Keywords stillbirth - periviable gestation - risk factor - race–ethnicity - preterm delivery
Periviable gestation commonly refers to deliveries at 20 to 25 weeks' gestation, a time frame that is considered to represent the border of viability.[1 ] These deliveries are rare but make a substantial contribution to infant mortality and long-term morbidity.[1 ] Current understanding of their causes is limited.
Most studies of deliveries at periviable gestation focus on live births, although exceptions do exist.[2 ]
[3 ] However, stillbirths (defined here as in utero fetal deaths delivered at 20 or more weeks' gestation) comprise a substantial proportion of deliveries at 20 to 25 weeks.[4 ] Stillbirth and live birth may even be considered competing outcomes during this time frame, due to variability in health care practices and reporting of these deliveries.[5 ] As such, understanding the proportion of live births and stillbirths during this time frame is important because it may impact neonatal mortality estimates and comparisons within and between institutions and populations.[6 ] In addition, stillbirth and preterm delivery (i.e., live-born delivery at <37 weeks' gestation) are known to share some of the same risk factors, including maternal medical conditions such as diabetes and hypertension, obesity, and a marked black–white disparity.[7 ] This finding may result in part from the fact that, at least in more developed countries, most stillbirths are delivered preterm—approximately 80% in the United States.[8 ] The extent to which these similarities in risk factors hold among stillborn and live-born deliveries at 20 to 25 weeks' gestation is uncertain. To fully understand delivery at periviable gestation, it is important to understand risk factors for stillbirth during this time window, and whether they are similar to risk factors for live birth. In general, there is a paucity of information about risk factors for stillbirth at periviable gestational age, and minimal attention is given to the proportion of deliveries at periviable gestational age that are stillbirths.
The objective of this study was to compare the prevalence of and risk factors for stillbirth and live birth at periviable gestational age (i.e., 200/7 –256/7 weeks' gestation), within a cohort of almost 2.5 million singleton births in California. We aimed to determine whether prevalence differs by race–ethnicity given the increased risk of stillbirth and early preterm delivery among blacks relative to whites.[9 ]
[10 ]
Methods
We examined deliveries that occurred in California from 2007 to 2011 at 310 hospitals. Data were derived from vital records, using files prepared by the California Office of State-wide Hospital Planning and Development that contain data from fetal death certificates, and linked data from live birth and infant death certificates. For California vital records, the definition of stillbirth is a fetal death delivered at ≥20 weeks' gestation; all deliveries meeting this definition receive a fetal death certificate. Gestational age at delivery was derived from vital records and based on best obstetric estimates.
Analyses included births to mothers who were non-Hispanic white, non-Hispanic black, Hispanic, or Asian; we excluded mothers who had other or missing race–ethnicity because some analyses focused on racial–ethnic differences. We excluded (1) infants who were nonsingletons, given that periviable delivery in that circumstance may have unique etiologies and (2) infants with gestational age <20 or >41 weeks' gestation (i.e., >416/7 weeks), and (3) infants with implausible birth weight for gestational age based on previously published criteria.[11 ] After these exclusions, our sample included 2,487,468 deliveries at 20 to 41 weeks' gestation (11,141 of which were stillbirths). There were 10,888 periviable births: 4,610 stillbirths, and 6,278 live births delivered at 20 to 25 weeks' gestation (i.e., 200/7 –256/7 weeks).
