Objective
The national primary cesarean delivery rate increased until 2004, but after 2004, this rate cannot be tracked using Vital Statistics data. Additionally, it is unknown whether changes in the primary cesarean delivery rate reflect changes in the rate of labor attempts, labor success, or both. Here, using hospital discharge data, we examined national trends in primary cesarean deliveries, labor attempts, and labor success among women without prior cesarean delivery between 1990 and 2010.
Study Design
This analysis of serial cross-sectional data from the National Hospital Discharge Survey used Joinpoint regression to assess trends over time and logistic regression with marginal effects to identify rates of change over time and adjust for demographic and clinical factors.
Results
The primary cesarean delivery rate declined 0.2 percentage points per year (95% confidence interval [CI], 0.1–0.3) between 1990 and 1999, increased 1.0 percentage point per year (95% CI, 0.8–1.2) between 1999 and 2004, and increased 0.3 percentage points (95% CI, 0.1–0.6) per year from 2004 until 2010. Between 1998 and 2005, the rate of labor attempts declined 0.4 percentage points (95% CI, 0.3–0.5) per year. No changes in the labor attempt rate occurred between 2005 and 2010. Labor success rates increased 0.2 percentage points (95% CI, 0.1–0.3) per year between 1990 and 1998 but then declined 0.5 (95% CI, 0.5–0.8) percentage points per year from 1998 to 2010. Adjusted results were similar.
Conclusion
The primary cesarean delivery rate continued to increase after 2004. Increases in the primary cesarean delivery rate after 2005 were driven by declines in labor success rates.
The rate of cesarean deliveries increased from 22.7% of all births in 1990 to 32.8% of all births in 2010. Many studies have explored reasons for changes in vaginal birth after cesarean rates. However, primary cesarean deliveries likely make up more than half of all cesarean deliveries in the United States.
The national primary cesarean delivery rate increased from 14.6% of births in 1996 to 20.6% in 2004, accounting for approximately 60% of the increase in total cesarean deliveries. This increase is concerning because cesarean delivery is associated with short-term complications such as intraoperative injury to the bowel, bladder or ureter, hemorrhage, infection, and thromboembolism. It also has longer-term consequences in subsequent pregnancies such as placenta accreta, uterine rupture, intraoperative injury of the uterus, bowel, ureter, or bladder from adhesions and an increased risk of hysterectomy. To address this increase in cesarean delivery, a recent national workgroup has examined indications for cesarean with the aim of reducing primary cesarean deliveries.
After 2004, because of changes in the collection of birth certificate data, primary cesarean delivery rates cannot be calculated for all states from the National Vital Statistics System and instead must be based on vital statistics from a subnational group of states. Because of this limitation, national trends in primary cesarean deliveries since 2004 are unknown. Previous studies have examined single hospital trends or hospital consortium data to estimate primary cesarean rates at a single point in time, but to our knowledge, no studies have examined national trends over time since 2004. Analysis of hospital discharge data provides a means of estimating national primary cesarean delivery rates since 2004, but this has not been done previously.
The rate of primary cesarean deliveries depends on the rate at which labor is attempted and the rate at which labor successfully leads to vaginal births in women without previous cesarean deliveries. Decreases in rates of both labor attempts and labor success contributed significantly to recent increases in repeat cesarean deliveries. However, previous studies have not examined how each of these factors might contribute to trends in the national primary cesarean delivery rate.
The aim of this study was to examine trends in the national rate of primary cesarean delivery, especially after 2004, and examine the impact of changes in the attempted labor and labor success rates on these trends.
Materials and Methods
This analysis of serial cross-sectional data used data from the National Hospital Discharge Survey (NHDS) from 1990 to 2010 to examine changes over time in the rates of primary cesarean delivery, attempted labor, and labor success. NHDS is an annual cross-sectional, nationally representative sample survey of hospital discharges from nonfederal, noninstitutional short-stay US hospitals, conducted by the National Center for Health Statistics (NCHS) from 1965 to 2010. The sampling strata of the NHDS include hospital size and geography (census region and metropolitan and nonmetropolitan statistical areas) to ensure broad geographic and hospital size representation in the data. For 1990 through 2007, NHDS collected data on an average of 309,000 hospital discharges per year from an average of 459 hospitals. Due to NCHS funding limitations, the sample of hospitals was reduced by half for 2008-2010, and data were collected from only 160,000 discharges from 205 hospitals.
NHDS has a 3-stage sampling design, and sample weights were calculated based on the reciprocal of selection probability and adjusted for nonresponse. Details of survey methodology and weighting procedures have been previously described. NHDS has received approval by the NCHS Research Ethics Review Board. This study used NHDS data and did not require separate institutional review board approval.
