The Impact of the COVID-19 Pandemic and Socioeconomic Deprivation on Admissions to the Emergency Department for Psychiatric Illness: An Observational Study in a Province of Southern Italy
Round 1
Reviewer 1 Report
In the phases, it is unclear who determined what a high or medium level of restriction is. What is a medium level of restrictions? I believe that restrictions are either there or they are not. There is a precise definition of what is medium or high level of restrictions?
Change the variable named Sex to Gender.
There is something strange in the tables, the only gender modalities is Male, where are the females?
Add a 0 before the dot for the p-value of the variable Discharge in Table 1
Add a 0 before the dot for Class of Deprivation=VL
Please provide post-hoc test for chi-squared in Class of Deprivation (Table1)
In Table 2 you need to put a zero before the . of all p-values
On the territory taken into consideration, what is the level of deprivation? it is known that in southern Italy there is much lower ID than for example in northern Italy. Wouldn't it be more correct to construct deprivation classes not given by the data but based on the general deprivation of the territory?
Based on what were the age groups defined? The authors must have a valid reason for transforming a continuous variable.
Is it not obvious that people with psychiatric illness are more deprived because of their condition?
The deprivation index is generally composed of 4 indicators of socioeconomic deprivation:
- low schooling index
- unemployment index
- index of rent
- index of housing occupancy
In this regard, please describe more about what ID is and what it looks like.
From this reflection, I think the interaction in the model between the phases and ID is not quite adequate.
Author Response
We thank the reviewer for his really valuable comments. We have modified the text trying to follow the reviewer's suggestions. Thanks to the suggested changes, we believe that the text has improved and we hope that this revised version of the manuscript will meet the approval of the reviewer, the editor and therefore the readers. Thank you
In the phases, it is unclear who determined what a high or medium level of restriction is. What is a medium level of restrictions? I believe that restrictions are either there or they are not. There is a precise definition of what is medium or high level of restrictions?
During the pandemic, the Italian government has imposed different levels of lockdowns based on the waves of the virus spreading. Periodically, on the basis of 20 indicators (the most important: the RT and the occupancy rates of hospital beds in intensive care), a national technical-scientific committee suggested the different levels of lockdown to the government. Therefore, in the methods (line 141-152) we have specified which restrictive measures correspond to the levels of restrictions indicated with "high", "medium" and "low".
Change the variable named Sex to Gender.
We changed in table 1 the variable name according to the reviewer.
There is something strange in the tables, the only gender modalities is Male, where are the females?
We do not report the number of women in table 1 because in our dataset people have only two genders and data relating to females was complementary to males. However, to better explain our data we added a row with female description.
Add a 0 before the dot for the p-value of the variable Discharge in Table 1
Thanks for your kindly suggestion. We modified it according to the reviewer.
Add a 0 before the dot for Class of Deprivation=VL
Thanks for your kindly suggestion. We correct the mistake
Please provide post-hoc test for chi-squared in Class of Deprivation (Table1)
We added in Table 1 the post-hoc pairwise comparison as the reviewer proposed.
In Table 2 you need to put a zero before the . of all p-values
Thanks for your kindly suggestion. We modified it according to the reviewer.
On the territory taken into consideration, what is the level of deprivation? it is known that in southern Italy there is much lower ID than for example in northern Italy. Wouldn't it be more correct to construct deprivation classes not given by the data but based on the general deprivation of the territory?
We thank the reviewer for his timely concern. If we had built the classes on the basis of the deprivation of the territory (for example the Apulia region which includes the municipalities of the province of Taranto) the deprivation index of the municipalities in question would be included in a single class since, as the referee says, the area in question has a homogeneous deprivation compared to the rest of the region. Since the intent was to verify the existence of a relationship between psychiatric access and deprivation, we thought it appropriate to create a classification of deprivation within the area in question.
Based on what were the age groups defined? The authors must have a valid reason for transforming a continuous variable.
Thanks for the question. We decided to divide the age of the sample in four categories for similar characteristics of lifestyle: people between twelve and twenty were considered young, people between twenty one and forty were considered young-adult, people between forty one and sixty were considered adults while people older than that age were considered senior.
In our study we want to study the incidence of access to an emergency department for an acute psychiatric disease by modelling the data with a Poisson Regression. With this model the risk of access in an Emergency Department is expressed as Rate Ratio (RR). We preferred to show the RR between two different classes of age, respect to the unitary increase of age for patients.
Is it not obvious that people with psychiatric illness are more deprived because of their condition?
The deprivation index is generally composed of 4 indicators of socioeconomic deprivation:
- low schooling index
- unemployment index
- index of rent
- index of housing occupancy
In this regard, please describe more about what ID is and what it looks like.
The deprivation index used in our work was performed to evaluate the socio-economic disadvantage of people who live in a census section. This index evaluates five disadvantages: poor education, home occupied by renters, job shortage, monoparental family and distribution of population in dwellings. The calculated index is therefore only a proxy variable of individual deprivation. As the referee reports, the greater deprivation of the psychiatric patient could be obvious but it is not obvious that the psychiatric patient lives in a more deprived geographical area. Furthermore, our study does not intend to relate the incidence of psychiatric illness to socioeconomic deprivation, but the use of emergency departments from psychiatric patients who live in a socioeconomic setting. In fact, as we found from the study, socioeconomic deprivation was correlated with emergency room visits by psychiatric patients, with higher access rates from less deprived areas.
