www.AJOG.

org
fewer at day 14 than at day 7. This observation suggests that in some participants, pain returned after the press needles had been removed. Thus, it is important for future investigators to explore whether extended continuous auricular acupuncture is needed to have a sustained effect, as well as the characteristics of acupuncture responders vs nonresponders. Because we did not observe any major local irritation, infection, or adverse outcome, we are condent that 1 week of auricular acupuncture can be safely administered to women with low back and posterior pelvic pain in the last trimester of pregnancy. For a majority of participants receiving therapeutic acupuncture, pain relief was substantial and led to improvement of functional status. Thus, this study supports that use of acupuncture at specic auricular points is a safe and effective nonpharmacologic treatment of an important clinical entity for which few, if any, effective treatments are currently available. In summary, women who received 1 week of continuous acupuncture treatment at 3 specic auricular points experienced signicantly greater reduction in pain than those receiving sham acupuncture or no treatment, although the treatment effect was not sustained in all. Although the long-term efcacy of auricular acupuncture as a treatment for PRLP remains inconclusive, it clearly

Obstetrics

Research

shows promise. A future, large-scale randomized control study is indicated. CLINICAL IMPLICATIONS

Auricular acupuncture using retained press needles for 1 week can reduce the pain and disability caused by lower back and posterior pelvic pain during pregnancy. Once the intervention was removed, pain returned in some study participants. Future studies should be conducted to determine whether extensive auricular acupuncture treatment can decrease the development of chronic low back pain after delivery. f

Effects of onset of labor and mode of delivery on severe postpartum hemorrhage

Iqbal Al-Zirqi, MD, MRCOG; Siri Vangen, MD, PhD; Lisa Forsn, PhD, MSc; Babill Stray-Pedersen, MD, PhD
OBJECTIVE: Our purpose was to study the impact of labor onset and

delivery mode on the risk of severe postpartum hemorrhage. STUDY DESIGN: This was a population-based study of 307,415 mothers who were registered in the Medical Birth Registry of Norway from 1999-2004. RESULTS: Severe postpartum hemorrhage occurred in 1.1% of all mothers and in 2.1% of those mothers with previous cesarean section delivery (CS). Compared with spontaneous labor, hemorrhage risk was higher for induction (odds ratio [OR], 1.71; 95% condence interval

[CI], 1.56 1.88) and prelabor CS (OR, 2.05; 95% CI, 1.84 2.29). The risk was 55% higher for emergency CS and half that for vaginal deliveries (OR, 0.48; 95% CI, 0.43 0.53), compared with prelabor CS. The highest risk was for emergency CS after induction in mothers with previous CS (OR, 6.57; 95% CI, 4.2510.13), compared with spontaneous vaginal delivery in mothers with no previous CS. CONCLUSION: Induction and prelabor CS should be practiced with caution because of the increased risk of severe postpartum hemorrhage.

Cite this article as: Al-Zirqi I, Vangen S, Forsn L, et al. Effects of onset of labor and mode of delivery on severe postpartum hemorrhage. Am J Obstet Gynecol 2009;201:273.e1-9.

B ACKGROUND AND O BJECTIVE

Severe postpartum hemorrhage is a major cause of maternal death worldwide. The morbidity that is associated with severe hemorrhage remains a major problem. The onset of labor and mode of de-

livery, especially delivery by cesarean section (CS), is a prominent risk factor that is associated with severe postpartum hemorrhage. However, it has been proposed that the induction of labor and the underlying indications of CS delivery

From the Division of Obstetrics and Gynecology, Faculty of Medicine (Dr Al-Zirqi), the Division of Obstetrics & Gynecology, National Resource Center for Womens Health (Dr Vangen), Rikshospitalet, the Norwegian Institute of Public Health (Dr Forsn), and the Division of Obstetrics & Gynecology, Faculty of Medicine, University of Oslo (Dr StrayPedersen), Rikshospitalet, Oslo, Norway.
This study was supported by the Norwegian Foundation for Health and Rehabilitation and the Norwegian Womens Public Health Association. 0002-9378/free 2009 Mosby, Inc. All rights reserved. doi: 10.1016/j.ajog.2009.06.007

might be the real causes of hemorrhage rather than the procedure itself. The estimation of hemorrhage risk for CS delivery that is performed before labor after adjustment for other risk factors may reveal the risk of severe hemorrhage that is related to the procedure itself. Our aim was to study the impact of labor onset and delivery mode on the risk of severe postpartum hemorrhage.

