Do the children born after in vitro fertilization or intracytoplasmic sperm injection have poor intellectual performance? A prospective cohort study

In total, 639 children were included in our study. The baseline characteristics were shown in Table  1 . Maternal age was significantly higher in IVF/ICSI group [median (interquartile range): 28 (26, 30) vs. 27 (26, 30)), P  = 0.03)]. Additionally, the other varieties, including infertility duration (years), body mass index, educational level, the incidence of pregnancy complications and the use of exogenous folic acid were comparable between two groups. Paternal age was also higher in IVF/ICSI group than for the control group [median (interquartile range): 29 (27, 31) vs. 28 (26, 31), ( P  = 0.005)]. In addition, fathers in IVF/ICSI group emerged with a significantly higher BMI [median (interquartile range): 24.2 (22.0, 27.1) vs. 23.6 (21.2, 26.6), ( P  = 0.04)] and a higher proportion of smoking [number (percentage): 158 (37.1) vs. 59 (27.7), ( P  = 0.02)]. The main cause of infertility of IVF/ICSI group was tubal factor [169 (39.7%)]. Obviously, the infertility of AID group was totally attributed to the male factor. Table 1 Characteristics of the study population AID group IVF/ICSI group P value No. of participants 213 426 Maternal characteristics  Age, years 27 (26, 30) 28 (26, 30) 0.03  Infertility duration, years 3 (2, 4) 3 (2, 5) 0.26  BMI, kg/m2 22.0 (20.6, 23.4) 22.0 (20.3, 23.9) 0.89 Education, n (%) 0.80  Junior high and below 118 (55.4) 224 (52.6)  Senior high school 51 (23.9) 109 (25.6)  College and above College and above 44 (20.7) 93 (21.8) Causes of infertility, n (%) -  Tubal infertility - 169 (39.6)  PCOS - 29 (6.8)  Endometriosis - 9 (2.1)  Male infertility 213 (100.0) 135 (31.6)  Unexplained infertility - 80 (18.7)  Mixed factor - 4 (10.9)  Hypertensive disorders of pregnancy, n (%) 10 (4.7) 12 (2.8) 0.22  Gestational diabetes, n (%) 5 (2.4) 11 (2.6) 0.86  Vitamin E during pregnancy, n (%) 187 (87.8) 375 (88.0) 0.93  Folic acid during pregnancy, n (%) 160 (75.1) 289 (67.8) 0.06  DHA during pregnancy, n (%) 206 (96.7) 417 (97.9) 0.37 Paternal characteristics  Age, years 28 (26, 31) 29 (27, 31) 0.005  BMI, kg/m 2 23.6 (21.2, 26.6) 24.2 (22.0, 27.1) 0.04  Smoking, n (%) 59 (27.7) 158 (37.1) 0.02  Drinking, n (%) 2 (0.9) 7 (1.6) 0.72 Education, n (%) 0.25  Junior high and below 90 (42.3) 178 (41.8)  Senior high school 80 (37.6) 139 (32.6)  College and above College and above 43 (20.2) 109 (25.6) Average monthly family income per capita, n (%) 0.20  <3000 130 (61.0) 261 (61.3)  3000–4999 71 (33.3) 125 (29.3)  ≥ 5000 12(5.6) 40 (9.4) ART characteristics Fertilization mode, n (%) -  IVF - 292 (68.5)  ICSI - 134 (31.5)  Embryo mode -  Fresh embryo - 312 (73.2)  Frozen embryo - 114 (26.8) Stage of embryo transferred, n (%) -  Cleavage embryo - 325 (78.5)  Blastocyst embryo - 89 (21.5) Neonatal characteristics  Age, months 85 (82, 95) 85 (80, 102) 0.70  Male, n (%) 92 (43.2) 198 (46.5) 0.43  Birthweight, g 3.5 (3.2, 3.9) 3.8 (3.2, 3.8) 0.63 Gestational age at birth, n (%) 0.64  28–36 weeks 6 (2.8) 15 (3.5)  37–42 weeks 207 (97.2) 411 (96.5) Caregiver characteristics Caregiver, n (%) 0.29  Parents 183 (85.9) 253 (82.6)  Grandparents 30 (14.1) 74 (17.4) Education, n (%) 0.07  Primary school or below 