Review Reports - Progressing Towards Strengthened Vaccination Programmes: Investigating COVID-19 Vaccine Usage in Public and Private Sectors in KwaZulu-Natal, South Africa

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This mixed-methods study investigates COVID-19 vaccine usage rates (VUR) for Pfizer-BioNTech and Janssen (J&J) vaccines across public (PUBS) and private (PRIVS) healthcare sectors in KwaZulu-Natal, South Africa, from May 2021 to July 2022. The quantitative phase retrospectively analysed VUR using data from the Stock Visibility System (SVS) and Electronic Vaccination Data System (EVDS), while the qualitative phase involved interviews with vaccination managers from both sectors in early 2024. The study aims to evaluate vaccine distribution, administration, and monitoring systems and to propose improvements for future vaccination programmes. Below are my comments:

The manuscript frequently uses undefined or inconsistently applied acronyms (e.g., PUBSR, PRIVSR, SP, VUR). This hinders readability and comprehension. Can the authors provide a clear and consistent glossary of terms and ensure all acronyms are defined upon first use?

The study claims to be “multicentre” but does not specify how many facilities were included or how they were selected. The sampling strategy is unclear. How were the facilities and districts selected? Was there any attempt to ensure representativeness across urban and rural settings?

VUR values exceeding 100% suggest serious data reporting issues (e.g., under-reporting of issued doses). The authors acknowledge this but do not adequately address its implications. How did the authors validate the data from SVS and EVDS? What steps were taken to address inconsistencies or missing data?

The manuscript mentions statistical significance (p < 0.05) but does not describe the statistical tests used or the rationale for their selection. Which statistical tests were applied? How were the data normalised or adjusted for confounding variables?

The qualitative findings are presented as percentages (e.g., “100% of PUBSR reported…”), which is unusual for thematic analysis and may misrepresent the depth of responses. Why were qualitative responses quantified in this manner? Was thematic saturation considered? How were themes derived and validated?

The study is confined to one province in South Africa. The findings may not be applicable to other regions or countries. How do the authors contextualise KZN within the broader South African or global vaccination effort?

The study highlights systemic issues like poor integration between SVS and EVDS, yet the recommendations are broad and lack specificity. What concrete, actionable steps do the authors propose for integrating these systems? How might these be funded and implemented in a resource-limited setting?
What public health or health systems frameworks informed this study? How do the findings contribute to theory or policy beyond descriptive reporting?
Tables 1 and 2 are poorly formatted and difficult to interpret. Headings are misaligned, and data presentation is confusing. Can the authors reformat these tables for clarity? Could visual aids (e.g., graphs, maps) better illustrate temporal or geographic trends?

Comments on the Quality of English Language

The English could be improved to more clearly express the research.

Author Response

Dear Reviewer

Many thanks for your insightful comments and assistance with improving the manuscript.

Comment 1

The manuscript frequently uses undefined or inconsistently applied acronyms (e.g., PUBSR, PRIVSR, SP, VUR). This hinders readability and comprehension. Can the authors provide a clear and consistent glossary of terms and ensure all acronyms are defined upon first use?

Response

Acronyms were added

Updated on Page 2 (yellow highlights)

Comment 2

The study claims to be “multicentre” but does not specify how many facilities were included or how they were selected. The sampling strategy is unclear. How were the facilities and districts selected? Was there any attempt to ensure representativeness across urban and rural settings?

Response

A total population sampling strategy was applied, as all facilities and Districts identified from the dataset were included in the analysis, eliminating the need for sample selection. There was no need to ensure representativeness across urban and rural settings; as all 11 Districts in KZN, including the urban and rural Districts were included in this study. Pfizer PUBS and PRIVS recorded a maximum of 511 and 40 vaccination facilities respectively. J&J recorded a maximum of 603 PUBS and 43 PRIVS vaccination facilities.

Updated on Page 3 under sampling strategy (yellow highlights)

Comment 3

VUR values exceeding 100% suggest serious data reporting issues (e.g., under-reporting of issued doses). The authors acknowledge this but do not adequately address its implications. How did the authors validate the data from SVS and EVDS? What steps were taken to address inconsistencies or missing data?

