Predictive statistical model for the factors associated with prenatal depression among pregnant adolescents in Maha Sarakham province, Thailand [version 2; peer review: 1 approved, 1 not approved]

Background: The rate of adolescent pregnancy has steadily increased in Thailand over the years and depression is considered an important psychological problem in many pregnant adolescents. The aims of this study were to determine the proportion of pregnant adolescents with depressive symptoms and to identify a predictive statistical model for the factors related to prenatal depression among adolescents at antenatal care clinics in 10 community health centers in Maha Sarakham Province. Methods: A cross-sectional study was conducted in 230 randomly selected participants who were pregnant adolescents who volunteered to participate in the research, those aged ≤19 years, and had an estimated gestational age of ≤18 weeks. Those with severe obstetrics complications or a psychiatric disorder were excluded. The subjects were asked to complete a demographic questionnaire and the Thai version of the Edinburgh Postnatal Depression Scale. The mean age of participants was 17.0 (±1.4) years. Results: Overall, 43.9% of the adolescents had a cut off score of ≥11, suggestive of prenatal depressive symptoms. Multivariate analysis identified that pregnant adolescents at a primary school education level were 6.5 times more likely to experience depression symptoms than those with a diploma. Those who were emotionally supported by their husband or partner compared to relatives had a five times higher chance of presenting depressive symptoms, and those with a perceived bad relationship with their husband or partner had seven times higher.


Introduction
Adolescents can develop health problems during pregnancy that adversely affect both their physical and psychological health (UNFPA, 2014a).Pregnant adolescents commonly experience depressive symptoms (Lancaster et al., 2010); the prevalence has been estimated to be approximately 20% (Abajobir et al., 2016;Nasreen et al., 2011;Roomruangwong & Epperson, 2011).Pregnant adolescents with depression and their infants are at a higher risk of increased morbidity (Tzilos et al., 2012), resulting in a greater risk for preterm labor and intrauterine fetal growth retardation (Field, 2017;Grote et al., 2010;Orr et al., 2007).Depressed pregnant adolescents also have a greater risk of suicidal ideations and suicide attempts than those without depression (Hodgkinson et al., 2010).
Risk factors predisposing adult pregnant women to depressive symptoms include daily maternal stress, loss of selfconfidence, and inadequate social support and emotional coping (Da Costa et al., 2000;Dennis et al., 2007).Risks factors for depression among pregnant adolescents include not only those listed for adult pregnant women but also include unplanned pregnancy, low education, non-marital status, alcohol and drug use, inadequate income, and a history of emotional, sexual, and physical abuse (Koniak-Griffin et al., 1996;Meltzer-Brody et al., 2013;Siegel & Brandon, 2014;Uthaipaisanwong et al., 2015).Because prenatal depression in adolescence is a significant factor for predicting postpartum depression (Dennis et al., 2007), the ability to identify prenatal depression at the earliest stage of pregnancy is particularly important for the health and well-being of mothers and infants and for public health in general.
Thailand has a high rate of adolescent pregnancy (Ministry of Public Health, 2017).Studies by Limlomwongse & Liabsuetrakul (2006), and Uthaipaisanwong et al. (2015) in Bangkok and the South of Thailand have shown that the incidence of prenatal depression among adolescents is 46% and 20.5%, respectively.Little research, however, has been conducted in northeastern Thailand, especially in Maha Sarakham province, where the rate of adolescent pregnancy is reported to be high (Ministry of Public Health, 2017).Therefore, the aim of this study was to estimate the proportion of prenatal depressive symptoms among adolescents in the Maha Sarakham Province, Thailand, and to identify factors associated with prenatal depression symptoms in adolescents.