Data from the Vital Statistics system could not be used for this analysis because, prior to the 2003 revision of the birth certificate, information on whether a woman underwent labor was not collected on the birth certificate. By 2010, only 33 states had adopted the 2003 revision of the birth certificate; these states are not considered a random sample or representative of the United States, making a calculation of trends in national estimates after 2004 impossible.
Cases were defined as discharges with International Classification of Diseases-9th Revision Clinical Modification (ICD-9-CM) codes indicating delivery. Determination of whether the patient underwent labor during the current delivery, had a vaginal or cesarean delivery, and whether the patient had a previous cesarean delivery or the other clinical conditions listed in Table 1 was made by the presence of the relevant ICD-9-CM codes ( Table 1 ). Deliveries with labor were identified using a modified list of codes used in previous studies ( Table 1 ). The ICD-9-CM codes used in the calculation of primary cesarean delivery, attempted labor, and labor success have been validated against clinical data in previous research.
| Category | ICD-9-CM Codes |
|---|---|
| Delivered | |
| Diagnosis code | V27 |
| Procedure codes | 72, 74.0-74.2, 74.4, 74.9 |
| Previous cesarean delivery | |
| Diagnosis code | 654.2 |
| Cesarean delivery | |
| Procedure codes | 74.0-74.2, 74.4, 74.9 |
| Labored | |
| Diagnosis codes | 650, 653.4-653.5, 653.8-653.9, 658.2, 658.3, 659.0-659.3, 660-662, 664, 665.1 |
| Procedure codes | 72.0-72.4, 73.01, 73.09, 73.1, 73.3-73.6, 73.93-73.99, 75.32, 75.38, 75.6 |
| Hypertension | |
| Diagnosis codes | 642.0-642.9 |
| Diabetes | |
| Diagnosis codes | 648.0, 648.8, 250 |
| Placenta previa | |
| Diagnosis codes | 641.0-641.1 |
| Preterm delivery | |
| Diagnosis codes | 644.2 |
| Multiple gestation | |
| Diagnosis codes | V27.2-V27.7, 651 |
| Genital herpes | |
| Diagnosis codes | 054.1 |
| Breech presentation | |
| Diagnosis codes | 652.1-652.2 |
Our study population included only those cases without a diagnosis code for previous cesarean delivery. The primary cesarean delivery rate was calculated for each year as the percent of discharges with a cesarean delivery divided by the total number of discharges with deliveries in our study population. The attempted labor rate in our study sample was calculated for each year as the number of discharges with no previous cesarean deliveries that also had ICD-9-CM codes for labor divided by the total number of discharges with deliveries and no previous cesarean delivery. Delivered cases with labor but with no procedure codes for cesarean delivery were considered to have had successful labor resulting in a vaginal birth. For each year, the rate of successful labor among women without a previous cesarean delivery was calculated as the number of vaginal births divided by the number of women who labored. Rates and associated SEs were estimated for each year from 1990 through 2010 using Stata 12.0 SE (StataCorp, College Station, TX) and adjusted for the complex design of the survey using Taylor series linear approximation.
Estimates of yearly primary cesarean rates and associated SEs were entered into a Joinpoint regression using the National Cancer Institute’s Joinpoint 3.4.3 software using year as the independent variable and primary cesarean rate as the dependent variable. Joinpoint was used to identify time points (joinpoints) in which linear trends changed. Joinpoint regression fits the simplest linear model with no changes in trend (a straight line) and, using a series of Monte Carlo permutation tests, tests whether 1 or more joinpoints (changes in linear trend) are statistically significant and should be added to the model. Similarly, the rates of attempted labor and labor success among women without previous cesarean delivery and associated SEs were entered into 2 additional Joinpoint regressions as dependent variables with year as the independent variable for each.
The trends identified in Joinpoint were further investigated using logistic regression in Stata 12.0 SE. We conducted 3 sets of logistic regressions. The first set used primary cesarean delivery as the dependent variable, the second set used attempted labor as the dependent variable, and the third set used labor success as the dependent variable. For each dependent variable, regressions were conducted for each time period identified as a separate trend by Joinpoint regression. For each time period identified by Joinpoint, 2 models were conducted: an unadjusted model with only year as the independent variable and an adjusted model.
All adjusted models included maternal age (greater than or less than 35 years), hospital bed size (less than 50, 50-99, 100-199, 200-299, 300-499, 500-999, or 1000 or more beds), US Census region (Northeast, Midwest, South, and West), and expected source of payment (private insurance, Medicare, Medicaid, uninsured, other/unknown). Adjusted models using labor success as the dependent variable also included factors that may affect labor success (preterm delivery, multiple gestation, hypertension, diabetes, large for gestational age, intrauterine growth restriction, and fetal anomalies) as independent variables. Adjusted models using attempted labor as the dependent variable and adjusted models using primary cesarean delivery as the dependent variable included clinical factors that may influence the decision to attempt labor (all variables included in the models for labor success in addition to placenta previa, genital herpes, and breech presentation). For each regression, marginal effects were calculated for the survey year variable.
