{"paper_id":"a69a47a1-eb8f-41bd-893f-2fd1feef1e9b","body_text":"Infertility is common in the United States (U.S.), with 1 in 6 couples unable to conceive a child after 12 months ( 1 ,  2 ). Existing infertility treatments are burdensome and expensive, with the overall cost of in vitro fertilization (IVF), the most common assisted reproductive technology (ART), often exceeding $40,000 ( 3 ). Nonetheless, the use of ART in the U.S. has more than doubled in the past decade, and is now approaching a half million cycles annually ( 4 ).\nWhile several nutrients (e.g., folic acid ( 5 ,  6 ), fatty acids ( 5 – 9 )), food groups (e.g., seafood ( 5 ,  8 ), soy ( 5 ,  7 ,  8 ), dairy ( 5 – 7 ,  10 ), whole grains ( 8 ,  11 )), and dietary factors (e.g., caffeine ( 5 ,  6 ,  12 ,  13 )) have been found to be associated with live birth, pregnancy loss, or semen quality, there remains insufficient evidence to support recommendations for any specific dietary pattern to improve the fertility outcomes of couples experiencing infertility. Nevertheless, there is strong interest among couples for dietary guidance ( 14 ). Recognizing this, experts in nutrition and fertility have proposed dietary patterns intended to improve fertility outcomes ( 15 ,  16 ). Yet, it remains unknown whether these “fertility diets” are more beneficial for improving clinical fertility outcomes than traditional, healthy dietary patterns. To this end, our objective was to estimate associations of fertility-focused and traditional healthy preconception dietary patterns with live birth and pregnancy loss among couples experiencing infertility. We hypothesized that greater adherence to traditional healthy or fertility-focused patterns during preconception by female partners, male partners, or both partners is associated with a higher likelihood of live birth and a lower risk of pregnancy loss.\n\nThis prospective cohort study was a secondary analysis of data from the Folic Acid and Zinc Supplementation Trial (FAZST), a double-blind, block-randomized, placebo-controlled trial of the effects of folic acid and zinc supplementation in males on semen quality and live birth among 2,370 couples planning infertility treatment (IVF and non-IVF) ( 17 ,  18 ), and the Impact of Diet, Exercise, and Lifestyle (IDEAL) on Fertility Study, a prospective cohort of the female partners of men enrolled in FAZST ( 19 ). Couples were recruited from four reproductive endocrinology and infertility care centers in the U.S. (Salt Lake City, Utah; Iowa City, Iowa; Chicago, Illinois; Minneapolis, Minnesota) from June 2013 to December 2017. Abstraction of pregnancy outcomes was completed in April 2019. Ineligible couples were those whose male partner was <18 years or female partner was <18 or >45 years, were pregnant at enrollment, were planning to use donor sperm or a surrogate, or whose male partner was unlikely to benefit from supplementation (e.g., obstructive azoospermia) or had a poorly controlled chronic disease. Females were followed for 9 months, with an additional 9 months for pregnancy outcomes. All couples were included in live birth analyses; pregnancy loss analyses were restricted to couples who experienced at least one pregnancy ( Supplemental Figure 1 ).\nProtocols were approved by institutional review boards at each site and a data coordinating center. A Data and Safety Monitoring Board oversaw the study. All participants provided written informed consent. Reporting of this paper followed STROBE guidelines.\nAt baseline, each partner completed a 125-item food frequency questionnaire ( 20 ), reporting usual dietary intake (servings/day) during the past 3 months. Daily energy, nutrient, and alcohol intake and glycemic load were calculated using the Nutrient Data System for Research from the University of Minnesota Nutrition Coordinating Center ( 21 ). Multivitamin use was collected in a separate questionnaire.\nScores for four  a priori  dietary patterns ( Table 1 ) were calculated for each partner: two traditional healthy dietary patterns (Healthy Eating Index-2015 [HEI-2015] and Mediterranean Diet) and two fertility-focused dietary patterns (Fertility Diet and Pro-fertility Diet). The scoring approaches are shown in  Supplemental Table 1 . Higher scores indicated greater alignment between reported dietary intake and the recommended intake for each component of the dietary pattern; we refer to this alignment as “adherence,” though participants were not assigned to any dietary pattern. The HEI-2015 reflects alignment with the 2015–2020 Dietary Guidelines for Americans ( 22 ). The Mediterranean Diet from Trichopoulou et al. reflects alignment with a traditional Greek diet ( 23 ). The Fertility Diet from Chavarro et al. includes dietary factors associated with anovulatory infertility in a cohort of female nurses ( 16 ). Finally, the Pro-fertility Diet from Gaskins et al. includes dietary factors associated with improved ART outcomes in women ( 15 ).