Analyses
First, we conducted analyses to describe the percentage of deliveries at 20 to 25 weeks' gestation that had the following outcomes: stillbirth, live birth, live birth who died <24 hours after delivery, live birth who died ≥24 hours after delivery (up to 1 year), and live birth who survived the first year. We analyzed live births separately depending on whether they did or did not live beyond the first 24 hours of life, given potential misclassification between these early infant and fetal deaths. We also examined the prevalence of each of these outcomes; for prevalence calculations, the denominator was all eligible deliveries that were stillborn or live born at 20 to 41 weeks' gestation. These analyses were conducted separately for each racial–ethnic group and for 2-week gestational time windows. For each outcome, we estimated racial–ethnic prevalence ratios (PRs) and their 95% confidence intervals (CIs), with non-Hispanic whites as the reference. Second, we examined the association of selected risk factors with each of the specified outcomes at 20 to 25 weeks' gestation. The comparison group for these analyses was live births delivered at term (37–41 weeks' gestation). These “term controls” represent the optimal outcome for all infants and include the vast majority of deliveries. This approach avoids selection bias due to conditioning on a collider. We used multivariable logistic regression analysis to generate odds ratios (ORs) and 95% CIs for these comparisons. Risk factors included maternal race–ethnicity, education, payer status, nativity, age, height, pre-pregnancy body mass index (BMI), smoking status during the first trimester, pre-pregnancy diabetes or hypertension, parity and interpregnancy interval, prior preterm birth, and prior stillbirth. BMI was computed as pre-pregnancy weight (kg) divided by height (m) squared and categorized using published guidelines.[12 ] The variables were based on data reported in vital records. Models included 3,545 stillbirths (77%) and 5,043 live births (80%) delivered at 20 to 25 weeks' gestation, and 2,048,890 term live births for comparison, who had complete data on covariates. As sensitivity analyses, we conducted logistic regression analyses that restricted the cases to deliveries at 22 to 25 weeks' gestation, to facilitate comparison with other studies of stillbirths that may be restricted to this more narrow gestational age range.
Results
[Table 1 ] shows the distribution of maternal characteristics among all deliveries, as well as among stillbirths and live births delivered at 20 to 25 weeks' gestation. A majority of births were to Hispanic and multiparous mothers, 26% of mothers had less than a high school education, and 48% had California state public insurance (Medi-Cal). A description of characteristics of subgroups of live births born at 20 to 25 weeks' gestation that were based on whether or not they survived the first year postdelivery is provided in [Supplementary Table 1 ], available in the online version.
Table 1
Characteristics of mothers who delivered a singleton stillborn or live-born infant in California from 2007 to 2011, overall and restricted to stillbirths and live births at 20 to 25 weeks' gestation
All live births and stillbirths at 20–41 wk (n = 2,487,468)
Stillbirths at 20–25 wk
(n = 4,610)
Live births at 20–25 wk (n = 6,278)
n
%
n
%
n
%
Race/Ethnicity
Non-Hispanic white
703,721
28.3
1,172
25.4
1,214
19.3
Non-Hispanic black
146,868
5.9
599
13.0
1,009
16.1
Asian
314,678
12.7
479
10.4
617
9.8
Hispanic
1,322,201
53.2
2,360
51.2
3,438
54.8
Education
Less than high school
622,197
25.5
1,302
29.8
1,705
28.4
High school graduation
647,743
26.6
1,174
26.9
1,889
31.5
Some college
573,286
23.5
1,186
27.2
1,565
26.1
College graduate
594,962
24.4
707
16.2
837
14.0
Missing
49,280
241
282
Payer
Medi-Cal
1,185,266
47.8
2,136
49.9
3,368
54.2
Private
1,140,323
46.0
1,835
42.8
2,342
37.7
Uninsured
58,632
2.4
182
4.3
276
4.4
Other
97,525
3.9
131
3.1
234
3.8
Missing
5,722
326
58
Country of birth
U.S. born
1,404,704
56.5
2,774
61.7
3,947
62.9
Foreign born
1,081,509
43.5
1,724
38.3
2,326
37.1
Missing
1,255
112
5
Age (y)
Mean (SD)
28.2 (6.3)
28.4 (7.0)
27.8 (6.7)
Missing
37
13
0
Height (in)
Mean (SD)
63.6 (2.9)
63.6 (3.1)
63.5 (2.9)
Missing
105,849
255
446
Pre-pregnancy BMI[a ]
Underweight
94,473
4.1
120
2.9
183
3.4
Normal
1,141,438
49.9
1,655
39.9
2,020
37.6
Overweight
588,258
25.7
1,156
27.9
1,479
27.5
Obese I
282,223
12.4
648
15.6
939
17.5
Obese II
113,733
5.0
340
8.2
456
8.5
Obese III
65,411
2.9
231
5.6
294
5.5
Missing
201,932
460
907
Smoking (first trimester)
No
2,394,719
97.7
4,261
95.5
5,854
96.0
Yes
56,879
2.3
200
4.5
245
4.0
Missing
35,870
149
179
Pre-pregnancy hypertension
12,245
0.5
84
1.8
82
1.3
Pre-pregnancy diabetes
11,045
0.4
61
1.3
47
0.8
Parity
Nulliparous
989,439
39.8
1,983
43.8
3,128
50.0
Multiparous
1,496,905
60.2
2,541
56.2
3,128
50.0
Missing
1,124
86
22
Interpregnancy interval[b ]
< 6 months
74,921
5.1
161
7.0
260
8.5
6–23 months
507,510
34.3
647
28.1
782
25.7
24–47 months
420,213
28.4
541
23.5
680
22.3
≥ 48 months
476,393
32.2
956
41.5
1,322
43.4
Missing
17,868
236
84
Prior preterm birth
15,830
0.6
124
2.7
262
4.2
Prior stillbirth
No
2,457,430
98.9
4,026
90.0
5,973
95.6
Yes
27,998
1.1
449
10.0
277
4.4
Missing
2,040
135
28
Abbreviations: BMI, body mass index; SD, standard deviation.
