Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Antepartum Membrane Stripping in GBS Carriers, Is It Safe? (The STRIP-G Study)

  • Doron Kabiri ,

    Affiliation Department of Obstetrics and Gynecology, Hadassah-Hebrew University Medical Center, Jerusalem, Israel

  • Yael Hants,

    Affiliation Department of Obstetrics and Gynecology, Hadassah-Hebrew University Medical Center, Jerusalem, Israel

  • Tom Raz Yarkoni,

    Affiliation Department of Obstetrics and Gynecology, Hadassah-Hebrew University Medical Center, Jerusalem, Israel

  • Esther Shaulof,

    Affiliation Department of Obstetrics and Gynecology, Hadassah-Hebrew University Medical Center, Jerusalem, Israel

  • Smadar Eventov Friedman,

    Affiliation Department of Neonatology, Hadassah-Hebrew University Medical Center, Jerusalem, Israel

  • Ora Paltiel,

    Affiliation Braun School of Public Health and Community Medicine, Hadassah-Hebrew University Medical Center, Jerusalem, Israel

  • Ran Nir-Paz,

    Affiliation Department of Clinical Microbiology and Infectious Diseases, Hadassah-Hebrew University Medical Center, Jerusalem, Israel

  • Wesam E. Aljamal,

    Affiliation Department of Obstetrics and Gynecology, Hadassah-Hebrew University Medical Center, Jerusalem, Israel

  • Yossef Ezra

    Affiliation Department of Obstetrics and Gynecology, Hadassah-Hebrew University Medical Center, Jerusalem, Israel



Stripping of the membranes is an established and widely utilized obstetric procedure associated with higher spontaneous vaginal delivery rates, reduced need for formal induction of labor and a lower likelihood of post-term pregnancy. Due to the theoretical concern of bacterial seeding during the procedure many practitioners choose not to sweep the membranes in Group B Streptococcus (GBS) colonized patients. We conducted ‘the STRIP-G study’ in order to determine whether maternal and neonatal outcomes are affected by GBS carrier status in women undergoing membrane stripping.

Study design

We conducted a prospective study in a tertiary referral center, comparing maternal and neonatal outcomes following membrane stripping among GBS-positive, GBS-negative, and GBS-unknown patients. We compared the incidence of composite adverse neonatal outcomes (primary outcome) among the three study groups, while secondary outcome measure was composite adverse maternal outcomes.


A total of 542 women were included in the study, of which 135 were GBS-positive, 361 GBS-negative, and 46 GBS-unknown status. Demographic, obstetric, and intra-partum characteristics were similar for all groups. Adverse neonatal outcomes were observed in 8 (5.9%), 31 (8.6%), and 2 (4.3%) in the GBS-positive, GBS-negative, and unknown groups, respectively (P = 0.530), (Odds Ratio between GBS-Positive vs. GBS-Negative groups 0.67 (95%, CI = 0.30–1.50)); while composite adverse maternal outcomes occurred in 9 (6.66%), 31 (8.59%), and 5 (10.87%) in the GBS-positive, GBS-negative, and unknown groups, respectively (P = 0.617).


Antepartum membrane stripping in GBS carriers appears to be a safe obstetrical procedure that does not adversely affect maternal or neonatal outcomes.


Stripping or sweeping of the fetal membranes is a widely utilized technique to hasten delivery, first described by James Hamilton in 1810 [1, 2]. The procedure is performed during a vaginal examination, by separating the chorioamniotic membrane from the lower uterine segment by a circular movement of the finger. This maneuver is thought to initiate a cascade of physiological reactions in which local production of prostaglandins promotes cervical ripening and may lead to the onset of active labor [3]. Several studies have reported that membrane stripping is associated with higher rates of spontaneous vaginal delivery, shorter induction-to-delivery interval, reduced likelihood of post-term pregnancy, and a decrease in the need for induction of labor [36]. Adverse effects commonly reported following the procedure are limited to transient maternal discomfort during the procedure, irregular uterine contractions, and clinically insignificant vaginal bleeding.

Streptococcus agalactiae (Group B Streptococcus (GBS)) is the leading major cause of early neonatal sepsis and an important pathogen associate with maternal peripartum infection [7]. Early onset neonatal GBS disease, defined as infection presenting in the first 7 days of life, has a broad clinical spectrum of ranging from mild illness that resolves spontaneously within several days, to severe respiratory disease, meningitis, neonatal sepsis, and death [7, 8]. It is estimated that 20–30% of all pregnant women are colonized with GBS [9].