26 (12.2) 64 (15.0)  Junior high 109 (51.2) 177 (41.6)  Senior high school or above 78(36.6) 185 (43.4) Continuous variables are expressed as median (25th,75th percentile); Categorical variable are expressed as number (percentage) AID Artificial insemination with donor sperm, IVF In vitro fertilization, ICSI Intracytoplasmic sperm injection, BMI Body mass index, PCOS Poly-cystic ovary syndrome Characteristics of the study population Continuous variables are expressed as median (25th,75th percentile); Categorical variable are expressed as number (percentage) AID Artificial insemination with donor sperm, IVF In vitro fertilization, ICSI Intracytoplasmic sperm injection, BMI Body mass index, PCOS Poly-cystic ovary syndrome Table  2 shows the difference in WISC Questionnaire scores between children conceived by AID and IVF/ICSI. The block design and digit span abilities in IVF/ICSI children were significantly lower than those of control group [Block design: Mean ± SD: 13.6 ± 6.5 vs12.6 ± 4.4; adjusted β (95%CI): -1.13 (-2.17, -0.08); Digit span abilities: 12.2 ± 4.2 vs. 11.2 ± 3.5, adjusted β (95%CI): -1.07 (-1.75, -0.39)]. The FSI, the VCI and the PRI were all comparable between two groups. Figure  1 presents the data relationship between them more intuitively. Table 2 Differences in WISC questionnaire scores between children conceived by AID and IVF/ICSI AID( n  = 213) Mean ± SD IVF/ICSI( n  = 426) Mean ± SD Crude β (95%CI) Adjusted β (95%CI) Similarities 11.2 ± 3.0 11.2 ± 2.9 0.06 (-0.45, 0.57) -0.04 (-0.56, 0.48) Comprehension 10.7 ± 2.5 10.7 ± 2.6 0.03 (-0.38, 0.44) -0.02 (-0.44, 0.39) Block design 13.6 ± 6.5 12.6 ± 4.4 -0.98 (-1.97, 0.003) -1.13 (-2.17, -0.08) Matrix reasoning 11.7 ± 3.6 11.5 ± 2.9 -0.25 (-0.81, 0.31) -0.35 (-0.93, 0.23) Digit span 12.2 ± 4.2 11.2 ± 3.5 -1.03 (-1.69, -0.36) -1.07 (-1.75, -0.39) Coding 14.0 ± 10.9 13.4 ± 10.1 -0.53 (-2.29, 1.23) -0.90 (-2.74, 0.95) VCI 104.7 ± 12.9 105.6 ± 13.0 0.90 (-1.33, 3.12) 0.47 (-1.78, 2.72) PRI 107.8 ± 12.5 108.0 ± 11.6 0.13 (-2.07, 2.32) 0.06 (-2.26, 2.39) FSI 107.7 ± 12.1 107.9 ± 11.4 0.15 (-1.98, 2.27) -0.07 (-2.27, 2.12) Adjusted for maternal age, paternal age, paternal BMI, paternal smoking AID Artificial insemination with donor sperm, IVF In vitro fertilization, ICSI Intracytoplasmic sperm injection, VCI Verbal Comprehension Index, PRI Perceptual Reasoning Index, FSI Full-Scale Index Differences in WISC questionnaire scores between children conceived by AID and IVF/ICSI Adjusted for maternal age, paternal age, paternal BMI, paternal smoking AID Artificial insemination with donor sperm, IVF In vitro fertilization, ICSI Intracytoplasmic sperm injection, VCI Verbal Comprehension Index, PRI Perceptual Reasoning Index, FSI Full-Scale Index Fig. 1 WISC questionnaire scores between children conceived by AID and IVF/ICSI. Similarities A , Comprehension B , Block design C , Matrix reasoning D , Digit span E , Coding F , VCI G , PRI ( H ) and FSI ( I ) are shown; n  = 213 (AID group); n  = 426 (IVF/ICSI group). Mean ± SD: Similarities 11.2 ± 3.0 (AID group) vs. 11.2 ± 2.9 (IVF/ICSI group); Comprehension 10.7 ± 2.5 (AID group) vs. 10.7 ± 2.6 (IVF/ICSI group); Block design 13.6 ± 6.5 (AID group) vs. 12.6 ± 4.4 (IVF/ICSI group); Matrix reasoning 11.7 ± 3.6 (AID group) vs. 11.5 ± 2.9 (IVF/ICSI group); Digit