Bengt Källén1
(1)
Tornblad Institute, Lund University, Lund, Sweden
13.1 Pharmacovigilance
The main type of pharmacovigilance – to monitor adverse events of drug use – is based on voluntary reporting of such events. This activity was discussed in Chap. 3 and has been valuable in some situations, notably when new drugs were introduced on the market. In case of a very strong teratogenicity (for instance of equal strength as that of thalidomide), already a few reports of severe malformations could result in the identification of the hazard, especially if the reported malformations showed similarities in their pathogenesis.
There is, however, a need for other ongoing monitoring of the effects of maternal use of drugs on pregnancy outcome, not relying on voluntary reporting. In 2003 Mitchell pointed out that the American FDA should develop a teratogen surveillance system. The Swedish Medical Birth Register gives opportunities for such monitoring. In contrast to most scientific studies which are limited in time, this register allows an ongoing monitoring of the outcome after drug use during pregnancy. Up till now this has been performed in the following way which certainly is not ideal but which perhaps can give the reader some ideas.
Data from the register concerning drug use are updated once a year. At this updating, the drug names as written down by the midwives who performed the interviews are transferred to ATC codes. The complete register is then analyzed. All ATC codes are tabulated and drugs or drug groups with at least 50 exposures are identified. For each drug (group) the number of congenital malformations as recorded in the Medical Birth Register is calculated, and the expected number is compared with the observed number with a Poisson distribution model or with chi-square tests if numbers are large enough. “Significant” differences (as increases or decreases) are identified. The process is repeated for a number of malformation groups: all malformations, relatively severe malformations, neural tube defects, cardiovascular defects, orofacial clefts, alimentary tract atresia, hypospadias, etc. No adjustment for possible confounders is made but only crude numbers and risks are given. A special interest is given to risk estimate changes from the previous years, appearance or disappearance of risks.
At a meeting with representatives from the Swedish National Board of Health and Welfare and the Medical Drug Agency, the list is scrutinized and discussed against the background of what is known from the literature. When something noteworthy is seen, a detailed investigation is made and more complete information on malformations is collected, and the possible effect of confounding is considered.
This system would miss a strong teratogenicity of a new drug, perhaps used by less than 50 women but with a high risk of congenital malformations (the thalidomide type). Furthermore, no data on other pregnancy outcomes or long-time effects are analyzed.
Specific investigations may also be initiated from literature reports or from adverse drug reaction reports. This type of monitoring should be a natural consequence of running a register. The authority which runs the register should not only allow scientists access to the data but should use the data for an ongoing monitoring of drug use during pregnancy.
Another system for “signal detection” was described from the Netherlands (de Jonge et al. 2013) where data from a malformation register with respect to drug exposure (based on pharmacy records, verified by telephone interviews after delivery) were compared with data from a population register (where drug exposure was identified from a prescription register). It is not clear if this has developed into an ongoing surveillance or if it only was a way to test a possible approach.
13.2 Information on the Risk with Drug Use During Pregnancy
In an ideal world, every published scientific article should give a clear and unequivocal answer to the question posed. As has repeatedly been pointed out in the text above, this is seldom the case in the real world. Each research result should be looked upon as a piece in a jigsaw puzzle – the complete answer is not obtained until many pieces have been added, and a picture begins to appear which at least resembles the truth. The uncritical spreading of findings through various media and the Internet can be harmful and cause unnecessary anxiety. On the other hand, an event similar to the thalidomide tragedy would need a rapid spread of information to prevent further damage. A definite problem is that if the media regularly cry “Wolf!”, finally a serious warning may not be believed by the public, as was the case in the legend of Aisopos.
One often sees a general recommendation – not to use drugs during pregnancy. This is a rather impractical rule. It is true that drugs which are not needed should not be used during pregnancy (or otherwise) and that, for instance, the use of street drugs by the pregnant woman should be strongly discouraged. The pregnant woman, however, must get adequate therapy when that is needed; the problem is to balance the need of the woman against the possible risk for the embryo or fetus. Lack of treatment may actually be a larger risk for the embryo than adequate drug use, for instance, at maternal asthma or epilepsy.