\nA combined score for each pattern was calculated as the sum of the two partners’ individual scores.\nThe outcomes were live birth and pregnancy loss abstracted from medical records. Pregnancy loss was defined as any loss before 20 weeks’ gestation, including a serum hCG measurement >5 mIU/mL followed by a decline. Couples could have experienced one or more pregnancies, one or more pregnancy losses, and a live birth.\nMultiple imputation by chained equations (m=10) was used for missing diet and covariate data. Outcome data was complete for all couples. Sample characteristics prior to imputation are shown in  Supplemental Table 2 . Diet data for participants with energy intake <500 (n=53 females, 43 males) or >6,000 kcal/day (n=1 female, 6 males) were considered implausible and set to missing prior to imputation.\nDescriptive statistics of sample characteristics, overall and according to quartiles of dietary pattern scores, were calculated separately for females and males. Baseline sociodemographic, lifestyle, and fertility information was collected via questionnaire, and included age, self-identified race and ethnicity, education, annual household income, employment, marital status, health and fertility insurance, previous live birth, physical activity ( 24 ), smoking, medical history, months trying to conceive, and previous infertility diagnoses. Body mass index (BMI) was calculated from height and weight measured at baseline using standard protocols. Spearman correlations estimated within-partner and -couple correlations of diet patterns.\nCrude and adjusted log-binomial regression models estimated risk ratios (RR) and 95% confidence intervals (CI) for live birth and pregnancy loss. For each dietary pattern, a separate model was created for each partner’s score and the combined score, modeled continuously (per 1-SD increase in score). Scores were also modeled in quartiles ( Supplemental Tables 3 – 4 ). Potential confounders were identified using directed acyclic graphs ( Supplemental Figure 2 ) and included total energy intake, age, race/ethnicity, education, physical activity, smoking, previous live birth, time trying to conceive, and previous infertility diagnosis. Optimal forms of ordinal and continuous variables, including diet pattern score variables, were determined by model fit statistics comparing linear, quadratic, and categorical forms. Although supplemental folic acid and zinc had no effect in the original trial ( 17 ), treatment assignment was additionally included in the adjusted models. For pregnancy loss, stabilized inverse probability weights were used to account for selection on couples who became pregnant ( 25 ) (n=1,054), constructed using factors associated with the probability of pregnancy, including age, race/ethnicity, and BMI of both partners as well as treatment assignment and dietary pattern score.\nIn sensitivity analyses, alternative adjustment sets were considered, given uncertainty around the roles of total energy intake and BMI in the relationship between dietary patterns and fertility. We considered total energy intake to be a determinant of dietary pattern scores and treated it as a confounder in our primary analyses. However, it could be considered a mediator, as dietary patterns characterized by greater intake of energy-dense foods (e.g., meat) may lead to higher total energy intake; therefore, we repeated the analyses without adjustment for energy intake. BMI was not considered a confounder in our primary analyses because its measurement occurred after the dietary recall period. However, if BMI is assumed to be stable, then BMI measured at baseline may approximate BMI prior to the dietary recall period, a potential confounder. Therefore, analyses were repeated with adjustment for baseline BMI.\nWe examined heterogeneity by BMI (<25 kg/m 2 , 25 to <30 kg/m 2 , ≥30 kg/m 2 ), infertility treatment (IVF, OI/IUI, none), and baseline infertility diagnosis (yes, no). For the combined diet analyses, rather than examining associations within all nine female-male BMI combinations, we restricted analyses to couples for whom both partners had a BMI ≥30 kg/m 2 . For analyses stratified by infertility diagnosis, we further subset the female sample to those with polycystic ovary syndrome (PCOS) or anovulation to examine whether potential fertility benefits of dietary patterns may be explained by improved ovulatory function.\nBecause our approach to modeling the combined score may not have captured potential synergies between the partners’ diets, we also considered a 5-category combined score variable: 1) both in 1 st  quartile; 2) both in 2 nd  or both in 3 rd  quartile; 3) discordant, female in higher quartile; 4) discordant, female in lower quartile; 5) both in 4 th  quartile.