span 12.2 ± 4.2 (AID group) vs. 11.2 ± 3.5 (IVF/ICSI group); Coding 14.0 ± 10.9 (AID group) vs. 13.4 ± 10.1 (IVF/ICSI group); VCI 104.7 ± 12.9 (AID group) vs. 105.6 ± 13.0 (IVF/ICSI group); PRI 107.8 ± 12.5 (AID group) vs. 108.0 ± 11.6 (IVF/ICSI group); FSI 107.7 ± 12.1 vs. 107.9 ± 11.4 (IVF/ICSI group). * p  < 0.05 for AID vs. IVF/ICSI group. AID, artificial insemination with donor sperm; IVF, in vitro fertilization; ICSI, intracytoplasmic sperm injection; VCI, Verbal Comprehension Index; PRI, Perceptual Reasoning Index; FSI, Full-Scale Index WISC questionnaire scores between children conceived by AID and IVF/ICSI. Similarities A , Comprehension B , Block design C , Matrix reasoning D , Digit span E , Coding F , VCI G , PRI ( H ) and FSI ( I ) are shown; n  = 213 (AID group); n  = 426 (IVF/ICSI group). Mean ± SD: Similarities 11.2 ± 3.0 (AID group) vs. 11.2 ± 2.9 (IVF/ICSI group); Comprehension 10.7 ± 2.5 (AID group) vs. 10.7 ± 2.6 (IVF/ICSI group); Block design 13.6 ± 6.5 (AID group) vs. 12.6 ± 4.4 (IVF/ICSI group); Matrix reasoning 11.7 ± 3.6 (AID group) vs. 11.5 ± 2.9 (IVF/ICSI group); Digit span 12.2 ± 4.2 (AID group) vs. 11.2 ± 3.5 (IVF/ICSI group); Coding 14.0 ± 10.9 (AID group) vs. 13.4 ± 10.1 (IVF/ICSI group); VCI 104.7 ± 12.9 (AID group) vs. 105.6 ± 13.0 (IVF/ICSI group); PRI 107.8 ± 12.5 (AID group) vs. 108.0 ± 11.6 (IVF/ICSI group); FSI 107.7 ± 12.1 vs. 107.9 ± 11.4 (IVF/ICSI group). * p  < 0.05 for AID vs. IVF/ICSI group. AID, artificial insemination with donor sperm; IVF, in vitro fertilization; ICSI, intracytoplasmic sperm injection; VCI, Verbal Comprehension Index; PRI, Perceptual Reasoning Index; FSI, Full-Scale Index In the subgroup analysis, the IVF/ICSI children were divided into two subgroups: fresh ET and frozen ET group and the results were shown in Table  3 . Compared with the control group, children born after frozen ET showed impaired performance in block design, reasoning ability, and coding [block design: Mean ± SD:12.0 ± 3.4 vs. 13.6 ± 6.5, adjusted β (95%CI): -1.62 (-2.71, -0.54); Matrix reasoning: Mean ± SD: 11.0 ± 2.7 vs. 11.7 ± 3.6, adjusted β (95%CI): -0.78 (-1.47, -0.08); coding: Mean ± SD: 12.2 ± 7.2 vs. 14.0 ± 10.9, adjusted β (95%CI): -2.12 (-4.09, -0.15)]. However, the fresh ET group did not detect differences in those varieties. Additionally, both fresh ET and frozen ET groups reported a lower ability in digit span than for the control children [digit span: Mean ± SD: 11.3 ± 3.7 and 10.9 ± 3.0 vs. 12.2 ± 4.2, adjusted β (95%CI): -0.97 (-1.69, -0.24) and − 1.30 (-2.12, -0.49)]. For FSI, different embryo transfer strategies did not have significantly positive or negative effects in 6-year-old children. Table 3 Differences in WISC questionnaire scores between children conceived by AID, ET and FET AID( n  = 213) Mean ± SD Fresh ET ( n  = 312) Frozen ET ( n  = 114) Crude β1(95%CI) Adjusted β1(95%CI) Crude β2(95%CI) Adjusted β2(95%CI) Similarities 11.2 ± 3.0 11.4 ± 2.9 10.9 ± 2.8 0.17 (-0.37, 0.71) 0.10 (-0.45, 0.64) -0.25 (-0.96, 0.46) -0.38 (-1.11, 0.34) Comprehension 10.7 ± 2.5 10.8 ± 2.5 10.5 ± 2.8 0.13 (-0.30, 0.56) 0.08 (-0.36, 0.51) -0.23 (-0.86, 0.39) -0.28 (-0.92, 0.36) Block design 13.6 ± 6.5 12.8 ± 4.7 12.0 ± 3.4 -0.75 (-1.79, 0.29) -0.95 (-2.04, 0.14) -1.62 (-2.71, -0.54) -1.62 (-2.78, -0.47) Matrix