Response

This was a retrospective study, with data from the COVID-19 vaccination programme (C19VP). This data was only reviewed about 2 years after the programme; therefore, missing data could not be accessed as the programme had already ended and SVS COVID-19 application was not functional. However, during Phase 1, SVS and EVDS were programmatic data systems that were reliant on facility-level reporting. Therefore, data validation was performed through checks at facility, District and Provincial level

Updated under data analysis and validation Page 4 (yellow highlights)

Comment 4

The manuscript mentions statistical significance (p < 0.05) but does not describe the statistical tests used or the rationale for their selection. Which statistical tests were applied? How were the data normalised or adjusted for confounding variables?

Response

Phase 1 data that was analyzed in this study represented usage rates of vaccines, rather than raw counts. These rates were calculated relative to the total doses administered within each sector (public vs private), thereby providing a form of inherent standardization that allowed direct comparison across the two groups. Prior to performing the independent samples t-tests, the underlying distribution of the usage rates was exam-ined using the Kolmogorov Smirnov test and Shapiro–Wilk test and visual inspection of Q–Q plots, which indicated that the data were sufficiently close to normal to justify the use of parametric methods. Furthermore, there was testing a hypothesis about the mean and the mean according to the central limit theorem is also normally distributed.

We acknowledge that our analysis did not adjust for potential demographic or geographic confounders (such as age distribution, provincial differences, or access to vaccination sites), as these data were not available within the dataset. This is recognised as a limitation of the study. However, our primary aim was to assess overall sectoral differences in usage rates rather than to establish causal relationships with underlying population characteristics. We have not used any form of regression modelling as well.

Updated under Statistical analyses page 5 (yellow highlights)

Comment 5

The qualitative findings are presented as percentages (e.g., “100% of PUBSR reported…”), which is unusual for thematic analysis and may misrepresent the depth of responses. Why were qualitative responses quantified in this manner?

Response

While qualitative data are typically presented thematically, percentages were included in this study to illustrate the relative frequency of particular responses. The quantification was not intended to replace the depth of thematic analysis, but rather to provide an indication of how widely each theme was reported across PUBSR and PRIVSR. This mixed presentation aims to balance nuance with transparency, enabling clearer interpretation of patterns. Nonetheless, it is acknowledged that percentages in qualitative work should be viewed as indicative rather than definitive measures of prevalence.

Not updated on manuscript

Comment 6

Was thematic saturation considered?

Response

The study included 12 respondents from the public sector (PUBSR) and 7 respondents from the private sector (PRIVSR) across KZN, a province with a diverse mix of rural and urban districts. Given the variation in COVID-19 vaccine supply chain experiences between districts as well as between the public and private sectors, thematic saturation was not considered an appropriate criterion for determining sample adequacy. While some common themes were identified, unique themes continued to emerge as interviews progressed, reflecting the heterogeneity of participant experiences. As such, the aim was not to reach saturation, but rather to capture the breadth and diversity of perspectives across different contexts.

Updated under data collection-page 4 (yellow highlights)

Comment 7

How were themes derived and validated?

Response

An inductive thematic content analysis approach was applied to the qualitative data. This process involved systematically identifying, analyzing, and reporting patterns within the interview transcripts without imposing pre-existing coding frames or theoretical perspectives. The analysis began with open coding, where meaningful segments of text were highlighted and assigned descriptive codes. These codes were then reviewed, compared, and grouped into broader categories to capture underlying ideas. Through an iterative process, categories were refined and consolidated into themes that reflected recurring and non-recurring patterns. NVivo software was used to facilitate this process by managing the transcripts, supporting systematic coding, and organizing codes and categories. While NVivo assisted in the efficient handling of large volumes of qualitative data, the identification and interpretation of themes remained a researcher-driven process. Themes were compared across public and private sector respondents checking for consistency and diversity. Furthermore, themes were cataloged in an Excel spreadsheet and aggregated for PUBSR and PRIVSR, represented as a proportion of the total PUBSR and PRIVSR.

Updated under Statistical analyses -page 5 (yellow highlights)

Comment 8

The study is confined to one province in South Africa. The findings may not be applicable to other regions or countries. How do the authors contextualise KZN within the broader South African or global vaccination effort?

