Public Knowledge and Attitudes Towards Clinical Trial Participation: A Mixed-Method Study in Bagamoyo District, Tanzania

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear authors,
I've attached my suggestions for the article. I wish you good work.
Sincerely. 

Comments for author File: Comments.pdf

Author Response

RESPONSES TO REVIEWER COMMENTS

We appreciate all the comments of each reviewer for improving this manuscript. We have tried to address all the issues and comments as suggested by the reviewers.

Reviewer: 1

Comments 1:  The literature review for this study was supported by only a few studies; existing studies related to this study could be added.

Response 1: Thank you for the valuable comments. The literature review has been improved by adding more information related to the topic

Comments 2: A theoretical or conceptual framework (as appropriate) for the study could be added. This title could provide a detailed explanation of public health knowledge and attitudes.

Response 2: Thank you for the valuable comments, the conceptual framework has been added (see line 149-171)

Comments 3: “Therefore, novel strategies that would increase clinical trial participant participation and retention are desperately needed. To increase community participation in research, it is necessary to develop such interventions with thorough understanding of the factors influencing recruitment and sustained participation within our community, as well as to improve the relationship between researchers and study participants [6], [7].” What are the strategies mentioned here? Are there any studies related to these strategies? Additional explanations could be added.

Response 3: Thank you for constructive comments. I have revised this sentence into a readable form.( see line 109-119

Comments 4: “Tanzania to address the awareness, attitudes, and perception related to interventions delivered in clinical trials, [17], [18].” The studies mentioned here could be explained in more detail regarding participants, scope, and results. Furthermore, international studies, not just Tanzania, could be included.

Response 4: The paragraph has been improved by adding more information and relevant references. (See line 90-97)

Comments 5: Because the scope of the study is broad, more sample participant responses could be included (Qualitative Findings).

Responses 5: We tried to analyse and include most of the interview responses related to the objective of this study. We have tried to add some quotes responses in the qualitative part. (See line 343-358)

Comments 6: The study's conclusions in the discussion section could be supported by further literature review. In this case, the discussion section could be structured so that the quantitative and qualitative results are discussed in two separate paragraphs or two headings.

Responses 6: Thank you for the nice comments. We think because the study used convergent parallel mixed method, separating quantitative and qualitative conclusions will affect the meaning of comparison between findings

Comments 7: The reference list can be formatted according to the journal's preferred format.

Response 7: The reference has been formatted according to the journal style

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript addresses an important public health topic – the community’s knowledge and attitudes toward clinical trial participation in Tanzania. The mixed-methods approach (survey with 394 participants and follow-up qualitative interviews) is appropriate for exploring both breadth and depth of perceptions. However, several issues need to be addressed.

 

• There is a notable inconsistency in how the sampling method is described versus how it was executed. In the Methods, the authors state that a “multistage stratified random sampling” was used to select participants (with stratification by village’s prior trial involvement and systematic random household selection). This suggests a rigorous probability sampling approach. However, in the Discussion’s limitations, the authors refer to “the use of convenience sampling” leading to an overrepresentation of females (74.4% of respondents). This contradiction is concerning. If the sampling was truly random at the household level, such a strong gender imbalance might indicate a response bias (e.g. perhaps women were more often present at home to be surveyed).

 

• The study is described as a convergent parallel mixed-methods design, but the execution appears more sequential, and certain details of the qualitative component are lacking. The Methods section indicates that after the quantitative survey, ten participants were purposively selected for in-depth interviews (IDIs) and focus group discussion (FGD). This actually sounds like a sequential explanatory design (quantitative followed by qualitative to explain results) rather than truly parallel/convergent. The manuscript should clarify the design to avoid confusion – if qualitative data were collected after quantitative, say it was a sequential mixed-methods approach (which is perfectly acceptable).