reasoning 11.7 ± 3.6 11.6 ± 2.9 11.0 ± 2.7 -0.09 (-0.68, 0.50) -0.22 (-0.83, 0.39) -0.70 (-1.38, -0.01) -0.78 (-1.47, -0.08) Digit span 12.2 ± 4.2 11.3 ± 3.7 10.9 ± 3.0 -0.93 (-1.64, -0.22) -0.97 (-1.69, -0.24) -1.30 (-2.11, -0.50) -1.30 (-2.12, -0.49) Coding 14.0 ± 10.9 13.9 ± 11.0 12.2 ± 7.2 -0.07 (-1.98, 1.84) -0.48 (-2.48, 1.53) -1.79 (-3.78, 0.20) -2.12 (-4.09, -0.15) VCI 104.7 ± 12.9 106.0 ± 12.5 104.6 ± 14.2 1.28 (-1.03, 3.58) 0.95 (-1.36, 3.25) -0.15 (-3.52, 3.22) -0.72 (-4.16, 2.71) PRI 107.8 ± 12.5 108.6 ± 11.8 106.3 ± 10.9 0.75 (-1.59, 3.09) 0.58 (-1.90, 3.06) -1.55 (-4.29, 1.19) -1.44 (-4.25, 1.37) FSI 107.7 ± 12.1 108.5 ± 11.3 106.3 ± 11.3 0.73 (-1.51, 2.97) 0.51 (-1.78, 2.81) -1.46 (-4.34, 1.42) -1.63 (-4.58, 1.32) β1: Fresh ET vs. AID. β2: Frozen ET vs. AID. Adjusted for maternal age, paternal age, paternal BMI, paternal smoking AID Artificial insemination with donor sperm, Fresh ET Fresh embryo transfer, Frozen ET Frozen embryo transfer, VCI Verbal Comprehension Index, PRI Perceptual Reasoning Index, FSI Full-Scale Index Differences in WISC questionnaire scores between children conceived by AID, ET and FET β1: Fresh ET vs. AID. β2: Frozen ET vs. AID. Adjusted for maternal age, paternal age, paternal BMI, paternal smoking AID Artificial insemination with donor sperm, Fresh ET Fresh embryo transfer, Frozen ET Frozen embryo transfer, VCI Verbal Comprehension Index, PRI Perceptual Reasoning Index, FSI Full-Scale Index Additionally, the IVF/ICSI population were divided into IVF and ICSI groups, which was presented in Table  4 . Children in the 6-year-old IVF cohort had a significantly poorer performance of block design and matrix reasoning than the control cohort [block design: Mean ± SD: 12.6 ± 4.6 vs. 13.6 ± 6.5, adjusted β (95%CI): -1.16 (-2.27, -0.04); matrix reasoning: Mean ± SD: 11.3 ± 2.9 vs. 11.7 ± 3.6, adjusted β (95%CI): -0.62 (-1.23, -0.01)], while these was not detected in ICSI cohort. Whereas, both IVF and ICSI differ from the control group in the digit span test [digit span: Mean ± SD: 11.2 ± 3.5 and 11.2 ± 3.5 vs. 12.2 ± 4.2, adjusted β (95%CI): -1.15 (-1.87, -0.43) and − 0.97 (-1.83, -0.11)]. However, other varieties and FSI did not find any significant differences between the IVF or ICSI groups and the control group. Table 4 Differences in WISC questionnaire scores between children conceived by AID, IVF and ICSI AID( n  = 213) Mean ± SD IVF ( n  = 292) ICSI ( n  = 134) Crude β1(95%CI) Adjusted β1(95%CI) Crude β2(95%CI) Adjusted β2(95%CI) Similarities 11.2 ± 3.0 11.4 ± 3.0 11.0 ± 2.7 0.19 (-0.36, 0.75) 0.06 (-0.51, 0.62) -0.23 (-0.88, 0.41) -0.18 (-0.84, 0.47) Comprehension 10.7 ± 2.5 10.6 ± 2.6 10.9 ± 2.5 -0.07 (-0.51, 0.38) -0.18 (-0.62, 0.27) 0.24 (-0.31, 0.79) 0.30 (-0.26, 0.86) Block design 13.6 ± 6.5 12.6 ± 4.6 12.6 ± 3.9 -1.00 (-2.04, 0.04) -1.16 (-2.27, -0.04) -0.96 (-2.07, 0.16) -1.09 (-2.23, 0.05) Matrix reasoning 11.7 ± 3.6 11.3 ± 2.9 11.9 ± 2.7 -0.43 (-1.02, 0.15) -0.62 (-1.23, -0.01) 0.14 (-0.54, 0.81) 0.13 (-0.57, 0.83) Digit span 12.2 ± 4.2 11.2 ± 3.5 11.2 ± 3.5 -1.01 (-1.71, -0.31) -1.15 (-1.87, -0.43) -1.06 (-1.91, -0.21) -0.97 (-1.83, -0.11) Coding 14.0 ± 10.9 12.5 ± 10.7 12.4 ± 8.8 -0.49 (-2.40, 1.42) -1.02 (-3.00, 0.95) -0.62 (-2.74, 1.50) -0.87 (-3.04, 1.30) VCI 104.7 ± 12.9 105.4 ± 13.0 106.0 ± 12.9 0.73 (-1.66, 