\nTo understand whether certain components of a dietary pattern explained any observed associations of the overall score with live birth or pregnancy loss, we conducted exploratory analyses of the components, adjusting for the confounders in the primary analyses plus the overall dietary pattern score. Crude and adjusted RRs per 1-point increase in the component score were estimated using log-binomial models.\nAnalyses were performed in SAS 9.4 (Cary, NC).\n\nAmong the 2,370 couples, 1,054 experienced a pregnancy, 287 had a pregnancy loss, and 821 had a live birth. The mean age (SD) was 31 (5.1) and 33 (5.9) years among females and males, respectively ( Table 2 ). Roughly half of females and males had a bachelor’s degree or higher, more than 80% identified as non-Hispanic white, and 46% had an annual household income ≥$75,000. The median time trying to conceive at enrollment was 19 months (interquartile range: 12–36). On average, females consumed more vegetables and less meat than males ( Supplemental Table 5 ).\nFemales in higher HEI-2015 quartiles were older, more physically active, less likely to smoke, less likely to have a previous live birth, and more likely to have fertility insurance than those in lower quartiles ( Table 2 ). They also had a lower BMI, higher educational attainment, higher annual household income, and had spent fewer months trying to conceive. Similar patterns were observed among males, except males in higher quartiles had lower physical activity and were more likely to have health insurance and an infertility diagnosis than those in lower quartiles. There were no differences in time trying to conceive across HEI-2015 quartiles among males. Similar trends were observed for the other dietary patterns ( Supplemental Tables 6 – 8 ).\nWithin-partner correlations between dietary pattern scores were weak to moderate ( Supplemental Table 9 ), with the strongest correlation between scores for the two traditional healthy dietary patterns (0.42 in both females and males), and between scores for the Fertility Diet and HEI-2015 (females: 0.43; males: 0.42). The two fertility-focused patterns were weakly correlated (females: 0.15; males: 0.16). Within-couple correlations were also relatively weak, with the strongest correlation between female and male scores for the HEI-2015 (0.39).\nAmong females, males, and couples combined, greater adherence to the traditional healthy dietary patterns was associated with a higher likelihood of live birth ( Figure 1 ). Adjusted for confounders, a 1-SD increase in a female’s score for the HEI-2015 (aRR=1.06 [95% CI 1.00–1.13]) or Mediterranean Diet (1.05 [0.99–1.12]) was associated with a higher likelihood of live birth. Similar associations were observed for males and couples, with a 1-SD increase in scores for the HEI-2015 (males: 1.05 [0.99–1.11]; combined: 1.05 [0.99, 1.12]) or Mediterranean Diet (males 1.05 [0.99, 1.11]; combined: 1.06 [1.00, 1.12]) associated with a higher likelihood of live birth.\nAdherence to the fertility-focused dietary patterns was not associated with live birth, except among females. Adjusted for confounders, a 1-SD increase in a female partner’s Fertility Diet score was associated with a higher likelihood of live birth (1.05 [0.99, 1.12]). This association did not exist among male partners or couples combined. There was no association between adherence to the Pro-fertility Diet pattern and the likelihood of live birth among females, males, or couples.\nPregnancy loss among couples who became pregnant was not associated with adherence by females, males, or couples to the traditional healthy dietary patterns or the fertility-focused patterns ( Figure 2 ). Removal of total energy intake from ( Supplemental Table 10 ) or the addition of BMI to ( Supplemental Table 11 ) the adjustment set did not meaningfully change the estimates.\nThe positive associations of HEI-2015 and Mediterranean Diet scores of females and males with the likelihood of live birth were strongest among those with a BMI ≥30 kg/m 2  ( Supplemental Table 12 ). The same was observed for the positive association of the Fertility Diet in females with live birth. All estimates were less precise than those measured in the entire sample. The positive associations of the combined HEI-2015 and Mediterranean Diet scores with live birth were stronger, though less precise, among couples for whom both partners had a BMI ≥30 kg/m 2 , relative to the association among all couples. Associations with pregnancy loss were null across all BMI categories ( Supplemental Table 13 ).\nIn analyses stratified by treatment type, the associations between all four dietary patterns, whether individual or combined, and live birth were null for couples who underwent IVF ( Supplemental Table 14 ). Associations with pregnancy loss were null across all treatment types ( Supplemental Table 15 ).