A recent Cochrane review confirmed that membrane stripping does not increase the risk of maternal and neonatal infection, however this review did not analyze the outcomes by GBS carrier state [10]. However, based on the theoretical increased risk of bacterial seeding, as well as concern for fast labor which would prevent the administration of adequate antibiotic prophylaxis after membrane stripping, some practitioners choose not to sweep the membranes in GBS colonized patients. Both the American College of Obstetricians and Gynecologists (ACOG) and the Centers for Disease Control (CDC) do not consider GBS colonization as a contraindication to membrane stripping. Their latest guidelines discuss the risks of membrane stripping in women colonized with GBS. They concluded that the subject has not been investigated in well-designed prospective studies and therefore data are insufficient to encourage or discourage this practice in women known to be GBS-colonized [1113].

Given the rare neonatal morbidity caused by GBS sepsis [14] as well as the lack of well-powered studies designed to address the safety of membrane stripping in known GBS carriers [15], approaches are inconsistent, [16, 17], and consensus has not been reached regarding whether antepartum membrane stripping in GBS carriers adversely affects maternal or neonatal outcomes.

In order to further elucidate the effect of antepartum membrane stripping in GBS carriers, we conducted a prospective study, the STRIP-G study, which examined whether the rates of adverse neonatal and maternal outcomes differ by GBS carrier status among women undergoing membrane stripping.


Study Design and participants

We conducted a prospective observational cohort study in the maternal-fetal unit of a tertiary center university teaching hospital with approximately 5,000 deliveries per year. The study population consisted of all women who underwent membrane stripping at this hospital between October 10, 2011 and December 31, 2013. All candidates for vaginal delivery with a singleton pregnancy and confirmed cephalic presentation between 37 0/7 and 41 6/7 weeks' gestational age were eligible. We excluded from this study women with a closed cervix, multiple gestation pregnancies, pregnancies with major fetal anomalies, and those who were not candidates for vaginal delivery (placenta previa, breech presentation, planned cesarean delivery). GBS-positive women were considered “exposed”, and the comparison groups consisted of women who were GBS-screening negative and those with unknown-GBS status. Detailed demographic data and medical, prenatal and antenatal history, was extracted by trained staff using the patients' electronic medical records. Postpartum, maternal and neonatal outcomes were collected along with maternal and newborn discharge summaries by trained study investigators.

The Hadassah Medical Organization Institutional Review Board approved the STRIP-G study and waived the requirement for informed consent (IRB N°: 0204-11-HMO). The obstetrical staff performed membrane stripping after explanation and obtaining patient’s consent to the procedure. Since the IRB considered this as an observational study, written informed consent was not required.

Treatment Protocol

During the study period and in the absence of contraindications, women anticipating vaginal delivery were offered membrane stripping. In our institution we followed the RCOG and NICE guidelines for membrane stripping [18, 19] offering membrane stripping to all women prior to formal induction of labor, to nulliparous women at 40 weeks, to all women at the 41 week antenatal visit, and whenever the cervix assessed at term. No further restrictions or risk-based selection was done prior to offering membrane stripping. The obstetrical staff, physicians or midwives performed stripping after explanation and obtaining patient’s consent to the procedure. An ultrasonographic scan was performed before the procedure to verify fetal presentation and biophysical profile. In addition, prior to the procedure, fetal heart rate tracing (for nonstress testing) was obtained to confirm fetal well-being. After membrane stripping, women were discharged for routine obstetric care according to a standard local protocol, until the onset of spontaneous labor. Induction of labor was performed if the participant reached 42 weeks’ gestation or earlier if there was any neonatal or maternal indication that justified induction of labor.

Screening for GBS is routinely offered as part of obstetrical surveillance. According to our protocol, intrapartum antibiotic prophylaxis (IAP) was administered to women with known vaginal/intestinal tract GBS colonization, documented urinary tract infection with GBS at any time during the present pregnancy, a history of an infant with GBS disease in any prior delivery, or unknown GBS status with ruptured membranes for 18 hours or more. For every woman with suspected chorioamnionitis (defined as intrapartum maternal fever greater than 38°C or signs of chorioamnionitis), blood and urine cultures were taken and broad-spectrum intravenous antibiotics were administrated, including coverage for group B streptococcus.


