Social determinants of health diagnosis codes in in-person and telehealth visits during pregnancy in the United States, 2020-2025: a repeated cross-sectional and case-control study.

backgroundSocial determinants of health (SDoH) diagnosis codes (also known as Z-codes) are used to document patients’ SDoH, which could affect a patient’s health. However, their use during pregnancy is unknown. Additionally, with the growth in telehealth use in recent years, how the Z-codes are documented across in-person outpatient/office and telehealth prenatal visits is undetermined. Therefore, the objective of this study was to document trends in prenatal Z-code diagnosis in telehealth (audio-only and audio-visual) and in-person outpatient/office visits, and the association between visit type and Z-codes during pregnancy.

methodsRepeated cross-sectional and case-control designs were employed using the Epic Cosmos database (July 2019–March 2025). All pregnancies with start dates on or after July 01 2020 and visits until March 31, 2025 were considered. Total in-person outpatient/office and telehealth visits (denominators) were summed, and visits with a Z-code were determined (numerators). Rates were expressed per 1,000 visits. Case-control analyses, stratified by provider type, were conducted to estimate the relationship between visit type and Z-code incidence. The first documentation of Z-codes (a wash-out period of 365 days) in each pregnancy episode was identified (case visits), and control visits were gathered. Multivariable logistic regressions were estimated and odds ratios (OR) were reported.

resultsA total of 3,435,577 pregnancies were included, totaling 33,376,363 in-person outpatient and telehealth visits. Overall average Z-code rates per 1,000 visits were: 8.64 (in-person), 6.71 (audio-only), and 12.35 (audio-visual). The rates increased on average for all three visit types over the study period, with rates slightly higher for audio-visual visits. In the provider sub-group of family practice/ob-gyn/other clinicians, audio-visual visits had no significant association with Z-code incident diagnosis (OR: 1.04 [0.99; 1.10]), while audio-only visits had lower odds (OR: 0.60 [0.47; 0.78]), compared to in-person visits. Among nursing/midwife provider sub-group, audio-visual visits were associated with 42% higher odds (OR: 1.42 [1.35; 1.49]), while audio-only visits were associated with 48% lower odds (OR: 0.52 [0.43; 0.63]), compared to in-person visits.

conclusionAudio-visual visits with nurses/midwives were better at capturing SDoH measures. The integration of audio-visual visits with these providers for routine follow-up during pregnancy could better capture patients’ SDoH.