a Underweight (<18.5 kg/m2 ), normal weight (18.5–24.9), overweight (25–29.9), and obesity class I (30.0–34.9), II (35.0–39.9), and III (≥40.0) based on World Health Organization International classification (http://apps.who.int/bmi/index.jsp?introPage=intro_3.html ).
b Multiparous mothers only.
The prevalence of delivery at periviable gestations was 3.4 per 1,000 deliveries among whites, 10.9 for blacks, 3.5 for Asians, and 4.4 for Hispanics ([Table 2 ]). The black–white PR for having a stillbirth at 20 to 25 weeks was 2.5 (95% CI: 2.2, 2.7); for live birth, it was 4.0 (95% CI: 3.7, 4.3), and similarly elevated for the subcategories of live birth ([Table 2 ]). Prevalences among Asians were similar to those among whites (PRs from 0.9 to 1.2). Prevalences for Hispanics were similar to those among whites for stillbirth (PR 1.1) and modestly elevated for live birth (PRs from 1.5 to 1.6). This pattern of results was similar within the 2-week gestational time periods.
Table 2
Distribution and prevalence of singleton stillbirths and live births at 20 to 25 weeks' gestation, by maternal race–ethnicity, gestational age at delivery, and survival (live births), and racial–ethnic prevalence ratios (relative to non-Hispanic whites), California, 2007 to 2011[a ]
Non-Hispanic white
Non-Hispanic black
Asian
Hispanic
N
% of births within group
Prevalence
n
% of births within group
Prevalence
Prevalence ratio (95% CI)
n
% of births within group
Prevalence
Prevalence ratio (95% CI)
n
% of births within group
Prevalence
Prevalence ratio (95% CI)
20–21 wk
All births
689
0.98
428
2.91
2.98 (2.64, 3.36)
310
0.99
1.01 (0.88, 1.15)
1593
1.2
1.23 (1.13, 1.35)
All stillbirths
498
72.3
0.71
261
61.0
1.78
2.51 (2.16, 2.92)
200
64.5
0.64
0.90 (0.76, 1.06)
1046
65.7
0.79
1.12 (1.00, 1.24)
All live births
191
27.7
0.27
167
39.0
1.14
4.19 (3.40, 5.16)
110
35.5
0.35
1.29 (1.02, 1.63)
547
34.3
0.41
1.52 (1.29, 1.80)
Live births—died < 24 hours after birth
167
24.2
0.24
152
35.5
1.03
4.36 (3.50, 5.43)
96
31.0
0.31
1.29 (1.00, 1.65)
494
31
0.37
1.57 (1.32, 1.88)
Live births—died ≥24 hours after birth
18
2.6
0.03
14
3.3
0.1
3.73 (1.85, 7.49)
12
3.9
0.04
1.49 (0.72, 3.10)
42
2.6
0.03
1.24 (0.71, 2.16)
Live births—survived first year
6
0.9
0.01
1
0.2
0.01
0.80 (0.10, 6.63)
2
0.6
0.01
0.75 (0.15, 3 0.69)
11
0.7
0.01
0.98 (0.36, 2.64)
22–23 wk
All births
818
1.16
551
3.75
3.23 (2.90, 3.60)
356
1.13
0.97 (0.86, 1.10)
1999
1.51
1.30 (1.20, 1.41)
All stillbirths
451
55.1
0.64
216
39.2
1.47
2.29 (1.95, 2.70)
162
45.5
0.51
0.80 (0.67, 0.96)
898
44.9
0.68
