room for improvement in the English language.

Abstract

• Line 16: “Population-Based Study” Why is it written in capital letters?
• Line 18: Please write non-NIP in full since this is the first time it is used in the abstract.
• Line 18: Much better to highlight as a “population cross-sectional survey”
• Line 19: Please include a more precise date: October to December 2024.
• Line 34: This variable is not clear “urban/rural residence”, is it a ratio? Please clarify.
• The abstract conclusion is too long and confusing. Please generate a conclusion based on the study’s findings and add the recommendations.

Introduction

• Please briefly talk about different types of non-NIP vaccines in China, focussing mainly on their efficacy.
• Lines 79-94 should be inserted in the discussion.
• Mainly focus on NIP and non-NIP vaccines from the point of view of knowledge and attitudes in China. Please review previous studies conducted in China.

Materials and Methods

• Line 106: See the abstract; it's much better to say a population cross-sectional survey.
• Line 106: See the abstract; please include the exact dates when the survey was undertaken.
• Clearly state explanatory and outcome variables.
• Line 132-134: Please clearly define the sociodemographic information as categorical or continuous. E.g., educational level: no education, primary, secondary, and higher education.
• Line 152-153: Descriptive statistics summarise demographic characteristics and outcome distributions. Clearly state whether percentage, mean (SD), or median (interquartile) was used.
• How did you check the multicollinearity between variables?
• What were the assumptions of using ordinal logistic regression models? How did you choose the variable to include in the model? How did you test that the models fit to the data? Please clearly state it, as the section of statistical analysis is too short for this study.
• Clearly show that normality fails; this is why the Kruskal–Wallis tests was used.

Results

• Line 187: Table 2, Univariate Analysis – is it univariable ordinal logistic regression? This is not clear.
• Tables 3 and 4: I am concerned about large odd ratios reported in this table. Please compute the model fitness and well as variance inflation factor (VIF) for multicollinearity.
• Improve how to report the study’s results by including figures such as a coeplot for statistically significant odd ratios.  
• Clearly state whether the odd ratios are crude or adjusted.

Discussion

• You just talk about the study’s limitations; please include the study’s strengths.
• Clearly highlight what public health policy this study may generate in the Chinese immunisation NIP based on non-NIP vaccine efficacy and factors influencing public perceptions of these vaccines.

Author Response

Comments 1: Line 16: "Population-Based Study" Why is it written in capital letters?

Response 1: Thank you for pointing this out. We agree with the comment. Therefore, we have revised the capitalization to ensure consistency with standard formatting conventions, changing "Population-Based Study" to "population-based study". This revision can be found on Line 16 of the revised manuscript.

 

Comments 2: Line 18: Please write non-NIP in full since this is the first time it is used in the abstract.

Response 2: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the abstract to spell out the full term at its first occurrence and provided the abbreviation in parentheses for clarity and consistency. This revision can be found on Line 18 of the revised manuscript: non-National Immunization Program (non-NIP) vaccines.

 

Comments 3: Line 18: Much better to highlight as a "population cross-sectional survey"

Response 3: Thank you for pointing this out. We agree with the reviewer. Therefore, we have revised the wording in the abstract to explicitly describe the study as "population cross-sectional survey" improving methodological clarity and accuracy. This revision can be found on Line 19 of the revised manuscript.

 

Comments 4: Line 19: Please include a more precise date: October to December 2024.

Response 4: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the abstract to specify the exact study period to improve temporal clarity. This revision can be found on Line 20 of the revised manuscript: from 20 October to 31 December 2024

 

Comments 5: Line 34: This variable is not clear "urban/rural residence", is it a ratio? Please clarify.

Response 5: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified this variable by explicitly defining it as a categorical residence type rather than a ratio, and by specifying its categories as urban community, town-center, or rural. This revision improves the clarity and interpretability of the variable definition. This revision can be found on Lines 38-39 of the revised manuscript: "residence type (urban community, town-center, or rural)".

 

Comments 6: The abstract conclusion is too long and confusing. Please generate a conclusion based on the study's findings and add the recommendations.

Response 6: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the abstract conclusion to improve clarity and conciseness by streamlining the wording while retaining the core findings and key recommendations derived from the study. The revised version focuses on the central phenomenon identified and presents targeted recommendations in a clearer and more concise manner. This revision can be found on Lines 41-47 of the revised manuscript: Overall, Shanghai residents exhibited suboptimal awareness and acceptance of non-NIP vaccines, with a clear 'high acceptance but low knowledge' phenomenon. To improve awareness, strategies should include strengthening healthcare providers' recommendations, implementing systematic educational campaigns. To enhance acceptance, efforts should focus on disseminating positive, evidence-based information; reinforcing provider guidance; expanding outreach and education; and optimizing payment mechanisms to improve economic accessibility.

 

Comments 7: Introduction: Please briefly talk about different types of non-NIP vaccines in China, focusing mainly on their efficacy. Response 7: Thank you for pointing this out. We agree with this comment. Therefore, we have added a concise description of the major types of non-National Immunization Program (non-NIP) vaccines currently used in China, with a primary focus on their demonstrated efficacy. Specifically, we summarized commonly used non-NIP vaccines and highlighted evidence from clinical trials and real-world studies showing their effectiveness in reducing disease incidence and related complications across different age groups. This revision can be found on Lines 66-72 of the revised manuscript: Common non-NIP vaccines in China include rabies vaccine, seasonal influenza vaccine, varicella (chickenpox) vaccine, rotavirus vaccine, enterovirus 71 (EV71) vaccine, and pneumococcal conjugate vaccines (PCV), among others. Among non-NIP vaccines, rabies, influenza, and varicella vaccines are typically the most frequently administered in China. Existing evidence from clinical trials and real-world studies has demonstrated that these vaccines are effective in reducing disease incidence and related complications across different age groups [3-6].

 

Comments 8: Lines 79-94 should be inserted in the discussion.

Response 8: Thank you for pointing this out. We agree with this comment. Therefore, we have substantially condensed the content in Lines 79-94 and relocated the original detailed discussion to the Discussion section, where it is more appropriate for interpretation and contextualization of the findings. This restructuring improves the logical flow and avoids redundancy between sections. This revision can be found in the revised manuscript, with the content moving toLines 520-530.

 

Comments 9: Mainly focus on NIP and non-NIP vaccines from the point of view of knowledge and attitudes in China. Please review previous studies conducted in China.

Response 9: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the relevant section to focus explicitly on National Immunization Program (NIP) and Non-National Immunization Program (non-NIP) vaccines from the perspective of knowledge and attitudes in China, and incorporated a concise review of prior empirical studies conducted in the Chinese context. The revised text summarizes evidence showing how vaccine-related knowledge, attitudes toward efficacy and safety, and socioeconomic considerations influence willingness to vaccinate, and situates these findings within China's immunization service framework. This revision can be found on Lines 151-169 of the revised manuscript: In the Chinese context, empirical studies have identified several determinants of vaccine awareness and acceptance. For example, a survey of the general adult population in China found that knowledge and positive attitudes toward vaccine efficacy and safety were significantly associated with willingness to vaccinate, suggesting that higher levels of disease and vaccine knowledge can enhance acceptance in large-scale immunization campaigns [32]. Moreover, systematic reviews of HPV vaccination in mainland China have shown that awareness of the vaccine, understanding of disease risk, perceived vaccine safety, and cost considerations are among the key predictors of willingness to accept non-NIP vaccines, indicating that both informational and socioeconomic factors play important roles in shaping vaccine decisions [33]. In Shanghai, immunization services are delivered through an integrated public health network coordinated by the Centers for Disease Control and Prevention and community health centers, where National Immunization Program (NIP) vaccines are administered according to the national schedule and non-NIP vaccines are offered on a voluntary, self-paid basis. The municipal government has implemented supplementary programs to increase immunization coverage and uptake among residents, including provision of varicella and pneumococcal conjugate vaccine (PCV) vaccination services for targeted populations, as well as other supplementary immunization measures.

 

Comments 10: Line 106: See the abstract; it's much better to say a population cross-sectional survey.

Response 10: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the wording to ensure consistency with the abstract by explicitly describing the study as a population-based cross-sectional survey. This revision can be found on Line 188 of the revised manuscript: this population-based cross-sectional survey.

 

Comments 11: Line 106: See the abstract; please include the exact dates when the survey was undertaken

Response11: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the Methods section to include the exact dates during which the survey was conducted, ensuring consistency with the abstract and improving temporal clarity. This revision can be found on Lines 199-200 of the revised manuscript: from 20 October to 31 December 2024.

 

Comments 12: Clearly state explanatory and outcome variables.