3.11) 0.01 (-2.39, 2.41) 1.27 (-1.69, 4.23) 1.51 (-1.49, 4.52) PRI 107.8 ± 12.5 107.4 ± 12.2 109.2 ± 10.0 -0.43 (-2.81, 1.95) -0.61 (-3.14, 1.92) 1.36 (-1.25, 3.97) 1.32 (-1.40, 4.03) FSI 107.7 ± 12.1 107.6 ± 11.6 108.4 ± 10.8 -0.10 (-2.36, 2.17) -0.62 (-2.94, 1.71) 0.67 (-2.10, 3.45) 0.98 (-1.83, 3.78) β1: IVF vs. AID; β2: ICSI vs. AID. Adjusted for maternal age, paternal age, paternal BMI, paternal smoking IVF In vitro fertilization, ICSI Intracytoplasmic sperm injection, VCI Verbal Comprehension Index, PRI Perceptual Reasoning Index, FSI Full-Scale Index Differences in WISC questionnaire scores between children conceived by AID, IVF and ICSI β1: IVF vs. AID; β2: ICSI vs. AID. Adjusted for maternal age, paternal age, paternal BMI, paternal smoking IVF In vitro fertilization, ICSI Intracytoplasmic sperm injection, VCI Verbal Comprehension Index, PRI Perceptual Reasoning Index, FSI Full-Scale Index Furthermore, infertility years, which refers to time-to-pregnancy (TTP), may have a negative effect on intellectual performance. Therefore, we further conducted an analysis. According to the 33rd percentile and the 66th percentile of TTP, the IVF/ICSI children were divided into 3 groups (≤ 2 years, 3–4 years and > 4 years), which were presented in Supplementary Table 1. Finally, all items in WISC Questionnaire were comparable among 3 groups. Given the differing performances of the IVF and ICSI groups on the WISC questionnaire, we conducted a subgroup analysis of the score distributions for both groups in Supplementary Table 2. After adjusting the possible confounders, we did not detect significant difference in scores between two groups other than Matrix reasoning, which were higher in ICSI group than IVF group [Matrix reasoning: Mean ± SD: 11.9 ± 2.7 vs. 11.3 ± 2.9, adjusted β (95%CI): 0.66 (0.07, 1.25)]. Supplementary Table 4 compares the scores of the offspring grouped according to different causes of infertility. The scores of ovulation disorders group slightly lower to the reference group in Coding [Coding: Mean ± SD: 11.03 ± 3.69 vs. 13.98 ± 10.90, adjusted β (95%CI): -2.72 (-4.52, -0.92)]. The same was observed in the tubal factor group in the item of matrix reasoning and digit span. Supplementary Table 5 compares the differences in WISC Questionnaire scores between male and female factors. The score of block design was lower in female factor group [block design: Mean ± SD: 12.60 ± 4.60 vs. 13.58 ± 6.47, adjusted β (95%CI): -1.12 (-2.22, -0.01)]. Consistent results were obtained in the score of digit span of male and female factor groups. Details of the sensitivity analysis were shown in Supplementary Table 2. After eliminated the confounder of repeat assessment, 537 children were included (173 in AID group and 364 in IVF/ICSI group), resulting in similar FSI scores for both the children conceived by AID and IVF/ICSI [Mean ± SD: 107.23 ± 12.13 vs. 108.19 ± 11.46, adjusted β (95%CI): 0.60 (-1.60, 2.80)]. Consistent with our preliminary analysis, we still found a poor performance in block design and digit span in IVF/ICSI children compared to the control group [block design: Mean ± SD: 12.80 ± 4.57 vs. 13.83 ± 7.00, adjusted β (95%CI): -1.15 (-2.17, -0.12) ; digit span abilities: 11.40 ± 3.62 vs. 12.32 ± 4.56, adjusted β (95%CI): -0.98 (-1.71, -0.25)].