Response

KwaZulu-Natal (KZN) is the second largest province in South Africa, with a population of approximately 12 million. All provinces in South Africa utilized the SVS and EVDS systems for reporting COVID-19 vaccinations; however, vaccine usage rates (VURs) were neither calculated nor monitored. Currently, there are no published studies documenting COVID-19 VURs in South Africa or globally. This restricts direct comparisons with other provinces or countries. Nonetheless, the experiences of KZN offer significant insights into the overall performance of vaccine supply chains and underscore the necessity of monitoring vaccine usage rates (VURs) in current and future vaccination programs. VUR reflect the proportion of issued doses that are actually administered and can pinpoint areas with reduced vaccination acceptance and wastage.

Updated under Introduction-Page 2 (yellow highlights)

Comment 9

The study highlights systemic issues like poor integration between SVS and EVDS, yet the recommendations are broad and lack specificity. What concrete, actionable steps do the authors propose for integrating these systems? How might these be funded and implemented in a resource-limited setting?

Response

This study identified that the lack of integration between the vaccination (EVDS) and stock management system (SVS) led to the absence of real time generation of automated VUR. SVS and EVDS integration could involve developing interoperable platforms for automatic sharing of stock and administration data, standardizing reporting protocols, conducting regular reconciliation audits, training facility staff on integrated data entry, and establishing real-time monitoring dashboards. Vaccine procurement costs comprise a significant share of immunization program costs in low- and middle-income countries, yet not all procured vaccines are administered (Mvundura M et al., 2023). Additionally, lower vaccination uptake increases morbidity and mortality, leading to further healthcare costs.

Updated under conclusion-Page 18 (yellow highlights)

Comment 10

What public health or health systems frameworks informed this study? How do the findings contribute to theory or policy beyond descriptive reporting?

Response

This study, informed by WHO's Monitoring Wastage at the country level and South Africa's monitoring systems like SVS and EVDS during the COVID-19 vaccination rollout, examined VURs between the PUBS and PRIVS sectors in KZN. The findings extend beyond descriptive reporting to understanding successes, challenges and opportunities for improvement. This research enhances theoretical understanding by demonstrating the impact of supply chain integration, monitoring, and workforce practices on vaccine availability and acceptance. From a policy perspective, the study provides evidence to inform interventions such as integrating stock management and patient administration/dispensing systems optimizing training, and improving monitoring systems, which may enhance the efficiency and effectiveness of current and future vaccination programmes both nationally and in similar low- and middle-income contexts.

Updated under conclusion page 19 (yellow highlights)

Comment 11

Tables 1 and 2 are poorly formatted and difficult to interpret. Headings are misaligned, and data presentation is confusing. Can the authors reformat these tables for clarity? Could visual aids (e.g., graphs, maps) better illustrate temporal or geographic trends?

Responses

Table 1 has been edited; we have included Figure 1. We removed Table 2 and included Figure 2 and 3.

Updated from Page 7 to 11

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This study investigated COVID-19 vaccine usage rates (VUR) for Pfizer and Johnson & Johnson (J&J) in public (PUBS) and private (PRIVS) healthcare sectors of KwaZulu-Natal, South Africa, from May 2021 to July 2022. A mixed-methods approach was used, combining retrospective quantitative analysis of stock (SVS) and vaccination (EVDS) data with qualitative interviews of vaccination managers.

The overall provincial VUR was 81.8% for Pfizer and 63.3% for J&J. Sector-specific analysis revealed a higher Pfizer VUR in the private sector (104.4%) compared to the public sector (78.7%), attributed to underreporting of issued doses. J&J VUR was higher in public (64.7%) than private (40.2%) facilities. Discrepancies and VURs exceeding 100% highlighted significant challenges in stock reporting and data integration between the SVS and EVDS systems. Qualitative findings emphasized operational hurdles like network connectivity, manual data capture, and varied reconciliation processes.

The study recommends implementing an integrated, real-time national vaccination recording system to improve data accuracy, enable evidence-based decision-making, reduce vaccine wastage, and strengthen future pandemic preparedness through enhanced public-private collaboration.