Additionally, these analyses were repeated with only singleton, term deliveries to assess unadjusted and adjusted trends over time in a more homogeneous population of births. Finally, these analyses were also repeated including additional ICD-9-CM codes that might, but did not necessarily, indicate labor (fetal distress, 656.3, and fetal heart rate abnormalities, 659.7).
No observations had missing data for hospital bed size or US Census region. Age was missing for 0.1% of observations. Values imputed for these observations by NCHS on the NHDS file were used. Expected source of payment had a value of unknown or other for 1.9% of observations. These were included in regressions as a separate category because this category may be informative, rather than represent only missing data.
From 1990 to 2007, there were 338-400 hospitals that contributed discharges with deliveries in the NHDS. From 2008 to 2010, there were 158-162 hospitals that contributed discharges with deliveries, as NCHS reduced the size of the survey. The estimated number of total deliveries for each year ranged from 3,738,000 in 2000 to 4,144,000 in 2008.
The NHDS reports rates per 100 deliveries, with multiple gestation births counted as a single delivery, whereas vital statistics reports rates per 100 births. Also, data from NHDS include only births that occur within hospitals, whereas Vital Statistics include out-of-hospital births. After accounting for these factors, the confidence interval of the NHDS estimate for deliveries in each year always included the number of births obtained from the Vital Statistics system (data not shown). Although regression analyses were conducted using unrounded estimates, the weighted estimates presented were rounded to the nearest 1000 cases to be consistent with NCHS practice. NCHS recommends rounding NHDS estimates to the nearest thousand to imply an appropriate level of precision for this survey. All estimates met criteria of statistical reliability of a relative standard error of less than 30% and a sample size of greater than 30 observations. Results within each analysis were considered significant at values of P < .05.
Results
Annual estimates from 1990 to 2010 for the number of deliveries by women without previous cesarean ranged from a high of 3,601,000 in 1990 to a low of 3,290,000 in 2000 ( Table 2 ). The number of primary cesarean deliveries between 1990 and 2010 ranged from 519,000 in 1995 to 781,000 in 2007 ( Table 2 ). The primary cesarean delivery rate reached low levels of 15.5% in both 1995 (95% confidence interval [CI], 14.7–16.3%) and 1999 (95% CI, 14.9–16.1%) and a high of 23.1% (95% CI, 21.7–24.5%) in 2010 ( Figure 1 ). Joinpoint regression identified 2 significant joinpoints, 1999 and 2004. The unadjusted trend between 1990 and 1999 showed a decline of 0.2 percentage point per year ( P < .01) ( Table 3 ). The unadjusted trend between 1999 and 2004 showed an increase of 1.0 percentage point per year ( P < .001). Between 2004 and 2010, there was an unadjusted increase of 0.3 percentage point per year ( P < .05). After adjustment for maternal age, hospital bed size, US Census region, expected source of payment, and clinical factors, trends were largely unchanged.