1.06 (0.95, 1.19)
All live births
367
44.9
0.52
335
60.8
2.28
4.37 (3.77, 5.07)
194
54.5
0.62
1.18 (0.99, 1.41)
1101
55.1
0.83
1.60 (1.42, 1.80)
Live births—died < 24 hours after birth
241
29.5
0.34
192
34.8
1.31
3.82 (3.16, 4.61)
127
35.7
0.4
1.18 (0.95, 1.46)
631
31.6
0.48
1.39 (1.20, 1.62)
Live births—died ≥24 hours after birth
60
7.3
0.09
65
11.8
0.44
5.19 (3.65, 7.37)
43
12.1
0.14
1.60 (1.08, 2.37)
223
11.2
0.17
1.98 (1.49, 2.63)
Live births—survived first year
66
8.1
0.09
78
14.2
0.53
5.66 (4.08, 7.86)
24
6.7
0.08
0.81 (0.51, 1.30)
247
12.4
0.19
1.99 (1.52, 2.61)
24–25 wk
All births
879
1.25
629
4.28
3.43 (3.10, 3.80)
430
1.37
1.09 (0.97, 1.23)
2206
1.67
1.34 (1.24, 1.44)
All stillbirths
223
25.4
0.32
122
19.4
0.83
2.62 (2.10, 3.27)
117
27.2
0.37
1.17 (0.94, 1.47)
416
18.9
0.31
0.99 (0.84, 1.17)
All live births
656
74.6
0.93
507
80.6
3.45
3.70 (3.30, 4.16)
313
72.8
0.99
1.07 (0.93, 1.22)
1790
81.1
1.35
1.45 (1.33, 1.59)
Live births—died < 24 hours after birth
65
7.4
0.09
43
6.8
0.29
3.17 (2.16, 4.66)
34
7.9
0.11
1.17 (0.77, 1.77)
186
8.4
0.14
1.52 (1.15, 2.02)
Live births—died ≥24 hours after birth
146
16.6
0.21
106
16.9
0.72
3.48 (2.71, 4.47)
67
15.6
0.21
1.03 (0.77, 1.37)
355
16.1
0.27
1.29 (1.07, 1.57)
Live births—survived first year
445
50.6
0.63
358
56.9
2.44
3.85 (3.35, 4.43)
212
49.3
0.67
1.07 (0.90, 1.25)
1249
56.6
0.94
1.49 (1.34, 1.66)
20–25 wk
All births
2386
3.39
1608
10.9
3.23 (3.03, 3.44)
1096
3.48
1.03 (0.96, 1.10)
5798
4.39
1.29 (1.23, 1.36)
All stillbirths
1172
49.1
1.67
599
37.3
4.08
2.45 (2.22, 2.70)
479
43.7
1.52
0.91 (0.82, 1.02)
2360
40.7
1.78
1.07 (1.00, 1.15)
All live births
1214
50.9
1.73
1009
62.7
6.87
3.98 (3.66, 4.33)
617
56.3
1.96
1.14 (1.03, 1.25)
3438
59.3
2.6
1.51 (1.41, 1.61)
Live births—died < 24 hours after birth
473
19.8
0.67
387
24.1
2.64
3.92 (3.43, 4.48)
257
23.4
0.82
1.22 (1.04, 1.41)
1311
22.6
0.99
1.48 (1.33, 1.64)
Live births—died ≥24 hours after birth
224
9.4
0.32
185
11.5
1.26
3.96 (3.26, 4.81)
122
11.1
0.39
1.22 (0.98, 1.52)
620
10.7
0.47
1.47 (1.26, 1.72)
Live births—survived first year
517
21.7
0.73
437
27.2
2.98
4.05 (3.57, 4.60)
238
21.7
0.76
1.03 (0.88, 1.20)
1507
26
1.14
1.55 (1.40, 1.71)
Abbreviation: CI, confidence interval.
a Prevalence reflects the number of births with the outcome of interest per 1,000 total births (i.e., live births plus stillbirths) delivered from 20 to 41 weeks' gestation. The reference for prevalence ratios is non-Hispanic whites. There were 703,721 total births to whites, 146,868 to blacks, 314,678 to Asians, and 1,322,201 to Hispanics.
