Objective
The purpose of this study was to examine the relationship between typically measured prenatal screening biomarkers and early-preterm birth in euploid pregnancies.
Study Design
The study included 345 early-preterm cases (<30 weeks of gestation) and 1725 control subjects who were drawn from a population-based sample of California pregnancies who had both first- and second-trimester screening results. Logistic regression analyses were used to compare patterns of biomarkers in cases and control subjects and to develop predictive models. Replicability of the biomarker early-preterm relationships that was revealed by the models was evaluated by examination of the frequency and associated adjusted relative risks (RRs) for early-preterm birth and for preterm birth in general (<37 weeks of gestation) in pregnancies with identified abnormal markers compared with pregnancies without these markers in a subsequent independent California cohort of screened pregnancies (n = 76,588).
Results
The final model for early-preterm birth included first-trimester pregnancy-associated plasma protein A in the ≤5th percentile, second-trimester alpha-fetoprotein in the ≥95th percentile, and second-trimester inhibin in the ≥95th percentile (odds ratios, 2.3–3.6). In general, pregnancies in the subsequent cohort with a biomarker pattern that were found to be associated with early-preterm delivery in the first sample were at an increased risk for early-preterm birth and preterm birth in general (<37 weeks of gestation; adjusted RR, 1.6–27.4). Pregnancies with ≥2 biomarker abnormalities were at particularly increased risk (adjusted RR, 3.6–27.4).
Conclusion
When considered across cohorts and in combination, abnormalities in routinely collected biomarkers reveal predictable risks for early-preterm birth.
All biomarkers that are used in routine aneuploidy screening are directly or indirectly associated with placental function, pregnancy maintenance, and/or other processes that are tied closely to preterm birth (eg, parturition, placental and trophoblast function, inflammation, immune system function). Thus, there is pathophysiologic evidence that supports the findings of a number of investigators who have reported an increased risk of preterm birth when ≥1 routinely collected screening markers are abnormally high and/or low (first-trimester nuchal translucency [NT], pregnancy-associated plasma protein-A [PAPP-A], and human chorionic gonadotropin [hCG], second-trimester alpha-fetoprotein [AFP], hCG, unconjugated estriol [uE3], and inhibin). Despite these observations, the standard of care for pregnancies with abnormal biomarkers is uncertain. One challenge in creating a set of standards is the absence of well-defined population-scale data that has investigated preterm delivery by important clinical subgroups (eg, early, spontaneous, medically indicated) in conjunction with biomarker patterns across trimesters.
Herein, we used data from the California Prenatal Screening Program and the California Perinatal Quality Care Collaborative (CPQCC) to investigate whether preterm birth (overall and by medically indicated and spontaneous labor subgroups) is associated with single and multiple biomarker abnormalities. Two independent population-scale sample sets of euploid singleton pregnancies were used: one population set was used to establish an association model, and one population set was used to determine whether the patterns could be recapitulated across cohorts.
Materials and Methods
Evaluation of early-preterm biomarker relationships was undertaken in 2 independent datasets; one set was used for model building (the “training” study set), and one set was used for model testing (the “testing” study set). The training study set included 345 early-preterm singleton cases (<30 weeks of gestation) and 1725 term singleton pregnancies (control subjects) with expected dates of delivery in September 2009 through December 2010. These cases and control subjects were drawn from 497,023 unique women who were participants in the California Prenatal Screening Program during this same time period. Cases and control subjects were restricted to pregnancies with ultrasound dating, maternal age between 12 and 60 years, no missing information on race/ethnicity, and sequential integrated screening results (that is, pregnancies with first-trimester NT, PAPP-A, and hCG measurements and second-trimester measures of AFP, hCG, uE3, and inhibin; n = 119,185). Cases and control subjects were also restricted to pregnancies with a linked newborn screening record (indicating a live birth between 20 and 44 completed weeks of gestation) without any history of diabetes mellitus or smoking and without chromosomal or neural tube defects in registries that are maintained by the Genetic Disease Screening Program. We identified 643 case pregnancies that had resulted in early-preterm birth between 22 weeks 0 days and 29 weeks 6 days of gestation and 83,039 control pregnancies with births ≥37 completed weeks of gestation. The final case determination was made after linkage of the case group to the CPQCC dataset. The CPQCC database stores clinical data on >90% of all neonates who receive neonatal intensive care in California. All newborn infants with a gestational age between 22 weeks 0 days and 29 weeks 6 days qualify for inclusion in the CPQCC, regardless of department of care within partner hospitals. This set of early-preterm pregnancies was ideal for more intensive analyses because of the availability of extensive data on pregnancies. The CPQCC dataset was used to make additional exclusions from the early-preterm case grouping ( Figure 1 ). The final case-control set included the 345 cases after CPQCC linkage and exclusions and 1725 control subjects who were selected randomly from the available 83,309 term pregnancies at a ratio of 5 control subjects for each case.