Response 12: Thank you for pointing this out. We agree with this comment. Therefore, we have revised Section 2.3 (Variables) to explicitly distinguish between explanatory variables and outcome variables, clearly stating the analytical role of each variable group. Specifically, sociodemographic characteristics and additional factors were defined as explanatory variables, while awareness and acceptance of non-NIP vaccines were defined as outcome variables, and measurement details were consolidated in a dedicated table for clarity. These revisions can be found on Lines 301, 304-305, 311, and 319 of the revised manuscript: sociodemographic information was specified as explanatory variables; awareness and acceptance were specified as outcome variables; additional factors were specified as explanatory variables; and Table 1 (Measurement and Response Options of Study Variables) was added to systematically present all variable definitions and response categories.

 

Comments 13: Line 132-134: Please clearly define the sociodemographic information as categorical or continuous. E.g., educational level: no education, primary, secondary, and higher education.

Response 13: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified that all sociodemographic variables were treated as categorical variables because the primary objective of this study was to identify influencing factors associated with different levels of awareness and acceptance. Categorization allows for clearer comparison across meaningful population subgroups and facilitates interpretation of associations in both univariate screening and multivariable ordinal logistic regression analyzes. This revision can be found on Lines 300-301 of the revised manuscript: All sociodemographic variables were treated as categorical variables.

 

Comments 14: Line 152-153: Descriptive statistics summarize demographic characteristics and outcome distributions. Clearly state whether percentage, mean (SD), or median (interquartile) was used.

Response 14: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the Statistical Analysis section to explicitly state the type of descriptive statistics used for summarizing demographic characteristics and outcome distributions, clarifying that frequencies and percentages were applied for categorical variables and ordinal outcomes. This revision improves transparency and aligns the descriptive analysis with the nature of the study variables. This revision can be found on Lines 362-365 of the revised manuscript: Descriptive statistics were first used to summarize participants' sociodemographic characteristics as well as the distribution of awareness and acceptance levels of non-NIP vaccines, including frequencies and percentages for categorical variables and ordinal outcomes.

 

Comments 15: How did you check the multicollinearity between variables?

Response 15: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified how multicollinearity among explanatory variables was assessed and revised the wording describing model diagnostics to ensure methodological accuracy. Specifically, multicollinearity was examined using variance inflation factors (VIFs), and no variables exceeded commonly accepted thresholds, indicating no concerning multicollinearity. In addition, we refined the description of model assumptions and fit, clarifying that a non-significant test of parallel lines supports the proportional odds assumption, while a significant likelihood ratio test indicates that the fitted model provides a better fit than the null model. This revision can be found on Lines 377-381 of the revised manuscript: The proportional odds assumption was evaluated using the test of parallel lines, while overall model fit was assessed using the likelihood ratio test. A non-significant result for the parallel lines test supports the proportional odds assumption, and a significant likelihood ratio test indicates that the model with predictors fits the data better than the null model.

 

Comments 16: What were the assumptions of using ordinal logistic regression models? How did you choose the variable to include in the model? How did you test that the models fit the data? Please clearly state it, as the section of statistical analysis is too short for this study.

Response 16: Thank you for pointing this out. We agree with this comment. Therefore, we have revised and expanded the Statistical Analysis section to explicitly state the assumptions of ordinal logistic regression, the criteria used for variable selection, and the procedures applied to assess model fit, as these details were not sufficiently described in the original version. Specifically, we added a clear statement that variables included in the multivariable ordinal logistic regression models were selected based on statistical significance in univariate Kruskal-Wallis tests as an initial screening step. In addition, we clarified that the proportional odds assumption was evaluated using the test of parallel lines, overall model fit was assessed using the likelihood ratio test, and multicollinearity was examined using variance inflation factors (VIFs). These revisions can be found on Lines 373-384 of the revised manuscript: Variables that showed statistical significance in the Kruskal-Wallis tests were subsequently entered into the multivariable ordinal logistic regression models for further analysis. The proportional odds assumption was evaluated using the test of parallel lines, while overall model fit was assessed using the likelihood ratio test. A non-significant result for the parallel lines test supports the proportional odds assumption, and a significant likelihood ratio test indicates that the model with predictors fits the data better than the null model. Multicollinearity among explanatory variables was assessed using variance inflation factors (VIFs), and variables with VIF > 10 were considered to have concerning multicollinearity; no explanatory variables exceeded this threshold. Lines 373-384 of the revised manuscript: Variables that showed statistical significance in the Kruskal-Wallis tests were subsequently entered into the multivariable ordinal logistic regression models for further analysis.

 

Comments 17: Clearly show that normality fails; this is why the Kruskal-Wallis tests was used.

Response 17: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the Statistical Analysis section to explicitly state that the dependent variables did not meet the assumptions of normal distribution and were measured on ordinal scales, which justified the use of non-parametric methods, including the Kruskal-Wallis test. This clarification directly explains the rationale for selecting non-parametric tests in the univariate analysis. This revision can be found on Lines 365-367 of the revised manuscript: Because the dependent variables (awareness and acceptance of non-NIP vaccines) were measured on ordinal scales and did not meet the assumptions of normal distribution, non-parametric methods were applied.

 

Comments 18: Line 187: Table 2, Univariate Analysis - is it univariable ordinal logistic regression? This is not clear.

Response 18: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified that the analyzes presented in Table 2 are based on univariate (single-factor) non-parametric comparisons using Kruskal-Wallis tests, rather than univariable ordinal logistic regression. To avoid ambiguity, we explicitly labeled these analyzes as "Kruskal-Wallis tests (univariate analysis)" in the Results section. This clarification ensures a clear distinction between univariate screening analyzes and subsequent multivariable ordinal logistic regression models. These revisions can be found on Lines 413 and 424 of the revised manuscript: Kruskal-Wallis tests (univariate analysis).

 

Comments 19: Tables 3 and 4: I am concerned about large odd ratios reported in this table. Please compute the model fitness and also variance inflation factor (VIF) for multicollinearity.

Response 19: Thank you for pointing this out. We agree with this comment. Therefore, we have further evaluated the fitness of the multivariable ordinal logistic regression models and assessed multicollinearity among explanatory variables by calculating variance inflation factors (VIFs), in response to concerns regarding the magnitude of the reported odds ratios. The results show that all VIF values were within acceptable ranges, indicating no concerning multicollinearity, and the model fit statistics support the adequacy of the fitted models. These additional results were incorporated into the revised tables to improve transparency and robustness of the regression analyzes. These revisions can be found on Lines 454 and 480 of the revised manuscript and are presented in Tables 5 and 6 (formerly Tables 3 and 4).

 

Comments 20: Improve how to report the study's results by including figures such as a coeplot for statistically significant odd ratios.

Response 20: Thank you for pointing this out. We agree with this comment. Therefore, we have improved the presentation of the study results by adding coefficient plots (coeplots) to visually display the adjusted odds ratios and corresponding 95% confidence intervals for statistically significant predictors identified in the multivariable ordinal logistic regression models. These figures enhance the interpretability of the regression results and allow readers to more intuitively compare the magnitude and direction of associations across predictors. The newly added figures can be found on Line 455 as Figure 1. Adjusted Odds Ratios and 95% Confidence Intervals for Predictors of Awareness of Non-NIP Vaccines, and on Line 481 as Figure 2. Adjusted Odds Ratios and 95% Confidence Intervals for Predictors of Acceptance of Non-NIP Vaccines, both generated using Stata.

 

Comments 21: Clearly state whether the odd ratios are crude or adjusted.

Response21: Thank you for pointing this out. We agree with this comment. Therefore, we have explicitly clarified that all odds ratios reported in the multivariable ordinal logistic regression tables are adjusted odds ratios. This clarification ensures that readers clearly understand that the reported associations account for the simultaneous effects of multiple covariates included in the models. These revisions can be found on Lines 454 and 480 of the revised manuscript, where in Tables 5 and 6 now explicitly state "Adjusted Odds Ratio (95% CI)".

 

Comments 22: You just talk about the study's limitations; please include the study's strengths.

Response 22: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the Discussion section to explicitly include the study's strengths in addition to its limitations by adding a dedicated subsection highlighting the public health relevance and methodological strengths of this study. These revisions can be found on Lines 685-716 of the revised manuscript in the newly added subsection: 4.5. Public Health Relevance and Study Strengths.

This study's multimethod analytical approach—starting with Kruskal-Wallis tests to identify potential predictors and followed by multivariable ordinal logistic regression—enabled robust identification of factors independently associated with both awareness and acceptance of non-NIP vaccines. The use of ordinal measures for awareness and acceptance and appropriate statistical models strengthened the validity of effect estimates, while consideration of a broad set of explanatory variables including sociodemographic characteristics, provider recommendation, education participation, and affordability provided a comprehensive view of determinants shaping vaccine perceptions. Although the cross-sectional design limits causal inference and self-reported data may be subject to recall bias, the findings illuminate key behavioral and informational barriers that can be targets for public health action.