 

• More information is needed about the qualitative sample and procedures: How were those 10 participants selected (criteria)? How many FGDs versus one-on-one interviews were conducted? The phrasing suggests possibly a single FGD plus some IDIs, but it’s unclear. Notably, quotes are attributed to participants (P# FGD) and also to “CHW IDI”, implying that a community health worker was interviewed. The Methods did not mention interviewing health workers; if key informants like a CHW were included, the authors should describe this in the inclusion criteria for qualitative interviews. Clarifying the qualitative sampling (e.g., “we conducted X FGDs with community members and Y IDIs, including Z community health workers”) is important for reproducibility. Moreover, claiming data saturation with only ten participants is debatable – the authors should justify how saturation was assessed or consider a more cautious phrasing. Providing details on how interviews were conducted (language, recording, transcription, translation, coding process) would enhance the credibility of qualitative findings.

 

• The outcome coding in the logistic regressions (Tables 3 and 5) is a bit confusing in text. For knowledge, the text states that “men were more likely to have low knowledge compared to women (AOR = 22.95, p = 0.001)”. This suggests that being male dramatically increases the odds of poor knowledge. However, in Table 3, the adjusted OR for gender shows Female = 22.95 with male as reference, implying females had much higher odds of knowledge (or conversely, males had higher odds of low knowledge). The authors should double-check and clarify the coding: Did an OR of 22.95 indicate the odds of high knowledge for females vs males, or odds of low knowledge? Such a large OR is striking – if accurate, it means gender is by far the dominant factor. If this is a result of how “low knowledge” was defined (perhaps most of the sample had low scores, making the few high-knowledge individuals predominantly female), it warrants explanation. The huge confidence interval (10.27–51.28) suggests instability, possibly due to the low number of knowledgeable respondents among men. The authors might consider providing the underlying proportions (e.g., what % of men vs women scored ≥60%) to contextualize this OR. Similarly, for attitudes, the adjusted OR for female gender is 7.61, indicating women had significantly higher odds of a positive attitude. These results are important, but the presentation is a bit redundant and confusing in places. For instance, lines 844–862 repeat the logistic findings and then say “See Table 4” when referring to the adjusted analysis, but actually the multivariable results are in Table 5. The text should be corrected to refer to the proper table, and ideally the narrative of results should be streamlined (currently the adjusted attitude results are described twice in slightly different wording).

 

• In Table 5, the crude OR for age 45–54 is given as 1.83 with 95% CI 1.11–3.00 but a p-value of 0.18. This appears inconsistent (such a CI would typically yield p < 0.05). It may be a typo (perhaps p = 0.018) or an error in calculation. The authors should verify all p-values and CIs for consistency. Likewise, ensure that the percentage values cited in text match the tables (most do, but it’s worth double-checking after any data updates).

 

• The method of scoring knowledge and attitude could be described in more detail for clarity. The cut-off of 60% for “poor” knowledge is mentioned, as well as using the mean to dichotomize attitude. It would help to state how many participants fell into each category (e.g., how many had “positive attitude” by that definition), to appreciate the base rates for the logistic regression.

 

• The study’s key finding is that knowledge about clinical trials was generally low, yet attitudes toward participation were largely positive. While this is a compelling result, there are some internal inconsistencies in the survey responses on attitudes that the authors should address. Notably, Table 4 shows that 90.9% of respondents agreed that “clinical trials are beneficial to society,” and 89.6% said they would be willing to participate given information. At the same time, a surprisingly high proportion – 53.8% – agreed that “clinical trials conducted harm society”, and 31.5% felt that humans in research are treated like “guinea pigs”. These statements indicate underlying fears or misconceptions despite the overall positive outlook. The manuscript currently highlights the positive attitude (willingness to participate, recognizing benefits) and notes some “moderate concerns” about trust and privacy, but it does not explicitly reconcile how over half of respondents can view trials as potentially harmful to society while still mostly supporting them. This is a logical consistency issue that should be discussed. It may be that respondents acknowledge both the potential benefits and risks of trials – which is reasonable – or it could reflect a subset of the population holding strongly negative views. The authors should clarify in the Discussion what the implication of 54% perceiving harm is. For example, is it possible the phrasing of that survey item was confusing? (The wording “Clinical trials conducted harms society” is slightly awkward and might have been misinterpreted.) Or do people perhaps agree that trials pose risks (thus “harm society” in a sense of danger) even though they recognize the overall importance? Providing a nuanced interpretation here would strengthen the paper. It might also be useful to mention in the Discussion that a sizable minority hold mistrustful views (e.g., nearly one-third think researchers treat participants like lab animals), which aligns with the qualitative findings of fear and misconceptions (e.g., blood “selling” myths).