Study subjects and data collection
This cross-sectional study was conducted from July to October 2017 with 230 pregnant adolescents.The participants were selected using simple randomization in order to avoid sampling bias.Every week, two to three pregnant adolescents were recruited from each antenatal clinic in 10 community hospitals in the Maha Sarakham province, Thailand by randomizing HN numbers (patient identification numbers were drawn as lots) among all of the pregnant adolescents who met the inclusion criteria.Inclusion criteria were 1) pregnant adolescents who volunteered to participate, 2) those ≤19 years of age, 3) and those with an estimated gestational age of ≤18 weeks.The exclusion criteria were pregnant adolescents with obstetric complications or those who had been diagnosed with a psychiatric disorder.Initially, a sample size of 209 adolescents was calculated (Hsieh et al., 1998) assuming a prenatal depression prevalence of 46% (Uthaipaisanwong et al., 2015) with an alpha error of 0.05 and a power of 80%.An extra 21 adolescents (10%) were added to compensate for possible incomplete data, giving a total sample of 230 participants.All participants were recruited in person during their first appointment at the antenatal clinic.After receiving informed consent, participants provided their demographic and socioeconomic information on a written questionnaire (available as Extended data; Jantasin et al., 2019b;Jantasin et al., 2019c) then completed the Thai version of the Edinburgh Postnatal Depression Scale (Pitanupong et al., 2007).
The Edinburgh Postnatal Depression Scale (EPDS) is an assessment tool available in Thailand to screen for depressive symptoms.Originally developed to detect possible depressive symptoms in the postpartum period (Cox et al., 1987), the EPDS has been successfully used to identify depressive symptons in prenatal women (Eberhard-Gran et al., 2001;Lee et al., 2004).In 2003, the EPDS was translated to Thai by Pitanupong et al. (2007).Three doctorally prepared Thai experts in midwifery reviewed the translation for content validity and approved all items without recommendations.Published psychometric testing supports the 10-item selfassessment questionnaire's validity and reliability among Thai pregnant women (Pitanupong et al., 2007;Vacharaporn et al., 2003).Preliminary testing for internal reliability using 30 people, not part of the study, showed a Cronbach's alpha = 0.76.
Each of the ten EPDS items has four response options ranging from 0 to 3. The response descriptors vary somewhat depending on the item's wording.The participants choose the descriptor most closely indicating their emotions during the prior seven days.Seven items require reverse scoring so that higher total scores are more suggestive of depressive symptoms.A total score can range between 0 and 30.The cutoff score for the EPDS that indicates the presence of depression symptoms has been reported between 10 and 14 (Bolton et al., 1998;Evans et al., 2001).We defined an EPDS score of 11 or greater as a positive screen for depressive symptoms (Uthaipaisanwong et al., 2015).

Statistical analysis
Data analysis were conducted with descriptive statistics to present demographic data.Multivariable analysis through unconditional multiple logistic regression was used to identify factors associated with prenatal depression among pregnant adolescents and to determine potential associations of sociodemographic information with prenatal depressive symptoms (EPDS ≥11=1 and <11=0).Crude analysis through simple logistic regression analysis was used.After that, all variables with p-values less than 0.25 (Hosmer & Lemeshow, 2000) were selected to include an initial regression model for the assessment of multicollinearity.Then, backward elimination for a fitted regression model was performed.The model was developed and its goodness-of-fit was evaluated by determining the predictive accuracy and cut-off that was suitable for clustering towards depressive symptoms among pregnant adolescents using a receiver operating characteristics (ROC) curve.Statistical significance was set at p<0.05.Analyses were done with Statistical Package for the Social Sciences (SPSS) software, version 18.0 (SPSS ® , Chicago, USA).

Ethical considerations
The study was approved by the Institutional Ethics Committee at Mahasarakham University, Thailand.(Ref No. 054/2560;21 July 2017).Adolescents of legal age offered their informed consent volunteering to participate in the study.Parents of adolescents younger than 18 years of age provided written informed consent prior to their child's participation.