In the daily clinical work, questions on drug use during pregnancy are common and have to be answered. It is not reasonable to expect that practitioners and midwives should be able and have the time to follow and interpret the scientific literature on the subject. They would have to rely on digested impartial information. There have been many efforts to supply the medical practitioner with evaluated information. The first effort was probably made in Sweden where in 1978 a classification system was introduced in the Swedish National Drug Formulary (FASS) (Berglund et al. 1984). The classification of a drug is made by the drug industry. The system consists of the following groupings (somewhat abbreviated):
1. A.
2. B.
3. C.
4. D.
Soon after, a similar classification was introduced in the USA (FDA Pregnancy Categories) as follows (also somewhat abbreviated):
1. A.
2. B.
3. C.
4. D.
5. X.
6. N.
Other such classifications have been made, for instance, in Australia.
The problem with these and similar classifications is that they are rather blunt and will sometimes be of little help in the practical situation. There are two major problems:
1. 1.
2. 2.
For the evaluation of the situation for the individual patient, more specific information than a general pregnancy labeling is needed. The evaluation of a clinical pharmacologist may deviate from the crude classification (Erdeljic et al. 2010). Such information can be given by a TIS organization when such is available. Other efforts have been made, e.g., the Internet-based Janusinfo (in Swedish) which summarizes data from the Swedish Medical Birth Register and the scientific literature on individual drugs (Nörby et al. 2013). The site is written for medical professionals but has been much used also by the general public. It refers to the legal and health system in Sweden and may not be directly applicable in other countries.
An important issue is at what time during pregnancy the question turns up. A consultation of a woman who is planning a pregnancy and wonders if she can use a specific drug is reasonably simple. One should try to change drug therapy if a drug with definite or suspected hazards for the pregnancy and infant is used, e.g., valproic acid. When less strong and perhaps uncertain effects exist, like SSRI drugs, it may be possible to try to stop the drug use or at least to select the drug with the best track record.
In the most common situation, the woman turns up for consultation when she is in early pregnancy and then tells that she has already used a specific drug. Is this harmful for the embryo? If the question concerns a drug which has been studied in large and well-performed studies and has appeared harmless, obviously clear information should be given that the use of it does not endanger the baby. In this situation there is a not uncommon complication – the patient may search for information on the Internet and find a study which indicates a risk. One then has to explain why one does not believe in the published results. Perhaps this book can help.
If, on the other hand, a risk has (more or less certainly) been demonstrated, its significance has to be evaluated. If an exposure has already occurred, the question should be reformulated to the following: Is the risk so large and the damage so serious that an interruption of the pregnancy should be considered? This is a relatively rare situation, but on the other hand, it is likely that many pregnancies are unnecessarily interrupted because of anxiety for low or nonexisting risks. In some instances, a detailed search for congenital malformations by fetal diagnosis may be recommended because of a drug use. A typical example is use of valproic acid which – among other effects – increases the risk for spina bifida some 10–20 times, and this malformation can often be detected prenatally and a pregnancy interruption can be performed if the woman so wishes. Other ill effects of this drug can often not be detected prenatally but may be less severe.
One sometimes hears the woman saying that she does not want to take any risk that the baby will be malformed, but she should know that the risk always exists and if this risk is marginally increased, it is of little practical importance.
In most instances, the question of a continued use of the drug during pregnancy does not refer to the risk for malformations but to other fetal risks. We can illustrate this with two examples.
A woman with essential hypertension has become pregnant when using an ACE inhibitor. The continued use of the drug during pregnancy represents a risk for intrauterine death of the fetus, and the drug should be stopped and other antihypertensive drugs should be used.
A woman with depression has become pregnant using an SSRI drug. Continued use of this drug during pregnancy will increase the risk for neonatal morbidity which as a rule is temporary. If she stops using the drug, the risk for a worsening of her depression is considerable and includes an increased risk for post-delivery depression. It may be better to accept the infant risk. If, however, her mental condition is such that a trapping down of the treatment can be made, this is to be preferred, among other things because of a possible effect on the long-term development of the child.
As pointed out above, studies of the effect of drugs on the long-term development of the child are difficult to perform in an adequate way. Such effects are important and are most likely to occur after the use of psychoactive drugs. If such exposures can be avoided, this is obviously best, but it is a difficult balance between the (often hypothetical) risk of the child and the sometimes strong need of the woman of drug therapy. Even though maternal use of opioids during late pregnancy increases the risk of neonatal abstinence syndrome (NAS), it is possible that long-term effects seen in children of opioid-abusing women less are due to intrauterine drug exposure than to postnatal environmental factors (Baldacchino et al. 2014; Sithisarn et al. 2012). The well-known consequences of high alcohol exposure during pregnancy, resulting in fetal alcohol syndrome which includes also long-term neuropsychiatric effects, is a warning that such risks may exist also with drug exposure. To balance these risks against the needs of the woman is very difficult and necessitates careful considerations and access to data from well-performed studies. There is also a need of competent interpretation of published data. I will finish with some suggestions for such an evaluation.
