Journal Pre-proof General Health in a cohort of children conceived after assisted reproductive technology in the UK: a population-based record-linkage study

61 Background: Assisted reproductive technology usage is increasing annually; however, data on long- 62 term child health outcomes including hospital admissions is limited. 63 Objectives: To examine the potential effects of assisted reproductive technology on any and cause- 64 specific hospital admissions not related to perinatal diagnoses. 65 Study design : Population-based record-linkage study utilizing a previously established cohort of 66 children born after assisted reproductive technology in the United Kingdom between 1997 and 67 2009 (n=63877), their naturally conceived siblings (n=11343), and matched naturally conceived 68 population (n=127544) controls linked to their postnatal health outcomes up to 31 st March 2016 to 69 provide robust risk estimates of the potential effects of assisted reproductive technology on any 70 and cause-specific hospital admissions not related to perinatal diagnoses. Additionally, comparison 71 of hospital admissions by type of treatment was performed. Cox regression was used to estimate 72 the risk of hospital admissions and negative binomial regression was used to compare the number 73 of hospital admissions per year. 74 Results: This study had 1.6-million-person years of follow-up [mean: 12.9 years; range 0 to 19 75 years], and the mean age at the time of first hospital admission was 6.5 years [range 0 to 19 years]. Singletons born after assisted reproductive technology had increased risk of any hospital admission compared to naturally conceived population controls [Hazard ratio: 1.08; 95% CI: 1.05 to 1.10] but not naturally conceived siblings [Hazard ratio: 1.01; 95% CI: 0.94 to 1.09]. We observed increased risk for diagnoses related to neoplasms and diseases of the respiratory, musculoskeletal, digestive, 80 and genitourinary systems, and lower risk of injury, poisoning, and consequences of external causes 81 compared to naturally conceived population controls. Children born after intracytoplasmic sperm 82 injection had a lower risk of hospital admission compared to those born after in-vitro fertilization, although no such differences were observed between children born after fresh and frozen embryo transfers. 85 Conclusions: Children born after assisted reproductive technology showed greater numbers of 86 hospital admissions compared to naturally conceived population controls. Attenuation of these 87 differences in relation to their naturally conceived siblings suggests that this could be partially 88 attributed to the influence of parental subfertility on child health or increased parental concerns as 89 well as an actual increase in morbidity in children born after assisted conception.


INTRODUCTION
The number and proportion of children born after assisted reproductive technology (ART) is 95 increasing annually with over 8 million children born after ART globally. 1 Families with ART 96 conceived children report potential general health risks to their children as their paramount 97 concern, 2 with health care providers and stakeholders focusing on the potential adverse short-and Scotland were also excluded. Cases that had withdrawn consent for their data to be used for 147 research and children born after donor ART (oocytes, sperm, or embryos) were excluded. Finally, 148 triplets and higher order births were also excluded as they are known to be associated with adverse 149 outcomes such as higher infant mortality, birth defects, premature birth, and low birthweight. 18 150 Outcome data 151 The ART, NCS, and NCP groups were linked to their post-natal health records up to 31 st March 2016 152 using the HES database containing details of all admissions at NHS hospitals in England 19 in a one-153 off linkage. Diagnoses were defined using ICD-10 (International classification of diseases 10 th 154 revision), and conditions that would have originated in the perinatal period; those related to 155 pregnancy, childbirth, and the puerperium; and other diagnoses that are vague and cover a range 156 of symptoms, findings, and social circumstances that do not represent specific diagnoses or 9 The primary outcome measures were risk of any hospital admission and 'relevant' (i.e., related to 160 the included diagnostic chapters) diagnostic specific hospital admissions, while the secondary 161 outcomes of interest were the mean overall and diagnostic chapter specific admission rates 162 (number of admissions per year per child).   (Table 3). Compared to the matched NCP, children born after ICSI had a somewhat 273 lower risk than that seen for IVF without ICSI.

274
Children born after ICSI were at a lower risk of hospital admission compared to those born after IVF be the result of a selection effect rather than due to the ART procedure itself.
Although the role of biological mechanisms is still unclear, potential explanations could include 291 perturbed fetal growth (pre-term birth and low birthweight) and subsequent child growth 17 as well 292 as social explanations such as increased parental concern, with parents of ART children viewing 293 their offspring as more vulnerable (higher risks to child) and subsequently being more likely to seek of the different causes is complex. However, we note that one category, admissions for injury and 300 poisoning, cannot reasonably be linked to ART causes (parental subfertility or ART treatment) and 301 can therefore be considered responsive to parental concern alone. In this context, we note that 302 such admissions are in fact reduced in ART children, suggesting that parental concern does not 303 necessarily drive increased admission to hospital.

304
The increased risk of congenital defects observed was consistent with a recent meta-analysis that When comparing ART type to NCP, an increased risk of hospital admission was seen for both IVF 313 and ICSI and also for fresh and frozen embryo transfers. The associations were relatively small, 314 suggesting relative increases of 5 to 8%. ICSI children were at a lower risk of hospital admission 315 compared to those born after IVF. There is no clear explanation for this finding as ICSI is an invasive 316 technique associated with male subfertility, frequently with a genetic background, and children 317 born subsequently have been found to have higher rates of congenital anomalies. 24 Therefore, showing ART increases preterm birth and SGA. 14, 30,31 Thus, where results from the two comparator 351 groups are similar, there is increased confidence in the findings despite differences in bias.

352
However, where there are differences, caution is needed in assuming that one of the comparisons 353 is likely to be the least biased.

354
The main limitations of this study include those related to identification of the study cohort itself carried out on the linkage process to minimize this. 17 Approximately 23% of children were also lost during the linkage process due to the weak/inaccurate identifier data on the HFEA register and the 360 high threshold used for matching; however, although unavailable for the study period explored 361 here, future studies may be able to avoid this loss to follow-up as the HFEA now record both the 362 mother and child's NHS number on their register. Gestational age/prematurity data could not be 363 included as outcomes or used to explore whether results for other outcomes were mediated 364 through gestational age because these data are not available for the NC children. They are not 365 recorded in the birth registration dataset and are incomplete in maternal HES data (>50% missing).

366
Finally, children born to women who underwent ART prior to 1997 could not be linked to any 367 hospital records due to unavailability of HES data, thus limiting our ability to examine health 368 outcomes in these children and to explore the effects of changes in ART techniques over that  ART: Assisted reproductive technology; NCP: Naturally conceived population controls; sART: ART with siblings; NCS: Naturally conceived siblings J o u r n a l P r e -p r o o f J o u r n a l P r e -p r o o f

ART vs NCS: Singletons
Hazard Ratio (95%CI) J o u r n a l P r e -p r o o f