Risk management of teratogenic medicines: A systematic review

To systematically identify studies of implementing risk management measures when prescribing teratogenic medicines for women of childbearing age and studies reporting risk perceptions of teratogenic medications.


| INTRODUCTION
A teratogen is a substance that can adversely affect the development of an embryo or a foetus if administered under specific conditions of dose, route of administration, gestational age, and genotype (Bánhidy, Lowry, & Czeizel, 2005). A wide range of substances have been recognized as teratogens, including some medications (Holmes, 2011; Twining's Textbook of Fetal Abnormalities, A. M. Coady & S. Bower, 2014). There is a need to ensure that potential teratogens are used as safely as possible by women of childbearing age, because the use of teratogenic medications is likely to be inevitable in many cases due to the unavailability of equally effective alternative treatment options (Honein, Moore, & Erickson, 2004;E. B. Schwarz, Parisi, Handler, et al., 2012).
To minimize foetal harm when prescribing potential teratogens, risk management programmes have also been developed for certain medications, with the manufacturer of isotretinoin launching the first pregnancy prevention programme aimed at preventing foetal exposure in 1988 (Honein et al., 2004;Mayall & Banerjee, 2014). Subsequently, the use of teratogenic medications has been increasingly controlled through the development of risk minimisation activities and programmes (Mayall & Banerjee, 2014). Elements to ensure safe use of teratogenic medications include certification of prescribers and dispensers, patient counseling regarding contraception use and monitoring patient contraception behaviors through regular pregnancy testing and use of contraception (FDA, n.d.; Mayall & Banerjee, 2014).
The development and implementation of teratogenic risk management programmes should also take into consideration patients' experience of using a medication (Bwire, Freeman, & Houn, 2011). The value of recognizing patients' experience of medication-taking as part of ensuring medications are used effectively and deliver intended outcomes is one of the principles of medicines optimisation, a model for informing pharmacy practice based on the aim of improving outcomes of medication use. The four guiding principles of medicines optimisation are: aim to understand the patient experience; evidence-based choice of medicines; ensure medicines use is as safe as possible and make medicines optimisation part of routine practice (Royal Pharmaceutical Society, 2013). Medicines optimisation is a patient-centred approach for achieving optimal use of medications by providing personalized care for each patient (Cutler, Fattah, Shaw, & Cutts, 2011). Conceptualized in terms of medicines optimisation, with the patient at the centre of healthcare, patients' views, opinions, and perceptions of taking a teratogenic medicine, and understanding of teratogenic risk, are therefore important factors when investigating the effectiveness of any risk management programme (Collins & Bonneh-Barkay, 2016;Widnes & Schjott, 2017). Moreover, because a key actor in ensuring evidence-based choice of medications are healthcare providers, these stakeholders' perceptions of teratogenic risk will play a part in understanding patients' experience of using the medication (Bwire et al., 2011). In fact, evidence from the literature suggests that the patient-physician relationship and teratogenic risk communication have a significant impact on patients' medication utilization (Widnes & Schjott, 2017). In this context, over-estimation of teratogenic risk may result in poor adherence to treatment during pregnancy, anxiety or pregnancy termination, while under-estimation of teratogenic risk can result in foetal exposure to the harmful effects of a teratogenic medication (Gils, Pottegard, Ennis, & Damkier, 2016;Sanz, Gomez-Lopez, & Martinez-Quintas, 2001;E. B. Schwarz, Maselli, Norton, & Gonzales, 2005).
A growing body of literature has investigated the implementation of pregnancy prevention measures while prescribing teratogenic medications to women of childbearing age (Brandenburg et al., 2017;Hayward et al., 2016;Leverenz et al., 2019;Paton et al., 2018;Uuskula et al., 2018). Additionally, research has focused on the perceived risk of teratogenic medications of various populations (Ceulemans et al., 2019;Gils et al., 2016;Petersen, McCrea, Lupattelli, & Nordeng, 2015). Yet to date, what is lacking is a systematic synthesis of data from a medicines optimisation perspective that explores teratogenic medication safety by systematically reviewing publications on the implementation of risk management (pregnancy prevention) measures when prescribing teratogens to women of childbearing age in combination with a review of patients' experience of using teratogenic medications in terms of reported perceptions of teratogenic risk.