In this cohort, 42% of the periviable deliveries were stillbirths (4,610/10,888), 22% (2,428) were live births who died <24 hours after delivery, 11% (1,151) were live births who died from 24 hours to 1 year after delivery, and 25% (2,699) were live births who survived at least 1 year. The percentage of deliveries at 20 to 25 weeks' gestation that were stillbirths was highest for whites, 49%, versus 37% for blacks, 44% for Asians, and 41% for Hispanics ([Table 2 ], [Fig. 1 ]). However, the percentage of deliveries that were live births who died within 24 hours after delivery was lowest for whites, 20%, versus 24% for blacks and 23% for Asians and Hispanics. When examining 2-week gestational age intervals, the percentage of deliveries that were stillborn was highest at 20 to 21 weeks (72, 61, 65, and 66% for whites, blacks, Asians, and Hispanics, respectively) and lowest for deliveries at 24 to 25 weeks (25, 19, 27, and 19%, respectively). Less than 1% of infants delivered at 20 to 21 weeks survived to 1 year of age, whereas a majority of infants delivered at 24 to 25 weeks survived.
Fig. 1 Distribution of outcomes of singleton deliveries at 20 to 25 weeks' gestation, by maternal race–ethnicity and gestational age at delivery, California, 2007 to 2011. Outcomes were stillbirth, live birth who died <24 hours after delivery, live birth who died ≥24 hours after delivery (up to 1 year), and live birth who survived the first year. NH, non-Hispanic.
We examined the potential associations of maternal characteristics with risk of stillbirth and live birth at 20 to 25 weeks' gestation, relative to having a liveborn delivery at term (i.e., 37–41 weeks' gestation), after adjustment for covariates ([Table 3 ]). When compared with non-Hispanic whites, elevated odds for stillbirth and live birth at 20 to 25 weeks' gestation were observed for blacks (OR: 2.0 for stillbirth and 3.4 for live birth), Asians (ORs: 1.3 and 1.7), and Hispanics (ORs: 1.1 and 1.5). Lower education, uninsured status, being U.S. born, older age, obesity, smoking, pre-pregnancy hypertension, nulliparity, extremes of interpregnancy interval, and prior preterm birth or stillbirth were all associated with increased risk of both stillbirth and live birth at 20 to 25 weeks' gestation. The magnitudes of the ORs were similar for both outcomes for most variables, with some exceptions. In concordance with unadjusted results reported earlier, racial–ethnic disparities were stronger for live birth than stillbirth. In addition, ORs were higher for live birth than stillbirth for nulliparity (2.6 vs. 1.9) and prior preterm birth (9.0 vs. 3.9), whereas the ORs were lower for live birth than stillbirth for prior stillbirth (3.3 vs. 7.7) and pre-pregnancy hypertension (1.8 vs. 2.7). These patterns of results were similar for the survival-based subgroups of live-born cases ([Supplementary Table 2 ], available in the online version). They were also very similar for analyses that were restricted to cases that occurred at 22 to 25 weeks' gestation (data not shown).
Table 3
Association of maternal characteristics with risk of delivering a singleton stillborn or live-born infant at 20 to 25 weeks' gestation, relative to delivering a live-born infant at term (37–41 weeks), California, 2007 to 2011[a ]
Odds ratio (95% confidence interval)
Stillbirth at 20–25 wk
(n = 3,545)
Live birth at 20–25 wk
(n = 5,043)
Race/Ethnicity
Non-Hispanic white
1.0 (ref)
1.0 (ref)
Non-Hispanic black
2.03 (1.79, 2.29)
3.43 (3.11, 3.78)
Asian
1.30 (1.13, 1.49)
1.70 (1.51, 1.92)
Hispanic
1.11 (1.01, 1.23)
1.47 (1.34, 1.60)
Education
Less than high school graduation
1.37 (1.25, 1.51)
1.07 (0.99, 1.15)
High school graduation
1.0 (ref)