First-trimester PAPP-A and total hCG were measured in serum samples that had been drawn between 10 weeks 0 days and 13 weeks 6 days of gestation. Second-trimester AFP, hCG, uE3, and inhibin were measured in serum samples that were drawn between 15 weeks 0 days and 20 weeks 0 days of gestation. NT measurements were done between 11 weeks 2 days and 14 weeks, 2 days of gestation by practitioners who were credentialed by the Nuchal Translucency Quality Review Program or Fetal Medicine Foundation. All serum samples were sent to 1 of 7 regional laboratories in California for testing with fully automated equipment (Auto DELFIA; Perkin Elmer Life Sciences, Waltham, MA). As part of routine prenatal screening, all biomarker levels were converted to biomarker multiples of the medians (MoM) to adjust for gestational age with log-linear or nonlinear regression methods, as appropriate; median analyte values were regressed against gestational age and were adjusted for maternal weight (as a proxy for blood volume) and self-reported race/ethnicity
Our analyses used logistic regression to calculate odds ratios (ORs). We estimated the odds that were associated with specific maternal characteristics in 3 early-preterm case groups (spontaneous labor, medically indicated, and combined groupings) compared with those in the term control grouping. Preterm groupings were based on information from the CPQCC. Maternal characteristics included self-reported white, Hispanic, black, Asian, and “other” race/ethnicity; maternal age <18, 18-34, and >34 years, and an obesity proxy variable (maternal weight by race/ethnicity grouping at weeks of gestation at initial first-trimester serum screen >95th percentile). Such a proxy was needed for this dataset given the absence of data on height and/or body mass index. Weight percentiles were computed with the full sample of pregnancies with screening during the same period (n = 119,185 pregnancies).
Logistic regression models also were used to compare the odds associated with first- and second-trimester biomarker abnormalities in early-preterm cases relative to term control subjects and to build final predictive models. For NT, a measurement at ≥3.5mm was considered abnormal ; for serum markers, MoM of ≤5th or ≥95th percentile were considered abnormal based on the distribution of markers in the full sample of screened pregnancies (n = 119,185). All maternal characteristics and biomarker groupings that were found to be associated with early-preterm birth (by spontaneous labor, medically indicated, and combined grouping) in crude analyses ( P < .10) were entered into the full model. Final models included maternal characteristics and biomarkers that remained significant ( P < .05) for early-preterm birth when backward stepwise methods were applied.
The testing study set was used to test replicability of the biomarker early-preterm relationships revealed by the models in the training study set. The testing study set included all California singleton pregnancies with ultrasound dating that received screening from January through September 2011 for whom there was no indication of diabetes mellitus, smoking, aneuploidy, or neural tube defects in Genetic Disease Screening Program records (n = 76,588 pregnancies). Other detailed information about obstetric risks and reasons for early-preterm birth was not available for this population. As such, evaluation was restricted to a focus on the combined early-preterm model. Generalizability was evaluated by logistic binomial regression methods to examine the frequency and associated adjusted relative risks (RRs) and 95% confidence intervals (CIs) for early-preterm birth and all preterm births (<37 weeks of gestation) in pregnancies with identified abnormal markers compared with those without these markers. The adjusted models included all characteristics that were found to be significantly more or less frequent in the combined early-pre term grouping compared with the term grouping.
The analyses were done with Statistical Analysis Software (version 9.2; SAS Institute Inc, Cary, NC) and were based on data received by the Genetic Disease Screening Program by Sept. 1, 2012. The methods and protocols were approved by the Committee for the Protection of Human Subjects within the Health and Human Services Agency of the State of California and the Institutional Review Board of Stanford University.