 

• The authors appropriately state that the findings are specific to Bagamoyo and not necessarily generalizable to all of Tanzania. However, some statements in the manuscript still use broad language like “general population” or “the general community” without qualification. It would be safer to consistently refer to “the study population” or “the community in Bagamoyo district” when drawing conclusions. For instance, the title and aim say “Pwani Region” but in reality the sample was from Bagamoyo district (which is one part of Pwani). Unless the study truly covered multiple districts in Pwani (it appears focused on Bagamoyo), the title might better reflect that specificity to avoid overstating the coverage.

 

• The authors assumed a 21% awareness rate based on a prior study (ref. 19) and calculated a sample of 384 with a 95% CI and 5% margin of error. In the equation provided, they actually used p = 0.5 in the formula (which yields the maximum sample size of 384) despite citing 21%. This is a bit confusing. It would be clearer to either use p=0.21 in the formula (and show the design effect adjusted result ~384) or explain that they used a conservative approach (p=0.5) to ensure sufficient sample. Also, the term “P – standard of deviation” in the formula legend is incorrect; it should be “P – expected proportion” or “prevalence”. Please correct that terminology.

 

• It’s stated that a KAP (Knowledge, Attitude and Practice) survey was used. However, the results only report on knowledge and attitude items, and there is no explicit “Practice” section (apart from a question on willingness to participate, which is more of an intention than a measured practice). To avoid confusion, consider clarifying that the survey focused on knowledge and attitudes (with a question on participation willingness).

 

• When presenting demographic percentages in text, use consistent precision. For example, “293 women (74.4%) and 101 men (25.6%)” is fine. Later, age group proportions are given (e.g., 35.5% were aged 18–44) – ensure the age brackets mentioned in text match the ones in Table 1 (it looks like the text combined 18–44 as one category for a summary, which is okay).

 

• Table 1: Check that the totals and percentages align (they do sum to 394, but e.g., education category in text said 67.5% primary vs Table 1 shows 55.1% had primary only, with an additional 12.4% having no education – perhaps the text combined those as “primary or less” to get 67.5%. Just be clear on what is being described). Also, “Residency” categories (Bago, Kiwangwa, etc.) presumably correspond to the four villages – consider labeling that row as “Village” for clarity.

 

• Table 3 & 5 (Logistic regression tables): These are clearly structured. A minor formatting point: in Table 3’s heading, it says “Knowledge about CTs (low/high)”, and in Table 5 “Attitude toward CTs (Negative/positive)”. It might help to add a footnote or note in caption clarifying which outcome is coded as the event (e.g., “low knowledge = score <60%” etc.). Also, make sure the text consistently refers to these tables by the correct numbers (as noted, currently the text references to “Table 4” for regression which should be Table 5). Renumbering or rechecking cross-references is needed.

Author Response

RESPONSES TO REVIEWER COMMENTS

We appreciate all the comments of each reviewer for improving this manuscript. We have tried to address all the issues and comments as suggested by the reviewers.

Comments 1: There is a notable inconsistency in how the sampling method is described versus how it was executed. In the Methods, the authors state that a “multistage stratified random sampling” was used to select participants (with stratification by village’s prior trial involvement and systematic random household selection). This suggests a rigorous probability sampling approach. However, in the Discussion’s limitations, the authors refer to “the use of convenience sampling” leading to an overrepresentation of females (74.4% of respondents). This contradiction is concerning. If the sampling was truly random at the household level, such a strong gender imbalance might indicate a response bias (e.g. perhaps women were more often present at home to be surveyed)

Response 1: The contradictions related to sampling method as explained in the method section and discussion section has been rectified (see line 423-425)

 