\nThe positive association between live birth and HEI-2015 score among females and couples without an infertility diagnosis was stronger than it was among those with a diagnosis ( Supplemental Table 16 ). The opposite pattern emerged for the Mediterranean and Fertility Diets; associations were stronger among females and males with an infertility diagnosis than it was among those without. However, all estimates were imprecise. Associations of HEI-2015, Mediterranean Diet, and Fertility Diet scores with live birth among females with PCOS or anovulation were slightly stronger, though less precise, than associations among females with any infertility diagnosis. There were no differences in associations with pregnancy loss ( Supplemental Table 17 ).\nIn analyses using the 5-category combined HEI-2015, Mediterranean, and Fertility Diet scores, compared to couples whose partners were both in the lowest quartile, couples with any other combination had a higher likelihood of live birth ( Supplemental Table 18 ). There were no associations with pregnancy loss.\nIn exploratory analyses of the dietary pattern components, for the HEI-2015, the likelihood of live birth was greater with lower intake of refined grains by couples combined ( Supplemental Table 19 ). For the Mediterranean Diet, the likelihood of live birth was greater with moderate alcohol intake, relative to no or high alcohol intake, by males and couples combined ( Supplemental Table 20 ). For the Fertility Diet, the likelihood of live birth was greater with more frequent multivitamin use by females ( Supplemental Table 21 ).\n\nIn this cohort of couples seeking infertility treatment, greater preconception adherence to traditional healthy dietary patterns was associated with a modestly higher likelihood of live birth. Adherence to fertility-focused dietary patterns did not appear to confer the same benefits; only among females was greater adherence to the Fertility Diet associated with a similarly higher likelihood of live birth. Notably, these positive associations were not observed among couples who underwent IVF. Adherence to neither traditional healthy dietary patterns nor fertility-focused patterns was associated with pregnancy loss. This is one of the only studies that has considered the preconception dietary patterns of both female and male partners in relation to clinical fertility outcomes, and it serves as an important step toward addressing the insufficient dietary guidance for couples with infertility.\nA key finding was that greater adherence to two traditional healthy dietary patterns, the HEI-2015 and the Mediterranean Diet, was associated with a modest increase in the likelihood of live birth. This association persisted regardless of whose diet was under consideration – that of the female partner, male partner, or both partners in combination. To date, few studies have examined male preconception dietary patterns in relation to clinical fertility outcomes, with most focusing on semen quality ( 5 ,  8 ), including a 2025 meta-analysis of eight observational studies that reported a positive association between a Mediterranean Diet and semen quality ( 26 ). While the body of literature on female preconception dietary patterns in relation to clinical fertility outcomes is larger, the evidence is mixed. Greater adherence to a Mediterranean Diet and the HEI-2010 was associated with greater fecundability in a cohort of 5,803 healthy U.S. females, but not in a cohort of 3,429 healthy Danish females ( 27 ). In the prospective EARTH Study, adherence to a Mediterranean Diet pattern, but not the alternative HEI, was positively associated with live birth among 357 females undergoing ART ( 15 ). However, an updated study of 612 females in the EARTH Study reported no associations of eight preconception dietary patterns, including the HEI-2015 and three versions of a Mediterranean Diet, with live birth following ART ( 28 ). Finally, in a cohort of 244 Greek females undergoing IVF, adherence to a Mediterranean Diet was associated with a higher likelihood of live birth ( 29 ), but no association between a Mediterranean Diet and live birth was observed in a cohort of 474 Italian females undergoing IVF ( 30 ). To our knowledge, there has only been one previous study examining both female and male partners’ adherence to a traditional healthy dietary pattern in relation to a clinical fertility outcome, which observed that preconception adherence to Dutch dietary recommendations by female partners, but not males, was associated with higher likelihood of ongoing pregnancy 6 months after IVF ( 31 ). Importantly, previous studies have been small in size and largely restricted to couples undergoing IVF at a single center. In contrast, the multi-center FAZST/IDEAL study is one of the largest preconception cohorts of couples experiencing infertility, and fewer than 20% of enrolled couples underwent IVF. Consequently, our findings may be more generalizable to the broader population of U.S. couples seeking infertility treatment and, therefore, more informative for guiding couples toward dietary patterns that may improve their likelihood of having a live birth.