Parental health status and infant outcomes: Upstate KIDS Study.

Overall, mothers rated their general health as very good or excellent (71%). Table 1 presents the prevalence of parental chronic disease. Among men, prevalence of chronic diseases was (in descending order): hypertension (10%), asthma (10%), mood/anxiety (8%), and autoimmune and diabetes (3% each), and hypo- or hyper-thyroidism (2%), with all other diseases having a reported prevalence of 1%. A slightly different pattern emerged for maternal chronic diseases (in descending order): mood/anxiety (19%), asthma (16%), hypo- or hyper-thyroidism (9%), autoimmune disorders (8%), hypertension (4%), and types 1 or 2 diabetes (2%). All other diseases had a reported prevalence of 1% or less. Given the cohort was oversampled on births conceived with infertility treatment, 51% of women reported having ever been sexually active for a year without contraception and becoming pregnant, or the current duration definition of infertility. This prevalence decreased to 32% and 24% for lifetime or current infertility, respectively, when basing on TTP. It is important to note that missing data were minimal for the current duration definition of infertility (2%), but notably larger for lifetime (51%) and current (46%) definitions, as a large percentage of pregnancies were reported as being unplanned or without a reported TTP. In terms of pregnancy outcomes, mean gestation was 37.5 (±2.8) weeks, birth weight was 3,027 (±749) grams, birth length was 49.1 (±4.5) centimeters, and head circumference was 33.2 (±2.7) centimeters. Chronic disease in either parent was generally negatively associated with gestation and reduced birth size, though few findings achieved significance. In unadjusted models ( Table 2 ), maternal hypertension was significantly associated with gestation reflecting a decrement of <1 week (β −0.71; 95% CI −1.10, −0.32), whereas maternal kidney disease was negatively associated with head circumference (−1.23; −2.26, −0.20). In adjusted models ( Table 3 ), however, maternal hypertension was negatively associated with gestation (−0.64; −1.03, −0.25) and birthweight (−151.98; −262.30, −41.67), as was asthma and birthweight (−75.01; −130.40, −19.62) and kidney disease and head circumference (−1.09; −2.17, −0.01). Conversely, paternal autoimmune diseases were positively associated with head circumference (0.87; 0.15, 1.60). Maternal cancer was associated with a smaller PI (−0.17; −0.33, −0.02), and paternal history of mood/anxiety was positively associated (0.05; 0.001, 0.09) with PI. Relative to other diseases in the unadjusted analysis, infertility was uniformly negatively associated with all infant outcomes and irrespective of definition. Moreover, only the association between lifetime infertility and head circumference failed to achieved significance ( Table 2 ). After adjustment, infertility remained significantly associated with birthweight (β −62.18; 95% CI −103.78, −20.58), length (−0.33; −0.60, −0.06) and head circumference (−0.35; −0.67, −0.03) when using the current duration definition, but not when relying on definitions restricted to planned pregnancies with a reported TTP >12 months ( Table 3 ). Infant adiposity as measured by the PI reflected two signals: a negative association with maternal cancer/HIV and a positive association with paternal mood/anxiety. Infertility remained significantly associated with shorter gestation and diminished birth size in sensitivity analyses that excluded infertility treatment from the models. Moreover, significant negative associations were observed for all infant outcomes except head circumference when infertility was defined as a TTP >12 months for either past or the index pregnancy. A new association emerged for maternal hyper- or hypo-thyroidism and reduced head circumference (−0.48; −0.94, −0.01).

We utilized the Upstate KIDS Cohort Study that comprises 4,886 women who had a live birth registered in Upstate New York (57 counties excluding 5 New York City boroughs) between 2008 and 2010, with available reproductive and medical histories data obtained from parents upon enrollment of their infants (n=5,845) at approximately 4 months of age. Consistent with the Study’s goal to investigate infertility and its treatment on children’s growth and development through 3 years of age, infants conceived by infertility treatment and multiples were over-sampled from the New York State’s Perinatal Data System (live birth registry), and included 3,807 singletons and 2,038 twins. A complete description of the methodology has been described elsewhere ( 39 ). The New York State Department of Health and the University at Albany Institutional Review Board (IRB) approved the study (NYSDOH IRB #07–097; UAlbany #08–179) and served as the IRB designated by the National Institutes of Health for this study under a reliance agreement. Parents provided written informed consent prior to data collection. Mothers were queried about their reproductive and medical histories and that of the child’s father’s using a standardized self-administered questionnaire. Specifically, women were queried whether either parent had been diagnosed by a doctor or health care provider as having any chronic diseases, i.e., autoimmune disorders (fibromyalgia, chronic fatigue syndrome, rheumatoid arthritis, Crohn’s disease, lupus erythematosus, and irritable bowel syndrome); asthma; hypo- or hyper-thyroidism; type 1 or 2 diabetes; hypertension; cancer/HIV; cardiovascular disease; kidney disease; or liver disease. Infertility was defined in three ways to capture the women’s full reproductive history and given new thinking about how best to measure infertility. We opted to investigate three infertility definitions to minimize potential reporting errors associated with definitions that encompass pregnancy intendedness or wantedness. Our three study definitions were operationalized as: 1) ever being sexually active for a year or more without the use of contraception and becoming pregnant, which is the current duration definition ( 40 – 42 ); 2) ever requiring 12+ months to become pregnant, which is defined as lifetime infertility; and 3) requiring 12+ months to become pregnant with index pregnancy, which is defined as current infertility. Previous research has demonstrated that prevalence varies by definition of infertility ( 44 ). Infant outcomes included gestation defined as completed weeks’ gestation and birth weight (gm) as obtained from birth certificates, while birth length (cm) and head circumference (cm) were obtained from maternal report as these are not reported on birth certificates. From these data, we calculated the ponderal index (PI) [(birthweight (gm)/length (cm 3 )] x100 to assess infant adiposity ( 44 ).