Comments 2: The study is described as a convergent parallel mixed-methods design, but the execution appears more sequential, and certain details of the qualitative component are lacking. The Methods section indicates that after the quantitative survey, ten participants were purposively selected for in-depth interviews (IDIs) and focus group discussion (FGD). This actually sounds like a sequential explanatory design (quantitative followed by qualitative to explain results) rather than truly parallel/convergent. The manuscript should clarify the design to avoid confusion – if qualitative data were collected after quantitative, say it was a sequential mixed-methods approach (which is perfectly acceptable)

Response 2: Thank you for the valuable observation and comments. The clear explanation has been added to avoid confusion. (See line 187-196)

Comments 3: More information is needed about the qualitative sample and procedures: How were those 10 participants selected (criteria)? How many FGDs versus one-on-one interviews were conducted? The phrasing suggests possibly a single FGD plus some IDIs, but it’s unclear. Notably, quotes are attributed to participants (P# FGD) and also to “CHW IDI”, implying that a community health worker was interviewed. The Methods did not mention interviewing health workers; if key informants like a CHW were included, the authors should describe this in the inclusion criteria for qualitative interviews. Clarifying the qualitative sampling (e.g., “we conducted X FGDs with community members and Y IDIs, including Z community health workers”) is important for reproducibility. Moreover, claiming data saturation with only ten participants is debatable – the authors should justify how saturation was assessed or consider a more cautious phrasing. Providing details on how interviews were conducted (language, recording, transcription, translation, coding process) would enhance the credibility of qualitative findings

 

Responses 3: Thank you for the nice comments. More information has been provided in the qualitative part for reproducibility (see line 247-251)

 

Comments 4: The outcome coding in the logistic regressions (Tables 3 and 5) is a bit confusing in text. For knowledge, the text states that “men were more likely to have low knowledge compared to women (AOR = 22.95, p = 0.001)”. This suggests that being male dramatically increases the odds of poor knowledge. However, in Table 3, the adjusted OR for gender shows Female = 22.95 with male as reference, implying females had much higher odds of knowledge (or conversely, males had higher odds of low knowledge). The authors should double-check and clarify the coding: Did an OR of 22.95 indicate the odds of high knowledge for females vs males, or odds of low knowledge? Such a large OR is striking – if accurate, it means gender is by far the dominant factor. If this is a result of how “low knowledge” was defined (perhaps most of the sample had low scores, making the few high-knowledge individuals predominantly female), it warrants explanation. The huge confidence interval (10.27–51.28) suggests instability, possibly due to the low number of knowledgeable respondents among men. The authors might consider providing the underlying proportions (e.g., what % of men vs women scored ≥60%) to contextualize this OR. Similarly, for attitudes, the adjusted OR for female gender is 7.61, indicating women had significantly higher odds of a positive attitude. These results are important, but the presentation is a bit redundant and confusing in places. For instance, lines 844–862 repeat the logistic findings and then say “See Table 4” when referring to the adjusted analysis, but the multivariable results are in Table 5. The text should be corrected to refer to the proper table, and ideally the narrative of results should be streamlined (currently the adjusted attitude results are described twice in slightly different wording).

Responses 4: Thank you for constructive comments; we have reversed the explanation regarding women and males’ to clearly describe what is presented in table 3(see line 271-282). The text that represented the wrong table has been corrected (see line296, 304)

 

Comments 5: In Table 5, the crude OR for age 45–54 is given as 1.83 with 95% CI 1.11–3.00 but a p-value of 0.18. This appears inconsistent (such a CI would typically yield p < 0.05). It may be a typo (perhaps p = 0.018) or an error in calculation. The authors should verify all p-values and CIs for consistency. Likewise, ensure that the percentage values cited in text match the tables (most do, but it’s worth double-checking after any data updates)

Response 5: Re-analysis was done new table replaced the old with error (line 305)

Comments 6: The method of scoring knowledge and attitude could be described in more detail for clarity. The cut-off of 60% for “poor” knowledge is mentioned, as well as using the mean to dichotomize attitude. It would help to state how many participants fell into each category (e.g., how many had “positive attitude” by that definition), to appreciate the base rates for the logistic regression.

