Variability of the Edinburgh Postnatal Depression Scale for Antenatal Depression Screening in Spanish: a Cross-Sectional Study.

Background Depression during pregnancy is a prevalent problem with signicant potential health effects on women and children. The most widely used screening instrument is the Edinburgh Postnatal Depression Scale (EPDS). Two validations of the EDPS in Spanish have been published. Given the differences in cut-off point, sensibility and specicity among the Spanish EPDS validations, the aim of this study is to provide further information regarding the topic through a validation of the EPDS is Spain with a different sample. Methods Cross-sectional study. 194 women being followed up for normal pregnancy in Primary Healthcare in Spain completed the EPDS and had a semi-structured interview with a clinical psychologist using the DSM-5 manual criteria for diagnosing depression. The ROC curve was calculated to determine the optimal cut-off point, sensitivity and specicity. Results The age range of the participating pregnant women was from 18 to 42 years of age. The optimal cut-off point is 11/12, with a sensitivity of 93.33% and a specicity of 91.62%. The ROC curve has an area of 0.936.


Background
Perinatal depression is a major global public health problem 1 . However, until relatively recently, more importance has been being given to postpartum depression than to depression during pregnancy (also called antenatal or antepartum depression) 2 . It was not until its last edition that the DSM-5 manual speci cally addressed pregnancy, as opposed to just postpartum as one of the speci ers for depressive disorders 3 .
Depression during pregnancy affects 7% of women in high income countries 4 , and an even higher percentage in low-and middle-income countries; about 25% 5 . Its negative effects include increased risk of postnatal depression and post-traumatic stress syndrome in women 6 , premature delivery, and failure to start breastfeeding 7 , as well as increased risk of behavioral problems and antisocial behavior in children 8 .
As this is a prevalent problem with signi cant potential health effects on women and children, the screening of pregnant women is recommended 9,10 . The most widely used instrument is the Edinburgh Postnatal Depression Scale (EPDS) 4 . According to the review by Kozinsky & Dudas 11 , this scale has been   validated in various countries for the screening of antenatal depression, with cut-off points between 4/5   and 13/14, sensitivity values between 64 and 87%, speci city values between 73 and 96%, positive predictive values between 22% and 75%, and negative predictive values between 92% and 100%. There is a validated version in Spanish in Chile 12 , with a cut-off point of 12/13, sensitivity of 76.3%, speci city of 93.2%, positive predictive value of 85.3%, and negative predictive value of 88.3%.
However, given the variability in the cut-off points and in the values of sensitivity, speci city and predictive values, which can be related to cultural aspects, it is recommended that the instrument is validated in all countries in which it is to be used, regardless of whether it is validated in the same language or not 11 .
A recent Spanish validation of the EPDS in Spain has been published 13 . The cut-off point in this case was 10, with a sensitivity of 72.4%, speci city of 79.3%, PPV of 18.2%, NPV of 97.8%.
Given the differences in cut-off point, sensibility and speci city among the Spanish EPDS validations, the aim of this study is to provide further information regarding the topic through a validation of the EPDS is Spain with a different sample.

Design, population and sample
Cross-sectional study. The target population was all pregnant women in the province of Segovia (Spain).
The inclusion criteria were that the date of the subject's last menstrual period (LMP) was between July 15, 2016 and June 15, 2017, and that she was also being attended to for the follow-up of her pregnancy in any of the public primary healthcare centers in Segovia; where follow-up is free of charge for all women who request it. The exclusion criteria were the language barrier to communication between women and health workers, and the inability to read or write in Spanish.
The recruitment period was from January 25 to July 28, 2017. Participants were recruited through the midwives' consultations in the following primary healthcare centers: Segovia I, Segovia II, Cuellar, El Espinar, Villacastín, Carbonero, Riaza, La Sierra, Sepúlveda, Cantalejo, Sacramenia and Nava de la Asunción. In the Segovia III and Segovia Rural primary healthcare centers, recruitment took place in the medical and/or nursing practices.
After giving their consent, the women completed a self-administered questionnaire with sociodemographic data (age, country of birth, level of education, and employment status). They were then given appointments to attend the primary healthcare center within seven days. On the day of their appointments, they completed the Edinburgh Postnatal Depression Scale (EPDS), and subsequently met with a clinical psychologist.