The identified associations between knowledge, attitudes, and acceptance suggest clear directions for public health policy. Tailored communication strategies that emphasize evidence-based information on non-NIP vaccine effectiveness and safety, combined with enhanced provider recommendation and structured educational campaigns, could mitigate misconceptions and improve vaccine confidence. Furthermore, aligning financing mechanisms to reduce perceived financial barriers may help translate positive attitudes into actual uptake. These insights can inform efforts to refine China's immunization framework, particularly in considering how non-NIP vaccines might be better integrated into population health planning to reduce disease burden and promote equitable access.

 

Comments 23: Clearly highlight what public health policy this study may generate in the Chinese immunization NIP based on non-NIP vaccine efficacy and factors influencing public perceptions of these vaccines.

Response 23: Thank you for pointing this out. We agree with this comment. Therefore, we have further clarified the public health policy implications of this study by explicitly linking our findings on non-NIP vaccine efficacy and factors influencing public perceptions to potential directions for China's immunization policy and NIP optimization.  These additions emphasize the relevance of our findings for refining China's immunization framework and supporting more equitable and effective vaccine delivery. These revisions can be found on Lines 685-716 of the revised manuscript in the subsection : Public Health Relevance and Study Strengths.

This study's multimethod analytical approach—starting with Kruskal-Wallis tests to identify potential predictors and followed by multivariable ordinal logistic regression—enabled robust identification of factors independently associated with both awareness and acceptance of non-NIP vaccines. The use of ordinal measures for awareness and acceptance and appropriate statistical models strengthened the validity of effect estimates, while consideration of a broad set of explanatory variables including sociodemographic characteristics, provider recommendation, education participation, and affordability provided a comprehensive view of determinants shaping vaccine perceptions. Although the cross-sectional design limits causal inference and self-reported data may be subject to recall bias, the findings illuminate key behavioral and informational barriers that can be targets for public health action.

The identified associations between knowledge, attitudes, and acceptance suggest clear directions for public health policy. Tailored communication strategies that emphasize evidence-based information on non-NIP vaccine effectiveness and safety, combined with enhanced provider recommendation and structured educational campaigns, could mitigate misconceptions and improve vaccine confidence. Furthermore, aligning financing mechanisms to reduce perceived financial barriers may help translate positive attitudes into actual uptake. These insights can inform efforts to refine China's immunization framework, particularly in considering how non-NIP vaccines might be better integrated into population health planning to reduce disease burden and promote equitable access.

Reviewer 3 Report

Comments and Suggestions for Authors

An interesting study: A few observations:

1) Were you able to gauge the specific reasons for non-VIP vaccines by participants?

2) Were prior vaccination taken into account...I did not see it asked

3) Were attitudes towards vaccination included?

4) What model(s)/theory(s) were used for the creation of the survey and the interpretation of the results? 

Author Response

Comments 1: Were you able to gauge the specific reasons for non-VIP vaccines by participants?

Response 1: Thank you for this insightful question. We agree with this comment. Therefore, we have clarified in the Methods and Discussion sections that the present study did not include direct items asking participants to explicitly report their specific reasons for not receiving Non-National Immunization Program (non-NIP) vaccines; instead, the survey was designed to examine awareness, acceptance, and a set of explanatory factors that have been widely shown to be associated with vaccine hesitancy and low uptake. Specifically, we measured indirect but policy-relevant factors such as perceived affordability of vaccine price, recommendation by healthcare personnel, awareness of vaccine-related adverse events, and participation in vaccine education activities, which together provide insight into potential economic, informational, and provider-related influences on non-NIP vaccine acceptance. These clarifications have been added on Lines 314-318 in the Methods section and Lines 487-490 in the Discussion section of the revised manuscript: The survey did not include direct items soliciting participants' self-reported reasons for not receiving non-NIP vaccines; instead, we measured a set of explanatory variables that have been shown to be associated with vaccine hesitancy and low uptake, including perceived affordability, recommendation by healthcare personnel, awareness of adverse events, and participation in vaccine education (Lines 314-318). While this study did not directly elicit participants' reasons for not vaccinating, associated factors identified suggest economic, informational, and provider-related influences on non-NIP vaccine acceptance (Lines 487-490).

 

Comments 2: Were prior vaccination taken into account...I did not see it asked

Response 2: Thank you for raising this important point. We agree with this comment. Therefore, we have clarified that the present survey did not include a direct item capturing participants' prior vaccination history (e.g., previous receipt of specific non-NIP vaccines or overall immunization records), as the primary focus of this study was to assess residents' awareness and acceptance of non-NIP vaccines and to identify sociodemographic, informational, and psychosocial factors associated with these perceptions rather than actual vaccination behaviors. We have now explicitly acknowledged this as a limitation and noted that future studies should incorporate prior vaccination status to more comprehensively examine the relationship between awareness, acceptance, and real-world vaccine uptake. This clarification has been added to the Limitations section on Lines 692-694 of the revised manuscript: Finally, although we included a broad set of explanatory variables, this study did not capture certain potentially relevant variables, such as vaccination status, which should be addressed in future research.

 

Comments 3: Were attitudes toward vaccination included?

Response 3: Thank you for raising this important observation. We agree with this comment. Therefore, we have clarified that in the present study we conceptualized and operationalized participants' overall attitudinal disposition toward Non-National Immunization Program (non-NIP) vaccines through the construct of "acceptance," which was measured using a five-point Likert scale reflecting respondents' degree of endorsement or willingness to vaccinate. While "attitude" and "acceptance" are conceptually related but not identical, our survey design distinguished between awareness (knowledge and understanding) and acceptance (supportive stance and behavioral inclination), and did not include separate multi-item psychometric scales explicitly labeled as vaccine attitudes . This approach is supported by population-based vaccination studies in which acceptance or willingness to vaccinate has been operationalized as an indicator of individuals' attitudinal orientation toward vaccination, especially in the context of vaccination decision-making and behavioral intention. We have now clarified this conceptualization in the Methods section. Future research will further explore vaccination attitudes using validated attitude scales to better distinguish underlying belief components from behavioral intent.

 

Comments 4: What model(s)/theory(s) were used for the creation of the survey and the interpretation of the results? 

Response 4: Thank you for raising this important question. We agree with this comment. Therefore, we have explicitly clarified the theoretical frameworks that informed both the development of the survey instrument and the interpretation of the study findings. Specifically, in the Methods section we have stated that the questionnaire was conceptually guided by the Knowledge-Attitude-Practice (KAP) model, which explains how knowledge influences attitudes and subsequently shapes health-related behaviors, and by key constructs of the Health Belief Model (HBM), including perceived susceptibility, severity, benefits, barriers, and cues to action, to support the organization of awareness and acceptance measures. In addition, in the Discussion section we have further interpreted our findings through the lens of the HBM, explaining how factors such as healthcare provider recommendation and participation in educational activities function as cues to action that enhance perceived benefits and reduce perceived barriers, thereby increasing vaccine acceptance. These revisions can be found on Lines 255-260 in the Methods section and Lines 601-613 in the Discussion section of the revised manuscript: We also referenced the Knowledge-Attitude-Practice (KAP) model, which conceptualizes how knowledge influences attitudes and, in turn, shapes health behavior, and constructs from the Health Belief Model (HBM), which emphasizes perceived susceptibility, severity, benefits, barriers, and cues to action in health decision-making, to support the conceptual organization of awareness and acceptance constructs in the questionnaire (Lines 255-260). From a theoretical perspective, the observed determinants of non-NIP vaccine acceptance align with constructs of the Health Belief Model (HBM), a well-established framework in health behavior research that explains preventive health actions such as vaccination (Lines 601-613).

Reviewer 4 Report

Comments and Suggestions for Authors

The topic is interesting. The paper is very well organized and well written. Before publication, some observations should be included and some issues clarified.

Introduction

It is very well developed. However, the authors should briefly explain the vaccination program developed in Shanghai. They should also include data on published studies on attitudes toward vaccination in other geographical areas besides Malawi and Ethiopia (lines 81-88).

Materials and Methods

It is well designed. The authors comment in the limitations section that ‘Although stratified random sampling covered multiple communities and districts, selection bias may still be present.’ The authors should comment on some characteristics of the selected districts.

A validated questionnaire was used. The authors should briefly explain the validation process.

A specific section on ‘Ethical Considerations’ should be included, including the approval of the relevant ethics committee.