| METHODS
The protocol for this systematic review was registered with PROSPERO (International Prospective Register of Systematic Reviews) registration number CRD42019142944 (Shrouk, Steinke, & Willis, 2019). The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement has guided the write-up of this review paper (Moher, Liberati, Tetzlaff, Altman, & Group, 2009).

| Information sources
Five electronic databases were systematically searched (MEDLINE, Cumulative Index to Nursing and Allied Health Literature (CINAHL), Scopus, Embase, and International Pharmaceutical Abstracts (IPA)). The search strategy was based on Medical Subject Headings (MeSH) and free text keywords in each database. Search terms included: pregnancy prevent*, risk manag*, teratogen*, risk perception and perceiv*. No limits were applied to publication dates. Papers not written in English and conference abstracts were excluded. Reference lists of included articles were manually screened to identify additional papers for inclusion in the review. The full search strategy is available in Appendix A.
2.2 | Inclusion/exclusion criteria 1. Risk management: the review aimed to establish the implementation of risk management for teratogenic medications used by women of childbearing age. Therefore, papers were included in the review if they reported the use of at least one teratogenic medication by females of childbearing age and the implementation of at least one risk minimisation measure consistent with the Food and Drug Administration (FDA) Risk Evaluation and Mitigation Strategies (REMS) drug safety programme designed to address safety concerns. Most of the included medicines were established teratogens. However, some included medicines were considered potential teratogens like deferoxamine and deferiprone, cyclophosphamide, and mood stabilizers. Safety of teratogenic medications use was further monitored against Elements to Assure Safe Use (ETASU), see Table 1 (Arnold, 2019;Dabrowska, 2018).
Papers focused on the following were excluded from the review: pregnancy rates while using a teratogenic medication, pregnancy outcomes after exposure to a teratogenic medication, emergency contraception, contraception due to a medical condition (medical conditions that require contraception regardless of the use of drugs), abortion or pregnancy termination, side effects of contraceptives, venous thromboembolism, teratogenic risk of contraceptives, prescription patterns during pregnancy and need for contraceptive due to HIV infection. 2. Perceptions of teratogenic risk: to address the aim of reviewing publications reporting how teratogenic risk of medications is perceived, papers included in the review were those reporting perceived teratogenic risk as numerical value. Studies reporting the perception of risk of non-teratogenic medications were excluded from the review

| Study selection
All papers identified through the database search and through manual search of reference lists were checked to remove any duplicates. Following this, study selection was carried out through three phases. First, all titles were screened against the inclusion and exclusion criteria. Second, abstracts from articles selected in the first phase were screened against the same criteria. Third, full texts of articles retrieved from the second phase were reviewed to check for eligibility for inclusion in the review.

| Data extraction
Data were extracted by the three authors and synthesized into summary tables presenting information about study characteristics, methods and outcomes of interest. For the part covering perceptions of teratogenic risk, the authors aimed to examine whether the results of the included studies indicated a true (proper) estimation of the teratogenic risk or not. If not, the perceived estimations of the teratogenic risk were examined to see if they were higher than the true risk (over-estimated) or lower than the true risk (under-estimated). Therefore, the numeric value of the perceived teratogenic risk extracted from the results section of each paper and presented as a percentage was compared to the true value of the teratogenic risk that was extracted either from the methods or the results section. The true value of the teratogenic risk was described in the included studies as the risk of causing "congenital malformations" or "birth defects" and was also presented as a percentage to allow comparison with the perceived risk. No description of specific outcomes was included and the risk was described in general terms. The values were based on the available literature on the risk of the included medicines. The authors followed the method shown in Figure 1 to assign the perceived teratogenic risk of every medication included in the review into one of three categories: properly-estimated, over-estimated, or under-estimated.

