From Pandemic to Prevention: Insights from COVID-19 Vaccination Attitudes in Individuals with Schizophrenia

Round 1

Reviewer 1 Report

The paper describes an important topic. As you note in the introduction, there is a gap in the literature that this study aims to fill. Overall it is a solid work, but I have a few major points for you to consider:

– I recommend being cautious when you compare results from this study with general population taken from other studies. The best fix would be to run the same survey on a control group without psychiatric diagnoses to shrink bias.

– You use a questionnaire developed for general-population and, as far as I can see, it has not been validated in people with schizophrenia. Also there is a lack of information about your 20-person pilot – it involved diagnosed patients or healthy people?

– Lines 221-224 say the regression is non-significant. That implies two points:
a. avoid strong conclusions based on those, I have on my mind lines 250-253
b. if the global test is non-significant, you shall not cherry-pick individual predictors

– In questions with three answer possibilities (yes/hesitant/no) you can use chi2 to quick check, but ordinal logistic regression will be better to show gradient. Avoid standard linear regression, because results are not continuous, and "hesitant" is not in halfway between "yes" and "no".

There are some minor points:

– You state that general-population mortality is below 10% (reference 12), yet the paper you cite actually reports almost 12%. If you used a different source you shall cite it.

– Table 3 and the text use parallelly one and two decimal places, this is inconsistent. If you want to keep two decimals, please note in Table 3 that 79/141 is 56.03%, not 56.02%.

– The differences within groups would be clearer in a table. Also you shall add a column for each statistical test (p-value) and, ideally, Cramer's V to show effect size.

– Explain missing values, for example why 13 people have no recorded sex and 15 no age? You can test the gaps are MCAR or MAR, also consider MICE rather than list-wise deletion.

– In "Limitations" add no validation of the questionnaire in this patients group and information that data were collected before vaccines were available, so you measured intention only, not behaviour.

Typographical:
– missing comma in number in line 215
– "COVD-19" in keywords
– "de vaccine" in line 323

Author Response

• Comment 1: – I recommend being cautious when you compare results from this study with general population taken from other studies. The best fix would be to run the same survey on a control group without psychiatric diagnoses to shrink bias

Response 1: We agree with the comments and have therefore explicitly highlighted this issue in both the 'Limitations' and 'Future Lines of Research' sections [Lines 328-335]

• Comment 2: You use a questionnaire developed for general-population and, as far as I can see, it has not been validated in people with schizophrenia. Also there is a lack of information about your 20-person pilot – it involved diagnosed patients or healthy people?

Response 2: We add the missing information that the participants of the pilot were people with schizophrenia to confirmed the simplicity, readability, and usability of the survey [Lines 127-128]

• Comment 3: Lines 221-224 say the regression is non-significant. That implies two points:
  a. avoid strong conclusions based on those, I have on my mind lines 250-253
  b. if the global test is non-significant, you shall not cherry-pick individual predictors

Response 3: In order to better adjust to this result, the wording was modified [Lines 247-253]. To moderate the Cherry Pick, the Monferroni-Holmes correction was used, and the regressions were adjusted and reduced only to the model that included psychopathology, as this was the most representative of the population studied.

• Comment 4: In questions with three answer possibilities (yes/hesitant/no) you can use chi2 to quick check, but ordinal logistic regression will be better to show gradient. Avoid standard linear regression, because results are not continuous, and "hesitant" is not in halfway between "yes" and "no".

Response 4: We would like to express our special gratitude for this correction. The ordinal regression model was employed following her [Lines 223-234]. And, as recommended by other reviewers, we have introduced an alternative model to assess the association of psychopathology.

• Comment 5: You state that general-population mortality is below 10% (reference 12), yet the paper you cite actually reports almost 12%. If you used a different source you shall cite it.

Response 5: Was corrected [line 36]

• Comment 6: Table 3 and the text use parallelly one and two decimal places, this is inconsistent. If you want to keep two decimals, please note in Table 3 that 79/141 is 56.03%, not 56.02%.

Response 6: Was corrected [lines 151 and 193]

• Comment 7: The differences within groups would be clearer in a table. Also you shall add a column for each statistical test (p-value) and, ideally, Cramer's V to show effect size

Response 7: We are in agreement that presenting group differences in a table can sometimes enhance clarity. However, in this instance, we elected to present the results in the current format to maintain consistency with the structure and flow of the manuscript, as well as to ensure readability for a broader audience. With regard to the statistical details, the key p-values have been included in the text to highlight significant findings. While Cramer's V is a valuable measure of effect size, we prioritised parsimony in reporting, as the main focus was on the significance and direction of the relationships rather than their strength.

• Comment 8: Explain missing values, for example why 13 people have no recorded sex and 15 no age? You can test the gaps are MCAR or MAR, also consider MICE rather than list-wise deletion

Response 8: They are MCARs and we have decided to keep them as missing data, indicating this [line 114]

• Comment 9: In "Limitations" add no validation of the questionnaire in this patients group and information that data were collected before vaccines were available, so you measured intention only, not behaviour.

Response 9: We pointed this out [lines  333-340]

• Comments 10:  Typographical:– missing comma in number in line 215, – "COVD-19" in keywords, – "de vaccine" in line 323

Response 10: All corrected,

 

 

Reviewer 2 Report

[Only minor technicalities]

1) This study measures vague intent to vaccinate and vague fears about side effect of vaccine which had not been at that time, while is not measuring actual vaccination. On its own its not a problem, just reflection of period when it had been conducted. However, in such case when compared with other studies it should be in a bit more explicit on what was measured. As while you mentioned divergence, one could not tell whether it’s genuine discrepancy or you are confronting study on intent at the end of 2020 with study on vaccination status in 2022.