Results

It is well written, and the tables are of great importance to make the findings clear and facilitate the reading of the study.

Discussion

The authors should explain the relationship between positive attitudes and actual knowledge and the existing key barrier (lines 250 and 251).

The limitations section is well explained by the authors.

The conclusions are clear and well written.

References

The number of references is high and they are up-to-date.

Author Response

Comments 1: Introduction: It is very well developed. However, the authors should briefly explain the vaccination program developed in Shanghai. They should also include data on published studies on attitudes toward vaccination in other geographical areas besides Malawi and Ethiopia (lines 81-88).

Response 1: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the Introduction to briefly explain the immunization program implemented in Shanghai and to expand the literature review by including published studies on vaccination attitudes and acceptance from geographical areas beyond Malawi and Ethiopia, particularly evidence from China. Specifically, we added a concise description of Shanghai's immunization service system and summarized empirical studies conducted in China that have identified key determinants of vaccine awareness and acceptance, thereby providing broader contextual and international relevance. This revision can be found on Lines 151-169 of the revised manuscript: In the Chinese context, empirical studies have identified several determinants of vaccine awareness and acceptance. For example, a survey of the general adult population in China found that knowledge and positive attitudes toward vaccine efficacy and safety were significantly associated with willingness to vaccinate, suggesting that higher levels of disease and vaccine knowledge can enhance acceptance in large-scale immunization campaigns [32]. Moreover, systematic reviews of HPV vaccination in mainland China have shown that awareness of the vaccine, understanding of disease risk, perceived vaccine safety, and cost considerations are among the key predictors of willingness to accept non-NIP vaccines, indicating that both informational and socioeconomic factors play important roles in shaping vaccine decisions [33]. In Shanghai, immunization services are delivered through an integrated public health network coordinated by the Centers for Disease Control and Prevention and community health centers, where National Immunization Program (NIP) vaccines are administered according to the national schedule and non-NIP vaccines are offered on a voluntary, self-paid basis. The municipal government has implemented supplementary programs to increase immunization coverage and uptake among residents, including provision of varicella and pneumococcal conjugate vaccine (PCV) vaccination services for targeted populations, as well as other supplementary immunization measures.

 

Comments 2: Materials and Methods: It is well designed. The authors comment in the limitations section that 'Although stratified random sampling covered multiple communities and districts, selection bias may still be present.' The authors should comment on some characteristics of the selected districts.

Response 2: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the Materials and Methods section to more clearly describe the characteristics of the selected districts and to explain how the stratified sampling strategy ensured representation of key variations across Shanghai, thereby addressing potential concerns regarding selection bias. Specifically, we clarified that all 16 administrative districts were first stratified based on geographic and demographic characteristics, and that the selected districts included both highly urbanized central areas and more socioeconomically diverse suburban areas, capturing variation in population density, economic conditions, and healthcare access. This revision can be found on Lines 202-214 of the revised manuscript: First, all 16 administrative districts in Shanghai were stratified into categories (e.g., central urban vs suburban) to form strata, based on geographic and demographic characteristics. We chose five districts based on statistical efficiency and logistical feasibility given our target sample size, consistent with prior surveys, where a similar number of primary sampling units was sufficient to achieve both adequate representation and manageable field operations. The selected five districts included both highly urbanized central areas (such as districts with high population density and advanced health service infrastructure) and more socioeconomically diverse suburban areas, ensuring that the sample captured key variations in demographic, economic, and healthcare access characteristics across Shanghai's municipal population. Within each stratum, the five districts were selected using simple random sampling, in which each district had an equal probability of being chosen from the list of all eligible districts within that stratum.

 

Comments 3: Materials and Methods: A validated questionnaire was used. The authors should briefly explain the validation process.

Response 3: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the Materials and Methods section to briefly but clearly describe the questionnaire validation process, including content validity, pilot testing, and reliability and construct validity assessments. Specifically, we added details on expert consultation to establish content validity, pre-testing in a pilot survey to assess item clarity and comprehensibility, and statistical evaluation of internal consistency and construct validity using Cronbach's α, the Kaiser-Meyer-Olkin (KMO) measure, and Bartlett's test of sphericity. This revision can be found on Lines 266-295 of the revised manuscript: Through this expert review process, content validity of the questionnaire was established by ensuring that all items were relevant, clear, and aligned with the study objectives. In addition to expert consultation workshops, the complete questionnaire was pre-tested in a pilot survey with a small sample of community residents representative of the target population. The questionnaire consisted of 49 questions covering sociodemographic characteristics, awareness and acceptance constructs, and potential influencing factors. Completion of the questionnaire took approximately 15 minutes on average. The purpose of this pre-test was to evaluate item clarity, relevance, response burden, and respondent comprehension. Trained field staff conducted the pilot using the same structured interview mode planned for the main survey. Findings from the pre-test indicated that most items were comprehensible; however, several questions were reworded to improve interpretability and reduce ambiguity. Minor modifications focused on simplifying phrasing and clarifying specific response options without changing the conceptual content of the items. The final questionnaire achieved a Cronbach's α of 0.731, indicating acceptable internal consistency reliability. To evaluate construct validity, the Kaiser-Meyer-Olkin (KMO) measure and Bartlett's test of sphericity were conducted, with a KMO value of 0.716 and a significant Bartlett's test (P < 0.001), supporting acceptable construct validity of the questionnaire.

 

Comments 4: Materials and Methods: A specific section on 'Ethical Considerations' should be included, including the approval of the relevant ethics committee.

Response 4: Thank you for this important comment. We agree that ethical considerations should be clearly stated. Accordingly, we have revised the Methods section to explicitly describe the ethical approval for this study, including the approving ethics committee and the corresponding approval information. The relevant revisions have been made in the Methods section (lines 246-249) of the revised manuscript: The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Institutional Review Board (Ethics Committee) of the Shandong Center for Disease Prevention and Control (protocol code SDJK(K)2024-049-01; date of approval: 9 October 2024).

 

Comments 5: Discussion: The authors should explain the relationship between positive attitudes and actual knowledge and the existing key barrier (lines 250 and 251).

Response 5: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the Discussion section to explicitly explain the relationship between positive attitudes toward vaccination and actual vaccine-specific knowledge, and to clarify how gaps in knowledge constitute a key barrier that may prevent favorable attitudes from translating into informed vaccination behavior. Specifically, drawing on the Knowledge-Attitude-Practice (KAP) framework, we clarified that while positive attitudes provide motivational support for immunization, adequate vaccine-specific knowledge is essential for interpreting nuanced information, assessing risk-benefit tradeoffs, and making context-appropriate vaccination decisions, and that insufficient knowledge can limit the conversion of positive attitudes into consistent vaccine uptake. This revision can be found on Lines 508-530 of the revised manuscript: The Knowledge-Attitude-Practice (KAP) paradigm conceptualizes vaccination behavior as a process in which knowledge influences attitudes, which in turn shape behavioral intentions and practices. This pattern suggests that positive attitudes toward vaccination may reflect a general confidence in vaccines as valuable preventive health tools, while specific knowledge about non-NIP vaccines represents a distinct cognitive dimension that informs practical decision-making. Although favorable attitudes provide motivational support for immunization, adequate vaccine-specific knowledge is necessary for individuals to interpret nuanced information, assess risk-benefit tradeoffs, and make context-appropriate vaccination decisions. Consequently, gaps in factual understanding can act as a key barrier that limits the extent to which positive attitudes translate into consistent and informed vaccine uptake in real-world settings. These findings were consistent with a substantial body of research both in China and internationally.

Reviewer 5 Report

Comments and Suggestions for Authors

This study uses data from a survey conducted during “October to December 2024” in Shanghai to explore residences’ knowledge and attitudes toward non-National Immunization Program (non-NIP) vaccines. The following points will help strengthen the manuscript.