| Quality assessment
The critical appraisal tool developed by Hawker et al. (Hawker, Payne, Kerr, Hardey, & Powell, 2002) was used to assess the quality of included papers. The tool includes nine questions to assess the abstract and title, introduction and aims, method and data, sampling, data analysis, ethics and bias, results, transferability or generalisability, and implication and usefulness. Each question has four options: good, fair, poor or very poor, scored from 1 (very poor) to 4 (good). This scoring method therefore allows for a total score to be calculated ranging from 9 to 36 for each paper. Based on the total score for each paper, four quality categories were applied to each paper as follows: high quality (score of 30-36), medium quality (score of 24-29), and low quality (score of 9-24).
Results of the quality assessment were not used as inclusion/exclusion criteria.

| Data analysis
Extracted data were presented in summary tables. In addition, the following analyses were carried out: 1. Risk management: For each risk minimisation measure reported, prevalence of implementing that measure was calculated as the proportion of a study population reported to be using a measure. This was calculated as follows: Prevalence of risk minimisation measure implementation/ 100 patients = total number of patients implementing the measure / total number of patients using the teratogenic medication × 100 2. Perceptions of teratogenic risk: The perceived teratogenic risk for each medication was assigned into one of three categories: properly-estimated, over-estimated, or under-estimated (as shown previously in Figure 1). Categorized results were presented in tabular form.

| Risk management
A total of 55 studies were included in the review as shown in Figure 2. Characteristics of the included studies are shown in Table 2 and summarized in Appendix B. Table 3 presents a summary of the risk minimisation measures reported by studies of prescribing teratogenic medications for women of child bearing age. No studies reported on all aspects of risk management included in the current review. Prevalence of teratogenic counseling ranged from 9.5% (Mulryan et al., 2018) to 99.3% (Brandenburg et al., 2017), contraceptive counseling from 6.1% (E. B. Schwarz et al., 2005) to 98% (Brandenburg et al., 2017), pregnancy F I G U R E 1 Method for determining category for perceived teratogenic risk testing before starting treatment from 0% (Chave et al., 2001;Mulryan et al., 2018) to 95.1% (Cheetham et al., 2006), pregnancy testing during treatment from 12.7% (Raguideau et al., 2015) to 100% (Hayward et al., 2016), contraception use before starting treatment from 15.7% (Uuskula et al., 2018) to 94% (Brandenburg et al., 2017), and contraception use during treatment from 1.7% (Tsur et al., 2008) to 100% (Ozyurt & Kaptanoglu, 2015).

| Perceptions of teratogenic risk
A total of 6,000 articles were initially screened. Of those, 141 were removed because of duplication, 5,725 were excluded based on title screening, 68 were excluded based on abstract screening, and 59 were excluded based on full text screening leaving a total of seven articles to be included in the review (see Figure 3). Characteristics of the seven included papers are shown in Table 4.
A proper estimation of the teratogenic risk was reported for thalidomide (by general practitioners and obstetric/gynecologists) (Gils et al., 2016) misoprostol (by pregnant and nonpregnant women) (Pons Eda et al., 2014). An under-estimation of the teratogenic risk was reported for warfarin and retinoids (by general practitioners and obstetric/gynecologists) (Gils et al., 2016). And over-estimation of the teratogenic risk was reported for thalidomide (by pregnant and nonpregnant women, healthcare professionals, and medical students) (Damase-Michel et al., 2008;Lupattelli et al., 2014;Nordeng et al., 2010;Petersen et al., 2015;Sanz et al., 2001), for valproate, lithium, isotretinoin, and warfarin (by healthcare professionals) (Damase-Michel et al., 2008), for phenytoin and warfarin (by pregnant and nonpregnant women, healthcare professionals, and medical students) (Sanz et al., 2001), and for etretinate (by nonpregnant women, healthcare professionals, and medical students) (Sanz et al., 2001). Details are presented in Table 5.

