Results
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. More than half of the included studies (n=29; 52.7%) were conducted in the USA [3, 17-20, 30-53], nine (16.4%) originated from the UK [21, 54-61], and the rest were from Canada (n=3; 5.5%) [62-64], Netherlands (n=3; 5.5%)[65-67], Poland (n=2;3.6%) [68, 69], Ireland (n=1;1.8%) [70], Belgium (n=1;1.8%) [71], Estonia (n=1;1.8%) [22], France (n=1;1.8%) [72], Iran (n=1;1.8%) [73], Turkey (n=1;1.8%) [74], Israel (n=1;1.8%) [75], Uganda (n=1;1.8%) [76], and Saudi Arabia (n=1;1.8%) [77]. Publication dates ranged from 1988 to 2019, with nearly two thirds (n=35, 63.6%) published after 2010 [3, 18-22, 34-39, 41-43, 46, 48-53, 58, 61, 65, 67-74, 76, 77].
Data sources included medical records (n=26; 47.3%) [17, 19, 21, 22, 31, 32, 34, 36, 37, 40, 41, 43, 44, 46, 49, 54, 57, 58, 60, 61, 65-67, 70, 72, 78], patient surveys (n=21; 38.2%) [18, 20, 30, 33, 35, 39, 45, 47, 48, 52, 53, 55, 56, 62, 64, 69, 71, 74-77] or a combination of patient surveys and medical records (n=5; 9.1%) [38, 50, 63, 68, 73]; patient logs [42], reproductive life plans [51], and physician surveys [59] were used less frequently (n= 3; 5.4%). Most studies (n=16; 29.1%) investigated the use of multiple teratogenic medications [17, 32, 36-40, 43, 44, 48-51, 60, 65, 78]. Sixteen studies (29.1%) reported on the use of isotretinoin [22, 30, 31, 33, 34, 47, 53, 63, 64, 66, 67, 71, 73-75, 77], five (9.1%) reported on the use of antiepileptic or anticonvulsant medications [46, 54, 56, 58, 68], four (7.3%) reported on the use of arthritis or lupus medications [20, 35, 52, 69], four (7.3%) reported on the use of valproate/valproic acid [21, 41, 61, 70], and three (5.5%) reported on the use of thalidomide or lenalidomide [18, 45, 59]. Each of the following medications was reported by one study (1.8%) in the review: acitretin [72], chemotherapy for breast cancer [55], deferoxamine and deferiprone [62], cyclophosphamide [19], isotretinoin and oral contraceptives [42], mood stabilisers [57] and warfarin [76]. A risk management programme or a pregnancy risk classification system was reported in 28 studies (50.9%) [17, 18, 21, 30-32, 34, 36, 37, 39-45, 47-51, 53, 54, 56, 65, 66, 71]. More than half of the studies (n=31; 56.4%) were of medium quality [17, 18, 21, 30-34, 36, 41, 43, 45, 46, 51, 54-57, 59, 61-63, 65, 67-72, 74, 77], while 23 (41.8%) were of high quality [19, 20, 22, 35, 37-40, 42, 44, 47-50, 52, 53, 58, 64, 66, 73, 75, 76, 78] and one study (1.8%) was of low quality [60].
Teratogenic counselling was reported in 19 studies (34.5%) [18, 21, 30, 38, 39, 41, 46, 50, 54, 56-58, 61, 64, 68, 70, 75-77], contraceptive counselling in 22 studies (40%) [17, 18, 21, 30, 35, 38, 40, 43, 46, 50, 52-55, 57, 58, 64, 68, 70, 71, 77], pregnancy testing before starting treatment in 10 studies (18.2%) [30, 31, 47, 48, 59, 64, 70-72, 75], pregnancy testing during treatment in eight studies (14.5%) [19, 44, 57, 64, 71, 72, 75, 77], contraception use before starting treatment in four studies (7.3%) [18, 22, 30, 55], and contraception use during treatment in 35 studies (63.6%) [18, 20-22, 32-37, 40-42, 44-46, 49-51, 55, 57, 60, 62-69, 73-76, 78]. No studies reported on all aspects of risk management included in the current review, see Table 3.
Prevalence of teratogenic counselling ranged from 9.5% [70] to 99.3% [18], contraceptive counselling from 6.1% [17] to 98% [18], pregnancy testing before starting treatment from 0% [59, 70] to 95.1% [31], pregnancy testing during treatment from 12.7% [72] to 100% [19], contraception use before starting treatment from 15.7% [22] to 94% [18], and contraception use during treatment from 1.7% [75] to 100% [74].
Perceptions of teratogenic risk
A total of 6000 articles were initially screened. Of those, 141 were removed because of duplication, 5725 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.
Two studies out of seven (28.6%) included multiple countries [24, 79], and the rest were from Denmark (n=1; 14.3%) [15], Norway (n=1; 14.3%) [80], France (n=1; 14.3%) [81], Spain (n=1; 14.3%) [16], and Brazil (n=1; 14.3%) [82]. All studies had a cross sectional design and were published after the year 2000. Four studies out of seven (57.1%) utilised online questionnaires for data collection [15, 24, 79, 80], two studies used questionnaires filled during a continuous educational course [16, 81], and one study collected data within prenatal services in primary care [82]. Data were collected using questionnaires in all studies. A numeric scale was used to measure the perception of teratogenic risk in five studies out of seven (71.4%) [15, 24, 79, 80, 82] and a visual analogue scale was used in two studies (28.6%) [16, 81]. Five studies (71.4%) [15, 24, 79, 80, 82] were of high quality and two (28.6%) were of medium quality [16, 81].
A proper estimation of the teratogenic risk was reported for thalidomide (by general practitioners and obstetric/gynaecologists) [15], for etretinate (by pregnant women) [16], and for misoprostol (by pregnant and non-pregnant women) [82]. An under-estimation of the teratogenic risk was reported for warfarin and retinoids (by general practitioners and obstetric/gynaecologists) [15]. And over-estimation of the teratogenic risk was reported for thalidomide (by pregnant and non-pregnant women, healthcare professionals, and medical students) [16, 24, 79-81], for valproate, lithium, isotretinoin, and warfarin (by healthcare professionals) [81], for phenytoin and warfarin (by pregnant and non-pregnant women, healthcare professionals, and medical students) [16], and for etretinate (by non-pregnant women, healthcare professionals, and medical students) [16]. Details are presented in Table 5.