The prospective cohort study was performed at the Center for Reproductive Medicine affiliated to Shandong University from January 2013 to December 2020. All children conceived by IVF/ICSI were eligible for this study. AID was performed on women whose husbands had been diagnosed with male azoospermia. Children conceived by AID were enrolled as controls. Exclusion criteria were as follows: (1) patients underwent Pre-implantation genetic testing (PGT) treatment; (2) multiple pregnancies; (3) uncompleted WICS-IV questionnaire. Propensity score matching (PSM) was performed to control the possible confounders. Confounding factors, including age and sex of children, maternal age, family income and education of parent and caregiver, were chosen for matching. The IVF/ICSI children and AID children were randomly matched in a 2:1 ratio by using the nearest neighbor matching. After calculating the propensity score of each child, the IVF/ICSI children were matched in a 2:1 ratio to those AID group. A total of 639 singletons aged 6–14 years old were included in our study (213 in AID group, and 426 in IVF/ICSI group). Firstly, control ovarian hyperstimulation (COH) was performed for the infertility parents, individually using a gonadotropin-releasing hormone (GnRH) agonist, GnRH antagonist or gonadotropin. The specific steps of COH were shown in previous literature [ 15 ]. After oocytes retrieval, IVF or ICSI was used for fertilization. Embryo transfer protocols, including fresh ET and frozen ET. In fresh ET cycles, embryos were transferred 3 or 5 days after oocytes retrieval. The frozen embryo consists of the frozen embryo at the cleavage stage or blastocyst stage. In frozen ET cycles, three frozen ET regimes were used, including spontaneous ovulatory (natural) cycles, hormone therapy (HT) frozen ET cycles and ovulation induction frozen ET cycles. After fresh ET or frozen ET, the luteal phase support will be applied. AID is mainly used for azoospermia. During the natural cycle or induced ovulation cycle, the average diameter of the dominant follicle was > 18 mm, the endometrial thickness exceeded 7 mm, so AID treatment was performed. The donor sperm came from Shandong Province sperm bank. The donor’s sperm was injected via intrauterine insemination (IUI) by experienced surgeons. The AID treatment is generally performed twice a cycle, before ovulation and after ovulation. After the AID treatment, routine luteal phase support was established [ 16 ]. The WISC-IV is used to evaluate the intellectual performance of children. The WISC-IV comprises six core subtests and two index scores standardized for sex and age. The six core subtests included Similarities, Comprehension, Block design, Matrix reasoning, Digit span and Coding. The two index items include VCI and PRI. VCI assesses the ability of verbal reasoning and acquired knowledge, PRI measures perceptual organization and logical reasoning ability. Finally, the full-scale index (FSI) could be analyzed. The primary outcome is Full-Scale Index (FSI) in WISC-IV questionnaire. According to previous cohort study, the FSI of ICSI and spontaneous conceived groups were 112 and 107, with standard deviations of 14.8 and 13.6 separately. The statistical power of our study was calculated via our sample size in each group ( n  = 213 for AID group and n  = 426 for IVF/ICSI). The two-side alpha level was 0.05. After calculating, the power was nearly 100% for FSI. Continuous variables proven to be non-normal distribution are expressed as median (25th–75th percentile). Non-normally distributed continuous variables were tested by Wilcoxon rank-sum test. Categorical variables were assessed by Pearson’s chi-square test. Considering some children completed the WISC-IV questionnaire twice or three times (31 children in AID group, 54 children in IVF/ICSI group), generalized estimating equation was performed to compare the WISC-IV scores between two groups. We conducted subgroup analysis according to the types of embryo transfer (fresh ET and frozen ET group) and ART (IVF and ICSI). Considering that the neurodevelopment of the offspring may be attributed to multiple factors, we divided the IVF/ICSI group into different reasons of infertility for analysis. To evaluate the robustness of our findings, the sensitive analysis based on the results from the children’s first WISC assessment was conducted. All data was analyzed by SAS software (version 9.4; SAS Institute, Cary, NC, USA). P-value < 0.05 was considered as statistically significant difference.