Response 6: The method and calculation of the knowledge and attitude score has been revised into a correct method that reflect the analysis. Mean score cutoff was used as indicated in the tables (see line 198-207

Comments 7: The study’s key finding is that knowledge about clinical trials was generally low, yet attitudes toward participation were largely positive. While this is a compelling result, there are some internal inconsistencies in the survey responses on attitudes that the authors should address. Notably, Table 4 shows that 90.9% of respondents agreed that “clinical trials are beneficial to society,” and 89.6% said they would be willing to participate given information. At the same time, a surprisingly high proportion – 53.8% – agreed that “clinical trials conducted harm society”, and 31.5% felt that humans in research are treated like “guinea pigs”. These statements indicate underlying fears or misconceptions despite the overall positive outlook. The manuscript currently highlights the positive attitude (willingness to participate, recognizing benefits) and notes some “moderate concerns” about trust and privacy, but it does not explicitly reconcile how over half of respondents can view trials as potentially harmful to society while still mostly supporting them. This is a logical consistency issue that should be discussed. It may be that respondents acknowledge both the potential benefits and risks of trials – which is reasonable – or it couldreflect a subset of the population holding strongly negative views. The authors should clarify in the Discussion what the implication of 54% perceiving harm is. For example, is it possible the phrasing of that survey item was confusing? (The wording “Clinical trials conducted harms society” is slightly awkward and might have been misinterpreted.) Or do people perhaps agree that trials pose risks (thus “harm society” in a sense of danger) even though they recognize the overall importance? Providing a nuanced interpretation here would strengthen the paper. It might also be useful to mention in the Discussion that a sizable minority hold mistrustful views (e.g., nearly one-third think researchers treat participants like lab animals), which aligns with the qualitative findings of fear and misconceptions (e.g., blood “selling” myths)

Response 7: Thank you very much for these constructive comments. I have made an Improvement on the attitude paragraph has been made (line 287-294) more information has been added in the discussion section (line 395-404)

Comments 8: The authors appropriately state that the findings are specific to Bagamoyo and not necessarily generalizable to all of Tanzania. However, some statements in the manuscript still use broad language like “general population” or “the general community” without qualification. It would be safer to consistently refer to “the study population” or “the community in Bagamoyo district” when drawing conclusions. For instance, the title and aim say “Pwani Region” but in reality the sample was from Bagamoyo district (which is one part of Pwani). Unless the study truly covered multiple districts in Pwani (it appears focused on Bagamoyo), the title might better reflect that specificity to avoid overstating the coverage

Responses 8: The statement that uses broad language in the manuscript have been revised into specific Similar to the tile of this manuscript.

 

Comments 9: The authors assumed a 21% awareness rate based on a prior study (ref. 19) and calculated a sample of 384 with a 95% CI and 5% margin of error. In the equation provided, they used p = 0.5 in the formula (which yields the maximum sample size of 384) despite citing 21%. This is a bit confusing. It would be clearer to either use p=0.21 in the formula (and show the design effect adjusted result ~384) or explain that they used a conservative approach (p=0.5) to ensure sufficient sample. Also, the term “P – standard of deviation” in the formula legend is incorrect; it should be “P – expected proportion” or “prevalence”. Please correct that terminology.

Response 9: The terminology in the sample calculation and the description of P in the formular have been corrected (see line 128-139)

 

Comments 10: It’s stated that a KAP (Knowledge, Attitude and Practice) survey was used. However, the results only report on knowledge and attitude items, and there is no explicit “Practice” section (apart from a question on willingness to participate, which is more of an intention than a measured practice). To avoid confusion, consider clarifying that the survey focused on knowledge and attitudes (with a question on participation willingness)

Response 10: The statement has been added to make clarity (see line 130-132)

Comments 11: When presenting demographic percentages in text, use consistent precision. For example, “293 women (74.4%) and 101 men (25.6%)” is fine. Later, age group proportions are given (e.g., 35.5% were aged 18–44) – ensure the age brackets mentioned in text match the ones in Table 1 (it looks like the text combined 18–44 as one category for a summary, which is okay).