1.Lines 49-103: Please consider explaining what types of vaccines are covered under non-NIP group and which ones are part of NIP. Which vaccines in the non-NIP group are “highly cost-effective”? For which populations are the non-NIP vaccines applicable to … Which ones of these population-specific non-NIP vaccines are “important supplements or alternatives to NIP vaccines”? If some of them are “alternatives to NIP” vaccines, what is their percentage in the non-NIP group? Why are residents expected to pay for the alternatives…
2.Lines 49-103: How does the “vaccination coverage for non-NIP vaccines” compare to the coverage for NIP vaccines?
3.Line 104: Please start this section with a setting section and provide characteristics of this setting.
4.Line 104: Did authors consider using a guideline for reporting and conducting this survey research?
5.Lines 105-118: Please be very specific about the sampling strategy used in this study. What type of a “stratified random sampling” is this? Which survey sampling strategy reference (book or article) did authors use? Please provide a specific survey sampling strategy reference here.
6.Lines 105-118: The sampling strategy explanations follow with a mention of “districts”. How many “districts” are there in Shanghai in total? 
7.Lines 105-118: Why “5 districts” were chosen? Based on what? Further, please align this section with the setting section above so that the readership will understand the characteristics of those districts and their relevance for this study.
8.Lines 105-118: What type of a “random” selection of those “5 districts” were conducted?
9.Lines 105-118: What do authors mean by “communities”? Which “4 communities”?Based on what? Again, what type of a “random” sample is this?
10.Lines 105-118: Why “35-40 residents per community” were chosen? Again, what type of a “random” sample is this?
11.Lines 105-118: What do authors mean by “residents”? Where and how were they “recruited”? 
12.Lines 105-118: Please relate the formula reported here to the revised description of the sampling strategy as mentioned above. Thanks for providing a reference. However, as above, it is preferable to use a survey sampling strategy reference. 
13.Lines 105-188: Where were these survey instruments “distributed”?
14.Lines 105-188: In survey research, this section should include the number of attempts made to contact responders… the approach to the calculation of response rates need to be provided… the evidence of the representativeness of the study sample of the population it is targeting should be presented… It is concerning to this reviewer that none of these information is provided. As before, it is not even clear what type of a sampling strategy was used.
15.Lines 119-150: What was the mode of survey administration? Which procedures were followed in this mode while administering the survey?
16.Lines 119-150: Please provide a copy of the survey instrument in an Appendix.
17.Lines 119-150: Please explain if the complete survey instrument was pre-tested other than getting input during “expert consultation workshops”? What were the findings of the pre-test? What, if any, modifications were carried out, why…
18.Lines 119-150: Authors seem to suggest that at least a fraction of non-NIP vaccines are “alternatives” (lines 53-55). Part of the non-NIP vaccines may be for different populations in the society. Which non-NIP vaccines were respondents thinking about when answering the survey questions?
19.Lines 119-150: Please be precise and explain how many questions were there in the survey instrument in total and how long did it take to complete it?
20.Lines 152-157: The statistical analysis reported here needs to incorporate detail based on the revised sampling strategy description above.
21.Lines 152-157: Were there any missing information? If yes, what remedies were used?
22.Lines 160-166: Thanks for the information. However, in survey research, first table should include population and sample characteristics side by side so that there is information about how representative the study sample is of the population. For example, is Shanghai “resident” population ’32% male’ and ’68% female’? 
23.Lines 160-166: If the sample is not representative, which techniques were used as a remedy? In this study, authors indicate that “the study was representative…” (lines 379-383) but they do not provide any evidence that the sample is, in fact, representative.
24.Lines 160-166: Please make sure to support assertions made with a reference (like the annual income percentages provided for 2023 here). Is this information for Shanghai? Please be specific.

 

Author Response

Comments 1: Lines 49-103: Please consider explaining what types of vaccines are covered under non-NIP group and which ones are part of NIP. Which vaccines in the non-NIP group are "highly cost-effective"? For which populations are the non-NIP vaccines applicable to … Which ones of these population-specific non-NIP vaccines are "important supplements or alternatives to NIP vaccines"? If some of them are "alternatives to NIP" vaccines, what is their percentage in the non-NIP group? Why are residents expected to pay for the alternatives…

Response 1: Thank you for pointing this out. We agree with this comment. Therefore, we have substantially revised the Introduction to clearly distinguish vaccines included in the National Immunization Program (NIP) from those classified as non-National Immunization Program (non-NIP) vaccines, and to systematically explain the types of non-NIP vaccines, their target populations, their cost-effectiveness evidence, and their role as important supplements or alternatives to NIP vaccines. Specifically, we clarified which vaccines are covered under the NIP, summarized commonly used non-NIP vaccines in China, identified non-NIP vaccines with favorable cost-effectiveness profiles, described population-specific indications, and explained why certain non-NIP vaccines function as alternatives to NIP vaccines and are self-paid under the current immunization financing framework. This revision can be found on Lines 53-113 of the revised manuscript: In China, vaccines are broadly categorized into National Immunization Program (NIP) vaccines and non-National Immunization Program (non-NIP) vaccines. NIP vaccines are funded and provided free of charge by the government as part of routine public health services and are administered according to the national immunization schedule, such as hepatitis B vaccine, Bacillus Calmette-Guérin (BCG) vaccine, inactivated poliovirus vaccine (IPV), attenuated oral poliovirus vaccine (OPV), diphtheria-tetanus-pertussis (DTP) vaccine, measles-mumps-rubella (MMR) vaccine, live and inactivated Japanese encephalitis vaccines, group A and group A/C meningococcal polysaccharide vaccines, and hepatitis A vaccines (both live and inactivated). The routine vaccination coverage rate for NIP vaccines among eligible children in China has been consistently maintained above 90%, reflecting long-standing success in preventing major childhood infectious diseases.

Non-National Immunization Program vaccines refer to vaccines not included in China's National Immunization Program and administered voluntarily at individuals' expense[2]. Common non-NIP vaccines in China include rabies vaccine, seasonal influenza vaccine, varicella (chickenpox) vaccine, rotavirus vaccine, enterovirus 71 (EV71) vaccine, and pneumococcal conjugate vaccines (PCV), among others. Among non-NIP vaccines, rabies, influenza, and varicella vaccines are typically the most frequently administered in China. Existing evidence from clinical trials and real-world studies has demonstrated that these vaccines are effective in reducing disease incidence and related complications across different age groups [3-6]. Compared with vaccines included in China's National Immunization Program (NIP), vaccination coverage for non-NIP vaccines was substantially lower. Relevant research indicated that the full primary series coverage for 13-valent pneumococcal conjugate vaccine (PCV13) was approximately 5.1% in 2019, rotavirus vaccine three-dose coverage was about 1.8%, and Haemophilus influenzae type b (Hib) vaccine three-dose coverage was approximately 25.0%; only first-dose varicella vaccine approached higher levels (around 67.1%) in selected settings, reflecting localized inclusion in some municipal programs rather than national free provision. These figures illustrated that non-NIP vaccine uptake remained far below the near-universal coverage of NIP vaccines and exhibited substantial regional and product-specific disparities [7].

A growing body of economic evaluation studies in China suggests that several non-NIP vaccines, particularly PCV, HPV, Hib, and influenza vaccines, demonstrate favorable cost-effectiveness profiles compared with standard thresholds based on GDP per capita, supporting their potential public health value in appropriate populations.[8, 9]. Different non-National Immunization Program (non-NIP) vaccines are indicated for distinct population groups based on disease risk and prevention objectives. For example, pneumococcal conjugate vaccines (PCV) and Haemophilus influenzae type b (Hib) vaccines are primarily used in infants and young children to prevent invasive bacterial infections such as pneumonia and meningitis, rotavirus vaccines are administered to young children to prevent severe diarrhoeal disease, human papillomavirus (HPV) vaccines target adolescents and young adults to prevent HPV-associated cancers, and seasonal influenza vaccines are recommended for children, older adults, pregnant women, and other high-risk groups to reduce influenza-related morbidity and mortality. Some non-National Immunization Program (non-NIP) vaccines in China are regarded as alternative vaccines because they can replace vaccines already included in the National Immunization Program (NIP) by using formulations with broader antigenic coverage or combined components. For example, the pentavalent combination vaccine (DTaP-IPV-Hib) integrates diphtheria, tetanus, acellular pertussis, inactivated poliovirus, and Haemophilus influenzae type b antigens into a single product, thereby substituting for multiple separate NIP vaccines while reducing the number of injections. Similarly, quadrivalent meningococcal conjugate vaccines (ACWY) provide broader serogroup protection than monovalent group A meningococcal vaccines, and nine-valent human papillomavirus (HPV) vaccines extend protection beyond bivalent formulations by covering additional oncogenic HPV types.

The extent to which these alternative non-NIP vaccines are used varies across regions and vaccine types, depending on local supply, service practices, and population demand. These alternative non-NIP vaccines are typically self-paid because they are not included in the publicly funded NIP schedule. From the perspective of vaccine recipients, the primary motivations for choosing these self-paid alternatives are their favorable safety profiles, and the reduced number of injections achieved through combination vaccines.

 

Comments 2: Lines 49-103: How does the "vaccination coverage for non-NIP vaccines" compare to the coverage for NIP vaccines?