Our study found that the full-scale scores were comparable between children born after IVF/ICSI and offspring conceived by AID. However, the block design and digit span scores were significantly lower in the IVF/ICSI group. Some studies have found that ART children perform better in intellectual development. Aukje M et al. focused on motor performance and physical development in 103 five-year-old children and found a better performance in ICSI children [ 14 ]. Likewise, Marco et al. also conducted a cohort study which included 396 children and reported a more positive cognitive development in ART children compared to average population [ 17 ]. This specific phenomenon may be associated with ART parents’ characteristics, including higher education and income level, which are underlying confounders of these studies. We have adjusted the possible confounders related to economic education level and found no significant difference in full-scale scores between IVF/ICSI and AID offspring. Our results are in line with some previous studies. Meijerink et al. accessed mental development in 103 younger ICSI children at the age of 5 years, with the use of the Dutch Wechsler Preschool and Primary Scale of Intelligence test [ 14 ]. Compared with general population, they did not detect abnormal condition in ICSI cohort. Punamäki et al. analyzed the mental developmental outcomes of 225 children 7–8 years old born after IVF/ICSI who were matched with 278 spontaneously conceived controls [ 18 ], and did not detect any differences between the two groups. Leunens et al. recruited 151 8-year-old children born after ICSI and compared their cognitive and motor development with the control group. Wechsler Intelligence Scale for Children-Revised (WISC-R) and Movement Assessment Battery for Children (ABC) were used for assessment. Moreover, they also found a comparable outcome between the two groups. Although there is no noticeable difference in FSI, the score of digit span was still low in IVF/ICSI group. The possible mechanism may involve parental factors, metabolic disorders and epigenetic changes. First, compared with NC children, ART children are more susceptible to metabolic disorders, such as hyperlipemia [ 19 ]. Cui and Zhou et al. conducted a prospective study to investigate metabolic changes in offspring conceived by IVF or ICSI, and recruited 380 children born from IVF/ICSI and 380 NC children, all aged 6–10 years. The results showed that IVF/ICSI children have a worse cardiovascular metabolic condition. Rosenberg et al. reported that hyperlipidemia in children may have an adverse effect on psychological development [ 20 ]. Second, longer parental time-to-pregnancy (TTP) is proven to be associated with offspring neurodevelopmental delays and difficulties. Compared with NC parents, patients who underwent ART mostly have a longer TTP. Maria C et al. conducted a national cohort study aimed to examine associations of parental TTP and conception by ART with neurodevelopmental traits up to 8 years of age. Their research has proven this link between long infertile time and impaired intellectual development. They have found that a longer time-to-pregnant could lead to impairment of language, motor and prosocial skills in children [ 21 ]. However, our study did not detect this discrepancy. This may be attributable to the limited sample size following subgroup stratification. Cohort studies with larger sample sizes were need to explore the association between TTP and the neurodevelopment of children. Another possible factor is epigenetic changes. Epigenetic changes refer to heritable changes in gene function, which ultimately lead to phenotypic changes, in the absence of changes in DNA sequence. The application of ART has been demonstrated a strong link with epigenetic changes in genes of offspring both in animals and humans. Animal studies have reported epigenetic alterations in imprinted genes after embryo culture or superovulation [ 22 , 23 ]. Moreover, in humans, ovulation stimulating drugs are proven to alter gene expression in human granulosa cells. Brannian et al. found differentially expressed genes in granulosa cells which were stimulated by recombinant FSH, including A kinase anchor protein 11 (AKAP11), bone morphogenetic protein receptor II (BMPR2), epidermal growth factor (EGF) and so on [ 24 ]. AKAP11 was proven to be associated with a higher risk of schizophrenia [ 25 ]. Thus, the change in AKAP11 may bring in varying degrees of brain development disorders. And EGF is a clinical biomarker of prognosis and diagnosis of psychological disorders [ 26 ]. Therefore, the changes in gene expression in ART children may be associated with poor intellectual performance. As well as, some ART processes, such as manipulation of oocytes and embryo culture, cause changes in gene expression patterns in early embryos [ 27 ]. The CX43 gene has been shown to be expressed differently between in-vivo and in-vitro embryo cultures in the bovine. The CX43 gene is the predominant gap junction protein and plays a role in many aspects of brain development and physiology, such as intercellular communication and the release of neuroactive substances [ 28 ]. Therefore, the change in gene expression may affect the neurodevelopment of ART offspring. In the analysis of embryo transfer strategies, offspring conceived by ET and frozen ET have shown different intellectual performance. Children born after frozen ET showed impaired performance in block design, reasoning ability, and coding. This subgroup analysis was in line with previous studies. Li and Zhou et al. recruited 326 ET children and 326 frozen ET, and the intellectual performance was evaluated by WISC-IV. They found that the latter one had a poor performance compared to ET group. Previous study reported that this may be related to the culture of embryos from cleavage to blastocyst stage [ 29 ]. Further, the children conceived by cleavage-stage embryo transfer and blastocyst-stage embryo transfer showed comparable levels of intelligence. Therefore, the potential mechanism may relate embryos culture from cleavage to