Results
Early-preterm cases and term control subjects in the training study set were mostly Hispanic (43.2% and 40.5%, respectively) and were between 18 and 34 years old (72.8% and 72.1%, respectively). Pregnancies that resulted in early-preterm birth were more likely than term control subjects to be black, regardless of early-preterm grouping (spontaneous labor, medically indicated, or combined; OR, 2.5 and 4.3; Table 1 ).
| Characteristic | Early-preterm birth (<30 weeks of gestation) | Term control subjects, n (%) | |||||
|---|---|---|---|---|---|---|---|
| Spontaneous labor a | Medically indicated b | Combined | |||||
| n (%) | OR (95% CI) | n (%) | OR (95% CI) | n (%) | OR (95% CI) | ||
| All | 222 | 123 | 345 | 1725 | |||
| Race/ethnicity | |||||||
| White | 57 (25.7) | Reference | 40 (32.5) | Reference | 97 (28.1) | Reference | 598 (34.7) |
| Hispanic | 95 (42.8) | 1.4 (1.0–2.0) | 54 (43.9) | 1.2 (0.8–1.8) | 149 (43.2) | 1.3 (1.0–1.7) | 699 (40.5) |
| Black | 29 (13.1) | 4.3 (2.6–7.1) | 12 (9.8) | 2.5 (1.3–5.0) | 41 (11.9) | 3.6 (2.3–5.5) | 71 (4.1) |
| Asian | 23 (10.4) | 1.2 (0.7–1.9) | 0.9 (0.5–1.8) | 13 (10.6) | 36 (10.4) | 1.1 (0.7–1.6) | 207 (12.0) |
| Other | 18 (8.1) | 1.3 (0.7–2.2) | 4 (3.3) | 0.4 (0.1–1.1) | 22 (6.4) | 0.9 (0.6–1.5) | 149 (8.6) |
| Maternal age at term, y | |||||||
| <18 | 4 (1.8) | 3.2 (0.9–10.6) | 0 | — | 4 (1.2) | 2.2 (0.7–7.2) | 9 (0.5) |
| 18-34 | 170 (76.6) | Reference | 81 (65.9) | Reference | 251 (72.8) | Reference | 1243 (72.1) |
| >34 | 47 (21.2) | 0.7 (0.5–1.0) | 42 (34.2) | 1.4 (0.9–2.0) | 89 (25.8) | 0.9 (0.7–1.2) | 472 (27.4) |
| Maternal weight (percentile) c | |||||||
| <5th | 11 (5.0) | 1.1 (0.6–2.2) | 1 (0.8) | — d | 12 (3.5) | 0.8 (0.4–1.4) | 78 (4.5) |
| 5-95th | 195 (87.8) | Reference | 111 (90.2) | Reference | 306 (88.7) | Reference | 1568 (90.9) |
| >95th | 15 (6.8) | 1.5 (0.9–2.7) | 10 (8.1) | 1.8 (0.9–3.6) | 25 (7.3) | 1.6 (1.0–2.6) | 79 (4.6) |
a Includes 88 pregnancies with premature rupture of the membranes;
b Includes pregnancies with indication of “fetal distress” (n = 25), maternal hypertension (n = 60), placental problems (n = 29), cardiac disease (n = 1), HELLP (hemolysis, elevated liver enzymes, and low platelet count syndrome) syndrome (n = 2), prolapsed cord (n = 3), fetal bradycardia (n = 1), pulmonary edema (n = 2), “1-month urinary tract infection with chorioamnionitis” (n = 1), and “severe intrauterine growth restriction” (n = 1); premature rupture of the membranes was indicated in 13 of these pregnancies;
c By race/ethnicity grouping at weeks of gestation at initial testing;
Cases in the training study set (with spontaneous or medically indicated preterm delivery) were significantly more likely than control subjects to have a first-trimester PAPP-A MoM ≤5th percentile and second-trimester AFP and/or an inhibin MoM of ≥95th percentile ( Table 2 ). Medically indicated cases were more likely to have an hCG MoM in the first and second trimester than were at ≥95th percentile and a second-trimester uE3 MoM at ≤5th percentile ( Table 2 ).