13.3 Concluding Remarks: How to Evaluate a Published Study
For the evaluation of studies related to maternal drug use and pregnancy or neonatal complications, some important things have to be considered.
Is the study large enough to be able to detect a moderately increased risk for a serious outcome? To study malformation risks in a material of one or two hundred exposed infants can only reveal very strong risk increases, however carefully the study is made. Such small studies are of limited value and cannot prove a lack of teratogenic properties in the drug.
Does the study have the power to investigate effects on specific malformations? Even a study comprising thousands of exposed infants may not be large enough to study effects on rare malformations like spina bifida or gastroschisis.
Are the data obtained retrospectively or prospectively related to the outcome studied? Retrospective studies where drug exposure is ascertained after the appearance of the harm (e.g., birth of a malformed child) are apt to falsely identify or exaggerate risks. Much consideration should not be paid to this type of study.
Is drug exposure ascertained from prescription registers? Exposure data from prescription registers may bias the risk estimates toward null and may therefore be unable to identify moderate risks, but they can hardly give falsely increased risks.
Have adequate considerations been taken of possible confounding? The fact that adjustment has been made for a number of different variables does not necessarily mean that significant confounders have been adjusted for. Are the variables which have been adjusted for really confounders and not mediators? Is underlying disease a likely cause of the outcome studied?
How many statistical tests have been performed? Is a finding the result of a “fishing party” or is it a test of a hypothesis which was set up in advance of the data collection?
In Studies of Congenital Malformations
Is the malformation rate in a control group reasonable? If the rate of major malformation is below 2 % or above 5 %, ascertainment is probably inadequate. A very low rate indicates a poor ascertainment, while a very high rate indicates inclusion of minor or uncertain anomalies which may hide effects on truly major malformations.
Are malformations described in adequate detail so possible effects on specific conditions can be identified? Studies which report only unspecified congenital malformations or birth defects are of less value than studies which describe specific malformation types. Are the malformation descriptions based on information from specialists (pediatricians, child pathologists, geneticists, etc.) or from general practitioners or parents?
References
Baldacchino A, Arbuckle K, Petrie DJ, McCowan C (2014) Neurobehavioral consequences of chronic intrauterine opioid exposure in infants and preschool children: a systematic review and meta-analysis. BMC Pscyhiatry 14:104. http://www.biomedcentral.com/1471-244X/14/104
Berglund F, Flodh H, Lundberg P, Prame B, Sannerstedt R (1984) Drugs used during pregnancy and breast-feeding. A classification system for drug information. Acta Obst Gynecol Scand Suppl 126:1–55CrossRef
De Jonge L, Zetstra-van der Woude PA, Bos HJ, de Jong-van den Berg LTW, Bakker MK (2013) Identifying associations between maternal medication use and birth defects using a case-population approach. An exploratory study on signal detection. Drug Saf 36:1069–1078CrossRefPubMed
Edeljic V, Francetic I, Makar-Ausperger K, Liklic R, Radafic-Aumiler M (2010) Clinical pharmacology consultation: a better answer to safety issues of drug therapy during pregnancy? Eur J Clin Pharmacol 66:1037–1046CrossRef
FDA Pregnancy Categories. www.drugs.com/pregnancy-categories.html
Janus Info. www.janusinfo.se/beslutsstöd/lakemedel-och-fosterpaverkan/
Mitchell AA (2003) Systematic identification of drugs that cause birth defects – a new opportunity. N Engl J Med 349:2556–2559CrossRefPubMed
Nörby U, Källén K, Eiermann B, Korkmaz S, Winbladh B, Gustafsson LL (2013) Drugs and birth defects: a knowledge database providing risk assessments based on national health registers. Eur J Clin Pharmacol 69:889–899CrossRefPubMed
Sithisarn T, Granger DT, Bada HS (2012) Consequences of prenatal substance use. Int J Adolesc Med Health 24:105–112CrossRefPubMed