Table 2.

Key outcome-specific considerations for study evaluation^a.

Category	Outcome	Outcome ascertainment	Exposure (relevant time window)	Population selection	Confounding (variables to consider)	Analysis
Pregnancy outcomes	Preterm birth	● Method used to estimate gestational duration (e.g., early pregnancy dating with ultrasound preferred to birth certificate data)	● Must be measured during pregnancy (preferred) or preconception.	● Timing of study entry (1st trimester preferred).	● Age (u-shaped), gender, pre-pregnancy BMI/adiposity, lifestyle factors (physical activity, diet, smoking, alcohol), SES, co-exposures, pregnancy interval, maternal reproductive and clinical factors	● Outcome treated as dichotomous or three-level variable preferred.
	Spontaneous abortion	● Ascertainment of early loss (through daily urine samples) preferred ● For clinical loss (> 5 weeks gestation), prospective ascertainment of pregnancy and loss preferred to recall or medical records	● Must be prior to outcome, ideally around conception	● Pregnancy planning couples preferred for identifying early loss. ● Population-based or clinic-based with discussion of catchment area preferred to convenience sample ● Recruitment prior to end of first trimester preferred	● Age, smoking, alcohol, parity, gravidity, SES, BMI/adiposity, fertility treatment, pregnancy interval, access to healthcare, co-exposures. If paternal exposure is the focus, maternal covariates should also be considered.	● Outcome treated as dichotomous variable preferred, with survival analysis for prospective designs.
Male reproductive	Reproductive hormones	● Time of collection (morning preferred) ● Information on laboratory assays, quality control procedures	● Can be measured concurrent with outcome.	● Appropriate comparison group (e.g., similar referral patterns)	● Age, SES, smoking, alcohol use, BMI/adiposity, time of day, co-exposures	● Outcome treated as continuous variable preferred.
	Semen parameters (concentration, motility, morphology)	● Semen analysis on one or more samples. ● Manual ascertainment of morphology preferred. ● Abstinence time should be collected and considered. ● Time between collection and analysis (influences motility) ● Information on laboratory methods, quality control procedures	● For adult exposures, exposure measured concurrent with outcome is acceptable. ● Must be before identification of infertility problem unless exposure does not change over time	● Selection at setting other than infertility clinic preferred. ● Sample not limited to volunteers with known fertility problems unless similar selection in comparison group.	● Age, smoking, BMI, chronic disease status, abstinence time, co-exposures	● Outcome treated as continuous or dichotomous (with justified cut-points) variable.
Male or female reproductive	Pubertal development	● Tanner staging or physical measurements (e.g., testicular volume) by trained physician or investigator preferred	● Must be measured before pubertal onset (prenatal or childhood appropriate)	● Population must span the relevant age range (approximately 11–15 years)	● Age, race/ethnicity, SES, diet, obesity, family history of early or late pubertal onset, co-exposures	● Outcome may be treated as time to event (e.g., puberty onset) or ordinal (e.g., Tanner stage)
	Time to pregnancy	● Prospective measurement of couples discontinuing contraception is preferred. ● If recall is used, should be number of cycles or months, and recall time < 10 years is preferred. ● Designs where recall is only among pregnant women or where data is collected in categories (< 3 mo., 3–6 mo., etc.) are Poor. ● Assessment of time to pregnancy from medical records (e.g., notation about history of infertility) not acceptable.	● Must be prior to conception unless exposure does not vary over time or with pregnancy. ● For women: prior to attempt to conceive, including in utero ● For men: Period of spermatogenesis for attempted conception	● Couples with no known fecundity impairments strongly preferred. ● Sample not limited to couples that achieved pregnancy. ● Population-based sampling preferred. Other sampling should be unbiased (e.g., not infertility clinic).	● Age (maternal and paternal), smoking, BMI, SES, co-exposures	● Outcome treated as time to event preferred.
Diabetes-related	Diabetes	● Clear definition of diabetes (e.g., ADA); self-report not sole criteria ● Fasting requirements ● Information on laboratory assays, quality control procedures	● Must be measured before diabetes onset.	● Exclude diabetics at baseline ● Cohort followed for sufficient period to allow development of disease	● Age, gender, BMI/adiposity, lifestyle factors (physical activity, diet, smoking, alcohol), SES, co-exposures	● Outcome treated as dichotomous or three-level variable preferred.
	Insulin resistance	● Absence of diabetes ● HOMA-IR and HOMA-β, or fasting insulin and fasting glucose ● Fasting requirements ● Information on laboratory assays, quality control procedures	● Can be measured concurrent with outcome.	● Exclude individuals with known diabetes	● Age, gender, BMI/adiposity, lifestyle factors (physical activity, diet, smoking, alcohol), SES, co-exposures	● Outcome treated as continuous variable preferred.
Asthma and allergies	Asthma	● Self-report using validated questionnaires for physician diagnosed asthma incidence or outcome prevalence (e.g., ATS, ISAAC) or medical record review/physician confirmation for incidence; ● Validation of questionnaires in same population preferred.	● Up to two years prior to ascertainment of incidence or concurrent with prevalence or related outcomes	● Population-based sampling preferred; adults or children > 4 years of age. Consider healthy worker effect in occupational studies.	● Age, sex, race, socioeconomic status, season of outcome measurement (for prevalence and related outcomes), geographic location, co-exposures	● Outcome is dichotomous; effect modification by age (child vs adult) preferred if appropriate
	Allergic sensitization and allergy outcomes	● Ascertainment of sensitization via skin prick test or specific immunoglobulin E levels in blood using validated methods; use of standardized cut-points, distinguishing between food and environmental allergens, and testing a panel of 5–10 allergens preferred. ● Ascertainment of site-specific symptoms and outcomes using standardized questionnaires, distinguishing food and other allergy; preferably validated in same population. Ascertainment via self-report of medication use must clarify type of medication.	● Must precede sensitization or symptom onset; or concurrent with prevalence or symptom ascertainment	● Studies among children preferred for sensitization or symptom onset (aged > 4 years). ● Self-reported symptoms preferred over restricting to physician diagnosed allergic disease.	● Age, sex, race, socioeconomic status, season of outcome measurement (for prevalence and related outcomes), geographic location, co-exposures	● Stratified analyses of children and adults; age analyzed as effect modifier for studies of sensitization

Abbreviations: ADA (American Diabetes Association), HOMA-IR (Homeostatic model assessment of insulin resistance), HOMA-β (homeostatic model assessment of β-cell function), ATS (American Thoracic Society), ISAAC (International Society of Arthritis and Allergies in Children).

^aClassification of the considerations in this table into levels (Good, Adequate, Deficient, and Critically deficient) requires additional considerations that apply to all outcomes and outcome-specific considerations. Rationale for the considerations are discussed in the full outcome-specific evaluation criteria (supplementary materials).