Response 11: This has been revised for consistence

Comments 12: Table 1: Check that the totals and percentages align (they do sum to 394, but e.g., education category in text said 67.5% primary vs Table 1 shows 55.1% had primary only, with an additional 12.4% having no education – perhaps the text combined those as “primary or less” to get 67.5%. Just be clear on what is being described). Also, “Residency” categories (Bago, Kiwangwa, etc.) presumably correspond to the four villages – consider labeling that row as “Village” for clarity.

Response 12: This has been revised for Consistence

Comments 13: Table 3 & 5 (Logistic regression tables): These are clearly structured. A minor formatting point: in Table 3’s heading, it says “Knowledge about CTs (low/high)”, and in Table 5 “Attitude toward CTs (Negative/positive)”. It might help to add a footnote or note in caption clarifying which outcome is coded as the event (e.g., “low knowledge = score

Response 13: The Caption has been added clarifying the mean score used to categorize the knowledge levels. In the new tables we have included the univariate analysis for more understanding

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Overall

This study, conducted using mixed methods to comprehensively investigate public knowledge and attitudes toward clinical trials in the Bagamoyo region of Tanzania, is of crucial public health importance. In particular, this study's attempt to identify barriers and facilitators in a specific region, amid global challenges in clinical trial recruitment and retention, is highly commendable.

However, as you point out, several fundamental improvements are needed to establish this paper as a scientific contribution. In particular, the novelty of the research is weakly emphasized, the rationale for the methodology chosen and the data processing process are insufficient, and the presentation of results and discussion of conclusions require deeper consideration and a more self-critical perspective. We believe that improving these points will more clearly communicate the inherent value of this study.

(1) Positioning the Paper and Clarifying its Novelty

While this paper is based on research in the field of health sciences, the novelty of this study is not sufficiently emphasized in the abstract and introduction to the intended journal's readership (in this case, health and environment, or the broader public health field).
The introduction concludes by stating, "Several studies have been conducted in Tanzania to address the awareness, attitudes, and perceptions related to interventions delivered in clinical trials, [17], [18]. However, there remains a gap in research specifically understanding the community knowledge and attitudes regarding participation in clinical trials." While this statement suggests the existence of a research gap, it does not actively highlight the uniqueness of this study, such as "why conducting this mixed methods research in Bagamoyo at this time is novel." The abstract also merely summarizes the results. This should be improved.

(2) Clarifying the Appropriateness of the Research Method and the Reasons for Its Selection

The methodology section states, "A convergent parallel mixed-methods study was conducted..." and "Bagamoyo was chosen because of its clinical trial facility, which we believed our data could be valid and reliable...," describing the methodology and location. However, the rationale for why mixed-methods was essential to answering this research question is weak. The analysis section states, "Binary logistic regression to examine the degree of correlation between the primary outcome variables... and the independent variables...," but provides no explanation for why logistic regression was chosen rather than other multivariate analysis methods. This makes the method appear to have been chosen simply out of convention, giving the impression of a lack of methodological rigor.

(3) Insufficient Explanation of the Dataset and Preprocessing

In the analysis section, the authors state, "The categorization of knowledge and attitude scores was based on assigning 1 point to each correct or favorable response...while "No"/"Disagree" or "Not Sure" responses were scored as 0." "Attitude scores were also dichotomized as positive or negative, using the mean score as the cutoff point..." "The overall knowledge and attitude were categorized using Bloom's cutoff point..."

These statements do not provide any rationale for their decisions. In particular, the decision to treat "Not Sure" as the same score as "No" seems arbitrary and could significantly affect the results. Furthermore, the method for determining the mean cutoff and the specific source of Bloom's cutoff point are unclear.

(4) Recognition of Issues in the Presentation of Results and Conclusions

The conclusion section lacks a discussion that takes into account the limitations of this study.
Tables 2 and 4 list the frequency and percentage of responses for each question. While a format like `228(60.3)` is concise, when there are many items listed, it is difficult to intuitively grasp which items have particularly low knowledge or divided attitudes.
The conclusions (Section 5. Conclusions) summarize the results and provide recommendations (e.g., the need for educational intervention), but do not mention the study's limitations (Strengths and Limitations) discussed in Section 4.1, particularly factors that significantly threaten the generalizability of the conclusions, such as "convenience sampling" and "overrepresentation of females (74.4%)."
This risks making the conclusions sound overly optimistic and assertive. Scientific conclusions should be stated carefully and within the limits of the strength of the evidence.