I have a few concerns regarding this manuscript:

1. There is a significant discrepancy in the reported Vaccine Usage Rates (VUR) between the Abstract and the Results section. In the Results section (Page 4, Table 1), the authors reported that Pfizer: PUBS VUR = 78.7%, PRIVS VUR = 104.4%; J&J: PUBS VUR = 64.7%, PRIVS VUR = 40.2%. However, the Abstract also states that "The analysis indicated a vaccine uptake rate (VUR) of 81.8% for Pfizer and 63.3% for the J&J C19V". The reader might be left confused about which values represent the true VUR.

2. The term "vaccine uptake rate" is often used interchangeably with "vaccine usage rate," which may not be a standard term and can contribute to confusion.

3. Several VUR values exceed 100% (e.g., Pfizer PRIVS = 104.4%, J&J PRIVS = 208% in August 2021). The authors correctly note this is due to non-reporting of issued doses, but this should be explicitly highlighted in the Abstract and Results to avoid misinterpretation.

4. The qualitative sample is relatively small (n=19 total respondents). While informative, caution should be exercised in generalizing these perspectives.

Author Response

Dear Reviewer

Many thanks for your insightful comments and assistance with improving the manuscript.

Comment 1

There is a significant discrepancy in the reported Vaccine Usage Rates (VUR) between the Abstract and the Results section. In the Results section (Page 4, Table 1), the authors reported that Pfizer: PUBS VUR = 78.7%, PRIVS VUR = 104.4%; J&J: PUBS VUR = 64.7%, PRIVS VUR = 40.2%. However, the Abstract also states that "The analysis indicated a vaccine uptake rate (VUR) of 81.8% for Pfizer and 63.3% for the J&J C19V". The reader might be left confused about which values represent the true VUR.

Response

Vaccine usage rate as per WHO= Number of doses administered/Number of Doses Issued (WHO, 2005)

The analysis indicated that KZN Province recorded a total vaccine usage rate (VUR) of 81.8% for Pfizer and 63.3% for the J&J C19V. A deep analysis into this showed that Pfizer PUBS and PRIVS recorded a VUR of 78.7% and 104.4% respectively and J&J recorded a VUR of 64.7% for PUBS and 40.2% for PRIVS.

Updated under Results -Page 6 (yellow highlights)

Comment 2

The term "vaccine uptake rate" is often used interchangeably with "vaccine usage rate," which may not be a standard term and can contribute to confusion.

Response

Vaccine usage rate as per WHO= Number of doses administered/Number of doses issued (WHO, 2005)

Vaccine uptake rate is the number of people vaccinated with a certain dose of the vaccine in a certain time period, which can be expressed as an absolute number or as the proportion of a target population. (Rikitu Terefa et al., 2021)

**RIKITU TEREFA, D., SHAMA, A. T., FEYISA, B. R., EWUNETU DESISA, A., GETA, E. T., CHEGO CHEME, M. & TAMIRU EDOSA, A. 2021. COVID-19 Vaccine Uptake and Associated Factors Among Health Professionals in Ethiopia. Infect Drug Resist, 14, 5531–5541.

**WHO 2005. Monitoring Vaccine Wastage at Country Level May 2005 ed (Page 2)

Updated under definitions – Page 2 (yellow highlights)

Comment 3

Several VUR values exceed 100% (e.g., Pfizer PRIVS = 104.4%, J&J PRIVS = 208% in August 2021). The authors correctly note this is due to non-reporting of issued doses, but this should be explicitly highlighted in the Abstract and Results to avoid misinterpretation.

Response

The analysis revealed that Pfizer PUBS and PRIVS both exhibited higher VUR in comparison to J&J. Notably, Pfizer PRIVS VUR exceeded 100% due to non-reporting and non-recording of issued doses. This underscores the necessity for integrated electronic, real-time monitoring of VURs in current and future vaccination programs

PRIVS reported VUR>er 100% (n=1-29) indicating that doses administered exceeded issue doses and reflecting non reporting of issued doses on SVS and suboptimal monitoring processes. (yellow highlights)

Updated under abstract, results and discussion.

Comment 4

The qualitative sample is relatively small (n=19 total respondents). While informative, caution should be exercised in generalizing these perspectives.