blastocyst stage. The study on mouse embryos have shown that in vitro culture may affect gene expression, reprogramming, embryo quality and finally affect fetal growth and cognitive development [ 30 ]. In the subgroup analysis, we found that score in Matrix Reasoning was higher in ICSI group compared with IVF group. However, there was no significant difference among other scores and FSI in the preliminary analysis. Previously, a large multicentre randomized controlled trial found no similar incidence rate of neonatal complications between IVF and ICSI groups, including preterm birth, large or small for gestational age, and congenital anomaly [ 31 ]. However, there have been no studies comparing the intellectual development of children conceived by IVF and ICSI. Thus, more researches in the future are needed to explore this issue. Our study has several strengths. First, the sample size was more significant than previous studies, so the results are more reliable. Second, it was elderly children that we focused on the intellectual development born after IVF and ICSI. In many countries, primary school begins from 6-year-old. At this age, children can perform some complex fine motor tasks such as writing [ 32 ]. So, the cognitive level of elderly children is more comparable to that of adults than infants, which enables us to evaluate intellectual development more accurately. Third, the WICS-IV questionnaire we used was a diagnosis utility in the clinic. It has accumulated a lot of data in clinical applications and is most commonly administered as standardized intelligence tests for children. Additionally, our study is a prospective cohort study and the baseline characteristics of parents and children we collected are more accurate. Therefore, we could adjust the possible confounders more comprehensively, making the results more convincing. There were still some limitations. The patients undergoing AID other than the NC population were selected for controls. This may lead to some possible biases. Women undergoing AID cannot completely represent the general population. The impact of AID operations could not be eliminated. Although little research is available, sperm cryopreservation has been suggested to be related to sperm DNA methylation [ 33 ], which may lead to neurodevelopmental changes. Besides, we only have the information of sociological rather than biological fathers in AID group. Therefore, we cannot match the baseline information to that of the biological father. Previous studies proved that the intellectual performance of children cannot be simply identified by genetic information [ 34 ]. And environmental interventions can markedly alter intellectual trajectories despite underlying genetic predispositions. In the future, large cohort studies conducted in NC children are needed to verify our findings. In conclusion, our study found that although the scores of some items were unfavorable, the full-scale scores were comparable between the IVF/ICSI group and the control group. The prevalence of infertility is increasing, and more and more children will be born after IVF/ICSI, their long-term intellectual development health needs to be evaluated in the future.

Since 45 years ago, almost 8 million children have been born with the help of assisted reproductive technologies (ART) [ 1 ]. Compared with the NC population, the prevalence of multiple pregnancies, preterm birth, and low birth weight was higher in women undergoing IVF/ICSI [ 2 ]. A meta-analysis reported a 30% increase in the risk of congenital defects in children born after ICSI [ 3 ]. In recent years, concern has been raised about the long-term health outcomes of children conceived by IVF/ICSI. Some long-term follow-up studies showed unfavourable endocrine outcomes in offspring born after IVF/ICSI. They have been detected comparable short-term health outcomes such as birth weight and neonatal length but an increased incidence of long-term cardiovascular and metabolic disorders, including higher blood pressure, fasting glucose and above-average total body fat composition [ 4 ]. Previous studies reported that some maternal and neonatal complications, such as hypertensive disorders of pregnancy (HDP) and fetal growth restriction, could affect child neurodevelopment, including mental and motor function [ 5 , 6 ]. However, until now, a limited number of follow-up studies have investigated the intellectual development of children conceived by IVF/ICSI. The evidence is limited and inconsistent. Previous studies reported a higher risk of impaired academic performance and social-emotional development among children and adolescents [ 4 , 7 , 8 ]. For cognitive development, some researchers have found a lower intelligence score of ICSI offspring in a cohort study of 83 children, based on the Revised Amsterdam Child Intelligence Test (short form) [ 9 , 10 ]. A meta-analysis reported ART children have a 35% increased risk of autism [ 11 ]. ART children are proven to be more likely to suffer from attention-deficit and hyperactivity disorder [ 8 , 12 ]. However, L.Leunens et al. concluded that compared with NC school-aged children, ICSI counterparts did not suffer any delay in cognitive and motor abilities [ 13 ]. Aukje M et al. performed a cohort study that included five-year-old children and found a higher score in cognitive performance but poor motor performance in ICSI children [ 14 ]. The results of previous studies are inconsistent and the sample size was small. Most studies focused on children younger than 5 years old. Compared with other intelligence tests, WISC-IV has a more complex structure and could better reflect children’s intelligence overall picture and solve the problems of Intelligence Quotient (IQ) variation. Currently, there are very few studies using WISC-IV to evaluate the intellect performance of children born after IVF or ICSI. Therefore, the aim of our prospective longitudinal cohort study is to evaluate the intellectual performance of children born after IVF/ICSI at the age of 6–14 using the WISC-IV questionnaire.

Supplementary Material 1. Supplementary Material 1.