Growth, psychomotor development and morbidity up to 3 years of age in children born after IVF

Transcription

1 Human Reproduction Vol.18, No.11 pp. 2328±2336, 2003 DOI: /humrep/deg445 Growth, psychomotor development and morbidity up to 3 years of age in children born after IVF S.Koivurova 1,2,5, A.-L.Hartikainen 2, U.Sovio 4, M.Gissler 3, E.Hemminki 3 and M.-R.JaÈrvelin 1,4 1 Department of Public Health Science and General Practice, University of Oulu, P.O.Box 5000, University of Oulu, 2 Department of Obstetrics and Gynecology, University Hospital of Oulu, P.O.Box 24, Oulu, 3 National Research and Development Center for Welfare and Health, P.O.Box 220, Helsinki, Finland and 4 Department of Epidemiology and Public Health, Imperial College, Norfolk Place W2 1PG, London, UK 5 To whom correspondence should be addressed at: Department of Obstetrics and Gynecology, University Hospital of Oulu, P.O.Box 24, Oulu, Finland. sari.koivurova@oulu. BACKGROUND: To examine the long-term child outcome after IVF until the age of 3 years in Northern Finland, we conducted a population-based cohort study. METHODS: First, a cohort of 299 IVF children born in 1990±1995 was compared with a cohort of 558 controls representing the general population in terms of a multiple birth rate of 1.2%, randomly chosen from the Finnish Medical Birth Register (FMBR) and matched for sex, year of birth, area of residence, parity, maternal age and social class (full sample analyses). Second, IVF singletons (n = 150) were compared with singleton controls (n = 280). Third, a plurality matched control cohort (n = 100) for IVF twins (n = 100) was randomly chosen, matched as above, from the FMBR and analysed separately. Infant mortality rate was compared with the national rate from the FMBR. RESULTS: Infant mortality in the IVF group was >2-fold higher compared to the national rate in the general population. The risk (OR, 95% CI) of low weight and height, below the lowest quartile, at 1 year of age (1.6, 1.1±2.2; 1.6, 1.1±2.4) and 2 years of age (1.5, 1.1±2.4; 1.7, 1.2±2.5) was signi cantly higher in the IVF group when compared with the general population control group. No statistically signi cant differences were found in the psychomotor development between the cohorts. Cumulative incidence of different diseases up to 3 years of age was signi cantly higher among IVF children in the full sample and singleton analyses (OR, 95% CI: 2.3, 1.7±3.2; 2.1, 1.3±3.3 respectively) especially regarding respiratory diseases (3.5, 1.9±6.5; 3.1, 1.0± 9.4) and diarrhoea (3.7, 2.2±6.2; 5.7, 2.6±12.7), but not in twin comparisons. CONCLUSIONS: The growth of IVF children was behind that of control children during the rst 3 years of life, but their psychomotor development was similar. Their postnatal health was worse, probably re ecting the problems in the neonatal period. Key words: child development/growth/ivf/morbidity Introduction The growing population of children born after IVF makes it important to study their long-term outcome, but so far only a few follow-up studies have been performed. It is already known that IVF pregnancies carry a higher risk for preterm birth, low birthweight and smallness for gestational age being mainly due to the high proportion of multifetal pregnancies resulting from IVF (Bergh et al., 1999; Koivurova et al., 2002a; Ludwig and Diedrich, 2002). The same is also true among singleton IVF pregnancies (Wang et al., 1994; Tanbo et al., 1995; Verlaenen et al., 1995; Schieve et al., 2002). As preterm birth and low birthweight in spontaneously conceived pregnancies are considered risk factors for childhood (Gissler et al., 1999) and adult (Jones et al., 1998; JaÈrvelin 2000) morbidity, it is important to study whether IVF children are at speci c risk. Previous follow-up studies, with few exceptions, have been quite short in their follow-up time and have suffered from a lack of statistical power, appropriate control groups or a proper control for confounding factors. Those studies following children up to the age of 1 to 13 years have not found any signi cant pathological features concerning growth and physical development of IVF children (Brandes et al., 1992; Olivennes et al., 1996; Olivennes et al., 1997; Wennerholm et al., 1998). Also the cognitive and behavioural development assessed at the age of 3±7 years has been reported to be normal among IVF children (Cederblad et al., 1996; Montgomery et al., 1999). Furthermore, the scholastic performance of IVF children has been reported to be encouraging (Olivennes et al., 1997). Even less is known about the morbidity of IVF children beyond the rst year of life. A French study found no speci c 2328 Human Reproduction 18(11) ã European Society of Human Reproduction and Embryology 2003; all rights reserved

2 Growth, psychomotor development and morbidity in IVF children medical or surgical illnesses in IVF children from cryopreserved embryos aged 1±9 years (Olivennes et al., 1996). Swedish researchers compared the prevalence of chronic diseases in children from cryopreserved IVF, standard IVF and spontaneous pregnancies up to 18 months of age and no differences were found between the groups (Wennerholm et al., 1998). However, in another recent Swedish population-based study, an increased risk for neurological problems, especially cerebral palsy, among IVF children was observed, being largely related to multiplicity (StroÈmberg et al., 2002). These observations and the growing population of IVF children indicate that more information about the long-term outcome after IVF is needed. The aim of this population-based cohort study was to compare the growth, the main developmental milestones and morbidity up to 3 years of age in children born after IVF with spontaneously conceived matched controls. The a-priori hypothesis was that, compared with the general population, IVF children grow more slowly, reach developmental milestones later and have a higher morbidity, because they are more often born preterm and have a poorer perinatal outcome than spontaneously conceived children (Koivurova et al., 2002a). Materials and methods A cohort study with a group of IVF-exposed children and two groups of unexposed naturally conceived controls was conducted. The exposed children were derived from the register at the IVF outpatient clinic