1.0 (ref)
Some college
1.05 (0.95, 1.15)
0.88 (0.82, 0.95)
College graduate or more
0.60 (0.53, 0.68)
0.51 (0.46, 0.56)
Payer
Medi-Cal
0.94 (0.86, 1.02)
1.10 (1.02, 1.18)
Private
1.0 (ref)
1.0 (ref)
Uninsured
1.89 (1.58, 2.27)
2.58 (2.24, 2.97)
Other
0.73 (0.60, 0.88)
0.96 (0.83, 1.11)
Maternal country of birth
U.S. born
1.0 (ref)
1.0 (ref)
Foreign born
0.79 (0.73, 0.86)
0.74 (0.69, 0.79)
Maternal age (per year)
1.03 (1.02, 1.04)
1.03 (1.02, 1.04)
Maternal height (per inch)
0.99 (0.98, 1.01)
0.99 (0.98, 1.00)
Pre-pregnancy BMI[b ]
Underweight
0.81 (0.65, 0.99)
1.01 (0.86, 1.19)
Normal
1.0 (ref)
1.0 (ref)
Overweight
1.28 (1.18, 1.39)
1.36 (1.27, 1.46)
Obese I
1.40 (1.27, 1.55)
1.74 (1.60, 1.89)
Obese II
1.86 (1.63, 2.11)
2.05 (1.84, 2.28)
Obese III
2.04 (1.75, 2.37)
2.08 (1.83, 2.37)
Smoking (first trimester)
No
1.0 (ref)
1.0 (ref)
Yes
1.48 (1.25, 1.76)
1.50 (1.30, 1.73)
Pre-pregnancy diabetes
No
1.0 (ref)
1.0 (ref)
Yes
1.74 (1.27, 2.38)
1.27 (0.92, 1.73)
Pre-pregnancy hypertension
No
1.0 (ref)
1.0 (ref)
Yes
2.74 (2.14, 3.53)
1.79 (1.40, 2.31)
Parity and interpregnancy interval
Nulliparous
1.92 (1.74, 2.13)
2.60 (2.38, 2.84)
< 6 months
1.55 (1.28, 1.87)
1.87 (1.59, 2.19)
6–23 months
1.0 (ref)
1.0 (ref)
24–47 months
0.95 (0.84, 1.08)
1.00 (0.89, 1.12)
≥ 48 months
1.28 (1.14, 1.44)
1.45 (1.31, 1.61)
Prior preterm birth
No
1.0 (ref)
1.0 (ref)
Yes
3.85 (3.10, 4.77)
8.95 (7.77, 10.3)
Prior stillbirth
No
1.0 (ref)
1.0 (ref)
Yes
7.67 (6.80, 8.66)
3.33 (2.90, 3.83)
Abbreviation: BMI, body mass index.
a All variables in the table are adjusted for each other; the comparison group for each model included 2,048,890 singleton live births delivered at 37 to 41 weeks' gestation.
b See [Table 1 ] for category definitions.
Discussion
In this cohort of California births delivered at periviable gestational age (20–25 weeks), a substantial proportion were stillbirths—close to two-thirds of deliveries at 20 to 21 weeks and approximately one-fifth of deliveries at 24 to 25 weeks. Risk factor associations for stillbirth and live birth during this early time window were similar, which supports the concept that at least some aspects of the underlying etiologies of both outcomes may be similar. Stillbirth and live birth are not typically included in the same studies, and studies of deliveries at periviable gestational age are usually (but not always[2 ]
[3 ]) restricted to live births.[1 ]
[13 ]
[14 ] Based on our findings, we recommend that assessments and studies of deliveries at periviable gestation should incorporate both live births and stillbirths whenever possible. A greater effort to include both outcomes would likely provide a more thorough understanding of what causes delivery at periviable gestation, and how to prevent it. Prior studies have noted potential similarities in etiologies of stillbirth, preterm delivery, and infant mortality in general and advocated for the importance of better integration of investigations of perinatal outcomes.[8 ]
[15 ]
[16 ] However, we are unaware of prior studies that have specifically compared risk factors among deliveries within the periviable gestational age range.
Our findings have parallels to other cohorts of deliveries at periviable gestational age that included stillbirths and live births. In particular, the EPICure cohort from the United Kingdom (which included 3,133 deliveries at 22–26 weeks' gestation in 2006) and the EXPRESS cohort from Sweden (which included 1,011 deliveries at 22–26 weeks' gestation in 2004–2007) both reported that approximately 30% of births at 22 to 26 weeks' gestation were stillbirths.[2 ]
[3 ] In our study, 33% of deliveries at 22 to 25 weeks' gestation were stillbirths (2,605/7,868).
