Reference: Ann Intern Med. 2022 Mar 29 early online
Clinicians routinely assess risk of medication use in the preconception and antenatal period for females of reproductive age. Medications associated with birth defects are monitored and discontinued if there is a small but significant risk. Studies of the risk of paternal prescription and use of medication are limited, however. A group of Danish investigators conducted a nationwide prospective cohort study to examine the association between paternal use of pharmacological therapy for diabetes and birth defects, leading to widespread and alarming media headlines.
National registry data on live singleton births from 1997-2016 was linked to parental health data and prescription fills from patients hospitalized or seen by specialists between 1995-2018. Researchers gathered information on date of conception, occurrence of birth defects and health outcomes during the first year of life, and parental health diagnoses and prescriptions. The authors excluded infants of mothers with diabetes or prescriptions for diabetic therapy. The primary outcome of interest was the diagnosis of one or more birth defects in the first year of life.
Of the over 1.1 million recorded live births analyzed, 3.2% (36,585) had a major birth defect. A total of 7,029 offspring were exposed to diabetic medications through paternal use in the three months prior to conception (the physiologic window for semen development). The majority were prescribed insulin therapy (75.4% [5,298]), while rates of metformin and sulfonylurea prescription were lower (20.6% [1,451] and 9.2% [647], respectively). Most (84%) taking insulin had type 1 diabetes. Patients designated as “fathers” on oral diabetic therapy had a higher median age and prescription rates for cardiovascular medicines and antidepressants compared with those on no therapy or insulin. The rate of a major birth defect in the insulin-exposed group was 3.3%, compared with 5.2% in the metformin-exposed and 5.1% in the sulfonylurea-exposed groups. After adjusting for birth year, paternal age, education, income level, maternal age, education, and smoking status, the adjusted odds ratio (aOR) for birth defects was 1.40 (95% CI 1.08-1.82) in the metformin group and 1.34 (95% CI 0.94-1.92) in the sulfonylurea group compared with the insulin group. In an analysis that included fathers younger than age 40, the aOR was 1.37 (95% CI 0.97-1.94). After adjusting for cardiovascular and antidepressant prescriptions, preconception diabetes diagnosis, and parental birth defects, analyses still showed an association between paternal metformin prescription and major birth defects. In an attempt to determine if unmeasured confounding was present, the authors calculated an E-value. An E-value is the minimum risk ratio of a potential cofounder, beyond the measured covariates, that could explain the association between an exposure (metformin) and an outcome (birth defects, in this case). A smaller E-value therefore implies that little unmeasured confounding would be needed to explain away a given finding. There is no defined threshold for a significant E-value; it must be assessed in the overall context of the study results. The E-value to reduce confidence interval estimate for the association between metformin and birth defects to the null was 1.37.
Media headlines would lead you to believe that we should start asking about preconception paternal metformin use and warning parents-to-be about the potential hams. But should we be? It seems this study is really comparing apples to oranges. Most of those receiving insulin therapy had type 1 diabetes, which represents a population with a different set of risk factors and a different set of genetics than those with type 2 diabetes. The study lacks a clear, separate analysis of these two populations. Next, general practitioner data was not included, only those referred to a specialist. This can lead to referral bias, reflecting a potentially sicker population on more medications. Additionally, when including only those live births with fathers younger than 40, the association between metformin and birth defects crossed the null. Finally, the E-value of 1.37 seems low to us, which may indicate that some unmeasured confounder is influencing the results. This article raises concern about paternal metformin use, but our real concern is the potential overreaction of patients and providers resulting in discontinuation of this commonly used, effective, and affordable medication for a small risk that might not even exist.
For more information, see the topic Metformin for Diabetes in DynaMed.
DynaMed EBM Focus Editorial Team
This EBM Focus was written by Carina Brown, MD, Assistant Professor at Cone Health Family Medicine Residency. Edited by Alan Ehrlich, MD, Executive Editor at DynaMed and Associate Professor in Family Medicine at the University of Massachusetts Medical School; Katharine DeGeorge, MD, MS, Deputy Editor at DynaMed and Associate Professor of Family Medicine at the University of Virginia; Dan Randall, MD, Deputy Editor at DynaMed; Nicole Jensen, MD, Family Physician at WholeHealth Medical; Vincent Lemaitre, PhD, Senior Medical Writer at DynaMed; and Sarah Hill, MSc, Associate Editor at DynaMed.