Measurements
Age was recorded as o cial age in years.
The level of education was classi ed into primary (9 years of educational training), secondary (12 years of educational training) and university (university degree).
The employment status was classi ed as working, unemployed (people who are looking for work), students and housewives.
The Edinburgh Postnatal Depression Scale (EPDS) is a 10-item self-report questionnaire. Each question is scored 0-3 (resulting range 0-30) and completion takes around 5 min 14 . As previously explained, it has been validated for antenatal depression screening in several countries and languages 11 . A translated version in Spanish is available and validated for postnatal depression 15 .
As the gold standard for diagnosing depression during pregnancy, a clinical psychologist conducted a semi-structured face-to-face interview with each woman using the DSM-5 manual criteria 3 . The interview was single-blind (the psychologist did not know the outcome of the Edinburgh Postnatal Depression Scale) and lasted approximately 30 minutes. Based on the interview, the clinical psychologist classi ed the women participants into two categories: "Depressed" and "Not depressed". The category depressed included any type of depression according to the DSM-5 manual, and not just major depression.

Statistical analysis
The socio-demographic data are expressed as absolute and relative frequencies in the case of the qualitative variables country of origin, level of education and employment status, and as mean (standard deviation) in the case of the quantitative variable age.
Little's test was conducted to determine whether the women who lled in the sociodemographic data and subsequently did not complete the Edinburgh Postnatal Depression Scale or attend the interview with the clinical psychologist had different characteristics or were randomly distributed.
The ROC curve was calculated to determine the optimal cut-off point, in the Edinburgh Postnatal Depression Scale (EPDS), for diagnosing antenatal depression. Using the data from the interview with the psychologist and the Edinburgh Postnatal Depression Scale (EPDS) scores, we calculated the sensitivity and speci city of the test with its 95% con dence intervals. In addition, the positive (PPV) and negative (NPV) predictive values were calculated.
Data analyses were performed using the IBM SPSS Statistics 23.0 for Windows (IBM, Armonk, New York).
Of the 219 women who completed the sociodemographic data, 25 (11.4%) did not complete the EPDS or the interview with the clinical psychologist. Little's test had a value of 0.659, indicating that there was randomly lost data.
194 women completed the EPDS and the interview with the clinical psychologist. Table 1 describes the socio-demographic characteristics of these women. The age range of the participating pregnant women was from 18 to 42 years of age. Of the women who were not born in Spain, the majority are European

5.3
In the interview with the psychologist, 7.7% of the women were detected as having antenatal depression.  The COR curve has an area of 0.936 (95% CI: 0.87-1.00). Table 2 shows the sensitivity and 1-speci city for the different scale scores.

Discussion
This cut-off point falls within the upper range of those found in the review for antenatal depression 11 , ranging from 4/5 to 13/14. It is slightly lower (11/12 vs. 12/13) than that of the Spanish version validated in Chile 12 , con rming the differences between countries despite sharing a common language 11 .
Likewise, compared to the Spanish version validated in Spain 13 , the cut-off point is slightly higher ( In search of an explanation for these differences between the Spanish versions of the EPDS in Spain, we nd that all the women studied in the published validation 13 were from one location (Orense hospital) using DSM-IV, while in our research, all the women were from another location (Segovia, primary care centers) using DSM-5.

Conclusion
Differences in cut-off point, sensibility, speci city and predictive values among the three Spanish EPDS validations (Chile and two in Spain), as well as between the two Spanish EPDS validations in Spain (Orense and Segovia) were found.
Given the differences we propose to continue searching for an optimal cut-off point in the Spanish version of the EPDS. In our opinion, further research is necessary in order to use the EPDS for antenatal depression screening with Spanish-speaking pregnant women, even within Spain. A multicentric study including pregnant women from all over the country would be suitable.

Declarations
Ethics approval and consent to participate The study was developed in accordance with the international ethical recommendations for human research and clinical trials of the Declaration of Helsinki. This research obtained a favorable report from the Ethics Committee of the Commission of Research of the Health Management of Segovia as part of the process to obtain the research grant, le GRS 1287/B16. Written informed consent was obtained from all individual participants included in the study. All women who met the criteria for depression in the interview with the clinical psychologist were referred to their doctor for assessment and treatment.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. conducted the statistical analysis. JGP and EMM redacted the rst version of the manuscript. The other authors provided signi cant suggestions. All authors read and approved the nal manuscript.