in the University Hospital of Oulu and from the Infertility Clinic of the Family Federation of Finland in Oulu, in which all IVF treatments in the provinces of Oulu and Lapland are performed. Pre-study sample size calculations were based on the approximation of a frequency of on average 15% for developmental disorders including neurological signs (gross and ne motor, speech and other disabilities) among the unexposed populations (Hadders-Algra and Touwen, 1990). For 80% power, 0.05 alpha-error, a risk ratio 1.6 for the outcome between the groups and with a ratio of 1 (exposed):2 (unexposed), a sample size of at >238 exposed and 476 unexposed children was required for comparison between IVF and general population cohorts. For twins, corresponding sample size calculations for 80% power, 0.05 alpha-error and a risk ratio 1.6 for the outcome between the groups with a ratio of 1:1 assuming a frequency of developmental disorders among unexposed multiples of an average 30%, a sample size of >125 per group was required. There were 306 liveborn IVF children (154 singletons and 152 children from multiple pregnancies of whom 123 were twins, 25 triplets and four quadruplets) born during 1990±1995 and three stillborn fetuses (one twin and two triplets). The children were born after conventional fresh oocyte IVF and one singleton was conceived by ICSI. To examine growth, psychomotor development and morbidity, two separate unexposed control groups from the Finnish Medical Birth Register (FMBR, including all births after completion of the 22nd gestational week or with birthweight of >500 g), were to be chosen. (i) 618 controls (i.e. 2:1, 23309, at this point we missed information on the three stillborn fetuses among the IVF population) were to be selected at random and matched in the following order: sex, year of birth, area of residence (i.e. the provinces of Oulu and Lapland), parity, maternal age and socioeconomic status de ned by the occupation of the mother (upper white collar, lower white collar, blue collar, entrepreneurs and farmers, students, housewives and unknown). This control group (control I) represents the general Table I. Age-related psychomotor developmental milestones collected at the child welfare clinics on the nationally standardized health card Developmental milestones 1 month (nurse) Lies on the stomach, raises head occasionally Fists mainly closed Incidental sounds 2 months (nurse) Lies on the stomach, head up Fists mainly open Utters sounds Response on a smile 3 months (GP and nurse) Keeps head up when lifted from the hands Babbles Selective smile 4 months (nurse) Lifts head and trunk when leaning to the elbows Reaches for a handed object Reciprocal babble Delight at the sight of mother 6 months (nurse) Raises head and trunk with straightened arms Takes support with the legs when lifted to a standing position Grips an object with the st Imitates sounds, laughs out aloud 7 months (nurse) Crawls or tries to crawl Removes a toy from one hand to another and to the mouth Babbles syllables Is shy of strangers Observations are made by general practitioners (GP) or nurses. All items answered either yes or no. population of similar maternal age, parity and social class and includes multiple births in the same proportion as the normal general population (1.2%) (Hartikainen, 2001). The analyses where all IVF children are compared with the control I group are hereafter called `full sample' analyses to distinguish these from the analyses between singletons or twins. (ii) A second control cohort for multiples was randomly chosen (1:1, 152:152), and matched for plurality in addition to the matching criteria above. Consequently, we were able to do the strati ed analyses by plurality, i.e. exposed singletons had their own singleton controls derived from the control I group [hereafter control S (singleton) II], and multiple births had their own controls [control T (twin) II]. In the nal study population, described earlier in detail (Koivurova et al., 2002a), the numbers vary somewhat because not all data were available for every child (i.e. an exposed child may have only one control instead of the planned two). This mixed study design was formulated so as to evaluate the effect of multiplicity (full sample analyses) and the effect of IVF technology (analyses strati ed by plurality) at the same time. Municipal child welfare clinics (CWC) are located in health centres and are the primary units used for growth and developmental followup of children from neonates to the age of 6 years. CWCs are run by public heath nurses supervised by general practitioners (GPs). This service is free of charge, available to all children and used by 99% of them. Nurses screen the children's growth, development and medical problems; refer children with delays or medical problems to GPs, who refer the children to paediatric outpatient clinics if needed. According to national recommendations, check-ups at 3 months, 1, 2 and 3 years are performed by GPs with more detailed medical examinations and the rest of the check-ups are performed by nurses. At each check-up, height and weight are measured, developmental tests (Table I) are 2329

3 S.Koivurova et al. performed according to a nationally standardized health regime used throughout the country and marked on special cards. The developmental tests are modi ed after Bayley scales (Bayley, 1993) and serve as a screening method for developmental disabilities at the population level. A pilot study, conducted to nd out the availability and the coverage of the health records, showed that data on developmental milestones were missing in an average of 10±15% due to incomplete notes made by the doctors or nurses. However, the missing information was distributed randomly and was similar in both IVF and control groups. The data were collected by a resident physician (S.K.). The data on growth and psychomotor development up to 3 years of age were collected from health records at local CWCs and data on morbidity from hospital records on outpatient and inpatient care at paediatric wards. All diagnoses in hospital records were recorded by resident or specialized paediatricians and were based on the International Classi cation of Diseases (ICD-9 until 1995 and ICD-10 since 1996). Growth was measured by height and weight