M ATERIALS AND M ETHODS

We used data from the Medical Birth Registry of Norway on 307,415 women with pregnancies from 16 weeks of ges273

SEPTEMBER 2009 American Journal of Obstetrics & Gynecology

Research
FIGURE

Obstetrics

www.AJOG.org
livery. The prevalence was higher at prelabor CS delivery and induction than at spontaneous labor onset (Figure). Induction of labor increased the prevalence of severe postpartum hemorrhage at every mode of delivery. Comparing all 3 types of labor onset in separate models revealed that the risk of severe postpartum hemorrhage was higher for prelabor CS delivery and induction than for spontaneous labor onset. Compared with spontaneous labor onset in mothers with no previous CS delivery, prelabor CS delivery doubled the risk of severe postpartum hemorrhage, and induced labor onset increased the risk of severe postpartum hemorrhage by 75%. In mothers with previous CS delivery, the risk of severe postpartum hemorrhage was 28% higher for prelabor CS than for spontaneous labor onset. When modes of deliveries were compared, the risk of severe postpartum hemorrhage was signicantly halved for all vaginal deliveries, and emergency CS delivery had 55% higher risk, compared with prelabor CS delivery. Causes of postpartum hemorrhage were not identied in 60% of CS deliveries. The lowest prevalence of severe postpartum hemorrhage was observed at spontaneous vaginal delivery after spontaneous labor in multiparous women with no previous CS delivery (0.6%), although the highest prevalence was at emergency CS delivery after induction in mothers with previous CS delivery (4.7%). The risk for severe postpartum hemorrhage among primiparous women was higher for prelabor CS delivery than for spontaneous vaginal delivery, but it was highest for operative vaginal delivery after induction followed by emergency CS delivery. The risk among multiparous women with and with no previous CS delivery was higher for prelabor CS delivery than for vaginal deliveries, but it was highest for emergency CS delivery. Emergency CS delivery after induction in mothers with previous CS delivery was associated with the highest risk of severe postpartum hemorrhage (odds ratio, 6.57), compared with spontaneous vaginal delivery after spontaneous labor in multiparous women with no previous CS.

Prevalence of severe postpartum hemorrhage for the onset of different labor modes
3
2.4 Severe hemorrhage (%) 2.1 2.0 1.8 1.8 2.3

1.0

No previous CS 0.8 Previous CS

0
All deliveries n = 291,604 n = 15,811 Spontaneous labor n = 241,889 n = 8254 Induced labor n = 32,377 n = 1810 Prelabor CS n = 17,338 n = 5747

CS, cesarean section. Al-Zirqi. Labor and delivery effects on severe postpartum hemorrhage. Am J Obstet Gynecol 2009.

tation who gave birth between Jan 1, 1999-April 30, 2004. Severe postpartum hemorrhage, which is dened as a visually estimated blood loss of 1500 mL within 24 hours after delivery or the need for blood transfusion after delivery, regardless of the amount of blood loss, coded as yes or no. Severe postpartum hemorrhage was identied in a ticked box on the birth registration form. The explanatory variables included onset of labor, which was dened as spontaneous labor onset, induced labor onset, and prelabor CS delivery (CS performed before labor onset) and mode of delivery, which was dened as spontaneous vaginal delivery, operative vaginal delivery, emergency CS delivery (CS performed after labor onset), and prelabor CS delivery. Complete information on the onset of labor and mode of delivery was available. The confounding variables included demographic, medical, and obstetric factors. The demographic variables were age in years at the time of delivery, parity, and ethnicity. The medical variables included medical diseases before pregnancy. The pregnancy-related variables included multiple pregnancies, HELLP (hemolysis, elevated liver enzymes, and low platelet count) syndrome, preeclampsia, gestational diabetes mellitus, polyhydramnios, and gestational age 274