| Year | Deliveries with no previous cesarean delivery (95% CI) (sample size) | Primary cesarean deliveries (95% CI) (sample size) | Deliveries with attempted labor (95% CI) (sample size) | Successful labor (95% CI) (sample size) |
|---|---|---|---|---|
| 1990 | 3,601,000 (3,336,000–3,867,000) (27,692) | 606,000 (554,000–659,000) (4671) | 3,362,000 (3,113,000–3,610,000) (25,900) | 2,995,000 (2,768,000–3,222,000) (23,021) |
| 1991 | 3,543,000 (3,289,000–3,796,000) (27,781) | 607,000 (543,000–671,000) (4627) | 3,297,000 (3,064,000–3,531,000) (25,957) | 2,936,000 (2,729,000–3,143,000) (23,154) |
| 1992 | 3,454,000 (3,209,695–3,699,000) (26,230) | 579, 000 (526,000–632,000) (4443) | 3,229,000 (3,001,000–3,458,000) (24,523) | 2,875,000 (2,670,000–3,080,000) (21,787) |
| 1993 | 3,567,000 (3,143,000–3,990,000) (23,909) | 583,000 (480,000–686,000) (3925) | 3,338,000 (2,948,000–3,729,000) (22,370) | 2,984,000 (2,653,000–3,315,000) (19,984) |
| 1994 | 3,459,000 (3,178,000–3,739,000) (28,238) | 547,000 (487,000–606,000) (4490) | 3,233,000 (2,974,000–3,492,000) (26,449) | 2,912,000 (2,679,000–3,144,000) (23,748) |
| 1995 | 3,353,000 (3,088,000–3,619,000) (26,091) | 519, 000 (468,000–571,000) (3975) | 3,141,000 (2,893,000–3,389,000) (24,561) | 2,834,0000 (2,611,000–3,057,000) (22,116) |
| 1996 | 3,372,000 (3,085,000–3,659,000) (27,798) | 529,000 (475,000–583, 000) (4167) | 3,158,000 (2,887,000–3,429,000) (26,225) | 2,842,000 (2,598,000–3,087,000) (23,631) |
| 1997 | 3,355,000 (3,108,000–3,602,000) (29,725) | 523,000 (464,000–582,000) (4482) | 3,144,000 (2,915,000–3,374,000) (27,946) | 2,832,000 (2,631,000–3,033,000) (25,243) |
| 1998 | 3,488,000 (3,220,000–3,756,000) (30,631) | 546,000 (490,000–601,000) (4672) | 3,251,000 (3,004,000–3,497,000) (28,719) | 2,942,000 (2,720,000–3,165,000) (25,959) |
| 1999 | 3,373,000 (3,085,000–3,661,000) (28,906) | 522, 000 (469, 000–576,000) (4554) | 3,148,000 (2,879,000–3,418,000) (26,977) | 2,850,000 (2,610,000–3,090,000) (24,352) |
| 2000 | 3,290,000 (3,006,000–3,574,000) (31,301) | 532,000 (479,000–584, 000) (5087) | 3,062,000 (2,794,000–3,330,000) (29,186) | 2,759,000 (2,514,000–3,004,000) (26,214) |
| 2001 | 3,364,000 (3,069,000–3,660,000) (31,568) | 604,000 (541,000–666,000) (5566) | 3,100,000 (2,831,000–3,369,000) (29,246) | 2,761,000 (2,517,000–3,004,000) (26,002) |
| 2002 | 3,453,000 (3,170,000–3,735,000) (30,684) | 638, 000 (576,000–700,000) (5549) | 3,160,000 (2,899,000–3,421,000) (28,367) | 2,814,000 (2,585,000–3,044,000) (25,135) |
| 2003 | 3,487,000 (3,160,000–3,814,000) (29,280) | 674, 000 (597,000–751,000) (5662) | 3,194,000 (2,891,000–3,497,000) (26,797) | 2,813,000 (2,555,000–3,071,000) (23,618) |
| 2004 | 3,577,000 (3,251,000–3,903,000) (33,784) | 738,000 (667,000–809, 000) (7216) | 3,253,000 (2,958,000–3,549,000) (30,537) | 2,839,000 (2,575,000–3,102,000) (26,568) |
| 2005 | 3,469,000 (3,154,000–3,783,000) (33,608) | 753,000 (679,000–827,000) (7466) | 3,133,000 (2,848,000–3,419,000) (30,275) | 2,716,000 (2,463,000–2,969,000) (26,142) |
| 2006 | 3,513,000 (3,181,000–3,845,000) (33,805) | 732,000 (657,000–806,000) (7459) | 3,174,000 (2,871,000–3,477,000) (30,387) | 2,781,000 (2,510,000–3,052,000) (26,346) |
| 2007 | 3,516,000 (3,172,000–3,860,000) (33,540) | 781,000 (696,000–866,000) (7586) | 3,179,000 (2,866,000–3,493,000) (30,202) | 2,735,000 (2,463,000–3,007,000) (25,954) |
| 2008 | 3,499,000 (2,714,000–4,284,000) (13,712) | 769,000 (586,000–952,000) (3019) | 3,180,000 (2,462,000–3,898,000) (12,508) | 2,730,000 (2,120,000–3,340,000) (10,693) |
| 2009 | 3,374,000 (2,617,000–4,131,000) (12,553) | 755,000 (582,000–929,000) (2814) | 3,051,000 (2,364,000–3,737,000) (11,366) | 2,619,000 (2,023,000–3,214,000) (9739) |
| 2010 | 3,338,000 (2,602,000–4,075,000) (11,663) | 769, 000 (598,000–940, 000) (2619) | 3,017,000 (2,340,000–3,694,000) (10,534) | 2,569,000 (1,996,000–3,143,000) (9044) |
| Total | 72,444,000 (70,513,000–74,375,000) (572,499) | 13,307,000 (12,868,000–13,746,000) (104,049) | 66,807,000 (65,034,000–68,579,300) (529,032) | 59,137,000 (57,595,000–60,679,000) (468,450) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
