Design and selection of participants
The study data is part of a larger investigation into the development of two instruments designed to detect the needs of women during pregnancy, childbirth and postpartum. The study protocol has been published previously .
This is a cross-sectional study, collecting data in the Basque Health Service between 2019 and early 2020. The Osakidetza-Basque Health Service is the public healthcare system of the Basque Country, a region located in the north of Spain with a population of just over two million inhabitants. Osakidetza was created by the Health Department of the Basque Government in 1983, and all the public hospitals and primary care centres of the Basque region come under this organization, structured into 13 Integrated Health Organizations (IHO) spread throughout the Basque country. An OSI is a network consisting of several primary care centres, dispensaries for sparsely populated rural areas, and a referral hospital. More than 30,000 professionals work for Osakidetza, which could be considered the largest health organization in the Basque Country. For pregnancy, delivery and postpartum follow-up, each hospital coordinates with a set of primary care centres. On average, one midwife works in each primary care health centre.
The women were recruited from primary care centres, under the management of several IHOs, by their midwives during a pregnancy check-up, and also through snowball sampling. The midwife informed women over 18 years of age with a good knowledge of the Spanish about the study and, if the women wished to participate, the research team provided them with a link to the questionnaires on their mobile phone. The profile of the woman who goes to the midwife for pregnancy control is that of a woman whose pregnancy is low risk, as high-risk pregnancies are supervised by the obstetrician. Additionally, the women were able to share the link with other pregnant women. In the same app, an informed consent request was made which, once accepted, gave access to the questionnaires (the measurement instruments and a formulary of sociodemographic and clinical questions). The study was approved by the Basque Clinical Research Ethics Committee (PI2019110).
To assess Attitudes towards the Medicalization of Childbirth (ATMC) among pregnant women, a questionnaire of the same name created by Benyamini et al.  was used. The questionnaire consists of 15 items aimed at measuring women’s inclination to use medical technology and interventions during childbirth. Seven items expressed a positive ATMC and eight items expressed a negative attitude, and women were asked to rate their level of agreement with each statement on 1–5 Likert scales. The negative items were reversed and then a summary score was calculated as the average of all responses. A high score indicates a positive attitude towards medicalization.
To assess the locus of control style of the participants, the Spanish version of the Wallston Form ‘A Multidimensional Locus of Control’ was used, adapted by Tomás-Sábado and Montes-Hidalgo in 2016 . It consists of 18 items, with 6 response possibilities for each of them, scoring from 1 = completely agree, to 6 = completely disagree. The scale provides 3 relatively independent scores corresponding to the 3 factors considered: internality (items 1, 6, 8, 12, 13 and 17), which expresses beliefs that health depends on one’s own behaviour; and chance/luck (2, 4, 9, 11, 15 and 16) and other relevant people [3, 5, 7, 10, 14, 18] which both express beliefs in externality, that is, they consider that health depends on chance/luck or on the actions of other competent people, respectively.
To evaluate the coping strategies of the women, the Spanish adaptation by Lorén-Guerrero  of the scale called “Revised Prenatal Coping Inventory (NuPCI)” developed for the assessment of coping strategies over the course of pregnancy  was used. The scale consists of 42 items. Women are asked to report how often they used each method of coping “to try to manage the strains and challenges of being pregnant” during a given time frame on a scale ranging from 0 (never) to 4 (almost always). The items refer to general coping strategies along with others that are specific to pregnancy, and are divided into 3 scales or types of coping: a) the preparatory scale, with 15 questions , b) the avoidant coping scale, with 11 questions and c) the spiritual coping scale, with 6 questions.
Fear of childbirth was also evaluated, since some studies relate it to ATMC, using a Spanish version of Wijma’s 33-item questionnaire from 2005 , adapted by Ortega-Cejas et al. , which maintained the one-dimensional structure proposed by the author of the original test. The answers to each question appear as a 6-point Likert scale, the answers being (0 and 5 respectively) the opposite extremes of a certain feeling or thought when the woman imagines her labour and delivery. The minimum score is 0 and the maximum is 165. Scores above 85 indicate severe fear of childbirth [40, 41].