| DISCUSSION
Guided by principles of medicines optimisation (Royal Pharmaceutical Society, 2013), to our knowledge this is the first systematic review that synthesizes the available literature on the safe use of teratogenic medications. Additionally, this review extends our understanding of patients' experience of using teratogenic medications by systematically summarizing published studies that report perceptions of potential teratogens.

| Risk management
Measures to minimize foetal exposure to potential teratogens investigated in this review were based on components of Risk Evaluation and Mitigation Strategies (REMS) with Elements to Assure Safe Use (ETASU). These measures were: teratogenic counseling, contraceptive counseling, pregnancy testing before or at start of treatment, pregnancy testing while on treatment, use of contraception before or on starting treatment, and use of contraception during treatment. Since 2007, implementation of REMS with ETASU have been required by the FDA for medications with serious safety issues like teratogenic medications to ensure that the benefits of a medication outweigh the risks to patients (Leiderman, 2009 Ozyurt & Kaptanoglu, 2015;Pinheiro et al., 2013;Rao et al., 2000;Teichert et al., 2010;Tsur et al., 2008;Uuskula et al., 2018;Werner et al., 2014). This may be because of two reasons. Firstly, isotretinoin is a relatively old medication that has been in the market since 1982-1983, and has been prescribed under a pregnancy prevention programme since 1988 (Crijns, Straus, Gispen-de Wied, & de Jong-van den Berg, 2011). Secondly, it is one of the most cost-effective acne treatments used by patients from different age groups including women of childbearing age (Algoblan et al., 2019;Bérard et al., 2007;Honein et al., 2004). By contrast, it was observed that there were fewer publications on medications prescribed under more recent risk management programmes such as thalidomide, lenalidomide, and valproic acid (Atturu & Odelola, 2015;Brandenburg et al., 2017;Castaneda et al., 2008;Chave et al., 2001;Gotlib et al., 2016;Mulryan et al., 2018;Paton et al., 2018), indicating a need for further investigation of the safety of these medications in terms of adherence to risk management measures for such medications. Good practice guidance suggests that ensuring safe use of medications can have a number of positive effects on treatment outcomes. For teratogenic medications in particular, this includes reducing the incidents of medication-induced foetal harm and empowering patients to make the most of their treatment (Royal Pharmaceutical Society, 2013).
Of the studies included in the review, 11 publications used the FDA pregnancy labeling categories (A, B, C, D, X) (DiPietro Mager et al., 2018;Fritsche et al., 2011;Goyal et al., 2015;Mody, Farala, Wu, Felix, & Chambers, 2015;Ruiter et al., 2012;E. B. Schwarz et al., 2010;E. B. Schwarz et al., 2005;E. B. Schwarz et al., 2007;Stancil et al., 2016;Steinkellner et al., 2010). However, it is worth mentioning that the FDA requested removal of these categories from the labels of all prescription drugs and biological products in 2015. The pregnancy labeling categories (A, B, C, D, X) were replaced by the Pregnancy and Lactation labeling Rule (PLLR) since 2015 ("Content and format of labeling for human prescription drug and biological products; requirements for pregnancy and lactation labeling. Final rule," FDA, 2014; Pernia & DeMaagd, 2016). Reasons behind this change in labeling included concerns regarding the clarity of the content of the old pregnancy categories labeling, a possibility of misinterpreting the categories, inability to provide significant information about drug exposure during pregnancy, and inability to identify the consequences of stopping the use of needed drugs during pregnancy (Pernia & DeMaagd, 2016).  Results of teratogenic risk management implementation showed a wide variation across studies. Some studies reported surprisingly low rates of implementation. For example, only 9.5% of women of childbearing age using valproate received teratogenic counseling in the study by Mulryan et al. (Mulryan et al., 2018). Additionally, rates of pregnancy testing before starting treatment with valproate or thalidomide were as low as 0% in two studies (Chave et al., 2001;Mulryan et al., 2018), and pregnancy testing during treatment with acitretin was 12.7% in the study by Raguideau et al. (Raguideau et al., 2015). Low rates of contraceptive use were also reported. Uuskula et al. reported that 15.7% of women of childbearing age on isotretinoin treatment in their study used a contraceptive before starting treatment (Uuskula et al., 2018), and Tsur et al. reported that only 1.7% of women in their study group used contraception during treatment with isotretinoin (Tsur et al., 2008).

