Predictors of medication use during pregnancy: a cohort study

Background: Sociodemographic characteristics and health behaviours are associated with medication use in pregnancy, but it is unclear if they are independent predictors because women´s health status has hardly been accounted for. We aimed to identify predictors of use of medications and of iron/folic acid. Methods: This cohort included pregnant women recruited in a prenatal clinic in Trieste, Italy, from 2007 to 2009. Dispensations were obtained from the regional outpatient dispensation database through record linkage. We calculated the Odds Ratio (OR), with 95% confidence interval (95%CI), of ≥ 1 dispensation of (a) any medication and (b) iron/folic acid, using unconditional logistic regression. The final model adjusted for age, partner education, housing size, comorbidities. Findings: Of 767 women, 70.5% had ≥ 1 dispensation of any medication and 46.1% of iron/folic acid. Use of any medication was predicted by immigrant status of the woman (OR 1.21; 95%CI 0.57–2.53) or of her partner (1.51; 0.67–3.40), ≤ high school degree of the woman (1.11; 0.61–2.03) or of her partner (1.21; 0.75–1.95), unemployment (1.47; 0.72–2.98), smoking (1.25; 0.65–2.40), alcohol consumption (≥5 drinks/week: 2.78; 1.78–4.34), and obesity (1.33; 0.59–2.99). Use of iron and/or folic acid was predicted by ≤ high school degree (0.65; 0.40–1.08), smoking (0.80: 0.47–1.37), and obesity (0.62; 0.31–1.25). Conclusion: In this cohort, characteristics including education, immigrant and employment status, smoking, alcohol consumption, and obesity independently predicted medication use.


INTRODUCTION
Women frequently use medications during pregnancy.It has been estimated that 27% to 99% of women in developed countries use prescription medications during pregnancy [1] and about 67% used Over-the-Counter (OTC) agents [2].However, the evidence on the riskbenefit profile in pregnant women is limited for most medications.Thus pregnant women often have concerns about using medicines [3,4] and their compliance with needed medication may be influenced by the perception of medication-related risks: in Northern Europe, about 70% of women avoided taking medication for fear of foetal adverse effects [5].Younger women, those with a first pregnancy and those with lower educational level reported the highest perceived risk related to medication use [6].Women reported receiving information on medications and supplements as one of their most important learning needs [7].Clinicians, midwives and other personnel involved in providing healthcare to pregnant women play a key role in providing such information during antenatal visits as well as in medication management and counselling.It is thus important to assess predictors of medication use in order to identify those women who may benefit the most from active counselling or information initiatives.Moreover, such evidence is needed to plan interventions promoting safe medication use during pregnancy and to tailor such interventions to the specific characteristics of women.
Differences in medication use in pregnancy according to women´s sociodemographic characteristics and health behaviours have been reported in prior studies but results were inconsistent, probably because factors such as the societal context and the healthcare system also play a relevant role.Thus, it is important to address this topic in specific and different contexts and countries.In a number of studies, use of medications increased with increasing maternal age [2,8,9], while in others, younger childbearing women were more likely to report the use of medications for acute/short-term illnesses [2] and of anti-asthmatics [10], as well as to fill prescriptions for antibiotics [11].In some studies in Northern European countries, women with low education were more likely to use any medication [12] or antibiotics [12,13]; in an international survey, women with lower educational level were more likely to report the use of medications for chronic conditions [2].Medication use was conversely higher in more educated women in a large cohort in the USA [8] and in a cross-sectional study in Belgium [9].Use of medications was less likely in immigrant women in Western and Northern Europe [2,10], while a Belgian study found that use of medications was reported more frequently by non-immigrant women [9].In Germany, female welfare recipients and unemployed women were more likely to fill prescriptions for antibiotics than those in white-blue collar occupations [11].Regarding health behaviours, both smoking and alcohol consumption during pregnancy have been consistently associated with increased use of medication [2,10,14].
Women´s health status is a determinant of medication use during pregnancy.Women reporting health problems during pregnancy were more likely to use analgesics, antiinfectives and antihistamines than those who did not report problems [15].Sociodemographic characteristics and health behaviours have a complex relation with maternal health status as well as with health care utilization during pregnancy, such as prenatal care visits and ultrasound evaluations.For instance, maternal education has been inversely associated with hypertension and preterm delivery [16][17][18][19] as well as with obesity [20].A social gradient in health behaviours, such as smoking during pregnancy [16,21] has been reported as well.Few prior studies, however, accounted for women´s health in assessing the relation of medication use with sociodemographic characteristics and health behaviours.
We conducted a prospective cohort study to identify predictors of the use of prescription medications as well as of folic acid and iron, adjusting for comorbidities before and during pregnancy.