| Biomarker | Early-preterm birth (<30 weeks of gestation) | Term control subjects, n (%) | |||||
|---|---|---|---|---|---|---|---|
| Spontaneous labor | Medically indicated | Combined | |||||
| n (%) | OR (95% CI) a | n (%) | OR (95% CI) a | n (%) | OR (95% CI) a | ||
| All | 222 | 123 | 345 | 1725 | |||
| First trimester | |||||||
| Nuchal translucency ≥3.5 mm b | 1 (0.5) | — c | — | — | 1 (0.3) | — c | 2 (0.1) |
| Pregnancy-associated plasma protein A (multiple of the median percentile) d | |||||||
| ≤5th | 20 (9.0) | 2.1 (1.3–3.5) | 17 (13.8) | 3.2 (1.8–5.6) | 37 (10.7) | 2.5 (1.6–3.7) | 81 (4.7) |
| 6th-94th | 189 (85.1) | Reference | 102 (82.9) | Reference | 291 (84.4) | Reference | 1562 (90.6) |
| ≥95th | 13 (5.9) | 1.2 (0.7-2.3) | 4 (3.3) | 0.7 (0.3-2.1) | 17 (4.9) | 1.1 (0.6-1.8) | 82 (4.8) |
| Human chorionic gonadotropin (multiple of the median percentile) d | |||||||
| ≤5th | 7 (3.2) | 0.5 (0.3–1.2) | 8 (6.5) | 1.2 (0.6–2.6) | 15 (4.4) | 0.7 (0.4–1.3) | 98 (5.7) |
| 6th-94th | 203 (91.4) | Reference | 104 (84.6) | Reference | 307 (89.0) | Reference | 1549 (89.8) |
| ≥95th | 12 (5.4) | 0.9 (0.4–2.2) | 11 (8.9) | 2.1 (1.1–4.2) | 23 (6.7) | 1.6 (1.0–2.6) | 78 (4.5) |
| Second trimester | |||||||
| Alpha-fetoprotein (multiple of the median percentile) d | |||||||
| ≤5th | 8 (3.6) | 0.6 (0.3–1.3) | 6 (4.9) | 1.1 (0.5–2.6) | 14 (4.1) | 0.8 (0.5–1.4) | 95 (5.5) |
| 6th-94th | 189 (85.4) | Reference | 84 (68.3) | Reference | 307 (89.0) | Reference | 1556 (90.2) |
| ≥95th | 25 (11.3) | 2.8 (1.7–4.6) | 33 (26.8) | 8.3 (5.2–13.2) | 58 (16.8) | 4.5 (3.1–6.5) | 74 (4.3) |
| Human chorionic gonadotropin (multiple of the median percentile) d | |||||||
| ≤5th | 7 (3.2) | 0.5 (0.2–1.0) | 6 (4.9) | 0.9 (0.4–2.1) | 13 (3.8) | 0.6 (0.3–1.1) | 109 (6.3) |
| 6th-94th | 200 (90.1) | Reference | 95 (77.2) | Reference | 295 (85.5) | Reference | 1543 (89.5) |
| ≥95th | 15 (6.8) | 1.5 (0.9–2.8) | 22 (17.9) | 4.7 (2.8–8.0) | 37 (10.7) | 2.6 (1.7–4.0) | 73 (4.2) |
| Unconjugated estriol (multiple of the median percentile) d | |||||||
| ≤5th | 5 (2.3) | 0.5 (0.2–1.2) | 16 (13.0) | 2.9 (1.6–5.1) | 21 (6.1) | 1.3 (0.8–2.1) | 84 (4.9) |
| 6th-94th | 201 (90.5) | Reference | 103 (83.7) | Reference | 304 (88.1) | Reference | 1562 (90.6) |
| ≥95th | 16 (7.2) | 1.6 (0.9–2.8) | 4 (3.3) | 0.7 (0.3–2.0) | 20 (5.8) | 1.3 (0.8–2.1) | 79 (4.6) |
| Inhibin (multiple of the median percentile) d | |||||||
| ≤5th | 10 (4.5) | 0.8 (0.4–1.6) | 3 (2.4) | 0.6 (0.2–1.9) | 13 (3.8) | 0.7 (0.4–1.2) | 97 (5.6) |
| 6th-94th | 193 (86.9) | Reference | 80 (65.0) | Reference | 273 (79.1) | Reference | 1550 (89.9) |
| ≥95th | 19 (8.6) | 1.9 (1.1–3.3) | 40 (32.5) | 9.9 (6.4–15.5) | 59 (17.1) | 4.2 (2.9–6.1) | 78 (4.5) |
a Included in the models were black race/ethnicity (all groups) and maternal weight (percentile) >95th percentile (combined group; characteristics found to be more frequent in cases compared with control subjects [ Table 1 ]);
b Reference group was <3.5 mm;
c Not computed where cell frequency was <3;
d Cut points for multiples of the median percentiles (≤5th and ≥95th ) were pregnancy-associated plasma protein A, 0.38 and 2.45; human chorionic gonadotropin (first trimester), 0.52 and 1.99; alpha-fetoprotein, 0.59 and 1.68; human chorionic gonadotropin (second trimester), 0.42 and 2.24; unconjugated estriol, 0.62 and1.36; inhibin, 0.54 and 2.05.