Medications included
The true value of teratogenic risk (%) The perceived value of teratogenic risk (%) The perceived value compared to the true value of teratogenic risk  (Schmier & Halpern, 2004). Another form of self-reporting bias associated with the disclosure of sensitive data is the social desirability bias (Althubaiti, 2016). Social desirability bias might have led to an overestimated adherence to risk management and pregnancy prevention measures ("Social Desirability Bias,"). On the other hand, if data were extracted from the medical records, several issues like incomplete records, noncaptured data, and low quality data might have an effect on the research outcomes (Herrett et al., 2015;Zozus et al., 2015). Therefore, it is important to bear in mind the possible types of bias associated with each source of data. Another well recognized variable leading to variations in the implementation of risk management for the different teratogenic medications is the availability of risk management programmes. For certain medications like thalidomide, linaledomide, and isotretinoin, detailed risk management programmes that aim to prevent foetal exposure to the drug are in place (Honein et al., 2004;Mayall & Banerjee, 2014). However, for other teratogenic medications, managing their risk is limited to the use of product labeling and patient information leaflets rather than rigorous monitoring (Freeman, Bwire, Houn, Sheehan, & Backstrom, 2014). The effectiveness of drug labeling as a risk management tool has been a matter of debate as research suggests a lack of effect on physicians' prescribing behaviors or patients' understanding of instructions (Freeman et al., 2014).
Results of the current review can be considered as a compliance assessment of teratogenic risk management (whether through existing risk management programmes or through labeling recommendations) (Freeman et al., 2014). Based on the findings of this review, safety of the utilization of teratogenic medications is sub-optimal, and entails a risk of foetal exposure to the harmful effect of potential teratogens. Consequently, it is recommended that the implementation of the existing teratogenic risk management programmes be monitored more carefully, and the criteria for the optimal management of teratogenic risk for potential teratogens be reviewed and revised based on the available evidence.
Exploring the implementation of risk management for teratogenic medications can help to develop interventions designed to minimize foetal exposure to cytotoxic effects, and thus future research utilizing multiple data sources is needed. Drawing on the strengths of data extracted from medical records and patient reported data, mixed methods research that utilizes quantitative and qualitative methods could yield more rigorous results than research utilizing quantitative or qualitative methods alone (Shorten & Smith, 2017;Tisnado et al., 2006).
Consequently, results of this review raise two important issues. First, the review uncovers deficiencies in the implementation of risk management of teratogenic medications which constitutes a serious public health concern that needs further investigation. Second, it highlights a potential need to reinforce policies and regulations that aim to reduce foetal exposure to the cytotoxic effects of teratogenic medications.