Association between depressive symptoms and participant variables
Of the 230 participants, 101 adolescents (43.9%) met the EPDS cutoff score of ≥11 suggestive of prenatal depressive symptoms.The remaining 129 adolescents (56.1%) scored below the cutoff score (Table 1).
The mean age of the 101 adolescents was 17.0 (SD±1.4)years old.The majority of those were >15 years old (n=80, 80.0%) and 21 (20.0%) were ≤15 of age.In bivariate analysis, depressive symptoms were significantly correlated with receiving less education, being a single mother with one supporter, having a husband as a supporter, and having a poor relationship with a husband, their family and friends.Pregnant adolescents who only finished primary school were 7.3 times more likely to have prenatal depressive symptoms than those with education to diploma level.In terms of marital status, the adolescents who were single mothers were at 9.4 times higher risk of depressive symptoms than those living together with their partner, with or without being married.Pregnant adolescents who had one supporter were 2.8 times more likely to be at risk of depressive symptoms compared to those with three supporters.The adolescents who had a husband as a supporter tended to be 3.7 times higher at risk of depressive symptoms when compared to those who were supported by relatives.Moreover, pregnant adolescents who reported a bad relationship with their husband, family, and friends were 5.5 times, 2.2 times, 2.8 times higher at risk of depressive symptoms than those with good relationship, respectively (Table 1).Responses to each question from each participant are available as Underlying data (Jantasin et al., 2019a).

Linear regression analysis
According to Table 1, eight variables with a p-value of less than 0.25 were selected and included in the initial regression model.From the multiple logistic regression analysis, there were only three factors that were significantly associated with prenatal depressive symptoms: level of education, the perceived support from a husband or partner, and the perceived relationship with a husband or partner (Table 2).Those pregnant adolescents with a primary school education level were 6.5 times more likely to develop prenatal depressive symptoms compared to those with a diploma degree [AOR=6.5 (95% CI: 1.5 to 28.7)].The pregnant adolescents who perceived support from their husband or partner were five times more likely to have prenatal depressive symptoms compared to those with relatives' support [AOR=5.1 (95% CI:1.9 to 14.5)].The chance of developing prenatal depressive symptoms in those with a bad relationship with the husband was approximately seven times higher than pregnant adolescents who reported a good relationship [AOR=7.4(95% CI: 3.1 to17.8)].
Three variables in the final model influencing with prenatal depressive symptoms significantly were as follows: P(Y=1/X) = 1/[1+EXP(-1.99+1.88 Education; Primary + 0.85 Education; Secondary + 1.65 Supporter; Husband/partner + 0.45 Supporter; Their family + 2.00 Relationship with a husband/partner] To validate how well the model fits a set of observations, Hosmer-Lemeshow's goodness-of-fit test was used and the results showed that the model was well-fit data (p-value = 0.675).The results of testing the model's predictive power by creating the relational graph of sensitivity and 1-specificity showed that the model had a good predictive power with ROC of 0.73 (95% CI 0.67 to 0.79) (Figure 1) at the cut point of -0.11, a sensitivity of 57.43, a specificity of 81.40, and an accuracy of 71.30.