The black–white disparity in the occurrence of delivery at periviable gestation was strong, with a 4.2-fold increased risk of live birth and 2.5-fold increased risk of stillbirth among blacks. Disparities were also observed for Asians and Hispanics, but they were much more modest. The disparities remained after adjustment for a variety of maternal characteristics. The few prior studies that have examined racial–ethnic disparities in the prevalence of early stillbirth tend to agree with our findings for blacks and Hispanics but have not included Asians.[17 ] In addition, we observed that a greater proportion of deliveries at 20 to 25 weeks were live births than stillbirths among nonwhites, whereas proportions were similar among whites, which agrees with a recent investigation of U.S. births.[17 ] Extensive efforts have been aimed at understanding the black–white disparity in live-born preterm delivery, with minimal progress; fewer studies have examined contributors to disparities in stillbirth.[15 ]
[18 ] Given that the black–white disparity is so strong at these earliest gestational age,[14 ]
[19 ] studies focusing on this time period may be particularly important to advancing our understanding of it. The distribution of outcomes among the studied deliveries (i.e., stillbirth and live birth according to survival) also varied by race–ethnicity. This finding could derive from differences in underlying etiologic factors by race–ethnicity. Hospital-level variability in intervention and reporting practices related to periviable births may also contribute,[5 ]
[20 ]
[21 ] given that delivery hospital tends to vary nonrandomly by race–ethnicity. In addition, variability in intervention and reporting practices by race–ethnicity itself may contribute to observed disparities. Institutional assessments of newborn and obstetric quality of care do not typically include stillbirths in their denominator. This omission could introduce bias on performance evaluations related to newborn outcomes as well as disparities. Further investigation into such factors is important but beyond the scope of the current analysis.
Strengths of our study include its population-based design and large size. The generalizability of our findings beyond California is uncertain, but likely enhanced by the fact that one in eight U.S. births occurs in California, and the diversity of the cohort with respect to race–ethnicity and socioeconomic status. Generalizability to women with missing data are also uncertain; however, given that the percentage of women with any missing data was similar for those with an early stillbirth (23%) or live birth (20%), we do not expect that missing data compromised the internal validity of our results. Our list of studied risk factors is by no means exhaustive, and more detailed studies are needed for us to better understand causes, and how to prevent delivery at periviable gestational age. Our intent was to illustrate whether these available risk factors had similar associations with early stillbirth and early live birth. Analyses included data on deliveries from 2007 to 2011 because vital records were revised in 2007 (e.g., the obstetric estimate for gestational age at delivery and maternal weight and height were first included then), and 2011 was the most recent year available to us. It would be helpful to be able to report more recent data years, especially since survival of infants born at the studied gestational age has been increasing over time,[22 ] but we expect that the general patterns of associations that we observed would likely persist in more recent years. A potential limitation of the stillbirth data are that while the gestational age at delivery is known, the gestational age of the in utero demise is not; the potential impact of this lag time on our results is unknown. In addition, we did not incorporate data about reported cause of the stillbirths because a specific cause was not available for most cases. However, at these early gestational ages, causes may be more homogeneous than if stillbirths at all gestational age were included. We did not exclude infants based on the presence of congenital anomalies given that their reporting in vital records is incomplete and likely to vary for live births versus stillbirths. We focused on deliveries at 20 to 25 weeks given that 20 weeks is the minimal gestational age at which stillbirths are recorded, and it concurs with the definition of periviable birth that is often used in the United States.[1 ] As expected, the viability of deliveries at 20 to 21 weeks is low (less than 1% survived infancy), but we considered their inclusion to be important since in the United States, they are recorded. Our reporting of results that restricted the cases to later gestational age windows (i.e., 22–25 weeks) and 2-week time periods should facilitate comparison with other studies that restricted their cases to narrower gestational age ranges. Finally, we were unable to obtain placental pathology data, autopsy data, genetic data, or infectious workup data from the stillbirths. Prior studies by the Stillbirth Collaborative Research Network have highlighted the importance of a comprehensive evaluation of all stillbirths and we hope that future studies incorporate this type of information when possible.[23 ]
[24 ]
In conclusion, our results suggest that some aspects of the etiologies of stillbirth and live birth at periviable gestational age may overlap. Given this finding, the substantial proportion of deliveries that are stillborn during this gestational window, inclusion of stillbirths, as well as live births in ongoing studies of periviable gestations is important if we are to understand the true burden of delivery during this time period and its etiology. In addition, studies focused on deliveries at these earliest gestational age will further the understanding of racial–ethnic disparities in perinatal outcomes, given how strong they are during this time window. Investigators have previously advocated for more “enlightened” datasets[15 ] and a more “united front”[16 ] with regard to more frequent inclusion of stillbirths in studies of perinatal outcomes; our findings suggest that this message is particularly important for understanding delivery at periviable gestational age.