at 1, 2 and 3 years of age and psychomotor development was measured by the psychomotor abilities of the child at 1, 2, 3, 4, 6, 7, 8, 9, 12 and 18 months, 2 and 3 years of age in the CWC (Table I). The values of low weight and height at 1±3 years were de ned as the lowest quartile of this study population. To simplify the analysis of psychomotor development, summary variables of 1±3 months, 4±9 months, 1 year, 18 months, 2 years and 3 years were made. For morbidity, all diagnoses found in hospital records were taken into account, focusing on the chronic illnesses. A child was counted only once in the cumulative incidence of different diseases. Infant mortality rate was regarded as deaths during the rst year of life per 1000 live births. Conditional logistic regression for matched sets was used to calculate odds ratios (OR) with 95% con dence intervals (CI) for categorized variables. The percentages were calculated using matched sets; the denominator varied slightly due to some missing information among variables. As there were not enough eligible naturally conceived children from multiple pregnancies, we had to expand the area of residence. Even with the extended area we were not able to nd matched controls for triplets and quadruplets, hence they were excluded from analyses strati ed by plurality, but included in the full sample analyses. The ICSI child and the matched controls were excluded from the analyses, as were the four IVF children who died (at the age of 90 min to 3 months) and their matched controls from the analyses concerning growth, development and morbidity, because no CWC data were available due to early death. The control children who died were excluded likewise but their matched cases were excluded only if there were no other controls left in the set. Consequently, the nal study population consisted of 299 IVF children and 558 control children in the full sample comparisons, 150 IVF singletons and 280 control singletons, and 100 IVF twins and 100 control twins for the analyses strati ed for plurality. The SAS System Release 8.02 (TSO2MO) for Windows and SAS/STAT software were used for statistical analyses. Results Infant mortality There were four IVF children who died during infancy resulting in an infant mortality rate of 13.1/1000 live births. In the general population of Northern Finland in 1990±1995, the infant mortality rate was 5.2/1000 live births. The causes of death in the IVF group were severe prematurity in two of the cases (triplet and singleton; deaths at 90 min and 28 days of age), severe asphyxia during labour with prematurity in one Figure 1. Mean of the differences in weight between IVF and control children in matched sets at birth, 1, 2 and 3 years of age. Full sample IVF n = 299, Control I n = 558; Singletons IVF n = 150, Control SII n = 280; Twins IVF n = 100, Control TII n = 100. The dotted lines should be compared with the 0-reference line. a Full sample analysis where controls represented the general population (with the same rate of multiple births than among the normal general population) (control I): P < , 0.018, 0.011, b Singleton analysis (singleton IVF versus control SII): P = 0.127, 0.007, 0.017, c Twin analysis (twin IVF versus control TII): P = 0.514, 0.187, 0.437, case (singleton, death at 3 months of age) and multiple anomalies in one case (twin, amniotic band sequence, death at 2 h of age). Growth The means of the differences in weight and height between the IVF children and their matched controls are presented in Figure 1 and Figure 2. In the full sample analysis, IVF children were signi cantly shorter and lighter in weight than their controls (control I) during the whole follow-up period from birth to 3 years of age, though the difference was most prominent at birth. The IVF singletons were also signi cantly lighter in weight than control singletons (control SII) until the age of 3 years, but height was similar between the groups. For twins (compared to control TII) no signi cant differences were found with the exception of mean of the difference in height at 1 year of age in favour of the IVF twins (Figures 1 and 2.). The risk of low weight and height de ned as the lowest quartile of this study population at 1 (OR, 95% CI: 1.5, 1.1±2.2; 1.6, 1.1± 2.4 respectively) and 2 years of age (1.6, 1.1±2.4; 1.7, 1.2±2.5) was signi cantly higher and the risk of low height at 3 years of age was marginally higher (1.4, 0.95±2.1) in the IVF children in the full sample analyses. When the analyses were strati ed by plurality, no signi cant differences were found with the exception of the 2-fold risk of low height at 2 years of age among IVF singletons (1.9, 1.1±3.2) (Table II). 2330

4 Growth, psychomotor development and morbidity in IVF children Psychomotor development Neither the full sample analyses nor analyses strati ed by plurality showed any signi cant differences in the psychomotor development between IVF and control children in different age periods up to 3 years of age (Table III). Figure 2. Mean of the differences in height between IVF and control children inside matched sets at birth, 1, 2 and 3 years of age. Full sample IVF n = 299, Control I n = 558; Singletons IVF n = 150, Control SII n = 280; Twins IVF n = 100, Control TII n = 100. The dotted lines should be compared with the 0-reference line. a Full sample analysis where controls represented the general population (with the same rate of multiple births than in the normal general population) (control I): P < , 0.014, 0.021, b Singleton analysis (singleton IVF versus control SII): P = 0.178, 0.242, 0.069, c Twin analysis (twin IVF versus control TII): P = 0.539, 0.047, 0.125, Morbidity The cumulative incidence of respiratory diseases requiring hospital treatment or examinations at outpatient clinic was signi cantly higher in the IVF group than in the control groups in the full sample (OR 3.5, 95% CI 1.9±6.5) and the singleton (3.1, 1.0±9.4) analyses. Of the respiratory diseases, the risks of pneumonia (5.6, 1.1±27.8) and obstructive bronchitis (6.0, 2.6± 14.1) were signi cantly higher in the IVF group in the full sample analyses. This was also the case with obstructive bronchitis in the singleton analyses (5.1, 1.3±19.1). The risk of diarrhoea requiring hospital treatment was >3-fold in the IVF children in the full sample