Participants
The cohort included all pregnant women residents of the Friuli Venezia Giulia (FVG) region in Northeast Italy who had attended a prenatal visit between 20 and 22 weeks of gestation at the Institute for Maternal and Child Health IRCCS Burlo Garofolo in Trieste from April 3, 2007 to March 3, 2009.During the recruitment period, about 1,800 live births per year were recorded in Trieste and 9,000 in FVG [22].Exclusion criteria were: age ˂18 years, poor Italian language skills, twin or complicated pregnancies.Complicated pregnancies were defined as those with maternal abnormalities of the reproductive tract (such as uterine fibroids), pre-existing chronic illness (such as cancer, AIDS, severe heart disease, severe kidney disease, severe Crohn's disease or ulcerative colitis), or foetal congenital defects.
All women filled out a self-administered questionnaire inquiring on: date of birth, marital status (woman cohabiting with the partner or living alone), housing size (<50 m 2 , 50-100 m 2 , ≥100 m 2 ), smoking, alcohol consumption, comorbidities before and during pregnancy (diabetes, asthma, allergies, epilepsy, hypertension, vomiting, hypothyroidism, hyperthyroidism, lupus, rheumatic diseases, urinary tract infections, other infections, fevers, seizures, anaemia, cardiovascular diseases, neurological diseases), prior pregnancies (gravidity), number of prenatal visits and ultrasound examinations, height and weight before and during pregnancy, gestational age at birth and date of Epidemiology Biostatistics and Public Health -2018, Volume 15, Number 2 Medication use in pregnancy delivery.For the woman and (if applicable) for her partner, we collected information on the country of origin, level of education (degree achieved: less than high school, high school, university or higher) and occupational status (on prebirth maternity leave, employed, housewife, unemployed).
For each woman, we extracted the records of all dispensations between 2006 and 2012 from the outpatient prescription database of the FVG Region through record linkage using an individual identifier.This database records information on all prescribed, reimbursable medications dispensed to residents of FVG by pharmacies.All residents are registered with the Regional Health System, providing universal access to health care.A unique personal identifier links anonymized individual records.For each dispensation, the following information is recorded: date of redemption, active substance (description and Anatomical Therapeutic and Chemical -ATC -classification code [23]), brand, quantity, strength, dispensed form, number of units and number of refills.Information on the indication and the prescribed dosage regimen is not recorded.All dispensations from the estimated date of conception to the date of delivery were assumed to have occurred during pregnancy.The date of conception was estimated by subtracting gestational age at birth from the date of delivery.

Statistical analysis
We estimated the Odds Ratio (OR), with 95% confidence interval (95%CI), of having ≥ 1 dispensation for (a) any medication, (b) folic acid and iron, using unconditional logistic regression.The following variables were evaluated through uni-and multi-variate analysis: age at delivery (5 classes), education of women and of partners, occupational status of women and of partners, prior pregnancies, smoking, alcohol consumption, BMI before pregnancy (underweight below 18.5; normal weight 18.5-24.9;overweight 25.0-29.9;obesity 30.0 and more [24]), comorbidities before and during pregnancy (none, 1, 2+), country of origin of women and partners (Italy, other), marital status, number of visits and of ultrasound imaging, housing size.Variables with univariate p ≥0.20 were entered individually in bi-and multi-variable models and only those that explained the variability or modified the regression coefficient estimators were retained.The final model included terms for age, paternal education, housing size and comorbidities.The statistical analysis was performed with SAS © software, version 9.3 (SAS, Cary, NC, USA).

Ethics Committee review
The study protocol was approved by the Ethics Committees of the University Hospital of Udine and of the Institute for Maternal and Child Health of Trieste.Written informed consent for participation in the study was obtained.

RESULTS
The cohort included 767 women, 42.6% of whom were aged 30 to 34 years and mostly (91.4%) nonimmigrants (Table 1).More than 45% held a high school degree or had a partner with a high school degree, 74.0% were on maternity leave and 92.7% had an employed partner.Only 9.5% smoked and 5.1% had 5 or more alcoholic drinks per week, 72.2% had a normal BMI but 14.6% were overweight and 5.5% obese; 45.6% were pregnant for the first time, 31.3% reported no comorbidities but 30.9% had 2 or more comorbidities, 44.2% had 9 or more prenatal visits and 27.0% had 8 or more ultrasound examinations.
Compared to women aged 25 to 29 years, those between the ages of 30 and 39 years were more than twice and those 40 years or above more than four times more likely to have ≥ 1 dispensation of any medication (excluding iron and folic acid) (Table 3).The likelihood of having ≥ 1dispensation of any medication was increased in immigrant women (1.21; 95% CI 0.57-2.53)or those with an immigrant partner (1.51; 0.67-3.40), in women with less than a high school degree (1.11; 0.61-2.03)or with a partner having less than a high school (1.21; 0.75-1.95) or a high school degree (1.51; 0.97-2.35), in unemployed women (1.47; 0.72-2.98)or those with a living space ≤100 m2 (1.51; 1.02-2.25).
Women with prior pregnancies were less likely to have ≥ 1dispensation of any medication, while an increasing number of comorbidities and of prenatal ultrasound examinations increased the likelihood of having ≥ 1dispensation of any medication.