Discussion
There was an unexpectedly high proportion of pregnant adolescents with prenatal depression symptoms in the ten community hospitals in the Maha Sarakham Province, at 43.9%.This proportion is twice the number reported worldwide (Abajobir et al., 2016;Nasreen et al., 2011;Roomruangwong & Epperson, 2011).However, the findings are similar to the studies reported by Uthaipaisanwong et al. (2015) and Nirattharadorn et al. (2005), with the risk for prenatal depression from EPDS screening at 46% and 47.1%, respectively.Findings indicate that the underlying reasons are associated with the adolescents' low level of education, the husband or partner who supports her, and the relationship she has with him.Bunevicius et al. (2009) and Lydsdottir et al. (2014) also found that pregnant adolescents with a low level of education are more at risk of having depressive symptoms (see also Field (2017) and Osok ( 2016)).In Thailand, pregnancy in adolescents often leads to dropping out from school.Consequently, a low level of education limits the opportunities for better employment and earning a higher income, which results in increased stress levels of these pregnant adolescents (Kaewjanta, 2012).
We found that not only was the emotional support coming from a husband or partner (compared to the pregnant adolescent's family) a risk factor for depressive symptoms, but also a poor relationship with that person contributes to her depression.
This finding seemingly contradicts Stapleton et al. (2012) who found that a husband's support reduces stress and positively affects pregnant women emotions.However, the perceived emotional support experienced by a pregnant adolescent from her husband or partner needs to be interpreted simultaneously with the finding of the poor relationship she experiences with him.The perceived adverse support from a husband or partner and the negative relationship with him likely increases the adolescents' depressive symptoms.
From the predictive model of prenatal depression symptoms by determining ROC, it was found that level of education, the perceived support from a husband or partner, and the perceived relationship with a husband or partner could predict prenatal depression symptoms well.
The 95% CI of the narrow ROC reflects that the model had predictive accuracy.Although the sensitivity was not adequate, specificity was at a high level.The model of three variables could be applied in operational planning to prevent and solve the problem of pregnant adolescents with prenatal depression symptoms.
The current Thai cultural norm is not to provide support to adolescents who become pregnant at a young age; society does not consider them to be ready for motherhood or a marital relationship.Adolescents who become pregnant may prefer to hide their condition from the family.Some may need to leave or transfer to a new school.They may lose their job and income as a result of the pregnancy or because of the change in living arrangements.Frequently, adolescent couples lack the maturity and emotional intelligence to begin a marital relationship, which might lead to domestic violence, although the latter was not part of the study's investigation or findings.In more agreement with our findings are the reports (Muangpin et al., 2010; UNFPA, 2014a) that pregnant adolescents with depressive symptoms and other mental health problems are adversely affected by the lack of emotional support and a negative relationship with a husband.
This cross-sectional study does not permit a closer understanding of the long-term trajectory of adolescents who become pregnant and develop depressive symptoms, nor can it examine how their pregnancy might affect their adult lives.This information could come from longitudinal and qualitative studies.In addition, the sample in this study was confined to only one region of the country; therefore, the findings may not be generalizable to other parts of Thailand.Future research might be conducted in other areas, which should consider adult women who became pregnant in adolescence and have shown depressive symptoms and what meaning that held for them in their later lives.Similarly, the husbands or partners of the adolescents should be studied as to their perceptions of the experiences related to unexpected fatherhood with adolescent women.
In conclusion, the adolescent pregnancy rate in Thailand is increasing and adolescents are vulnerable to suffer from depressive symptoms during pregnancy.As a matter of public health concern, more attention is needed to help reduce known problems with pregnant adolescents.Efforts to keep adolescents in school or ensure they return to school after delivery of the newborn may be an important step in improving the mental health of adolescent mothers.Public advocacy and education to support mental health initiatives for individuals and couples should be encouraged as a matter of health policy.If couples have become isolated from their families, attempts to reconcile and reunite them by public health personnel may promote better mental health.
Data are available under the terms of the Creative Commons Zero "No rights reserved" data waiver (CC0 1.0 Public domain dedication).

Introduction:
Can the other add recent references on the risk factors for depression in pregnancy.

Method:
Why did you exclude pregnant adolescents with psychiatric disorders?Did you excluded adolescents who previously experienced depression?

Results:
"The adolescents who had a husband as a supporter tended to be 3.7 times higher at risk of depressive symptoms when compared to those who were supported by relatives."Do you mean pregnant adolescents who had only their husband as a supporter?
○ "The pregnant adolescents who perceived support from their husband or partner were five times more likely to have prenatal depressive symptoms compared to those with relatives' support [AOR=5.1 (95% CI:1.9 to 14.5)]."I really wonder how this was measured.
○ Discussion: "We found that not only was the emotional support coming from a husband or partner (compared to the pregnant adolescent's family) a risk factor for depressive symptoms, but also a poor relationship with that person contributes to her depression."This sentence should be revised for clarity.
○ "Frequently, adolescent couples lack the maturity and emotional intelligence to begin a marital relationship, which might lead to domestic violence, although the latter was not part of the study's investigation or findings."I would be more cautious in the language used (e.g.adolescent couples may lack…) or completely delete this sentence.
○ "In addition, the sample in this study was confined to only one region of the country; therefore, the findings may not be generalizable to other parts of Thailand."Please modify accordingly.