The possible confounding effect and modifier of the effect of other variables that could be related to both ATMC and the locus of control and coping have been considered. These variables were collected through an introductory form, composed of 22 items, on sociodemographic and clinical data completed in the app: age, parity, nationality (Spanish/immigrant), educational level (low/middle /high), paid work (yes/no) and the presence of certain prior risk factors (such as obesity, toxic habits, high age, history of prematurity or low birth weight, high birth weight, previous miscarriage or stillbirth, illness requiring periodic medical monitoring, regular medication, family history of births, etc.) with two possibilities of answering yes/no.
An analysis of the tests used was carried out to see if they presented adequate psychometric properties in our sample. All the tests used showed appropriate psychometric properties. In the case of ATMC, the original one-dimensional structure showed an appropriate adjustment in the adaptation made for our linguistic and cultural context (χ2 = 402.49, gl = 90, χ2/gl = 4.47, CFI = 0.94, TLI = 0.93, SRMR = 0.09). Internal reliability was high (omega index = 0.85, ordinal alpha = 0.80). In the Locus of Control Multidimensional Scale, the internality factor items formed a single dimension in our sample (χ2 = 28.92, df = 9, χ2/gl = 3.21, CFI = 0.98, TLI = 097, SRMR = 0.06) which presented acceptable internal consistency (omega = 0.75, ordinal alpha = 0.70). The chance/luck factor also showed a good fit between the one-dimensional solution and the data (χ2 = 42.26, df = 9, χ2/gl = 4.69, CFI = 0.98, TLI = 0.96, SRMR = 0.07) and presented high internal consistency (omega = 0.76, ordinal alpha = 0.75). The same was so with the factor “Other relevant people” (χ2 = 37.82, df = 9, χ2/gl = 4.20, CFI = 0.90, TLI = 0.91, SRMR = 0.08, omega = 0.75, ordinal alpha = 0.52). Regarding the NuPCI scale, a good fit was found in a) the preparatory coping scale (χ2 = 323.32, gl = 90, χ2/gl = 3.59, CFI = 0.95, TLI = 0.94, SRMR = 0.08, omega = 0.85, ordinal alpha = 0.85), b) the avoidant coping scale, (χ2 = 79.35, gl = 44, χ2/gl = 1.80, CFI = 0.97, TLI = 0.96, SRMR = 0.07, omega = 0.74, ordinal alpha = 0.79) and c) the spiritual coping scale, (χ2 = 9.76, gl = 9, χ2/gl = 1.08, CFI = 0.99, TLI = 0.99, SRMR = 0.05, omega = 0.81, ordinal alpha = 0.83). Finally, the W-DEC questionnaire also showed a good fit with our data (χ2 = 1315.53, df = 495, χ2/df = 2.66, CFI = 0.91, TLI = 0.90, SRMR = 0.11, omega = 0.92, ordinal alpha = 0.92.
For the descriptive analyses, means and standard deviations were used for the continuous variables, and absolute and relative frequencies for the categorical variables.
For the analysis of the dimensional structure of the measurement instruments, confirmatory factor analyses were carried out and for the evaluation of their internal consistency, the McDonald's omega and ordinal alpha indices were calculated.
To analyse the relationship between the possible predictor variables (sociodemographic factors, locus, NuPCI, and fear) and ATMC, bivariate analyses were performed using linear regression for continuous variables and ANOVA for categorical variables. Multivariate ANCOVA models were built with all the predictor variables, obtaining the estimators and their 95% confidence intervals. The best model was chosen following a stepwise backward strategy using likelihood ratio tests (with a significance criteria of P < 0.05).
All analyses were performed with SAS 9.4 and R.