With the use of a backward stepwise approach, we found that cases in the training study set (regardless of spontaneous or medically indicated) were more than twice as likely to have PAPP-A MoM at ≤5th percentile (OR, 2.0 and 2.4) and >3 times as likely to have AFP MoM ≥95th percentile (OR, 3.2 and 4.5). Inhibin MoM of ≥95th was statistically dropped from the spontaneous labor model but remained in the medically indicated and combined models (OR, 6.4 and 3.2, respectively; Table 3 ). The findings across models remained when cases were limited to those without premature rupture of membranes, hypertension, or small for gestational age (OR after exclusion, 2.1 and 6.1, respectively).
| Variable | Odds ratio a | 95% CI |
|---|---|---|
| Model no. 1: spontaneous labor | ||
| First-trimester PAPP-A MoM ≤5th | 2.0 | 1.2–3.5 |
| Second-trimester AFP MoM ≥95th | 3.2 | 2.0–5.3 |
| Model no. 2: medically indicated | ||
| First-trimester PAPP-A MoM ≤5th | 2.4 | 1.2–4.9 |
| Second-trimester AFP MoM ≥95th | 4.5 | 2.5–8.1 |
| Second-trimester inhibin MoM ≥95th | 6.4 | 3.8–10.9 |
| Model no. 3: combined | ||
| First-trimester PAPP-A MoM ≤5th | 2.3 | 1.4–3.6 |
| Second-trimester AFP MoM ≥95th | 3.6 | 2.4–5.5 |
| Second-trimester inhibin MoM ≥95th | 3.2 | 2.1–4.8 |
a Included in the models were black race/ethnicity (all groups) and maternal weight (percentile) >95th percentile (combined group; characteristics found to be more frequent in cases compared with control subjects [ Table 1 ]).
Estimated risks (OR and 95% CI) that were associated with a PAPP-A MoM of ≤5th percentile, an AFP MoM of ≥95th percentile, and/or an inhibin MoM of ≥95th percentile in cases in the training study set (OR, 2.4 and 3.2) were similar in magnitude to their counterparts (RR and 95% CI) that were observed in the testing study set that measured risks of early-preterm birth in pregnancies with ≥1 of these target biomarker patterns compared with pregnancies without any of these patterns (RR, 2.4 and 3.6; Table 4 ; Figure 2 ). Pregnancies with these patterns (considered as the combined grouping or in isolation or in combination with other markers) were also found to be at increased risk of preterm birth occurring at <37 weeks’ gestation (RR, 1.6 and 9.2; 95% CI, 1.3–12.3; Table 4 ). The 1 exception was for pregnancies with an isolated first-trimester PAPP-A MoM of ≤5th percentile (without second-trimester AFP and/or inhibin MoM ≥95th percentile). These pregnancies were not at increased risk for preterm birth <30 weeks of gestation (adjusted RR, 1.4; 95% CI, 0.7–3.0; Table 4 ).