Response 2: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the Introduction to explicitly compare vaccination coverage for Non-National Immunization Program (non-NIP) vaccines with that of National Immunization Program (NIP) vaccines, using available empirical data to illustrate the magnitude of the difference. Specifically, we added representative coverage estimates for several commonly used non-NIP vaccines and contrasted them with the near-universal coverage achieved by NIP vaccines, highlighting substantial gaps as well as regional and product-specific disparities. This revision can be found on Lines 72-82 of the revised manuscript: Compared with vaccines included in China's National Immunization Program (NIP), vaccination coverage for non-NIP vaccines was substantially lower. Relevant research indicated that the full primary series coverage for 13-valent pneumococcal conjugate vaccine (PCV13) was approximately 5.1% in 2019, rotavirus vaccine three-dose coverage was about 1.8%, and Haemophilus influenzae type b (Hib) vaccine three-dose coverage was approximately 25.0%; only first-dose varicella vaccine approached higher levels (around 67.1%) in selected settings, reflecting localized inclusion in some municipal programs rather than national free provision. These figures illustrate that non-NIP vaccine uptake remained far below the near-universal coverage of NIP vaccines and exhibited substantial regional and product-specific disparities [7].

 

Comments 3: Line 104: Please start this section with a setting section and provide characteristics of this setting.

Response 3: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the beginning of the Materials and Methods section to start with a clear description of the study setting and to provide key characteristics of the study location, including its population size, administrative structure, socioeconomic heterogeneity, and immunization service delivery system. This revision can be found on Lines 188-197 of the revised manuscript: This population-based cross-sectional survey was conducted in Shanghai, one of China's largest megacities with a population exceeding 25 million. Shanghai is administratively divided into 16 districts, including both urban and suburban areas, with heterogeneous socioeconomic profiles and varied access to health services. The city has implemented extensive immunization services through CDC and community health centers, providing National Immunization Program (NIP) vaccines free of charge and offering non-NIP vaccines on a voluntary, self-paid basis. The reporting and conduct of this cross-sectional survey followed the STROBE statement, which provides guidance for reporting observational epidemiological studies including sampling strategy, variable definition, and analytic approach.

 

Comments 4: Line 104: Did authors consider using a guideline for reporting and conducting this survey research?

Response 4: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified that the reporting and conduct of this survey were guided by an established reporting guideline for observational studies. Specifically, we stated that this cross-sectional study followed the STROBE statement, which provides standardized guidance on sampling strategy, variable definition, and analytic approach. This revision can be found on Lines 194-197 of the revised manuscript: The reporting and conduct of this cross-sectional survey followed the STROBE statement, which provides guidance for reporting observational epidemiological studies including sampling strategy, variable definition, and analytic approach.

 

Comments 5: Please be very specific about the sampling strategy used in this study. What type of a "stratified random sampling" is this? Which survey sampling strategy reference (book or article) did the authors use? Please provide a specific survey sampling strategy reference here.

Response 5: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the Materials and Methods section to provide a more detailed and transparent description of the survey period and the stratified multistage probability sampling design, including the selection of districts, communities, and individual participants, in order to clarify how representativeness and precision were ensured. This revision can be found on Lines 199-229 of the revised manuscript: This population cross-sectional survey was conducted in Shanghai from 20 October to 31 December 2024. We employed a stratified multistage probability sampling design, consistent with standard survey sampling methodology to improve representativeness and precision. First, all 16 administrative districts in Shanghai were stratified into categories (e.g., central urban vs suburban) to form strata, based on geographic and demographic characteristics. We chose five districts based on statistical efficiency and logistical feasibility given our target sample size, consistent with prior surveys, where a similar number of primary sampling units was sufficient to achieve both adequate representation and manageable field operations. The selected five districts included both highly urbanized central areas and more socioeconomically diverse suburban areas, ensuring that the sample captured key variations in demographic, economic, and healthcare access characteristics across Shanghai's municipal population. Within each stratum, the five districts were selected using simple random sampling. Within each selected district, communities were listed and randomly selected, and 35-40 residents per community were recruited from official household registries using simple random sampling, with the target number derived from the overall sample size calculation to ensure sufficient precision while accounting for design effect in a multistage sampling framework [34,35].

 

Comments 6: Lines 105-118: The sampling strategy explanations follow with a mention of "districts". How many "districts" are there in Shanghai in total? 

Response 6: Thank you for pointing this out. We agree with this comment. Therefore, we have explicitly stated that Shanghai has a total of 16 administrative districts, in order to clearly define the overall sampling frame and help readers better understand how the selected districts were drawn from the full set of eligible districts in the stratified multistage sampling design. This revision can be found on Lines 202-207 of the revised manuscript: First, all 16 administrative districts in Shanghai were stratified into categories (e.g., central urban vs suburban) to form strata, based on geographic and demographic characteristics. We chose five districts based on statistical efficiency and logistical feasibility given our target sample size, consistent with prior surveys, where a similar number of primary sampling units was sufficient to achieve both adequate representation and manageable field operations.

 

Comments 7: Lines 105-118: Why "5 districts" were chosen? Based on what? Further, please align this section with the setting section above so that the readership will understand the characteristics of those districts and their relevance for this study.

Response 7: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified the rationale for selecting five districts and explicitly aligned the sampling strategy with the setting characteristics of Shanghai to improve transparency and interpretability. Specifically, the number of districts was determined based on statistical efficiency and logistical feasibility given the target sample size, consistent with prior population-based surveys, and the selected districts were intentionally drawn from both highly urbanized central areas and more socioeconomically diverse suburban areas to capture key variations in demographic, economic, and healthcare access characteristics relevant to this study. This revision can be found on Lines 204-214 of the revised manuscript: We chose 5 districts based on statistical efficiency and logistical feasibility given our target sample size, consistent with prior surveys, where a similar number of primary sampling units was sufficient to achieve both adequate representation and manageable field operations. The selected five districts included both highly urbanized central areas (such as districts with high population density and advanced health service infrastructure) and more socioeconomically diverse suburban areas, ensuring that the sample captured key variations in demographic, economic, and healthcare access characteristics across Shanghai's municipal population. Within each stratum, the five districts were selected using simple random sampling, in which each district had an equal probability of being chosen from the list of all eligible districts within that stratum.

 

Comments 8: Lines 105-118: What type of a "random" selection of those "5 districts" was conducted?

Response 8: Thank you for pointing this out. We agree with this comment. Therefore, we have explicitly specified the type of random selection used for choosing the five districts, clarifying that a stratified framework was applied followed by simple random sampling within each stratum, such that all eligible districts had an equal probability of selection. This revision can be found on Lines 207-214 of the revised manuscript: The selected five districts included both highly urbanized central areas (such as districts with high population density and advanced health service infrastructure) and more socioeconomically diverse suburban areas, ensuring that the sample captured key variations in demographic, economic, and healthcare access characteristics across Shanghai's municipal population. Within each stratum, the five districts were selected using simple random sampling, in which each district had an equal probability of being chosen from the list of all eligible districts within that stratum.

 

Comments 9: What do authors mean by "communities"? Which "4 communities"? Based on what? Again, what type of a "random" sample is this?

Response 9: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified what is meant by "communities," specified how the four communities were selected within each district, and explicitly stated the type of random sampling used, in order to improve transparency of the multistage sampling process. This revision can be found on Lines 215-219 of the revised manuscript: Within each selected district, communities (administratively defined neighborhood units equivalent to residents' committees or residential communities, which serve as standard sampling units in Chinese household and health surveys) were listed, and simple random sampling was used to select four communities per district, ensuring that each eligible community had an equal probability of being chosen.

 

Comments 10: Lines 105-118: Why "35-40 residents per community" were chosen? Again, what type of a "random" sample is this?

Response 10: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified both the rationale for selecting 35-40 residents per community and the type of random sampling used at the individual level, in order to fully explain the final stage of the multistage probability sampling design. Specifically, the target number of respondents per community was determined based on the overall sample size calculation to ensure adequate statistical precision for key outcome estimates while accounting for the design effect inherent in multistage sampling, and individual participants were selected using simple random sampling from the official household registry so that each eligible resident had an equal probability of selection. This revision can be found on Lines 219-229 of the revised manuscript: Finally, 35-40 residents per community were randomly recruited to complete a questionnaire. Within each selected community, 'residents' refers to adult individuals listed in the community's official household registry (sampling frame) who met the inclusion criteria, and recruitment was conducted by drawing from this registry using a simple random sampling approach to ensure that every eligible resident in the community had an equal probability of being selected from the sampling frame. The target number of 35-40 residents per community was derived from the overall sample size calculation to ensure sufficient precision for estimating key outcome proportions (e.g., awareness and acceptance levels) while accounting for design effect in a multistage sampling framework [34,35].

 

Comments11: Lines 105-118: What do authors mean by "residents"? Where and how were they "recruited"? 