Response

We acknowledge that the qualitative sample is small, but the aim was to gain rich, in-depth perspectives. KZN is a resource constrained Province in South Africa with a mix of rural and urban health Districts and facilities. While these results may not be applicable to all populations, meaningful, diverse insights can be derived from vaccination managers in KZN; and can inform practice and guide future studies and immunisation programmes; particularly in low resourced environments.

Updated under limitations (yellow highlights)

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

I have gone through the work done by Viloshini et al., which addresses an important aspect of the COVID-19 vaccination program that has implications for future public health efforts. However, there are major issues which requires significant revision.

In abstract remove the duplicated words “stock on hand”.
Please check what VUR means. I feel there is a confusion, whether it is a “usage rate” or “uptake”.
Some clarification is required, in methods you have mentioned phase 1 was conducted at all public and private sector healthcare facilities that were reporting on SVS and EVDS, which means that facilities not using those systems were not included. However, later in the discussion you mentioned that outreach sites without stock permits did administer vaccines but didn’t record stock, so their data couldn’t contribute to VUR. This criterion should be clearly mentioned as a limitation in the methods.
Research districts comprised A, B, C, D, E, G, H, I, J, and K…….. it appears to miss one letter. Please check.
Why data for Pfizer were collected for 15 months and J&J for 14 months? Is this because J&J rollout later? Please mtnion this clearly.
How stock transfers between facilities were handled? Were transferred doses double-counted or excluded?
I feel that the phase 2 description could be more detailed.
Sampling strategy or inclusion criteria for interviewees needs clarification. Further, a brief description of the interview content or example topics is required. Please add.
There is a lack of detail on any statistical tests in Phase 1. Methods section does not describe any statistical test procedure. Statistical details are required to judge the robustness of the analysis. Currently, nothing is there.
How did you compare groups? What approach you have taken?
In my opinion results section is very confusing and hard to read and follow. Lot of numbers. Basically, the whole results section is very number-heavy. Authors must consider moving some of these details to tables or figures to improve readability. For example, trends over time could be shown in a line graph for each sector/vaccine, which would more clearly illustrate the result. A figure could make it easier to get the information you want to present about the trajectory and the points of divergence between public and private. Similarly, with the district-level differences can be shown through a bar chart or map
In Table 1 what you mean by (n %). This column is bit unclear. I believe n is the list of number of facilities that reported data and the percentage that number represents out of all vaccination sites? If I am right, its better to clear this, either by splitting it into two columns or adding a footer note n = ………?????? and % =………?????. This would help readers understand.
In table 2 and methods why you have used letters for districts? The paper is about aggregate program data and not about individual patients. You don’t have to hide the district names. I suggest replacing letters with the actual district names.
Both the tables should have footnotes for better explanation. There are lot of 0’s without any interpretation. You have to correlate the 0 value and explain in the footnote. Include all necessary legends or footnotes so that the tables are self-explanatory. Currently, without footnotes the tables are difficult to read and understand.

Author Response

Reviewer 3

Dear Reviewer

Many thanks for your insightful comments and assistance with improving the manuscript.

Comment 1

In abstract remove the duplicated words “stock on hand”.

Response

Removed stock on hand

Comment 2

Please check what VUR means. I feel there is a confusion, whether it is a “usage rate” or “uptake”.

Response

Vaccine usage rate as per WHO= Number of doses administered/Number of Doses Issued (WHO, 2005)

RIKITU TEREFA, D., SHAMA, A. T., FEYISA, B. R., EWUNETU DESISA, A., GETA, E. T., CHEGO CHEME, M. & TAMIRU EDOSA, A. 2021. COVID-19 Vaccine Uptake and Associated Factors Among Health Professionals in Ethiopia. Infect Drug Resist, 14, 5531–5541.

**WHO 2005. Monitoring Vaccine Wastage at Country Level May 2005 ed (Page 2) (yellow highlights)

Updated under definitions- Page 1 (yellow highlights)

Comment 3

Some clarification is required, in methods you have mentioned phase 1 was conducted at all public and private sector healthcare facilities that were reporting on SVS and EVDS, which means that facilities not using those systems were not included. However, later in the discussion you mentioned that outreach sites without stock permits did administer vaccines but didn’t record stock, so their data couldn’t contribute to VUR. This criterion should be clearly mentioned as a limitation in the methods.