Author Response

RESPONSES TO REVIEWER COMMENTS

We appreciate all the comments of each reviewer for improving this manuscript. We have tried to address all the issues and comments as suggested by the reviewers.

Comments 1: Positioning the Paper and Clarifying its Novelty While this paper is based on research in the field of health sciences, the novelty of this study is not sufficiently emphasized in the abstract and introduction to the intended journal's readership (in this case, health and environment, or the broader public health field). The introduction concludes by stating, "Several studies have been conducted in Tanzania to address the awareness, attitudes, and perceptions related to interventions delivered in clinical trials, [17], [18]. However, there remains a gap in research specifically understanding the community knowledge and attitudes regarding participation in clinical trials." While this statement suggests the existence of a research gap, it does not actively highlight the uniqueness of this study, such as "why conducting this mixed methods research in Bagamoyo at this time is novel." The abstract also merely summarizes the results. This should be improved

Response 1:  Introduction of this paper has been revised and improved

 

Comments 2: Clarifying the Appropriateness of the Research Method and the Reasons for Its Selection The methodology section states, "A convergent parallel mixed-methods study was conducted..." and "Bagamoyo was chosen because of its clinical trial facility, which we believed our data could be valid and reliable...," describing the methodology and location. However, the rationale for why mixed methods was essential to answering this research question is weak. The analysis section states, "Binary logistic regression to examine the degree of correlation between the primary outcome variables... and the independent variables...," but provides no explanation for why logistic regression was chosen rather than other multivariate analysis methods. This makes the method appear to have been chosen simply out of convention, giving the impression of a lack of methodological rigor.

Responses 2: Thank you for this comment. We did not include all the information to avoiding having too long paper. However, we value your comment.

Comments 3: Insufficient Explanation of the Dataset and Preprocessing In the analysis section, the authors state, "The categorization of knowledge and attitude scores was based on assigning 1 point to each correct or favorable response...while "No"/"Disagree" or "Not Sure" responses were scored as 0." "Attitude scores were also dichotomized as positive or negative, using the mean score as the cutoff point..." "The overall knowledge and attitude were categorized using Bloom's cutoff point..." These statements do not provide any rationale for their decisions. In particular, the decision to treat "Not Sure" as the same score as "No" seems arbitrary and could significantly affect the results. Furthermore, the method for determining the mean cutoff and the specific source of Bloom's cutoff point is unclear

Response 3: Thank you for your comments. More explanation has been provided regarding the analysis of the data and how the data was categorised. Categorising attitude score disagrees/not sure with 0 score was inevitable for easily comparison of data within the groups. (line 208-216)

 

Comments 4: Recognition of Issues in the Presentation of Results and Conclusions The conclusion section lacks a discussion that takes into account the limitations of this study. Tables 2 and 4 list the frequency and percentage of responses for each question. While a format like `228(60.3)` is concise, when there are many items listed, it is difficult to intuitively grasp which items have particularly low knowledge or divided attitudes. The conclusions (Section 5. Conclusions) summarize the results and provide recommendations (e.g., the need for educational intervention), but do not mention the study's limitations (Strengths and Limitations) discussed in Section 4.1, particularly factors that significantly threaten the generalizability of the conclusions, such as "convenience sampling" and "overrepresentation of females (74.4%)." These risks making the conclusions sound overly optimistic and assertive. Scientific conclusions should be stated carefully and within the limits of the strength of the evidence.

Response 4: In the Conclusion paragraph we have added the statement that acknowledging the limitation and the factor that affect the generalization of the findings (501-503)

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Dear authors,

Thank you for the corrections. I can say that the final version of the study has become more comprehensive and rich. I wish you success in your future studies.

Author Response

Thanks for your valuable comments and positive feedbacks which actually has played a role part in improving our manuscript.

Reviewer 2 Report

Comments and Suggestions for Authors

Thanks to the authors for their responses for raised issues. The manuscript is now substantially improved. 

