Higgins, JohnPetherbridge, Rachel2026-10-1920262026-08-102026Petherbridge, Rachel. 2026. Characterizing hematologic dynamics in pregnancy towards improving adverse obstetric outcome prediction. Doctoral Dissertation, Harvard University Graduate School of Arts and Sciences.32855851https://dash.harvard.edu/handle/1/42760600The utility of routine blood measurements to assess health status is undermined in pregnancy because pregnancy-specific hematologic changes remain incompletely characterized and are often not accounted for. Standards of care for pregnant individuals are therefore often reduced or altered compared to non-pregnant counterparts. In this work, I characterize longitudinal hematologic dynamics in pregnancy and use these dynamics to adjust two routine blood measurements --- the complete blood count (CBC) and hemoglobin A1c --- for improved use in pregnancy. Specifically, I assess unusual changes in CBC indices as markers for difficult-to-predict adverse pregnancy outcomes, and use personalized CBC measurements to improve hemoglobin A1c as a potential diagnostic for gestational diabetes. In Chapter 2, I define reference intervals and typical longitudinal changes in CBC indices during pregnancy in a large U.S. cohort based at an academic health system. I then tested for associations between extreme CBC values for gestational age or extreme longitudinal changes in CBC indices and obstetric complications, including preterm delivery, hypertensive disorders of pregnancy (including preeclampsia) and small for gestational age infants. I found that variations in CBC indices during pregnancy regularly exceeded expected biological variation of such indices outside of pregnancy. Using generalized estimating equations to fit logistic regressions, I additionally found that high or increased CBC values were associated with adverse outcomes of interest, and that these signals preceded outcomes by several weeks. Findings were validated in a large, out-of-sample cohort at the same academic health system. This was a surprising finding, as current standard of care focuses heavily on preventing low values of CBC indices, such as low hemoglobin. Because CBC indices are routine in pregnancy care, thresholds from our associations may be easily incorporated into clinical practice for stratified risk assessment of assessed adverse outcomes. Building on the finding that CBC indices capture clinically meaningful pregnancy physiology, Chapter 3 uses these indices to improve interpretation of hemoglobin A1c, a biomarker whose performance in pregnancy is limited by altered red blood cell dynamics. Hemoglobin A1c --- the percentage of glycated hemoglobin in the blood --- represents average blood glucose across several months. Measurements are pre-analytically stable and tolerant to short term illness, stress, and dietary changes. A1c is prognostic for adverse outcomes of diabetes outside of pregnancy, but has not been adopted for use in gestational diabetes --- diabetes specific to pregnancy --- diagnosis due to its sensitivity to changes in red blood cell dynamics and turnover. This is reflected in a weaker relationship between A1c and other measures of maternal glycemia, such as average glucose from continuous glucose monitoring (AG) and laboratory measured fasting plasma glucose (FPG). Gestational diabetes is therefore diagnosed with an oral glucose tolerance test (OGTT), requiring individuals to fast for eight hours, consume a bolus of glucose and have subsequent timed blood draws for several hours; OGTTs are labor intensive, suffer from day-to-day perturbations, and frequently cause nausea and vomiting. In Chapter 3, I adjust A1c measurements with individual CBC indices to address underlying red blood changes and improve A1c's relationship with AG and FPG. I compare the ability of my best resulting model to predict large for gestational age infants (an adverse outcome associated with gestational diabetes) to the OGTT. I found my model to be statistically non-inferior, indicating that adjusted A1c is a viable alternative to the OGTT. This represents the first attempt to adjust A1c for altered red blood cell kinetics for the purposes of improving its utility for gestational diabetes diagnosis in pregnancy. Work to further capture red cell kinetics for the purposes of both understanding basic pregnancy physiology and improving A1c as a biomarker is continued in Chapter 4; in this chapter, I propose an ordinary differential equation model of red cell kinetics with which I aim to uncover physiological mechanisms of red blood dynamic changes. Though this project is ongoing, I determine that increased red blood cell production characteristic of pregnancy is alone insufficient to explain observed laboratory measurements of red cell count and mean red cell age. Continued research in this space may uncover additional methods by which A1c can be adjusted for use in pregnancy. Taken together, this dissertation offers new insight into hematologic dynamics in pregnancy, and novel methods to predict adverse pregnancy outcomes. These methods are based on already-routine laboratory blood measurements, and therefore may be quickly implemented into clinical practice.application/pdfencomplete blood countgestational diabeteshemoglobin A1cPregnancyPathologyObstetrics and gynecologyEndocrinologyThesis or Dissertation2026-10-190000-0002-3236-5587