Response

EVDS was used by all vaccination sites; as this was the National system for recording vaccine administrations to clients. However, stock issued was only captured on SVS by sites that stored vaccines as per legislative prescripts. While outreach/community sites vaccinated clients and used EVDS for recording vaccinations; these sites did not store stock and therefore did not record stock issued on SVS. Stock issued was recorded by the distributing facility. COVID-19 vaccines particularly the Pfizer vaccines required specialized storage conditions and these community based/outreach sites did not have the required fridges/freezers for storage of these vaccines. However, since each outreach site and hospital distributing site was classified under a single District; the total stock administered and issued for each District was added. Thereafter; we calculated District VUR by dividing the total stock administered by total stock issued. Therefore, the outreach facilities stock administered data were actually included in the District VUR.

This study is recommending that all sites administering vaccines captured stock issued data for the real time calculation of VURs for prompt interventions.

Updated under discussion-page 15 (yellow highlights)

Comment 4

Research districts comprised A, B, C, D, E, G, H, I, J, and K…….. it appears to miss one letter. Please check.

Response

Updated F on manuscript under sampling strategy (yellow highlights)

Comment 5

Why data for Pfizer were collected for 15 months and J&J for 14 months? Is this because J&J rollout later? Please mention this clearly.

Response

Pfizer rollout started in May 2021; while J&J commenced in June 2021 in KwaZulu-Natal.

Updated under data collection -Page 4 (yellow highlights)

Comment 6

How stock transfers between facilities were handled? Were transferred doses double-counted or excluded?

Response

Separate fields were available on SVS for stock issued and transfers. Therefore. when stock was issued; it was captured under the stock issued field. Likewise, when stock was transferred, it was captured under the stock transferred field. A stock transfer was not included in this study. However, the findings from the qualitative study indicated that reconciliations were complicated with stock transfers and we noted these results.

No updates

Comment 7

I feel that the phase 2 description could be more detailed.

Response

Sampling Strategy

Phase 2: Vaccination managers who were coordinating C19V supply management were selected from the PUBS and PRIVS. Twelve and eight interview invitations were sent to PUBS and PRIVS vaccination managers, respectively. One vaccination manager from each PUBS district, the PUBS Provincial Pharmaceutical Services, and each PRIVS organisation was invited to participate. Only managers who provided consent by completing the consent form were included; those who did not consent were excluded. Participants were anonymized and interviews were confidential. Interviews were comprehensive, covering aspects of the vaccine supply chain, including planning and coordination, legislative requirements, training, human resources, wastage minimisation, cold chain management, adverse events, transportation, rural area challenges, successes, weaknesses, barriers, risks, risk mitigation, and public-private partnerships. For purposes of this paper, only stock management and monitoring were considered.

Data Collection

Phase 2: Vaccination managers received the interview questions upon providing consent. Interviews were conducted, recorded, and transcribed. A semi–structured interview guide ensured consistency, and clarifications and real-time member validation enhanced accuracy. All participants were fully informed about the anonymity of interviews and the reasons for the research, and how their data would be used. Due to variations in the COVID-19 vaccine supply chain experience between urban, rural districts, and public and private sectors, thematic saturation was not used to determine sample adequacy. The aim was to capture a breadth and diversity of perspectives across different contexts.

Data Analysis and Validation

Phase 2: Several strategies ensured the validity of qualitative data. Credibility was enhanced as respondents confirmed emerging interpretations during interviews. Dependability was supported through an audit trail maintained in NVivo and Excel. Transferability was promoted by systematically comparing themes between respondents from the public and private sectors.

Statistical Analysis

Phase 2: Qualitative Study: An inductive thematic content analysis approach was applied. This process involved systematically identifying, analysing, and reporting patterns within the interview transcripts without imposing pre–existing coding frames or theoretical perspectives. The analysis began with open coding, where meaningful segments of text were highlighted and assigned descriptive codes. These codes were then reviewed, compared, and grouped into broader categories to capture underlying ideas. Through an iterative process, categories were refined and consolidated into themes that reflected recurring and non-recurring patterns. NVivo software facilitated transcript management, coding, and organisation, but the identification and interpretation remained researcher-driven. Themes were compared across PUBSR and PRIVSR to ensure consistency and diversity, catalogued in Excel, and aggregated as proportions of total PUBSR and PRIVSR.