that was calculated by ultrasound measurements at 18 weeks of gestation. The labor-related variables included prolonged labor, augmentation by oxytocin, macrosomia, intrapartum pyrexia, and uterine rupture. Frequency analysis and cross tabulations were used to measure the prevalence and causes of severe postpartum hemorrhage. The association between severe postpartum hemorrhage and labor onset was analyzed with the use of 3 logistic regression models. The association between severe postpartum hemorrhage and mode of delivery was analyzed with the use of 4 logistic regression models, with prelabor CS delivery as reference. Finally, the association between severe postpartum hemorrhage and delivery mode after both spontaneous and induced labor was analyzed with the use of logistic regression in 3 separate groups: (1) primiparous women, (2) multiparous women with no previous CS delivery, and (3) mothers with previous CS delivery.

R ESULTS
Severe postpartum hemorrhage was observed in 3333 mothers of the total population, with a prevalence of 1.1%. Cases were identied in 330 mothers (2.1%) with previous CS delivery and in 3003 mothers (1.0%) with no previous CS de-

American Journal of Obstetrics & Gynecology SEPTEMBER 2009

www.AJOG.org C OMMENT
Prelabor CS delivery and induction of labor carried signicantly higher risk of severe postpartum hemorrhage, compared with spontaneous labor onset. Emergency CS delivery, especially after induction, had the highest risk of severe postpartum hemorrhage. The strength of the current study lies in the population-based design that included the total pregnant population of Norway over a 5-year period. Weaknesses of this study were the retrospective and observational design and the absence of objective measurement of blood loss. The nding that the risk of severe postpartum hemorrhage was higher for prelabor CS delivery than for spontaneous labor onset, even in those with previous CS delivery, is supported by other studies. This conrms that major surgery, even when planned, is not risk free. In contrast to our ndings, Landon et al found that the rate of transfusion was higher among mothers having trial of labor than among those having elective repeat CS delivery. To further address this topic, we suggest additional large-scale prospective studies that would use objective methods to measure blood loss at prelabor CS delivery and after spontaneous and induced trial of labor. Among all mothers, the risk of severe postpartum hemorrhage was signicantly higher for induction than for spontaneous labor onset but was not signicantly different between induction and prelabor CS delivery. This again conrms that induction is an important risk factor that should not be underestimated in daily practice. Large randomized controlled studies are needed to assess the role of induction where information about the ripeness of the cervix, induction method, and obstetric history are taken into account. The nding that emergency CS delivery after labor onset had signicantly higher risk of severe postpartum hemorrhage than prelabor CS delivery is also supported by other studies. However, we found that severe postpartum hemorrhage was signicantly lower for vaginal deliveries than for prelabor CS delivery. This is not surprising, because CS delivery results in acute blood loss before the uterine musculature can contract around the spiral arteries and the hysterotomy incision can be closed. The high percentage of unidentied causes of postpartum hemorrhage that are related to delivery by CS delivery may indicate that there were cases of uterine atony that were not recognized clinically. However, it may indicate that there were undocumented surgical causes of bleeding that were related to the procedure. A future prospective study that would document the exact causes of hemorrhage at CS delivery is needed. There was noticeable risk of severe postpartum hemorrhage for operative vaginal delivery after induction among

Obstetrics

Research

primiparous women. This was due to the high prevalence of genital trauma and uterine atony at this delivery mode that occurred most among primiparous women. Multiparous women with no previous CS delivery had the least prevalence of severe postpartum hemorrhage because of the signicantly higher prevalence of spontaneous vaginal deliveries. The nding that emergency CS delivery after induction in mothers with previous CS delivery had the highest risk of severe postpartum hemorrhage might result in considering prelabor CS delivery to be a better option for this group. However, we have to emphasize that this delivery mode occurred at a very low rate. Vaginal deliveries, which are known to have the lowest risk of hemorrhage, were the most frequent mode of delivery after trial of labor, even in mothers with previous CS. CLINICAL IMPLICATIONS

In the absence of clear medical indication, induction of labor and prelabor cesarean section (CS) delivery should be avoided because of the increased risk of severe postpartum hemorrhage. Prelabor CS delivery might be a better option if the probability of emergency CS delivery is high. Large-scale prospective studies are needed to document objectively the amount and precise causes of blood loss at CS delivery. f

SEPTEMBER 2009 American Journal of Obstetrics & Gynecology

275