The primary outcome measure was incidence of composite adverse neonatal outcomes based on the NICE criteria for early detection of neonatal sepsis [20], while secondary outcome measure was composite adverse maternal outcomes.

We did not choose early-onset neonatal GBS sepsis as the primary outcomes because the rarity of this event (incidence of 0.17–0.3 cases per 1000 births), that renders the assumption of bacterial seeding during membrane stripping difficult to demonstrate, and it is almost impossible to design a well-powered study that would be able to achieve statistical significance by using neonatal GBS sepsis as a primary endpoint. Following the assumption that a healthy neonate would not progress directly from a perfect state of health to full blown neonatal sepsis, the STRIP-G specialist group (which includes expert in perinatology, neonatology, microbiology and epidemiology), constructed a formal and structured scale aiming to detect a “compromised neonate” as a surrogate primary endpoint, based on the NICE criteria for early detection of neonatal sepsis [20]. The NICE criteria are produced by the National Institute for Health and Care Excellence in the United Kingdom to guide practitioners regarding which neonates should receive antibiotics for suspected early-onset GBS infection. According to the NICE guidelines, clinical indicators of possible early-onset neonatal infection include 19 clinical indicators and 4 ‘red flags’ indicators (Fig 1). Therefore, we performed a systematic review of medical records for each neonate whose mother participated in the study in order to identify any “compromised neonate”, defined as the presence of at least one ‘red flag’ or two clinical indicators of possible early-onset neonatal infection [20].

Fig 1. Clinical Indicators Of Possible-Early-Onset Neonatal Infection.

(Figure reproduced from “Antibiotics for early-onset neonatal infection”, NICE clinical guidelines 149, 2012).

Adverse maternal outcome was defined as one or more of the following: chorioamnionitis, suspected chorioamnionitis or endometritis (defined by peripartum maternal fever and uterine tenderness), urinary tract infection or sepsis in the peripartum period, prolonged hospital stay, and allergic reaction to antibiotics.

We collected peripartum data until discharge from the hospital, and compared maternal and neonatal adverse outcomes among GBS-screening positive group, GBS-screening negative group and GBS unknown group.

Statistical methods

Statistical analysis was performed by using the SPSS package (version 19, SPSS Inc., Chicago, IL, USA). Differences in continuous variables among the three groups were analyzed by the ANOVA. Differences in categorical variables were analyzed via the χ2 or Fisher's exact test. A two-sided P<0.05 was considered statistically significant. Risk differences between GBS-Positive and GBS-Negative groups and their standard error as well as odds ratios were calculated in WinPepi ®.

We based our statistical analysis on the primary outcome (compromised neonate). The sample size was thus based on the expected difference between the study and control groups in percentage of ‘compromised neonate’. For the assumption that the ratio between the size of the GBS-positive and the GBS-negative group would be 1:3, with significance level of 5% (2 sided), and the proportion of compromised neonatal rate in the population is 6–8%, we estimated that 125 GBS-Positive women and 375 GBS-Negative women would provide power of 80% to detect a risk ratio of 2.2.


Between October 2011 and December 2013, there were 7,681 candidates for vaginal delivery at our institution. The GBS distribution status is presented in Fig 2. During the study period 542 women enrolled and underwent membrane stripping. Of those women, 135 (24.91%) were GBS-screening positive (GBS positive group), 361 (66.60%) were GBS-screening negative (GBS negative group), and 46 (8.49%) did not undergo pre-labor screening for GBS (GBS unknown group).

There were no significant differences between the groups in terms of demographic and obstetric parameters (Table 1). There was a statistically but not clinically significance difference in gestational age at stripping between the groups. Intra-partum characteristics such as simplified Bishop score, induction of labor, fever during labor, analgesia during labor and mode of delivery, were similar among the groups (Table 2).

Table 1. Demographic and obstetric characteristics at recruitment.