In a large population-based cohort oversampled on infertility treatment, we found a generalized pattern of reduced gestation and infant birth size for various parental chronic diseases though significance was limited to maternal hypertension and asthma in final adjusted models with the inclusion of infertility treatment. Our findings also reflect a negative relation between infertility and birth size, but underscore the importance of how infertility is defined. Relative to other diseases, the current duration definition of infertility, defined as 12 months of sexual activity without contraception and pregnancy, was associated with a reduction in birthweight comparable to asthma (75 and 62 grams, respectively), but was the only disease associated with reductions of <1 cm for birth length and head circumference when accounting for infertility treatment. Cautious interpretation of our head circumference findings is needed in light of missing data, though we found no evidence suggesting it was associated with infertility treatment (data not shown). The absence of significant findings albeit uniformly negative associations between infertility definitions based upon TTP and infant outcomes may reflect recall errors, since self-reported TTP has been found to have bidirectional reporting errors relative to prospectively measured TTP and particularly for longer periods of recall ( 47 ). This finding is consistent with varying prevalence estimates of infertility based upon its definition ( 43 ). Given the uncertain underlying causal structure between health, fecundity, infertility treatment, and infant outcomes, our sensitivity analyses reflect consistent negative associations between infertility, irrespective of definition, and infant outcomes without changing associations observed for other diseases. It remains to be determined what infertility treatment is measuring, as options are many including possibly being a surrogate for diagnostic subtype, health insurance coverage, couples’ treatment preferences, provider practices, or underlying chronic pathophysiology. Our limited understanding about the underlying causal relation between health, fecundity and pregnancy outcomes requires concerted methodologic study to better inform how best to model plurality, infertility treatment and other exposures impacting fecundity and fertility to resolve existing data gaps. Such a model might also specify other possible common causes (beyond health status) of fecundity and health. We also recognize that self-reported medical and reproductive history may be inaccurate relative to medical records or other data sources, though previous findings support their use for gynecologic history, infertility treatment and pregnancy outcomes ( 32 , 48 – 51 ). Interpreting our findings within the context of the available literature suggests that despite reliance on self-reported disease status, prevalence estimates for chronic diseases are in the general range, though somewhat lower than those reported for reproductive aged couples. For example, hypertension and asthma are reported to affect 8% of similarly aged women ( 52 , 53 ) and anxiety approximately 18% of adults ( 50 ). Our infertility prevalence is notably higher than the 15% reported for the general U.S. population when utilizing the current duration approach ( 54 ) reflecting our design that oversampled births conceived with infertility. To our knowledge, our findings are the first to assess parental health status and infant outcomes including conceptualizing infertility as a disease and in the context of other chronic diseases. Our findings support the need to include reproductive health in the context of overall health status, and support its recent call for inclusion as the “sixth vital sign” ( 55 ), and the need to answer questions about global patterns of human fecundity ( 56 ). Such research is imperative for answering questions about the early origins of human fecundity, health across the lifespan and the health of future generations ( 54 , 55 ). To answer these questions, concerted efforts to delineate the underlying causal structure between health, fecundity and fertility and offspring’s health is urgently needed. Our findings reflect the importance of measuring not only fecundity impairments such as infertility but those more generally of health, and the importance of how infertility is defined. If corroborated, our findings have relevancy for clinical practice in that a simple question may be informative about women and infants’ health status and, perhaps, across the lifespan, viz., have you ever been sexually active without contraception and pregnancy for a year or more? The application of this definition for couples’ or men’s health remains to be established with causal inference techniques. Still, our findings require careful interpretation considering the unknown underlying causal structure, the potential for residual confounding given the observational nature of this cohort, limited power for assessing head circumference given the missing data for this outcome, and reliance on self-reported health history at approximately 4-month postpartum. To this end, the exact timing of disease onset is unknown as are any treatment related effects on infant outcomes. Our findings are responsive to a recent call for rethinking infertility and disease given the potential for shared genetic and molecular pathways ( 56 ), and the implications of fecundity across the lifespan ( 57 , 58 ).

In the descriptive phase, we inspected distributions for all data including atypical patterns of missing data relative to the study exposures and outcomes. We also estimated the prevalence of parental chronic diseases and mothers’ gynecologic diseases in relation to infant outcomes. In multivariable models using generalized estimating equations (GEEs) to account for twins ( 45 ) and for each maternal chronic disease, we estimated the change (beta (β) coefficient and 95% confidence intervals (CI)) in infant outcomes modeled continuously, and then adjusted for maternal age (years), preconception body mass index (BMI) as derived from self-reported pregnancy weight and height (kg/m 2 ) and use of infertility treatment for the index birth (yes/no). We did not adjust for gestation in models of infant birth size, given that it may be an intermediate variable and such adjustment may induce over-adjustment bias ( 46 ), nor did we adjust for plurality given that it follows and not precedes fecundity or fertility. In light of uncertain modeling assumptions about the causal structure between fecundity, treatment and health in light of treatment being a downstream factor resulting from underlying disease, we reran the adjusted models without infertility treatment as a sensitivity analysis that could help inform about the underlying causal structure. All analyses were conducted in SAS, version 9.4.