analyses (3.7, 2.2±6.2). This difference was also seen among singletons (5.7, 2.6±12.7). Different neurological signs were twice as common among the IVF children (eight febrile convulsions, eight muscular hypotensions, two cases of delayed motor development and one mental retardation) than among the controls representing the general population (control I) (2.2, 1.1±4.5). The risk of juvenile arthritis was higher, though non-signi cantly (5.5, 0.6±49.9), among the IVF children in the full sample analyses. A number of other illnesses existed in the IVF and control populations; there were signi cantly more children with at least one diagnosed illness in the IVF groups than in the general population-based control group I (2.3, 1.7±3.2) and in the Table II. Mean (SD) weights and heights and the percentage of low weight and low height, de ned as the lowest quartile threshold point in this study population, at 1, 2 and 3 years of age in IVF and control children Full sample analysis a Singletons b Twins b IVF Control I OR (95% CI) IVF Control SII OR (95% CI) IVF Control TII OR (95% CI) n = 299 n = 558 n = 150 n = 280 n = 100 n = 100 n (%) n (%) n (%) n (%) n (%) n (%) Mean weight (kg) Birth 2.9 (0.7) 3.5 (0.6) 3.4 (0.5) 3.5 (0.5) 2.6 (0.5) 2.6 (0.6) 1 year 9.8 (1.2) 10.0 (1.2) 9.8 (1.0) 10.1 (1.3) 9.7 (1.3) 9.5 (1.2) 2 years 12.6 (1.6) 12.9 (1.6) 12.6 (1.3) 13.0 (1.5) 12.5 (1.7) 12.4 (1.7) 3 years 14.8 (2.0) 15.2 (2.0) 14.9 (1.8) 15.3 (1.9) 14.6 (1.9) 14.6 (2.0) Low weight 1year, <9.0 kg 67 (23.7) 92 (17.5) 1.5 (1.1±2.2) 23 (16.8) 40 (15.3) 1.3 (0.7±2.3) 33 (33.3) 34 (34.7) 1.0 (0.5±1.7) 2 years, <11.5 kg 61 (21.7) 77 (14.6) 1.6 (1.1±2.4) 24 (17.4) 36 (13.6) 1.3 (0.7±2.2) 26 (27.1) 25 (25.5) 1.0 (0.5±1.9) 3 years, <13.5 kg 58 (21.6) 87 (18.0) 1.2 (0.8±1.8) 18 (14.1) 40 (16.3) 0.8 (0.4±1.5) 26 (27.4) 19 (21.8) 1.3 (0.6±2.7) Mean height (cm) Birth 47.8 (3.4) 49.9 (2.5) 49.7 (2.4) 50.1 (2.4) 46.4 (3.0) 46.0 (3.0) 1 year 75.8 (2.7) 76.4 (2.9) 76.3 (2.5) 76.6 (2.9) 75.5 (2.9) 74.5 (3.3) 2 years 87.9 (3.1) 88.4 (3.3) 88.1 (3.0) 88.6 (3.3) 87.7 (3.4) 87.0 (3.1) 3 years 96.1 (3.5) 96.8 (3.7) 96.5 (3.4) 97.1 (3.6) 95.6 (3.8) 95.3 (3.9) Low height 1 year, <74 cm 65 (23.0) 80 (15.3) 1.6 (1.1±2.4) 22 (16.1) 39 (14.9) 1.1 (0.6±2.0) 30 (30.3) 35 (35.7) 0.8 (0.4±1.4) 2 years, <86 cm 80 (28.6) 101 (19.2) 1.7 (1.2±2.5) 37 (27.0) 46 (17.4) 1.9 (1.1±3.2) 27 (28.1) 30 (30.9) 0.9 (0.5±1.5) 3 years, <94 cm 70 (26.2) 96 (19.8) 1.4 (0.95±2.1) 26 (20.5) 43 (17.6) 1.2 (0.7±2.2) 31 (32.6) 31 (36.1) 0.9 (0.5±1.6) Odds ratio (OR) with 95% con dence interval (CI) for low weight and height between IVF and control children. a The whole IVF population was compared with controls representing average population (with the same rate of multiple births than in normal general population) (Control I). b Analyses strati ed by plurality. Control SII = singleton controls; Control TII = twin controls. 2331

5 S.Koivurova et al. Table III. Failure, de ned as inability to perform at least one of the developmental tests required at that age, in psychomotor developmental tests at different age periods in the IVF and the control children and odds ratio (OR) for failure with 95% con dence interval (CI) between IVF and control children Full sample analysis a Singletons b Twins b IVF Control I OR (95% CI) IVF Control SII OR (95% CI) IVF Control TII OR (95% CI) n = 299 n = 558 n = 150 n = 280 n = 100 n = 100 n (%) n (%) n (%) n (%) n (%) n (%) 1±3 months 33 (30.0) 75 (26.7) 1.0 (0.5±1.9) 15 (21.7) 39 (27.1) 0.6 (0.2±1.4) 13 (46.2) 15 (50.0) 0.7 (0.1±4.0) 4±9 months 23 (56.1) 46 (50.6) 1.2 (0.3±4.9) 12 (48.0) 30 (56.6) 0.9 (0.1±5.5) 8 (72.7) 6 (66.7) ± 12 months 38 (21.6) 83 (22.4) 1.1 (0.7±1.8) 15 (15.2) 33 (17.7) 1.1 (0.5±2.2) 17 (28.8) 20 (36.4) 0.9 (0.3±2.4) 18 months 9 (6.3) 8 (2.9) 2.0 (0.6±6.9) 4 (5.1) 2 (1.4) 3.7 (0.4±35.8) 3 (7.3) 2 (5.9) ± 2 years 16 (10.4) 47 (14.2) 0.9 (0.4±1.9) 6 (7.8) 27 (15.6) 0.6 (0.2±2.1) 5 (9.1) 1 (2.3) 2.0 (0.2±22.1) 3 years 15 (10.2) 27 (8.6) 1.4 (0.6±3.1) 8 (10.4) 17 (11.0) 0.8 (0.2±2.4) 6 (11.1) 6 (12.2) 0.8 (0.2±3.4) a The whole IVF population was compared with controls representing average population (with the same rate of multiple births than among normal general population) (Control I). b Analyses strati ed by plurality. Control SII = singleton controls; Control TII = twin controls. Table IV. Cumulative incidence of different diseases a in the IVF and the control children up to 3 years of age and odds ratio (OR) of cumulative morbidity with 95% con dence interval (CI) between IVF and control children Full sample analysis b Singletons c Twins c IVF Control I OR (95% CI) IVF Control SII OR (95% CI) IVF Control TII OR (95% CI) n = 299 n = 558 n = 150 n = 280 n = 100 n = 100 n (%) n (%) n (%) n (%) n (%) n (%) Respiratory diseases 30 (10.3) 18 (3.3) 3.5 (1.9±6.5) 9 (6.2) 7 (2.5) 3.1 (1.0±9.4) 15 (15.2) 8 (8.0) 2.6 (0.9±7.3) Pneumonia 6 (2.1) 2 (0.4) 5.6 (1.1±27.8) 1 (0.7) 0 (0.0) ± 3 (3.0) 1 (1.0) 3.0 (0.3±28.8) Obstructive bronchitis 24 (8.2) 10 (1.8) 6.0 (2.6±14.1) 9 (6.1) 5 (1.8) 5.1 (1.3±19.1) 11 (11.1) 7 (7.0) 1.8 (0.7±5.0) Asthma 6 (2.1) 7 (1.3) 1.5 (0.5±4.6) 0 (0.0) 3 (1.1) ± 5 (5.1) 2 (2.0) 2.5 (0.5±12.9) Diarrhoea 44 (15.0) 25 (4.5) 3.7 (2.2±6.2) 24 (16.3) 9 (3.2) 5.7 (2.6±12.7) 12 (12.1) 13 (13.0) 1.0 (0.4±2.4) Atopic eczema 10 (3.4) 15 (2.7) 1.3 (0.6±3.0) 6 (4.1) 8 (2.9) 1.3 (0.4±4.0) 3 (3.1) 7 (7.0) 0.4 (0.1±1.7) Juvenile arthritis 4 (1.4) 1 (0.2) 5.5 (0.6±49.9) 2 (1.4) 1 (0.4) 2.6 (0.2±29.1) 2 (2.0) 0 (0.0) Ð Neurological signs d 19 (6.5) 20 (3.6) 2.2 (1.1±4.5) 4 (2.7) 12 (4.3) 0.7 (0.2±2.5) 10 (10.2) 4 (4.0) 3.0 (0.8±11.1) Any illness e 95 (32.7) 94 (17.0) 2.3 (1.7±3.2) 45 (30.8) 48 (17.3) 2.1 (1.3±3.3) 34 (34.7) 27 (27.0) 1.5 (0.8±2.9) a A child is counted only once in the incidence, repeated events are not taken into account. b The whole IVF population was compared with controls representing average population (with the same rate of multiple births than in normal general population) (Control I). c Analyses strati ed by plurality. d Includes diagnoses of febrile convulsions, muscular hypotension, delayed motor development, mental retardation and delayed speech development. e Includes any diagnosed illness or neurological sign. Control SII = singleton controls; Control TII = twin controls. singleton control group II (2.1, 