\nOur study contributes to a growing literature dedicated to fertility-focused dietary patterns. Unlike the traditional healthy dietary patterns, for which adherence by either or both partners was shown to be modestly beneficial, the fertility-focused patterns were not associated with a higher likelihood of live birth, with one exception: greater adherence to the Fertility Diet by females was associated with a modest increase in the likelihood of live birth. This association was stronger in the subset of 564 females with an infertility diagnosis than it was among females without a diagnosis, and stronger yet among females with PCOS or anovulation. This suggests that the increase in the likelihood of live birth may be explained by improved ovulatory function in females with anovulatory infertility at baseline. This aligns with findings from the study that first introduced the Fertility Diet, in which greater adherence to the Fertility Diet was prospectively associated with a lower risk of ovulatory disorder infertility among 17,544 women in the Nurse’s Health Study II ( 16 ). Notably, in the aforementioned study of 357 women undergoing ART in the EARTH Study, there was no association between adherence to the Fertility Diet and live birth; however, adherence to a Pro-fertility Diet was associated with a higher likelihood of live birth and lower odds of pregnancy loss ( 15 ). In our sample, adherence to the Pro-fertility Diet was not associated with live birth or pregnancy loss. This may be explained by between-sample differences in intake of the pattern’s underlying components. More likely, perhaps, the findings from the females in the EARTH cohort, among which 92% had a college degree and the average BMI was 24 kg/m 2  ( 15 ), do not generalize beyond a population of relatively healthy, high-socioeconomic status females undergoing ART.\nImportantly, the beneficial associations of dietary patterns with live birth were limited to the couples who underwent OI/IUI or no treatment; there was no such association among couples who underwent IVF. A similar finding was reported in a recent study of female and male supplement use in association with live birth in FAZST/IDEAL ( 32 ). This has important implications for the dietary guidance provided to couples experiencing infertility. For those undergoing the most invasive treatments, improvements in diet quality may have little or no effect on outcomes.\nFuture studies of the relationship of dietary patterns with clinical fertility outcomes should consider other dietary patterns, such as the ketogenic diet, paleo diet, and intermittent fasting, as well as the implications of GLP-1 therapies on these relationships.\nThis study had limitations. First, the couples in FAZST/IDEAL were predominantly non-Hispanic White and highly educated, potentially limiting generalizability to non-White couples and those of lower socioeconomic status. Notwithstanding, these couples underwent a variety of infertility treatments, not just IVF, overcoming a key limitation of previous studies and making these findings generalizable to a broader population of couples experiencing infertility. Second, the self-reported data used to assess dietary patterns are prone to measurement error ( 33 ,  34 ). However, any bias from dietary measurement error is expected to have been non-differential with respect to outcomes, which were identified prospectively. Moreover, despite their limitations, dietary recall instruments make assessment of diet in a large study population, like FAZST/IDEAL, possible. Finally, as this study was observational, we cannot rule out residual confounding. Nonetheless, we accounted for many potential confounding factors, and the overall quality of our data is high, having been collected with validated instruments in the context of a randomized trial with high rates of follow-up ( 18 ).\n\nAmong couples experiencing infertility, greater preconception adherence to traditional healthy dietary patterns by either or both partners may improve the likelihood of live birth. In contrast, a fertility-focused pattern may yield comparable benefits only when followed by a female partner. Because they benefit both partners and are likely easier to follow, traditional healthy patterns provide practical advantages over more specific “fertility diets” for couples experiencing infertility.","source_license":"CC-BY-4.0","license_restricted":false}