Under perfect conditions you should be comparing yourself with other studies measuring intent, while ultimately confronting that with studies on what was the actual end result.

2) “In addition, persons with schizophrenia do not typically par- 48
ticipate in preventive health care plans, and only a quarter of this group usually gets vac- 49
cinated against influenza and other common viruses [26–30].”
In describing your sample you say:
“Here, n = 86 who had been vaccinated against influenza were more likely to accept 174
vaccination against COVID-19 (67.19%), and n= 42 who had not been vaccinated were 175
more likely to be hesitant (32.8%).”
So you claim it’s usually around round ¼, while in you sample you get 2/3. Either you should correct that initial overgeneralizing statement, or you should put some limitation that due to some sampling problem you measured extraordinarily vaccine accepting group of people with schizophrenia, what makes your findings hard to generalize.

3) “In general, participants reported following health instructions, although other stud- 285
ies have noted the opposite [71]”

71: “Larson, H.J.; Clarke, R.M.; Jarrett, C.; Eckersberger, E.; Levine, Z.; Schulz, W.S.; Paterson, P. Measuring Trust in Vaccination: A 549 Systematic Review. Human Vaccines & Immunotherapeutics 2018, 14, 1599–1609, doi:10.1080/21645515.2018.1459252.”
Uhm… this study does not build you that point, check please whether you haven’t intended to cite something else. And also check the other citations.

Author Response

• Comment 1) This study measures vague intent to vaccinate and vague fears about side effect of vaccine which had not been at that time, while is not measuring actual vaccination. On its own its not a problem, just reflection of period when it had been conducted. However, in such case when compared with other studies it should be in a bit more explicit on what was measured. As while you mentioned divergence, one could not tell whether it’s genuine discrepancy or you are confronting study on intent at the end of 2020 with study on vaccination status in 2022.

  Under perfect conditions you should be comparing yourself with other studies measuring intent, while ultimately confronting that with studies on what was the actual end result.

Response 1: We have explicitly highlighted these aspects. We have done so in the limitations and in future research [lines 332-339]

• Comment 2) “In addition, persons with schizophrenia do not typically participate in preventive health care plans, and only a quarter of this group usually gets vaccinated against influenza and other common viruses [26–30].”
  In describing your sample you say:
  “Here, n = 86 who had been vaccinated against influenza were more likely to accept 174
  vaccination against COVID-19 (67.19%), and n= 42 who had not been vaccinated were 175
  more likely to be hesitant (32.8%).”
  So you claim it’s usually around round ¼, while in you sample you get 2/3. Either you should correct that initial overgeneralizing statement, or you should put some limitation that due to some sampling problem you measured extraordinarily vaccine accepting group of people with schizophrenia, what makes your findings hard to generalize.

Response 2: This is indeed the case, which is why we point out possible sample biases in limitations [lines 337-340]

• Comment 3) “In general, participants reported following health instructions, although other stud- 285
  ies have noted the opposite [71] ”71:“Larson, H.J.; Clarke, R.M.; Jarrett, C.; Eckersberger, E.; Levine, Z.; Schulz, W.S.; Paterson, P. Measuring Trust in Vaccination: A 549 Systematic Review. Human Vaccines & Immunotherapeutics 2018, 14, 1599–1609, doi:10.1080/21645515.2018.1459252.” Uhm… this study does not build you that point, check please whether you haven’t intended to cite something else. And also check the other citations.

Response 3: There were indeed erroneous or misplaced references, including the one noted above, which has been corrected.

• Comment 4): The English could be improved to more clearly express the research.

Response 4: Effectively. We have found and fixed a number of mistakes.

Reviewer 3 Report

Authors need to present the data collection process.

Also, authors need to present the research questions more clearly. 

Authors also need to present the references which are related to the COVID-19 using separate literature review section.

Authors need to present the data analysis instrument focusing on threshold.

Authors need to improve theoretical contribution of this work more. 

Authors need to present the missing reasons more.

Authors might be able to compare the main attributes based on demographic information.

Authors need to present the scales of this work more. 

Author Response

• Comment 1: Authors need to present the data collection process.

Response 1: We have added the data collection procees item [lines 117-122].

• Comment 2: Also, authors need to present the research questions more clearly. 

Response 2: We have reorganised and refocused the research questions in light of the improvements suggested by the reviewers [Lines 96-101]

• Comment 3:Authors also need to present the references which are related to the COVID-19 using separate literature review section.

Response 3: We have added a folder containing specific references related to COVD that can be found in the supplementary material. [lines 368-370] 

• Comment 4:Authors need to present the data analysis instrument focusing on threshold.

Response 5: We have added a repository containing the survey and raw data. We have also revised some aspects of the data analysis process and results.

• Comment 5Authors need to improve theoretical contribution of this work more.

Response 5: We have reorganised parts of the manuscript and focused on theoretical aspects that should influence practice [309–314].

• Comment 6:Authors need to present the missing reasons more.

Response 6: We have added an explanatory footnote clarifying the fact that it is missing completely at random and the deletion by list [line 114]

• Comment 7: Authors might be able to compare the main attributes based on demographic information.

Response 7: We modified the analyses, applying an ordinal regression to the usual socio-demographic variables, and added a psychopathological model (lines 225–236).

• Comment 8:Authors need to present the scales of this work more. 

Response 8: We have added a folder containing the complete survey to the supplementary material. <`lines 368-370]