Updated under methods (yellow updates)

Comment 8

Sampling strategy or inclusion criteria for interviewees needs clarification. Further, a brief description of the interview content or example topics is required. Please add.

Response

Sampling Strategy

Updated under methods (yellow updates)

Comment 9

There is a lack of detail on any statistical tests in Phase 1. Methods section does not describe any statistical test procedure. Statistical details are required to judge the robustness of the analysis. Currently, nothing is there.

Response

Statistical Analysis

Phase 1 data that was analysed in our study represented usage rates of vaccines, rather than raw counts. These rates were calculated relative to the total doses administered within each sector (public vs private), thereby providing a form of inherent standardisation that allowed direct comparison across the two groups. Prior to performing the independent samples t-tests, we examined the underlying distribution of the usage rates using the Kolmogorov Smirnov test and Shapiro–Wilk test and visual inspection of Q–Q plots, which indicated that the data were sufficiently close to normal to justify the use of parametric methods. Furthermore, we are testing a hypothesis about the mean and the mean according to the Central limit theorem is also Normally distributed.

We acknowledge that our analysis did not adjust for potential demographic or geographic confounders (such as age distribution, provincial differences, or access to vaccination sites), as these data were not available within the dataset. However, our primary aim was to assess overall sectoral differences in usage rates rather than to establish causal relationships with underlying population characteristics. We have not used any form of regression modelling as well.

Updated under methods (yellow updates)

Comment 10

How did you compare groups? What approach you have taken?

Response

We compared the public and private sector groups to determine how the findings differed or were similar. We used descriptive data for Phase 1; with statistical testing. A thematic comparison for phase 2 to see how strongly each theme showed up in each sector. This was followed by a comparative narrative in the discussion.

Updates under data collection -page 4 (yellow updates)

Comment 11

In my opinion results section is very confusing and hard to read and follow. Lot of numbers. Basically, the whole results section is very number-heavy. Authors must consider moving some of these details to tables or figures to improve readability. For example, trends over time could be shown in a line graph for each sector/vaccine, which would more clearly illustrate the result. A figure could make it easier to get the information you want to present about the trajectory and the points of divergence between public and private. Similarly, with the district-level differences can be shown through a bar chart or map.

Responses

The first paragraph in the results; Provincial VUR analysis for Pfizer and J&J are not in the tables and figures and cannot be removed.

Table 1 is still present. We included Figure 1 shows the VUR for Pfizer PUBS and PRIVS J&J PUBS and PRIVS for KZN Province; this is a subset of data from Table 1.

Doses issued and administered that are recorded in table 1 were removed from manuscript. Table 2 will be removed and replaced by Figure 2 and 3.

Updated under results-page 7 to 11

Comment 12

In Table 1 what you mean by (n %). This column is bit unclear. I believe n is the list of number of facilities that reported data and the percentage that number represents out of all vaccination sites? If I am right, its better to clear this, either by splitting it into two columns or adding a footer note n = ………?????? and % =………?????. This would help readers understand.

Responses

We did not want to split the columns as the table will expand considerably.

We included a footnote defining (n%)

Updated under Table 1

Comment 13

In table 2 and methods why you have used letters for districts? The paper is about aggregate program data and not about individual patients. You don’t have to hide the district names. I suggest replacing letters with the actual district names.

Responses

As mandated by the UKZN and Department of Health ethics committees; the data still needs to be anonymized.

No updates

Comment 14

Both the tables should have footnotes for better explanation. There are lot of 0’s without any interpretation. You have to correlate the 0 value and explain in the footnote. Include all necessary legends or footnotes so that the tables are self-explanatory. Currently, without footnotes the tables are difficult to read and understand.

Responses

Footnotes were added under Table 1.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

Manuscript is significantly improved by the authors and now can be accepted in its current form.