Response 11: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified the definition of "residents" and explicitly described where and how they were recruited, in order to improve transparency of the individual-level sampling process. Specifically, residents were defined as adult individuals listed in the official household registry of each selected community, which served as the sampling frame, and recruitment was conducted by drawing eligible individuals from this registry using a simple random sampling approach to ensure equal probability of selection. This revision can be found on Lines 220-225 of the revised manuscript: Within each selected community, 'residents' refers to adult individuals listed in the community's official household registry (sampling frame) who met the inclusion criteria, and recruitment was conducted by drawing from this registry using a simple random sampling approach to ensure that every eligible resident in the community had an equal probability of being selected from the sampling frame.

 

Comments 12: . Lines 105-118: Please relate the formula reported here to the revised description of the sampling strategy as mentioned above. Thanks for providing a reference. However, as above, it is preferable to use a survey sampling strategy reference.

Response 12: Thank you for pointing this out. We agree with this comment. Therefore, we have explicitly linked the sample size formula to the revised stratified multistage probability sampling strategy by clarifying that the formula provided a baseline estimate under a simple random sampling assumption, which was subsequently aligned with the multistage survey design to ensure adequate precision, and we cited a standard survey sampling reference to support this approach. This revision can be found on Lines 237-241 of the revised manuscript: This formula is the standard approach for estimating sample size for a population proportion under a simple random sampling framework, and in our stratified multistage probability sampling design, it provided a baseline estimate which was then aligned with the sampling strategy to achieve the desired precision

 

Comments 13: Lines 105-188: Where were these survey instruments "distributed"?

Response 13: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified where the survey instruments were administered by explicitly stating the physical locations at which questionnaires were distributed and completed, namely community health service centers and respondents' homes, to improve transparency of the data collection process. This revision can be found on Lines 321-326 of the revised manuscript: Survey administrators received standardized training before data collection. The survey was conducted primarily through face-to-face structured interviews. Trained interviewers administered the questionnaire in person at community health service centers or at respondents' homes, reading each item verbatim to the respondent and recording responses directly. Face-to-face administration was chosen to ensure clarity of item interpretation, minimize missing data.

 

Comments 14: Lines 105-188: In survey research, this section should include the number of attempts made to contact responders… the approach to the calculation of response rates needs to be provided… the evidence of the representativeness of the study sample of the population it is targeting should be presented… It is concerning to this reviewer that none of this information is provided. As before, it is not even clear what type of sampling strategy was used.

Response 14: Thank you for pointing this out. We agree with this comment. Therefore, we have substantially clarified the sampling strategy, participant recruitment process, response rate calculation, and the representativeness of the study sample in the Methods section, and we have also explicitly acknowledged the remaining limitations related to representativeness in the Limitations section. Specifically, we clarified that a stratified multistage probability sampling design was used, described the selection of districts, communities, and residents, and explained that 760 questionnaires were distributed with 753 valid questionnaires returned, noting that this figure reflects the proportion of valid responses among collected surveys rather than an overall response rate for the entire sampling frame; because recruitment was conducted through on-site, face-to-face interviews at the community level, eligible residents were approached directly during the survey period rather than through repeated contact attempts, and we further contextualized representativeness by explaining how the inclusion of both central urban and suburban districts was intended to capture key demographic and socioeconomic heterogeneity of Shanghai's population. In addition, we revised the Limitations section to explicitly state the degree and constraints of representativeness. These revisions can be found in the Methods section (Lines 202-225) and in the Limitations section (Lines 674-675) of the revised manuscript.

 

Comments 15: Lines 119-150: What was the mode of survey administration? Which procedures were followed in this mode while administering the survey?

Response 15: Thank you for pointing this out. We agree with this comment. Therefore, we have explicitly clarified the mode of survey administration and the standardized procedures followed during data collection. Specifically, we stated that the survey was administered primarily through face-to-face structured interviews conducted by trained interviewers, described the training of survey administrators, the interview setting, the step-by-step procedures for administering the questionnaire, informed consent, confidentiality assurances, on-site quality control, and post-collection data verification processes to ensure data accuracy and completeness. These revisions can be found on Lines 321-346 of the revised manuscript: Survey administrators received standardized training before data collection. The survey was conducted primarily through face-to-face structured interviews. Trained interviewers administered the questionnaire in person at community health service centers or at respondents' homes, reading each item verbatim to the respondent and recording responses directly. Face-to-face administration was chosen to ensure clarity of item interpretation, minimize missing data. The purpose and procedures of the study were explained to all participants, and informed consent was obtained. Before each interview, the survey team described the objectives of the study, assured participants of voluntary participation, and obtained written informed consent. Participants were informed that the survey would take approximately 10 minutes and that their responses would remain confidential. Interviewers adhered to a standardized protocol for introducing the study, reading each question consistently, and responding to participant queries without leading responses. Participants were assured that their responses would remain confidential and be used only for research purposes. All questionnaires were anonymized. During field administration, supervisors conducted spot checks to monitor adherence to the protocol, ensure completeness of responses, and address any operational issues. Any questionnaire with missing or ambiguous responses was reviewed with the participant before the end of the interview to minimize item non-response and improve data integrity. After data collection, responses were entered into a database by a designated clerk and independently verified for accuracy. The database was then checked by another researcher for completeness and logical consistency to ensure data quality. Double data entry procedures were implemented where feasible to further reduce data entry errors. Discrepancies identified during entry and verification were resolved through cross-checking with original paper forms or electronic records.

 

Comments 16: Lines 119-150: Please provide a copy of the survey instrument in an Appendix.

Response 16: Thank you for pointing this out. We agree with this comment. Therefore, we have provided the complete survey instrument as an appendix to the manuscript to enhance transparency and allow readers to fully assess the questionnaire content. Specifically, the full questionnaire used in this study has been added as Appendix A, including all items and response options administered to participants. This revision can be found on Lines 770-1095 of the revised manuscript: Appendix A. Appendix A.1 Shanghai Survey Questionnaire on the Current Status of Non-NIP Vaccine Procurement and Use (General Public).

 

Comments 17: Lines 119-150: Please explain if the complete survey instrument was pre-tested other than getting input during "expert consultation workshops"? What were the findings of the pre-test? What, if any, modifications were carried out, why…

Response 17: Thank you for pointing this out. We agree with this comment. Therefore, we have explicitly clarified that the complete survey instrument was pre-tested through a pilot survey in addition to expert consultation workshops, and we have described the objectives, procedures, main findings, and subsequent modifications resulting from the pre-test. Specifically, we reported that the questionnaire was piloted among a small sample of community residents representative of the target population to assess item clarity, relevance, response burden, and respondent comprehension, that the pilot was conducted using the same structured interview mode as the main survey, and that minor wording revisions were made to improve interpretability and reduce ambiguity without altering the conceptual content of the items. These revisions can be found on Lines 267-279 of the revised manuscript: In addition to expert consultation workshops, the complete questionnaire was pre-tested in a pilot survey with a small sample of community residents representative of the target population. The questionnaire consisted of 49 questions covering sociodemographic characteristics, awareness and acceptance constructs, and potential influencing factors. Completion of the questionnaire took approximately 15 minutes on average. The purpose of this pre-test was to evaluate item clarity, relevance, response burden, and respondent comprehension. Trained field staff conducted the pilot using the same structured interview mode planned for the main survey. Findings from the pre-test indicated that most items were comprehensible; however, several questions (e.g., those related to recent exposure to vaccine information and perceived affordability categories) were reworded to improve interpretability and reduce ambiguity. Minor modifications focused on simplifying phrasing and clarifying specific response options without changing the conceptual content of the items.

 

Comments 18: Lines 119-150: Authors seem to suggest that at least a fraction of non-NIP vaccines are "alternatives" (lines 53-55). Part of the non-NIP vaccines may be for different populations in society. Which non-NIP vaccines were respondents thinking about when answering the survey questions?

Response 18: Thank you for pointing this out. We agree with this comment. Therefore, we clarified that the questionnaire guidance already explicitly specified which Non-National Immunization Program (non-NIP) vaccines respondents should consider when answering the survey questions, including non-NIP vaccines intended for different population groups. This can be found in the questionnaire guidance: Non-NIP vaccines mainly include, but are not limited to, pneumococcal vaccines, meningococcal vaccines, Haemophilus influenzae type b (Hib) vaccine, hand-foot-and-mouth disease (EV71) vaccine, influenza vaccine, HPV vaccine, rotavirus vaccine, varicella (chickenpox) vaccine, pentavalent (five-in-one) combination vaccine, herpes zoster (shingles) vaccine, hepatitis B vaccine, and other common self-paid vaccines. 