1.3±3.3). Twin analysis showed no signi cant difference in this respect (Table IV). In all the comparison groups with the exception of the control twins, boys predominated slightly over girls in the cumulative incidence of illnesses (data not shown). Discussion We conducted a population-based cohort study on the growth, development and morbidity of IVF children until the age of 3 years, born in Northern Finland in 1990±1995 with matched controls, representing the general population in proportion of multiple births and also controls matched for plurality. Our unique study design made it possible to estimate the effect of IVF technology and plurality separately from each other on the long-term outcome of the child on which there are scarce data at the present time. We have previously demonstrated that the stillbirth, perinatal and neonatal mortality rates in this IVF population were ~2-fold 2332 higher compared to the national gures in Northern Finland (Koivurova et al., 2002a). The present study shows the same for infant mortality, which was comparable to that presented previously by the Bourn±Hallam group (Rizk et al., 1991). In a French study, the infant mortality rate was even higher, 20.8/ 1000 (Rufat et al., 1994). The elevated mortality during infancy in the IVF children probably re ects the neonatal complications that mainly arise from multiplicity and from the characteristics of infertile women (Koivurova et al., 2002a,b). We found six studies, three of them being large register studies (Table V), with control groups and control for confounding exploring children's growth, psychomotor development or morbidity after IVF. Table V shows the main features of these studies. Three of them investigated children's neuromotor development, three studies investigated growth and three studies morbidity. These studies, with the exception of a Swedish population-based register study (StroÈmberg et al., 2002), have mostly shown normal growth and development with no higher morbidity than among matched controls.

6 Growth, psychomotor development and morbidity in IVF children Table V. Comparison of the main studies with control groups and control for confounding concerning growth, development or morbidity of IVF children Year Study group Sample size and study design Matching criteria Child outcome Follow-up period 1992 Brandes et al. (Israel) 116 IVF children Birthweight IVF children grow and develop similarly to their non-ivf matched controls 116 control children Gestational age Prospective cohort study Birth order Order in multiple delivery Mode of delivery Sex Age Maternal age Maternal education 1996 Saunders et al. (Australia) 314 IVF children Plurality The growth of IVF children was normal at 2 years of age 150 control children Gestation Prospective population-based Date of birth cohort study 1998 Wennerholm et al. (Sweden) 255 FET children Maternal age Growth, development and chronic illness were similar between the groups 255 IVF children Parity 252 spontaneously conceived Single or twin children pregnancy Retrospective cohort study Date of delivery 1999 Bergh et al. (Sweden) 5856 IVF children Strati ed by: No increase in childhood cancer in the IVF group control children Maternal age Retrospective populationand Parity register-based cohort study Years of infertility Year of birth Multiple of pregnancies 2001 Klip et al. (The Netherlands) 9484 IVF children Strati ed by: IVF children carried no increased risk of cancer during childhood 7532 control children Maternal age Retrospective populationbased register/ Gestation questionnaire cohort study Birthweight Number of siblings Plurality 2002 StoÈmberg et al. (Sweden) 5680 IVF children Sex IVF children carried an increased risk of neurological disability, impairment or handicap, especially cerebral palsy control children Year of birth 4120 twin controls Birth hospital Retrospective populationand register-based cohort study 12±45 months 2 years 18 months Cancers detected at 3 months±2 years of age 6 years 18 months±14 years (age distribution) In our study, the IVF children were smaller than the controls representing the general population at birth in terms of weight and height. A catch-up growth was seen during the rst year of life, but in spite of that catch-up their growth was still behind that of the control children at the age of 3 years. This was also true among the IVF singletons (between the ages of 1±3 years). Our previous paper showed a signi cantly higher incidence of preterm birth and very low/low birthweight as well as higher neonatal morbidity among the IVF infants when compared with the controls representing the general population (Koivurova et al., 2002a). It is probable that poor neonatal outcome, not only accountable for the high proportion of multiple births among the IVF population, affects the growth during the rst years of childhood. Contrary to our results, previous follow-up studies have demonstrated normal growth in IVF children when compared with matched controls or a general population (Brandes et al., 1992; Saunders et al., 1996; Olivennes et al., 1997; Wennerholm et al., 1998) (Table V). However, direct comparisons between the studies are dif cult, because of the substantial variation of the follow-up times (range 1±13 years) and the matching criteria. Our results show that the growth of the IVF children during the rst 3 years of age is at a lower level than that of the control children, indicating an obvious need for further studies with longer follow-up periods. In our study, the psychomotor development up to 3 years of age, measured by the speci c age-related developmental milestones, did not differ between the IVF and the control children in spite of the different growth patterns observed in the full sample and singleton comparisons. The data collection 2333