| Variable | Preterm birth | |||
|---|---|---|---|---|
| <30 wk | <37 wk | |||
| n (%) | Adjusted relative risk (95% CI) a | n (%) | Adjusted relative risk (95% CI) a | |
| Total sample (n = 76,588) | 422 (0.6) | 4734 (6.2) | ||
| No target marker pattern b (n = 66,113) | 296 (0.5) | Reference | 3581 (5.4) | Reference |
| Any target marker pattern b (n = 10,475) | 126 (1.2) | 2.7 (1.8–3.8) | 1152 (11.0) | 2.1 (1.9–2.3) |
| Any low PAPP-A (n = 3828) | 42 (1.1) | 2.4 (1.7–3.4) | 430 (11.2) | 2.2 (1.9–2.3) |
| Any high AFP (n = 3772) | 62 (1.6) | 3.6 (2.7–4.8) | 498 (13.2) | 2.5 (2.3–2.7) |
| Any high inhibin (n = 3828) | 63 (2.1) | 3.6 (2.7–4.8) | 433 (11.4) | 2.2 (1.9–2.4) |
| Isolated low PAPP-A (n = 3385) | 23 (0.7) | 1.4 (0.7–3.0) | 330 (9.8) | 2.0 (1.6–2.4) |
| Isolated high AFP (n = 3048) | 34 (1.1) | 2.2 (1.2–4.2) | 345 (11.3) | 2.2 (1.8–2.6) |
| Isolated high inhibin (n = 3156) | 34 (1.1) | 2.5 (1.4–4.5) | 291 (9.2) | 1.6 (1.3–2.0) |
| Low PAPP-A, high AFP only (n = 232) | 6 (2.6) | 5.8 (1.5–2.9) | 45 (19.4) | 3.6 (2.2–5.8) |
| Low PAPP-A, high inhibin only (n = 162) | 7 (4.3) | 10.0 (2.6–38.3) | 34 (21.0) | 4.2 (2.4–7.3) |
| High AFP, high inhibin only (n = 443) | 16 (3.6) | 7.1 (2.7–18.7) | 87 (19.6) | 4.2 (3.0–6.0) |
| Low PAPP-A, high AFP, and inhibin (n = 49) | 6 (12.2) | 27.4 (12.8–58.4) c | 21 (42.9) | 9.2 (5.9–14.3) |
a Adjusted for black race/ethnicity;
b Any risk based on final backward stepwise logistic model for preterm birth at <30 weeks of gestation (black race/ethnicity, first-trimester PAPP-A ≤5th percentile (low PAPP-A level), second-trimester AFP ≥95th percentile (high AFP level), second-trimester inhibin ≥95th percentile (high inhibin level);
Pregnancies with ≥2 predictive biomarker patterns were found to be at particularly increased risk for early-preterm birth and for preterm birth in general (<37 weeks of gestation; adjusted RR, 3.6 and 27.4, respectively; Table 4 ). Pregnancies with all 3 abnormal biomarker patterns (first-trimester PAPP-A of ≤5th percentile, second-trimester AFP, and inhibin of ≥95th percentile; n = 49) were at the greatest increased risk at >24-fold increased risk of early-preterm birth (crude RR, 24.4; 95% CI, 12.8–58.4) and at a >9-fold increased risk of preterm birth (<37 weeks of gestation; adjusted RR, 9.2; 95% CI, 5.9–14.3). Nearly 1 in 2 pregnancies with this pattern ended in a preterm birth (<37 weeks’ gestation) ( Figure 3 ).
Examination of maternal and infant characteristics and diagnoses in the training study set (the set with more detailed clinical data is available) indicated that several factors were significantly more frequent among groups of pregnancies with ≥2 compared with 0 or 1 abnormal biomarkers (eg, black race/ethnicity, >34 years old at testing, and hypertension). There were also some characteristics and diagnoses that were significantly less frequent among pregnancies with ≥2 compared with 0 or 1 abnormal biomarkers (eg, premature rupture of membranes and intraventricular hemorrhage; Table 5 ).