Figure and tables:
Define the factors in the legend.

Minors comments:
Methods: "the EPDS has been successfully used to identify depressive.symptons in prenatal women"."Symptons" should be "symptoms".
○ "Data analysis were conducted with descriptive statistics to present demographic data."This sentence could be modified as follows: "Descriptive statistics were used to present demographic data".

Results:
"According to Table 1, eight variables with a p-value of less than 0.25 were selected and included in the initial regression model."This sentence should be corrected.Suggestion: "The initial regression model included eight variables selected with a p-value of less than 0.25 (see.

Felix Emeka Anyiam
University of Port Harcourt, Port Harcourt, Nigeria I have no further comments to make as the revised article is well done.
The article titled "Predictive statistical model for the factors associated with prenatal depression among pregnant adolescents in Maha Sarakham province, Thailand", is a cross-sectional study focused on adolescent pregnant mothers aged ≤19 years, and have an estimated gestational age of ≤18 weeks.The study is relevant, as it highlighted several risk factors associated with prenatal depression and some of which are: level of education, perceived support from husband and perceived relationship with husband.The study found a high prevalence of depression among this group of people which I think is a public health concern needing an urgent governmental intervention.

Report:
A. Title: The title for this study is not appropriate as I did not see a predictive statistical modeling.
Predictive modeling in statistics is more complex than what you have here.Sometimes, data mining and probability are applied in order to forecast outcomes.I think you should just leave the title as: Factors associated with prenatal depression among pregnant adolescents in Maha Sarakham province, Thailand 1.

B. Abstract:
The results section, the sentence with the words: "… and those with a perceived bad relationship with their husband or partner seven times higher."should read: "…and those with a perceived bad relationship with their husband or partner had seven times higher ." 1.

C. Introduction:
The statement, "Adolescents can develop health problems during pregnancy that adversely affect both physical and psychological health (UNFPA, 2014b)."should read, "Adolescents can develop health problems during pregnancy that adversely affect both their physical and psychological health (UNFPA, 2014b). 1.
I do not think your introduction reference should begin with UNFPA, 2014b but instead UNFPA, 2014a."a" should come before "b."Please do switch the reference.

2.
I am not sure what the phrase "...adult women" means in the sentence: "Risks factors for depression among pregnant adolescents include not only those listed for adult women…" as you did not list any risk factors for adult women.

3.
The statement, " …and a history of emotion, sexual, and physical abuse.." should read, " …and a history of emotional, sexual, and physical abuse.

4.
"Little research, however, has been conducted in northeastern Thailand, especially in Maha Sarakham province, where the rate of adolescent pregnancy is reported to be high." is a very strong statement and should be backed-up with literature.

D. Results:
Your Table 2  We discussed the title and prefer to preserve it as our statistical analysis (ROC = 73%) is plausible to forecast future outcomes.

B. Abstract:
The results section, the sentence with the words: "… and those with a perceived bad relationship with their husband or partner seven times higher."should read: "…and those with a perceived bad relationship with their husband or partner had seven times higher." 1.
Answer: we added the missing word according to the comment.

C. Introduction:
The statement, "Adolescents can develop health problems during pregnancy that adversely affect both physical and psychological health (UNFPA, 2014b)."should read, "Adolescents can develop health problems during pregnancy that adversely affect both their physical and psychological health (UNFPA, 2014b). 1.
Answer: we added the missing word according to the comment.I do not think your introduction reference should begin with UNFPA, 2014b but instead UNFPA, 2014a."a" should come before "b."Please do switch the reference. 1.
Answer: we revised the reference according to the comment.I am not sure what the phrase "...adult women" means in the sentence: "Risks factors for depression among pregnant adolescents include not only those listed for adult women…" as you did not list any risk factors for adult women 1.
Answer: We mean "pregnant adult women" and the risks factors are listed in the first sentence.We added the word 'pregnant' in front of 'adult women' to make it more clear.The statement, " …and a history of emotion, sexual, and physical abuse.." should read, " …and a history of emotional, sexual, and physical abuse.