7 S.Koivurova et al. regarding psychomotor development taken from the health records of CWCs was not complete: 10±15% of observations regarding speci c developmental milestones (Table I) was missing. However, missing data appeared randomly in the IVF and the control groups and the CWC do not necessarily know whether the child was born after IVF or not, consequently we assume that no severe selection or information bias existed. This is supported by the fact that practically all Finnish children use the CWC services, where they have their scheduled vaccinations. At the time of vaccination, the children also have a medical examination. The selection bias was also decreased by the fact that the two infertility clinics cover and keep the register of all infertility treatments in our target area. Our results on normal psychomotor development after IVF are in agreement with previous literature (Brandes et al., 1992; Raoul-Duval et al., 1994; Gibson et al., 1998). However, a follow-up of 3 years is still too short to detect all developmental delays, especially with this large scale screening measure that we used in the present study. Some researchers have found a high average achievement in developmental examination with a 12±30 month follow-up among IVF children and concluded this to be a result of exceptional motivation of the parents (Morin et al., 1989). It is of interest how much the family functioning and environmental stimuli enhance the child's development and whether these factors differ between families with children of assisted and spontaneous conception. In the framework of our study design, we were not able to investigate this further. An English study found a superior quality of parenting in families with children after assisted conception, even with gamete donation when compared to families with children after natural conception, but no differences in the children's emotions, behaviour or relationships with parents were found between the groups (Golombok et al., 1995). We found an increased cumulative incidence of different diseases diagnosed in outpatient or inpatient care, especially regarding respiratory diseases and diarrhoea, during the 3 year follow-up period among the children conceived by IVF when compared with the spontaneously conceived controls. Our results showed that the cumulative incidence of any illness in the IVF group (32.5%) was 2-fold higher compared with that of the full sample control group. In a Swedish 18 month follow-up study, the prevalence of chronic diseases was quite similar in the cryopreserved (18.0%) and the standard IVF (15.3%) and the spontaneously conceived (16.7%) children (Wennerholm et al., 1998). Another recent Swedish, large population-based register study noted an increased risk of developing neurological problems, especially cerebral palsy, after IVF (StroÈmberg et al., 2002). In the present study, with a much smaller number of children, but more detailed data collection, the IVF children had more neurological signs than the controls representing the general population, but there were no children with cerebral palsy in any of the groups. Neurological signs can be related to multiplicity and preterm birth following IVF. A few large studies have explored the incidence of childhood cancer in IVF children with no increased risk for malignant conditions (Doyle et al., 1998; Bergh et al., 1999; Bruinsma et al., 2000; Klip et al., 2001), though an increased risk for 2334 retinoblastoma among the IVF children has recently been suggested (Moll et al., 2003). Our study population is too small to detect cerebral palsy or childhood cancer that are rare at population level. Our diagnoses regarding morbidity were collected from hospital records on outpatient and inpatient care at paediatric wards in central or regional hospitals around Finland, meaning that these patients were given specialized paediatric examinations or treatment. Similar conditions in our country may also be treated to some extent in health centres by GPs or by private paediatricians, though eventually many of these diagnoses lead to referral to specialized hospital outpatient clinics. One might speculate that IVF children could be more easily referred to specialized paediatric care while control children may be treated by their own GP more often, thus biasing our results. Furthermore, IVF parents may seek medical help more often than other parents. In contrast to our work, an Australian study group stated in their paper that IVF infants do not over-utilize health care resources after the neonatal period (Leslie et al., 1998). On the other hand, IVF children are more often born preterm than other children, as is the case also in this study population (Koivurova et al., 2002a), and consequently they may be more susceptible to common infections, such as respiratory infections and gastroenteritis, than spontaneously conceived full term children. Preterm birth is a known risk factor for respiratory syncytial virus infections, for example (Law et al., 2002), or chronic lung disease (Kurkinen-RaÈty et al., 2000). This might explain the signi cantly higher prevalence of pneumonia in IVF children in the full sample analyses. Furthermore, it is possible that the increased incidence of neonatal morbidity (Koivurova et al., 2002a) and possible treatments with ventilators among IVF neonates may predispose them to respiratory infections later in childhood. Zycosity-chorionicity is an important factor on the outcome of twin pregnancies. Unfortunately, we do not have the information about zycosity of our twins. IVF technology alters the zycosity distribution in twin pregnancies and therefore zycosity acts as an intermediate factor on the causal pathway from IVF to child outcome, but does not cause any major bias in this study. When compared to the other relevant follow-up studies (Huber and Ludwig, 2002; Sutcliffe, 2002) (Table V) our results showed a poorer outcome for IVF children concerning childhood growth and morbidity. The different results can be explained by the dissimilarities in the study designs. In several of the studies presented in Table V, plurality or gestation are used as matching criteria, making it impossible to nd outcomes that are strongly affected by multiplicity or preterm birth. In this study, with a control group representing the general population in terms of multiple births, the differences became apparent. Furthermore, with the exception of large register studies, most studies suffer from a low power to detect differences between the comparison groups. In our study the statistical power is high, especially in the full sample analyses, as can be seen from the con dence intervals, but in the subgroups (especially twins) the power decreases, though it is still relatively high compared with other studies.