In the method section we described better the methodology for the calculation of the Deprivation Index using the criteria defined by Caranci et al.
From this reflection, I think the interaction in the model between the phases and ID is not quite adequate.
Thanks to the reviewer because it gives us the possibility to better explain our purpose.
In the previous answer we clarified the role of the deprivation index. This covariate was significant in the Poisson model used to estimate psychiatric admission rates in emergency departments, as well as phase. For that reason it was appropriate to put in the model the interaction between the phases of pandemic and the Deprivation Index to verify if the deprivation index is able to modify the effect of the public health restriction applied in a phase of pandemic.
Reviewer 2 Report
I would like to thank the authors for this work, as I feel it could make contributions to evaluate how socioeconomic deprivation modified the psychiatric admissions to A&E units during COVID-19 Pandemics. The study aims to identify changes in psychiatric admissions to Emergency Departments in the province of Taranto from February 2020 to December 2021, due to the restriction measures.
The introduction overall is sparse. The authors should report studies on the impact of isolation and socioeconomic deprivation on mental illness and suicide.
The data on deaths from Covid-19 between February 2020 and December 2021 is wrong. The authors indicated 1.300 deaths. According to the information provided by the Italian Ministry of Health there were 137.402.
The statements in lines 42-47 are not supported by references. It is necessary to quote the references or rephrase the paragraph according to the references.
It is suggested to rephrase the paragraph Lines 69-73. The paragraph does not unequivocally define the objectives of the study.
The methodology is well described and the statistical analysis is well structured.
A deprivation index based on a previous 10-year census may have little significance. Why didn't the authors use the deprivation index based on the 2021 census?
Author Response
We thank the reviewer for his really valuable comments. We have modified the text trying to follow the reviewer's suggestions. Thanks to the suggested changes, we believe that the text has improved and we hope that this revised version of the manuscript will meet the approval of the reviewer, the editor and therefore the readers. Thank you
I would like to thank the authors for this work, as I feel it could make contributions to evaluate how socioeconomic deprivation modified the psychiatric admissions to A&E units during COVID-19 Pandemics. The study aims to identify changes in psychiatric admissions to Emergency Departments in the province of Taranto from February 2020 to December 2021, due to the restriction measures.
The introduction overall is sparse. The authors should report studies on the impact of isolation and socioeconomic deprivation on mental illness and suicide.
Thanks for your kindly suggestion. In the Introduction paragraph of the paper we added studies which explain the effect of public health measures to contain the infection on menthal health and in the risk of suicide (line 59-67). Furthermore we explain better the effect of socioeconomic deprivation on mental illness and suicide, with other authors papers (line 75-77).
The data on deaths from Covid-19 between February 2020 and December 2021 is wrong. The authors indicated 1.300 deaths. According to the information provided by the Italian Ministry of Health there were 137.402.
Thanks for your kindly suggestion. We corrected the mistake.
The statements in lines 42-47 are not supported by references. It is necessary to quote the references or rephrase the paragraph according to the references.
We insert in the line 42-47 the missing references.
It is suggested to rephrase the paragraph Lines 69-73. The paragraph does not unequivocally define the objectives of the study.
Thanks for your advice. We rephrase the paragraph in the line 69-73 to better clarify the aims of the study.
The methodology is well described and the statistical analysis is well structured.
Thanks for your gentle opinion.
A deprivation index based on a previous 10-year census may have little significance. Why didn't the authors use the deprivation index based on the 2021 census?
Thanks for the question. We have chosen to use the Caranci Deprivation Index because first of all it is the national index of hardship used by the Italian government to analyse the areas in which to carry out redevelopment interventions. Second, the index is built with the maximum territorial detail available (census unit). Currently the updated version was released in 2020, with the national census data from 2011 [https://epiprev.it/articoli_scientifici/aggiornamento-e-revisione-dell-indice-di-deprivazione-italiano-2011-a-livello-di-sezione-di-censimento], and a newer version has not been produced after the National Census of the 2021. Furthermore in a short communication of 2016 [http://www.sossanita.it/doc/2016_11_AIE-DEPRIVAZIONE-INDICE.pdf], Caranci et al verified the high correlation between the deprivation index (DI) of 2011 and the DI of 2001 (r=0.95) and lower correlation between the DI of 2011 and 1991 (r=0.85). Is therefore evident that within a decade, the socio-economic condition changes little. On the basis of this assumption, we thought that 2011's DI explained better the socio-economic condition during the period under analysis (2019-2021).
Round 2
Reviewer 1 Report
Although the authors have provided an answer, it does not always reflect the question asked, particularly the first one: “high," “medium” and "low" are qualitative variables defined by the authors' description. The same as for age. To model these concepts mathematically, it is useful to use a fuzzy approach (fuzzy logic) because boundaries between high to medium and low are uncertain and do not best represent reality.
