Drugs During Pregnancy: Methodological Aspects 1st ed.

1. Introduction

Bengt Källén1


Tornblad Institute, Lund University, Lund, Sweden

The use of drugs during pregnancy may affect embryonic and/or fetal development. Most feared is perhaps an increase of the malformation risk. It should be realized that maternal drug use is a relatively rare cause of congenital malformations. In perhaps 25 % of such instances, genetic conditions explain the malformation, in a few percent nongenetic factors are identified, but in the remaining cases no direct explanation to the event exists. It is then easy to look for an explanation, e.g., by postulating effects of maternal drug use. As will be explained in the text, the most crucial evidence comes from epidemiological investigations. These, however, are often burdened by uncertainty, and it is easy to jump to conclusions. For the individual, risks from drug exposure are often so low that they hardly matter: if a woman has a 3 or a 4 % probability to have a malformed infant is of little significance for her, but if the drug is commonly used, even a weak effect may play a role as it can cause many malformed infants.

In the following text, examples are given from the Swedish Medical Birth Register. This register started in 1973 and was then based on documents summarizing the pregnancies. These were prepared after delivery by secretaries at the obstetric clinics (practically all births in Sweden take place in hospitals). Since 1982, data have instead been taken from copies of the original medical records which have the same format in all delivery units. With the introduction of computer-based medical records, transfer of information can be made electronically. The register contains much medical information and is supplemented with some data from Statistics Sweden. Information on factors of interest in early pregnancy (e.g., smoking, maternal weight, height, drug use) is based on interviews made by the midwives at the woman’s first visit to the prenatal care (usually in week 10–12) and is thus prospective related to possible complications during pregnancy and in the neonate. Very few women do not attend prenatal care which is free of charge. Beginning on July 1, 1994, the information on drug use obtained from midwife interviews and from medical records during prenatal care was included in the register. Outcome data were obtained from the delivery records and from the pediatric examination of the newborn – all newborn infants are examined by a qualified pediatrician. Further data on the infants born were obtained from the Hospital Discharge Register (part of the Patient Register) and also the Birth Defect Register (previously called the Register of Congenital Malformations).

A description with details of the content of the Medical Birth Register is available in http://www.socialstyrelsen.se/Publikationer2003/2003-112-3. Most presented data refer to the period 2005–2013, some to the period 1995–2013. For the period 1995–2013, there are about 1.9 million deliveries, for the period 2005–2013 about 962,000 deliveries.

Already relatively early in the text, the terms odds ratio (OR) and 95 % confidence interval (95 % CI) are used. Odds ratio is a measurement of association between exposure and outcomes. This can be illustrated with a 2×2 table (Table 1.1):

Table 1.1

A 2×2 table on exposure and outcome



No outcome





N1 + N2




N3 + N4


N1 + N3

N2 + N4

N1 + N2 + N3 + N4

N represents numbers

The odds for the exposed group is N1/N2 and the odds for the unexposed group N3/N4, and the odds ratio will be the quotient between the two: (N1/N2)/(N3/N4). This will be an estimate of the risk of the exposure. An alternative way is to calculate a risk ratio, where the risk for the outcome in the exposed group is N1/(N1 + N2) and the risk in the unexposed group is N3/(N3 + N4). The odds ratio will always differ more than the risk ratio from “no effect,” 1.0.

Odds ratios can be adjusted (adjusted odds ratios) for the possible influence of other factors in ways which will be discussed later in the book. The 95 % confidence interval (95 % CI) shows the likely interval within which the true OR or RR lies.