8 Growth, psychomotor development and morbidity in IVF children Twin analyses showed quite similar outcomes between the IVF and naturally conceived twins in this study. It is likely that the lack of differences is due to lower power in these analyses. During recent years, there has been an effort to reduce the number of multiple pregnancies by transferring only one embryo per cycle. Good treatment results with a reduction in multiple pregnancy rates have been achieved with elective single embryo transfer (ESHRE Campus Course Report, 2001; Martikainen et al., 2001) Since multiplicity is mostly accountable for the morbidity and mortality of IVF children, this effort should be encouraged. In conclusion, we observed dissimilarities in the growth patterns concerning weight between the IVF and the control children during the rst 3 years of life. However, psychomotor development of the children was comparable. Furthermore, the IVF children had diseases and needed hospital in- or outpatient treatment more often than their matched controls, especially with respiratory problems and gastrointestinal infections. These ndings probably arise at least partly from problems such as preterm birth and low birthweight, the well-known complications of assisted reproduction, as well as possible predisposing factors for childhood (Gissler et al., 1999) and adult morbidity (JaÈrvelin, 2000). Growth and health disorders during childhood in turn may have a negative impact on the later health status in adult life. Therefore it is important to note that IVF technology, with the in uence of multiplicity, also has long-term health effects in childhood, as was pointed out in this study. Acknowledgements We gratefully acknowledge the nancial support from The Alma and K.A.Snellman Foundation, Oulu, Finland; the National Social Insurance Institute, Finland and the Academy of Finland. References Bayley, N. (1993) Bayley Scales of Infant Development, 2nd edn. The Psychological Corporation, San Antonio. Bergh, T., Ericson, A., HillensjoÈ, T., Nygren, K.-G. and Wennerholm, U.-B. (1999) Deliveries and children born after in-vitro fertilisation in Sweden 1982±95: a retrospective cohort study. Lancet, 354, 1579±1585. Brandes, J.M., Scher, A., Itzkovits, J., Thaler, I., Sarid, M. and Gershoni- Baruch, R. (1992) Growth and development of children conceived by in vitro fertilization. Pediatrics, 90, 424±429. Bruinsma, F., Venn, A., Lancaster, P., Speirs, A. and Healy, D. (2000) Incidence of cancer in children born after in-vitro fertilization. Hum. Reprod., 15, 604±607. Cederblad, M., Friberg, B., Ploman, F., SjoÈberg, N.O., Sternqvist, K. and Zackrisson, E. (1996) Intelligence and behaviour in children born after invitro fertilization treatment. Hum. Reprod., 11, 2052±2057. Doyle, P., Bunch, K.J., Beral, V. and Draper, G.J. (1998) Cancer incidence in children conceived with assisted reproduction technology. Lancet, 352, 452±453. ESHRE Campus Course Report (2001) Prevention of twin pregnancies after IVF/ICSI by single embryo transfer. Hum. Reprod., 16, 790±800. Gibson, F.L., Ungerer, J.A., Leslie, G.I., Saunders, D.M. and Tennant, C.C. (1998) Development, behaviour and temperament: a prospective study of infants conceived through in-vitro fertilization. Hum. Reprod., 13, 1727± Gissler, M., JaÈrvelin, M.-R., Louhiala, P., Rahkonen, O. and Hemminki, E. (1999) Can children's health be predicted by perinatal health? Int. J. Epidemiol., 28, 276±280. Golombok, S., Cook, R., Bish, A. and Murray, C. (1995) Families created by the new reproductive technologies: quality of parenting and social and emotional development of the children. Child Dev., 66, 285±298. Hadders-Algra, M. and Touwen, B.C.L. (1990) Body measurements, neurological and behavioural development in six-year-old children born preterm and/or small-for-gestational-age. Early Hum. Dev., 22, 1±13. Hartikainen, A.-L. (2001) MonisikioÈinen raskaus. In Ylikorkala, O. and Kauppila, A. (eds), Naistentaudit ja synnytykset. Kustannus Oy Duodecim, Vammalan Kirjapaino Oy, Vammala, Finland, pp. 426±433. Huber, G. and Ludwig, M. (2002) Perinatal outcome and follow-up of children born from ART. In Ludwig, M. (ed.), Pregnancy and Birth after Assisted Reproductive Technologies. Springer-Verlag, Berlin, pp. 101±106. Jones, P.B., Rantakallio, P., Hartikainen, A.-L., Isohanni, M. and SipilaÈ, P. (1998) Schizophrenia as a long-term outcome of pregnancy, delivery, and perinatal complications: a 28-year follow-up of the 1966 North Finland general population birth cohort. Am. J. Psychiat., 155, 355±364. JaÈrvelin, M.-R. (2000) Congenital heart disease: fetal and infant markers of adult heart diseases. Heart, 84, 219±226. Klip, H., Burger, C.W., de Kraker, J. and van Leeuwen, F.E. (2001) Risk of cancer in the offspring of women who underwent ovarian stimulation for IVF. Hum. Reprod., 16, 2451±2458. Koivurova, S., Hartikainen, A.-L., Gissler, M., Hemminki, E., Sovio, U. and JaÈrvelin, M.