DISCUSSION
This cohort included mostly women with a higher education, who were employed or had an employed partner, and had received prenatal care.About 70% of women were dispensed at least one medication during pregnancy, a prevalence in the range of developed countries [1].Besides iron and folic acid, the most frequently dispensed medications were analgesic nonopioids and thyroid medications.This result supports the observation that women in this study were mostly healthy, as only one in three women had more than one comorbidity.Moreover, the majority of women did not smoke, reported no or only occasional alcohol consumption and had normal weight.However, the limited number of women who were smokers, had a high alcohol consumption and were obese were more likely to have medications dispensed during pregnancy, probably because of poorer health.Higher BMI has been associated with higher use of prescription medications in prior studies [13,25].Additionally, we found that women with obesity as well as smokers were also less likely to take folic acid and iron.These findings underscore the importance of identifying subgroups of women with specific lifestyle habits and conditions which require the highest degree of counselling or intervention on medication use during pregnancy including recommended preventive agents.
In our study, sociodemographic characteristics also predicted the use of medications.Women younger than 25 and above 30 years were more likely to have at least one prescription medication dispensed during pregnancy.This is in line with prior studies showing higher use of medications in the oldest and youngest age categories compared to the intermediate age [2,8,9,25,26].In FVG, the mean maternal age at delivery was 31.2 years in 2008 27, suggesting that midwives should carefully assess medication use during pre-natal visits, as many of their patients are likely to use at least one medication.
In our study, the use of any medication was more likely in women with lower education, or in those whose partner had lower education, and in unemployed women.These findings are consistent with some [2,11,12] but not all studies [8,9].However, we found that women with lower education were less likely to take folic acid and iron.In our study, immigrant women, or those with an immigrant partner were more likely to use medications as well as folic acid and iron.Conversely, use of medications was less likely in immigrant women in Western and Northern Europe [2,10].Our results have two important implications.First, medication use may be a reflection of poor health and indicate higher information needs on health and safe use of medications in pregnancy.Second, it underscores the importance of fine-tuning the delivery of this information to the educational and cultural background of women.Of note, we found that women currently employed during pregnancy were less likely to have dispensations of any medication than those in pre-birth maternity leave.The 'healthy worker effect' may partially explain this result, because women experiencing less health problems, and thus using medication less frequently, may remain employed during pregnancy.
Women who had prior pregnancies were less likely to use any medication as well as iron and folic acid.Concordantly, many studies have reported that use of prescription medication was lower in women with more than one prior pregnancy [28].However, other studies had different results: in one, women with prior pregnancies were more likely to report the use of medications for acute/short-term illnesses, but not for chronic or long-term    conditions [2]; in another study, use of medication was less frequent in women during their first pregnancy than in those with prior pregnancies [9].In our cohort, women with a higher number of prenatal care visits and ultrasound examinations were more likely to use medication, probably because the more intensive use of healthcare is an indicator of poor health.Concordantly, in a Dutch cohort, the number of General Practitioner visits was a strong predictor of OTC medication use [29].

Limitations and strengths
We assessed medication use through dispensation data.On the one hand, dispensations are only a proxy for medication use because dispensed medications may actually not be taken [30] or because of noncompliance and medication borrowing or sharing [31].On the other hand, however, dispensations represent objective information not affected by recall bias.Recall bias may limit the accuracy of medication use collected through questionnaires depending on data collection methods and questionnaire design [32][33][34] and it has been associated positively with women´s educational leve [35,36].
The prescription database used in our study covers the entire resident population, without any exclusion according to occupational or socioeconomic status.All women in the cohort were linked to dispensing records without omissions of population subgroups (e.g., unemployed or immigrant women).The potential for information bias is thus reduced.
As measures of socio-economic status, we collected information on education and occupational status, but not on household income.However, education as a measure of socioeconomic status captures both the dimension of knowledge and earning capacity through the professional position.
We accounted for the health status of the women, a strong determinant of medication use during pregnancy, through adjustment for comorbidities.Moreover, the study also evaluates the effect of characteristics of the partner, such as educational level, occupational and immigration status.

CONCLUSION
In this cohort, lower education and immigrant status of women and their partners, women's unemployment, smoking, higher alcohol consumption and being underweight or obese were independent predictors of the use of medications during pregnancy, adjusting for women´s age and comorbidities.Careful assessment of medication use is an important routine component of midwifery, particularly in women more likely to use medications.Medication assessment and interventions to promote safe use of medications should be carefully tailored to specific women´s backgrounds and cultural preferences.
Women´s lower education, smoking and obesity independently predicted lower use of folic acid and iron.Specific interventions should target women of childbearing age as well as pregnant women to promote folic and iron supplementation during pregnancy.These interventions offer also an opportunity to promote healthy lifestyle habits.

TABLE 1 .
Characteristics of the women included in the cohort.

TABLE 1 (
CONTINUED).Characteristics of the women included in the cohort.

TABLE 2 .
Women with at least one dispensation during pregnancy, by therapeutic class.