Author Response

Thank you so much for your productive feedbacks and comments. They helped us to improve this scientific work professionally.

Reviewer 3 Report

Comments and Suggestions for Authors

(1) Clarifying the Paper's Position and Novelty

Current Assessment and Points of Reference
The reviewer pointed out that the section "Why is this research unique?" was weak. In response, the final paragraph of the Introduction has been revised.

Points of Reference: "No studies have been conducted in Tanzania to address the knowledge level and attitude of the general community toward participation in clinical trials. Therefore, this study aimed to assess the level of knowledge and attitudes of Bagamoyo district toward participation in clinical trials with the goal of informing future engagement and educational strategies."

This statement merely states a "research gap"—"there is no research on this topic in Tanzania." The reviewer is looking for an explanation of why filling that gap has academic and social value, and why using mixed methods research in Bagamoyo, at this time, is unique. We suggest adding the following elements to the current statement to more actively emphasize "why this research is important."

By adding this information, wouldn't it be possible to send a strong message that "there is unique value in conducting this research in this location using this method," rather than "we are doing this because there is no other research available?"

(2) Regarding the validity of the research method and clarification of the reasons for its selection

Current evaluation and points of criticism
The reviewer raised concerns that the reasons for choosing mixed methods and logistic regression analysis were not explained. The current response, "We did not include all the information to avoid having a too long paper," could be interpreted by the reviewer as "we omitted it due to space constraints," which could give the impression of dishonesty. A concise explanation of the reasons for the selection is required.

Point of criticism 1
> "A convergent parallel mixed-methods study was conducted..."
This does not explain "why" mixed methods were necessary.
Point of criticism 2
> "...binary logistic regression to examine the degree of correlation between the primary outcome variables of interest (knowledge and attitude) and the independent variables..."
Again, there is no explanation here as to "why" logistic regression analysis was chosen instead of other multivariate analyses.

Suggested Improvements
May I suggest that you provide a concise, academic explanation for each choice?

(3) Regarding the insufficient explanation of the dataset and preprocessing

Current Assessment and Points of Reference
The reviewer pointed out the arbitrariness of the data classification, particularly the arbitrariness of assigning the same score of 0 to "Not Sure" as "No/Disagree." Your response, "Categorising... was inevitable for easy comparison," is a weak scientific justification, given the reasoning behind it. Could you please elaborate a bit more?

Additional Points from AI Research

Logistic regression analysis can be viewed not only as a model for explaining factors but also as a model for "predicting" knowledge and attitude levels. From the perspective of AI and machine learning, is it possible to demonstrate how well a model can classify individuals by evaluating its predictive performance (e.g., the area under the AUC-ROC curve)? Please consider this.

Author Response

On behalf of all authors,  I will always appreciate your gold contribution in our manuscripts. we have clarified the comments and addressed them all.

Comment 1: Clarifying the Paper's Position and Novelty. The reviewer is looking for an explanation of why filling that gap has academic and social value, and why using mixed methods research in Bagamoyo, currently, is unique. We suggest adding the following elements to the current statement to more actively emphasize "why this research is important.

Response: Thank you for this constructive feedback. We have added a paragraph to explain the rationale of this research topic in Bagamoyo District and maybe in Tanzania at large. (see line 117-126)

 

Comment 2: Regarding the validity of the research method and clarification of the reasons for its selection

Response: Thank you very much for this important observation. We have added a reason for choosing mixed method design and using binary regression in the analysis paragraph (See line 131-137 and 227-235)

Comment 3: Regarding the insufficient explanation of the dataset and preprocessing

Response: I appreciate this valuable comment.  The categorization of knowledge and attitude score, No”/“Disagree” or “Not Sure” responses was scored as 0 because our focus was to differentiate (favorable) responses from negative or non-favorable responses. Because “Not Sure” response implies that the respondent has not expressed a positive remark, they were combined with the "Disagree" responses or negative responses. In other words, despite their psychological differences, both "Disagree" and "Not sure" indicate the lack of a positive attitude. We have added reason for combining the score (See line 206,206, 215,216)

Author Response File: Author Response.pdf