-R. (2002a) Neonatal outcome and congenital malformations in children born after in-vitro fertilization. Hum. Reprod., 17, 1391±1398. Koivurova, S., Hartikainen, A.-L., Karinen, L., Gissler, M., Hemminki, E., Martikainen, H., Tuomivaara, L. and JaÈrvelin, M.-R. (2002b) The course of pregnancy and delivery and the use of maternal healthcare services after standard IVF in Northern Finland 1990±1995. Hum. Reprod., 17, 2897± Kurkinen-RaÈty, M., Koivisto, M. and Jouppila, P. (2000) Preterm delivery for maternal or fetal indications: maternal morbidity, neonatal outcome and late sequelae in infants. Br. J. Obstet. Gynecol., 107, 648±655. Law, B.J., Carbonell-Estrany, X. and Simoes, E.A. (2002) An update on respiratory syncytial virus epidemiology: a developed country perspective. Respir. Med., 96 (Suppl. B), S1±7. Leslie, G.I., Gibson, F.L., McMahon, C., Tennant, C. and Saunders, D.M. (1998) Infants conceived using in vitro fertilization do not over-utilize health care resources after the neonatal period. Hum. Reprod., 13, 2055± Ludwig, M. and Diedrich, K. (2002) Follow-up of children born after assisted reproductive technologies. Reprod. Biomed. Online, 5, 317±322. Martikainen, H., Tiitinen, A., TomaÂs, C., Tapanainen, J., Orava, M., Tuomivaara, L., Vilska, S., HydeÂn-Granskog, C., Hovatta, O. and Finnish ET Study Group (2001) One versus two embryo transfer after IVF and ICSI: a randomized study. Hum. Reprod., 16, 1900±1903. Moll, A. C., Imhof, S. M., Cruysberg, J.R.M., Schouten-van Meeteren, A.Y.N., Boers, M. and van Leeuwen, F.E. (2003) Incidence of retinoblastoma in children born after in-vitro fertilization. Lancet, 361, 309±310. Montgomery, T.R., Aiello, F., Adelman, R.D., Wasylyshyn, N., Andrews, M.C., Brazelton, T.B., Jones, G.S. and Jones Jr, H.W. (1999) The psychological status at school age of children conceived by in-vitro fertilization. Hum. Reprod., 14, 2162±2165. Morin, N.C., Wirth, F.H., Johnson, D.H., Frank, L.M., Presburg, H.J., Van de Water, V.L., Chee, E.M. and Mills, J.L. (1989) Congenital malformations and psychosocial development in children conceived by in vitro fertilization. J. Pediatr., 115, 222±227. Olivennes, F., Schneider, Z., Remy, V., Blanchet, V., Kerbrat, V., Fanchin, R., Hazout, A., Glissant, M., Fernandez, H., Dehan, M. et al. (1996) Perinatal outcome and follow-up of 82 children aged 1±9 years old conceived from cryopreserved embryos. Hum. Reprod., 11, 1565±1568. Olivennes, F., Blanchet, V., Kerbrat, V., Fanchin, R., Rufat, P. and Frydman, R. (1997) Follow-up of a cohort of 422 children aged 6 to 13 years conceived by in vitro fertilization. Fertil. Steril., 67, 284±289. Raoul-Duval, A., Bertrand-Servais, M., Letur-KoÈnirsch, H. and Frydman, R. (1994) Psychological follow-up of children born after in-vitro fertilization. Hum. Reprod., 9, 1097±1101. Rizk, B., Doyle, P., Tan, S.L., Rainsbury, P., Betts, J., Brinsden, P. and Edwards, R. (1991) Perinatal outcome and congenital malformations in invitro fertilization babies from the Bourn±Hallam group. Hum. Reprod., 6, 1259±1264. Rufat, P., Dehan, M., Olivennes, F., Frydman, R. and de Mouzon, J. (1994) Task force report on the outcome of pregnancies and children conceived by in vitro fertilization (France: 1987 to 1989). Fertil. Steril., 61, 324±330. Saunders, K., Spensley, J., Munro, J. and Halasz, G. (1996) Growth and 2335

9 S.Koivurova et al. physical outcome of children conceived by in vitro fertilization. Pediatrics, 97, 688±692. Schieve, L.A., Meikle, S.F., FeÂrre, C., Peterson, H.B., Jeng, G. and Wilcox, L.S. (2002) Low and very low birth weight in infants conceived with use of assisted reproductive technology. N. Engl. J. Med., 346, 731±737. StroÈmberg, B., Dahlquist, G., Ericson, A., FinnstroÈm, O., KoÈster, M. and Stjernqvist, K. (2002) Neurological sequelae in children born after in-vitro fertilisation: a population-based study. Lancet, 359, 461±465. Sutcliffe, A.G. (2002) A review of studies investigating outcome in IVF children. In Sutcliffe, A.G. (ed.), IVF Children: The First Generation: Assisted Reproduction and Child Development. Parthenon, London, pp. 21±36. Tanbo, T., Dale, P.O., Lunde, O., Moe, N. and AÊ byholm, T. (1995) Obstetric outcome in singleton pregnancies after assisted reproduction. Obstet. Gynecol., 86, 188±192. Verlaenen, H., Cammu, H., Derde, M.P. and Amy, J.J. (1995) Singleton pregnancy after in vitro fertilization: expectations and outcome. Obstet. Gynecol., 86, 906±910. Wang, J.X., Clark, A.M., Kirby, A., Philipson, G., Petrucco, O., Anderson, G. and Matthews, C.D. (1994) The obstetric outcome of singleton pregnancies following in-vitro fertilization/gamete intra-fallopian transfer. Hum. Reprod., 9, 141±146. Wennerholm, U.-B., Albertsson-Wikland, K., Bergh, C., Hamberger, L., Niklasson, A., Nilsson, L., Thiringer, K., Wennergren, M., Wikland, M. and Borres, M. (1998) Postnatal growth and health in children born after cryopreservation as embryos. Lancet, 351, 1085±1090. Submitted on January 9, 2003; resubmitted on May 21, 2003; accepted on July 22,