| Variable | Very preterm birth (<30 weeks) | P value b | ||
|---|---|---|---|---|
| No abnormal biomarker a | Single abnormal biomarker a | ≥2 abnormal biomarkers a | ||
| n (%) | n (%) | n (%) | ||
| Sample | 229 | 83 | 33 | |
| Maternal characteristics | ||||
| Race/ethnicity | ||||
| White | 62 (27.1) | 23 (27.7) | 12 (36.4) | .085 |
| Hispanic | 95 (41.5) | 40 (48.2) | 14 (42.4) | |
| Black | 30 (13.1) | 4 (4.8) | 7 (21.2) | |
| Asian | 26 (11.4) | 10 (12.1) | 0 | |
| Other | 16 (7.0) | 6 (7.2) | 0 | |
| Age at term, y | ||||
| <18 | 3 (1.3) | 1 (1.2) | 0 | .064 |
| 18-34 | 177 (77.3) | 54 (65.1) | 20 (60.6) | |
| >34 | 48 (21.0) | 28 (33.7) | 13 (39.4) | |
| Weight (percentile) c | ||||
| <5th | 10 (4.4) | 1 (1.2) | 0 | .245 |
| 5-95th | 205 (89.5) | 71 (85.5) | 30 (90.9) | |
| >95th | 13 (5.7) | 10 (12.1) | 2 (6.1) | |
| Pregnancy complication d | ||||
| Unknown | 2 (0.9) | 0 | 0 | 1.000 |
| None | 48 (21.2) | 8 (9.6) | 3 (9.1) | .026 |
| Hypertension | 39 (17.2) | 31 (37.4) | 22 (66.7) | < .001 |
| Premature rupture of membranes | 73 (32.2) | 24 (28.9) | 3 (9.1) | .024 |
| Bleeding | 47 (20.7) | 18 (21.7) | 10 (30.3) | .459 |
| Chorioamnionitis | 25 (11.0) | 3 (3.6) | 0 | .018 |
| Medically indicated preterm birth | 61 (26.7) | 38 (45.8) | 24 (72.7) | < .001 |
| Spontaneous labor | 168 (73.4) | 45 (54.2) | 9 (27.3) | < .001 |
| Administration of steroids/tocolytics | ||||
| Steroids | 187 (81.7) | 76 (91.6) | 28 (87.9) | .087 |
| Indomethacin | 69 (30.1) | 24 (28.9) | 9 (27.3) | .935 |
| Born at a hospital with an approved neonatal intensive care unit e | ||||
| Yes | 193 (84.3) | 73 (88.0) | 31 (93.9) | .277 |
| No. transferred in after birth | 36 (15.7) | 10 (12.1) | 2 (6.1) | |
| Infant characteristics | ||||
| Sex | ||||
| Male | 123 (53.7) | 42 (50.6) | 16 (48.5) | .791 |
| Female | 106 (46.3) | 41 (49.4) | 17 (57.5) | |
| Completed weeks of gestation | ||||
| 22-24 | 33 (14.4) | 5 (6.0) | 0 | .007 |
| 25-29 | 196 (85.6) | 78 (94.0) | 33 (100.0) | |
| Weight <1000 g | 104 (45.4) | 37 (44.6) | 18 (54.6) | .586 |
| Head circumference <10th percentile f | 10 (4.4) | 11 (13.3) | 6 (18.2) | .002 |
| Small for gestational age g | 7 (3.1) | 8 (9.6) | 10 (30.3) | < .001 |
| Symmetric h | 5 (2.2) | 6 (7.2) | 6 (18.2) | < .001 |
| Asymmetric i | 2 (0.9) | 2 (2.4) | 4 (12.1) | .004 |
| One-minute Apgar score, <4 | 58 (25.3) | 21 (25.3) | 9 (27.3) | .971 |
| Infant diagnoses | ||||
| Pneumothorax | 9 (3.9) | 2 (2.4) | 2 (6.1) | .552 |
| Patent ductus arteriosus | 117 (51.1) | 49 (59.0) | 15 (45.5) | .323 |
| Intraventricular hemorrhage | 59 (25.8) | 25 (30.1) | 3 (9.1) | .047 |
| Grade | ||||
| 1 | 21 (9.2) | 13 (15.7) | 2 (6.1) | .199 |
| 2 | 15 (6.6) | 5 (6.0) | 0 | .386 |
| 3 | 9 (3.9) | 3 (3.6) | 1 (3.0) | 1.000 |
| 4 | 14 (6.1) | 4 (4.8) | 0 | .447 |
| Cystic periventricular leukomalacia | 4 (1.8) | 1 (1.2) | 0 | 1.000 |
| Retinopathy of prematurity (any) | 61 (26.6) | 24 (28.9) | 7 (21.2) | .699 |
| Stage | ||||
| 1 | 21 (9.2) | 7 (8.4) | 3 (9.1) | 1.000 |
| 2 | 21 (9.2) | 8 (9.6) | 2 (6.1) | .915 |
| 3 | 19 (8.3) | 9 (10.8) | 2 (6.1) | .665 |
| Supplemental oxygen at 36 weeks of gestation | 54 (23.6) | 21 (25.3) | 9 (27.3) | .875 |
| Hospitalization at ≥40 weeks of gestation | 46 (20.1) | 13 (15.7) | 6 (18.2) | .674 |
| Death in the neonatal intensive care unit | 18 (7.9) | 7 (8.4) | 2 (6.1) | .951 |
| At <30 d | 13 (5.7) | 6 (7.2) | 0 | .346 |
| At ≥30 d | 5 (2.2) | 1 (1.2) | 2 (6.1) | .344 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