1.
Answer: we added the missing word according to the comment."Little research, however, has been conducted in northeastern Thailand, especially in Maha Sarakham province, where the rate of adolescent pregnancy is reported to be high." is a very strong statement and should be backed-up with literature. 1.
Answer: we added the reference according to the comment.

D. Results:
Your Table 2 has 4 p-values but you only have 3 variables here you are associating.Please confirm and correct. 1.
In Table 2, you also have 5 Adjusted OR.I think this should be just the 3 you mentioned in your text: 6.5, 5.1 and 7.4.

2.
Answer: we revised the table according to the comment.

E. Discussion:
"As a matter of public health..." should read, "As a matter of public health concern".1.
Answer: we added the missing word according to the comment.
The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias • You can publish traditional articles, null/negative results, case reports, data notes and more • The peer review process is transparent and collaborative • Your article is indexed in PubMed after passing peer review • Dedicated customer support at every stage • For pre-submission enquiries, contact research@f1000.com

Publisher Full Text | Free Full Text UNFPA: The state of Thailand's population 2013: Motherhood in childhood facing the challenge of adolescent pregnancy. Thailand
PubMed Abstract | Publisher Full Text | Free Full Text Tzilos GK, Zlotnick C, Raker C, et al.: Psychosocial factors associated with depression severity in pregnant adolescents.Arch Womens Ment Health.2012; 15(5): 397-401.PubMed Abstract | : Prang Priyatruk.2014a.Reference Source UNFPA: The
Finding predictors of depression or finding a certain prevalence of depression?Is the study powered to define predictors of depression?Could you describe how did you measure factors associated with prenatal depression?Can you list the factors and give the reference of the tools used (e.g. was perceived social support measured with a validated tool)?
assuming a prenatal depression prevalence of 46%(Uthaipaisanwong et al., 2015)with an alpha error of 0.05 and a power of 80%."Could you provide more information on the way you performed sample size calculation?Based on which effect did you compute the sample size calculation?○

the work clearly and accurately presented and does it cite the current literature? Partly Is the study design appropriate and is the work technically sound? Yes Are sufficient details of methods and analysis provided to allow replication by others? No If applicable, is the statistical analysis and its interpretation appropriate?
Table 1)."I cannot comment.A qualified statistician is required. ○Is

all the source data underlying the results available to ensure full reproducibility? Partly Are the conclusions drawn adequately supported by the results? Partly Competing Interests:
No competing interests were disclosed.

confirm that I have read this submission and believe that I have an appropriate level of expertise to state that I do not consider it to be of an acceptable scientific standard, for reasons outlined above.
This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

the work clearly and accurately presented and does it cite the current literature? Yes Is the study design appropriate and is the work technically sound? Yes Are sufficient details of methods and analysis provided to allow replication by others? Yes If applicable, is the statistical analysis and its interpretation appropriate? Partly Are all the source data underlying the results available to ensure full reproducibility? Yes Are the conclusions drawn adequately supported by the results? Yes Competing Interests:
has 4 p-values but you only have 3 variables here you are associating.Please 1.In Table 2, you also have 5 Adjusted OR.I think this should be just the 3 you mentioned in your text: 6.5, 5.1 and 7.4.No competing interests were disclosed.Public Health, Population Health, Global Health, Epidemiological Research, Biostatistics, Open Research Data, Open Science, Data Science, Non-communicable diseases, Maternal and Child Health, Vulnerable Populations, Population at Risk, Reproductive Health, Data Management, Community-Based Research, HIV/AIDS Prevention, TB Prevention, Health Management and Health Promotion

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard. think
you should just leave the title as:Factors associated