| Perceptions of teratogenic risk
To help patients get the most from their treatment, it is important that their experience of medication use be explored and understood. In recent years, there has been an increasing interest in research on the perception of teratogenic risk (Sanz et al., 2001). This is corroborated by results of the current review, which shows that all papers included were published only in the last two decades. Additionally, the relatively small number of studies included in the review (seven studies) indicates that the study of perceptions of teratogenic risk is an important area for further research.
Two methods were used to measure the perception of teratogenic risk of participants, and those were either a numeric scale (Gils et al., 2016;Lupattelli et al., 2014;Nordeng et al., 2010;Petersen et al., 2015;Pons Eda et al., 2014) or a visual analogue scale (Damase-Michel et al., 2008;Sanz et al., 2001). One major issue regarding the use of a numeric scale to estimate the risk is its dependence on numeracy skills of participants ( , 2014). Evidence from the literature shows that correct estimation and understanding of health related risk information is significantly correlated with an ability to understand numbers and mathematical concepts (Peters et al., 2007;Rothman, Montori, Cherrington, & Pignone, 2008). The second method to measure the perception of teratogenic risk was the use of a visual analogue scale. There is an ongoing debate on the utility of visual analogue scales in measuring risk perception. Some argue that responses of participants to questions including a visual analogue scale tend to cluster around the middle point of the scale and might over-estimate the risk when it is low (Pons, Guimarães, Knauth, & Pizzol, 2014;Sanz et al., 2001), while others suggest that a visual analogue scale can provide a wide range of responses that can be chosen by research participants (Harland, Dawkin, & Martin, 2015). Pons et al. investigated the level of agreement between a visual analogue scale and a numeric scale in estimating the teratogenic risk. In their research, they concluded that there was no agreement between the two methods in estimating teratogenic risk. . Furthermore, it is recommended that future research exploring perceptions of teratogenic risk needs to utilize qualitative methods in addition to quantitative research. This is one way to overcome the ongoing controversy regarding how to reliably measure perception of teratogenic risk and will provide a deeper understanding of how risk is perceived (Shorten & Smith, 2017). It is clear from the results of the review that teratogenic risk of medications tends to be over-estimated (Damase-Michel et al., 2008;Lupattelli et al., 2014;Nordeng et al., 2010;Petersen et al., 2015;Sanz et al., 2001), while proper estimation (Gils et al., 2016;Pons Eda et al., 2014) or under-estimation (Gils et al., 2016) occurs less frequently. Yet while there is agreement in the literature about the difficulty of understanding the teratogenic risk of medications due to scientific uncertainty (Polifka, Faustman, & Neil, 1997;Twigg, Lupattelli, & Nordeng, 2016), a realistic perception of teratogenic risk is needed by women in childbearing age to adhere to their therapy (Lennon, 2016).
Over-estimating the teratogenic risk of medications might be due to several factors. For women, pregnancy is viewed as a sensitive phase of their lives which can be easily adversely affected by exposure to a number of teratogens (such as alcohol) and including medications. In addition, pregnancy entails a significant responsibility to the mother to keep her foetus as safe as possible. These attitudes are further emphasized by social norms and cultural beliefs and can affect women's ideas about medications (Twigg et al., 2016;Widnes & Schjott, 2017;Widnes, Schjott, Eide, & Granas, 2013). On the other hand, for health care professionals and particularly for physicians, exaggerating the teratogenic risk of medications can be a result of inadequate knowledge, which in turn might be the result of insufficient training and education provided for physicians (Damase-Michel et al., 2008), or the lack of relevant resources being utilized when needed (E. B. Schwarz et al., 2009). Furthermore, physicians' fear of legal liability or possible accusation of malpractice if anything goes wrong while prescribing a potential teratogen might underpin this over estimation of the teratogenic risk of medications (E. B. Schwarz et al., 2009). Subsequently, future research needs to focus on understanding how teratogenic risk is conceptualized and the reasons behind the tendency to exaggerate it.
The strength of this review relies in being the first attempt to shed light on the current status of implementing risk management measures when teratogenic medications are prescribed to women of childbearing age. It utilizes the principles of medicines optimisation, a paradigm that aims to help patients get the best outcomes from using medicines. However, this systematic review has some limitations. First, title and abstract screening were only carried out by one researcher which means that there is a possibility of missing publications. Second, for the section on perceptions of teratogenic risk, the number of included articles was relatively small, which is justified by the limited publications in this area.

| CONCLUSION
Considerable variation in the implementation of risk management measures when prescribing teratogenic medications to women of childbearing age is reported in the literature. Factors contributing to this variation require further investigation to understand barriers and facilitators of teratogenic medication risk management within a health system. Further studies of risk management of teratogenic medications, which take these factors into account, will need to be undertaken.
Additionally, a common tendency to over-estimate the risk of teratogenic medications was observed. To achieve the best possible therapeutic outcomes of using teratogenic medications, there is a need to explore the reasons behind this over-estimation. Understanding how teratogenic risk is conceptualized can usefully inform medicines optimisation so that patients derive the intended outcomes of a prescribed medication.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.