 

Comments 19: Lines 119-150: Please be precise and explain how many questions were there in the survey instrument in total and how long did it take to complete it?

Response 19: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified the total number of questions included in the survey instrument and the average time required for completion. This clarification can be found on Lines 269-272 of the revised manuscript: The questionnaire consisted of 49 questions covering sociodemographic characteristics, awareness and acceptance constructs, and potential influencing factors. Completion of the questionnaire took approximately 15 minutes on average.

 

Comments 20: Lines 152-157: The statistical analysis reported here needs to incorporate detail based on the revised sampling strategy description above.

Response 20: Thank you for pointing this out. We agree with this comment. Therefore, we have revised and expanded the Statistical Analysis section to explicitly incorporate details consistent with the stratified multistage probability sampling strategy described above. Specifically, we clarified procedures for data management and missing data handling, noted that key sociodemographic stratification variables were inspected and adjusted for in preliminary and multivariable analyzes to account for potential design effects, and provided a more detailed description of the analytical workflow, including the use of non-parametric tests for ordinal outcomes, variable selection for multivariable ordinal logistic regression, assessment of proportional odds assumptions, model fit evaluation, and multicollinearity diagnostics. These revisions can be found on Lines 348-384 of the revised manuscript: Data were organized in Excel and analyzed using SPSS 29.0. During data processing, we examined all variables for missing values; overall, the proportion of missing data was minimal across key variables. Where item non-response occurred, respondents were re-contacted using the contact information provided in the questionnaire to clarify or complete missing responses when feasible; when missing data remained after follow-up, those records were treated as incomplete and excluded from specific analyzes (complete-case analysis). Given the stratified multistage probability sampling design, key sociodemographic stratification variables (e.g., district type and community strata) were inspected in preliminary analyzes, and models were adjusted for these variables to account for potential design effects on association estimates. To visually summarize the multivariable ordinal logistic regression results, adjusted odds ratios and corresponding 95% confidence intervals were plotted as coefficient plots using Stata 19.0 after multivariable analysis. Descriptive statistics were first used to summarize participants' sociodemographic characteristics as well as the distribution of awareness and acceptance levels of non-NIP vaccines, including frequencies and percentages for categorical variables and ordinal outcomes. Because the dependent variables were measured on ordinal scales and did not meet the assumptions of normal distribution, non-parametric methods were applied, with Kruskal-Wallis H tests used for subgroup comparisons. Variables showing statistical significance in univariate analyzes were subsequently entered into multivariable ordinal logistic regression models. The proportional odds assumption was evaluated using the test of parallel lines, overall model fit was assessed using likelihood ratio tests, multicollinearity was examined using variance inflation factors, and all statistical tests were two-sided with a significance threshold of P < 0.05.

 

Comments 21: Lines 152-157: Were there any missing information? If yes, what remedies were used?

Response 21: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified the extent of missing information and the remedies applied during data processing. Specifically, we reported that the overall proportion of missing data was minimal, described procedures for re-contacting respondents to complete missing items when feasible, and explained that remaining incomplete records were excluded from specific analyzes using a complete-case approach. This clarification can be found on Lines 348-354 of the revised manuscript: During data processing, we examined all variables for missing values; overall, the proportion of missing data was minimal across key variables. Where item non-response occurred, respondents were re-contacted using the contact information provided in the questionnaire to clarify or complete missing responses when feasible; when missing data remained after follow-up, those records were treated as incomplete and excluded from specific analyzes (complete-case analysis).

 

Comments 22: Lines 160-166: Thanks for the information. However, in survey research, the first table should include population and sample characteristics side by side so that there is information about how representative the study sample is of the population. For example, is Shanghai's 'resident' population '32% male' and '68% female'?

Response 22: Thank you for pointing this out. We agree with this comment. Therefore, we have added contextual population-level socioeconomic indicators for Shanghai to facilitate comparison between the study sample and the target population and to better inform readers about the representativeness of the sample. Specifically, we supplemented the Results section with official statistics on average income and per capita disposable income of Shanghai residents, including corresponding figures for urban and rural populations, to provide an external benchmark against which the sample characteristics can be interpreted. This revision can be found on Lines 397-407 of the revised manuscript: For context on income levels in the overall Shanghai population, the average income of residents in Shanghai in 2024 was approximately RMB 88,366 according to the 2024 Shanghai Statistical Bulletin on National Economic and Social Development. Urban resident average income was approximately RMB 93,095, while rural resident average income was approximately RMB 45,644. These figures provide a reference for comparing the income distribution of the study sample with the broader economic conditions in Shanghai's resident population. For context on socioeconomic status in the overall Shanghai population, the per capita disposable income of Shanghai residents in 2024 was approximately RMB 88,366. Urban resident per capita disposable income was approximately RMB 93,095, while rural resident per capita disposable income was approximately RMB 45,644 [37-40].

 

Comments 23: Lines 160-166: If the sample is not representative, which techniques were used as a remedy? In this study, the authors indicate that "the study was representative…" (lines 379-383) but they do not provide any evidence that the sample is, in fact, representative.

Response 23: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified that no post-survey statistical weighting or calibration techniques were applied to formally correct for potential non-representativeness, and we have revised the manuscript to avoid overstatement by explicitly framing representativeness as partial and design-based rather than definitive. Specifically, representativeness in this study was addressed at the sampling design stage through a stratified multistage probability sampling strategy covering both central urban and suburban districts to capture key sociodemographic heterogeneity, but this approach does not guarantee full population representativeness. We therefore revised the Limitations section to explicitly acknowledge that, in the absence of weighting or population benchmarking, residual selection bias may remain and the findings should be interpreted with caution when generalizing to the target population. These revisions can be found in the Limitations section on Lines 674-675 of the revised manuscript: Although this study has a certain degree of representativeness and an adequate sample size, no post-stratification or weighting techniques were applied, and selection bias cannot be completely excluded when generalizing the findings to the entire Shanghai population.

 

Comments24: Lines 160-166: Please make sure to support assertions made with a reference (like the annual income percentages provided for 2023 here). Is this information for Shanghai? Please be specific.

Response24: Thank you for pointing this out. We agree with this comment. Therefore, we have clarified that the income distribution percentages reported in the Results section refer to the study sample, and we have explicitly supplemented population-level income data specific to Shanghai with authoritative references to support contextual assertions. Specifically, we added officially published statistics on average income and per capita disposable income of Shanghai residents, with separate figures for urban and rural populations, to clearly distinguish sample characteristics from population benchmarks and to ensure transparency regarding data sources. This revision can be found on Lines 397-407 of the revised manuscript: For context on income levels in the overall Shanghai population, the average income of residents in Shanghai in 2024 was approximately RMB 88,366 according to the 2024 Shanghai Statistical Bulletin on National Economic and Social Development. Urban resident average income was approximately RMB 93,095, while rural resident average income was approximately RMB 45,644. These figures provide a reference for comparing the income distribution of the study sample with the broader economic conditions in Shanghai's resident population. For context on socioeconomic status in the overall Shanghai population, the per capita disposable income of Shanghai residents in 2024 was approximately RMB 88,366. Urban resident per capita disposable income was approximately RMB 93,095, while rural resident per capita disposable income was approximately RMB 45,644 [37-40].

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Dear Authors,

 

Thank you for the improvement of the manuscript.

Author Response

Comments 1: Dear Authors, Thank you for the improvement of the manuscript.

Response 1: Thank you, We sincerely appreciate your time and effort in reviewing our manuscript and recognizing the improvements made. Your positive feedback is greatly appreciated and motivates us to further refine our work.

Reviewer 2 Report

Comments and Suggestions for Authors

Thanks for responding to my comments. Please improve the quality of the coeplot. Furthermore, include variables with statistically significant p-values. 

Author Response

Comments 1: Thanks for responding to my comments. Please improve the quality of the coeplot. Furthermore, include variables with statistically significant p-values.

Response 1: Thank you for pointing this out. We agree with the comment. Thank you for your suggestion. We have improved the quality of the coeplot by completely redrawing the figure using R version 4.4.1 with the RStudio platform to ensure higher resolution and clearer visual presentation. In addition, the plot has been updated to prominently include and annotate variables with statistically significant p-values, enhancing interpretability of the results. These revisions are reflected in the revised manuscript(lines 454 and 480) .

Reviewer 5 Report

Comments and Suggestions for Authors

Thanks to authors.

Author Response

Comments 1: Thanks to authors.

Response 1: Thank you, We sincerely appreciate your time and effort in reviewing our manuscript and recognizing the improvements made. Your positive feedback is greatly appreciated and motivates us to further refine our work.