We observed no cases of neonatal sepsis, death or serious neonatal morbidity in the study. The ‘compromised neonate’ rate was 5.9% (8/135), 8.6% (31/361), and 4.3% (2/46) in the GBS-positive, GBS-negative, and GBS-unknown group, respectively (P = 0.530) (Fig 3). The Odds Ratio between GBS-Positive and GBS-Negative groups was 0.67 [P = 0.33 (95%, CI = 0.30–1.50)]. Neonatal outcomes and the incidence of clinical indicators for possible-early-onset of neonatal infection are summarized in Table 3. The significant difference of hospital stay and NICU admission are due to the GBS-Unknown group.

Fig 3. ‘Compromised Neonate’ Rate and Composite Adverse Maternal Outcomes Rate.

Ninety-four women (69.63%) from the GBS-screening positive group received adequate antibiotic treatment, defined as IAP at least four hours prior to delivery, while forty-one women (30.37%) received inadequate antibiotic treatment. In a subgroup analysis, when stratified by IAP adequacy in the GBS-screening positive group, the risk for ‘compromised neonate was 5.3% (5/94) and 7.3% (3/41) for adequate and inadequate IAP treatment, respectively (P = 0.651).

There were no significant differences among the groups in the frequency of adverse maternal outcomes (Table 4). The composite adverse maternal outcomes rate was 6.66% (9/135), 8.59% (31/361), and 10.87% (5/46) in the GBS-positive, GBS-negative, and GBS-unknown groups, respectively (P = 0.617) (Fig 3). The Odds Ratio between GBS-Positive and GBS-Negative groups was 0.76 [P = 0.46 (95%, CI = 0.35–1.64)]. There were no cases of maternal death or serious maternal morbidity in our study.


Main findings

We performed a prospective cohort study to evaluate the effect of antepartum membrane stripping in GBS carriers on adverse perinatal outcomes. In this cohort study, adverse neonatal and maternal outcomes did not differ between GBS-Positive patients and others in the study population undergo membrane stripping.

Strengths and Limitations

A systematic review of 2,797 women in 22 randomized trials found that in comparison to women without membrane stripping, the risk of maternal and neonatal infection does not increase after stripping of the membranes [10], however this review did not analyze the outcomes by GBS carrier state which comprised 19–26% of all pregnancies. Given the rare neonatal morbidity following membrane stripping, only a few studies have addressed the safety of this procedure in known carriers of GBS. One randomized controlled trial presented in poster form evaluated GBS colonization associated with stripping of the membrane, and concluded that the procedure does not increase maternal colonization [15]. Another randomized controlled trial by Hill et al [21], reported no significant difference in the rate of chorioamnionitis in GBS-Positive patients who underwent stripping versus the non-stripping group, however the numbers in this subgroup were limited. After a thorough search of the literature using the keywords "Group B streptococcus”, “Streptococcus agalactiae”, “GBS in pregnancy”, “Membrane sweeping” and “Membrane stripping”, we were unable to find any prospective, retrospective, or controlled data to suggest that stripping of the membranes in GBS colonized patients is associated with an increased risk of maternal or neonatal infection.

While we performed this study to examine whether membrane stripping increased the risk of neonatal complications in GBS positive women, we found a non-significant risk reduction. This may represent no difference or possibly a protective effect of membrane stripping in an appropriately sized sample. One hypothesis for the possible protective effect may be related to the intrapartum antibiotic prophylaxis in the GBS positive group. Another explanation for the safety of membrane stripping in GBS carriers may be related to the assumed protective effect of the amniotic sac, which provides a barrier between the fetus and the external environment during membrane stripping

In this study the GBS-unknown group had similar maternal and neonatal infection rate as the other groups. This interesting group of patients, in the absence of clinical intra-partum indication for IAP, would not generally receive antibiotic during labor and delivery. If membrane stripping would have jeopardized mothers and neonates by increasing the risk of infection, this group would potentially be the most vulnerable, since they do not receive any antibiotic protection at all. However, the fact that they did not experience a higher rate of adverse outcomes may support our conclusion that membrane stripping is a safe procedure for the mother and neonate.

Some physicians may choose not to strip the membranes in GBS colonized patients because of the concern of fast labor associated with stripping and inadequate antibiotic prophylaxis. As part of the “STRIP-G Study”, we conducted another study comparing adequate antibiotic prophylaxis between GBS carriers who did and did not undergo membrane stripping. In that study we found that antepartum membrane stripping in GBS carriers did not affect the adequacy of antibiotic prophylaxis for GBS [22].

Since the incidence of GBS sepsis is very low in our population, it is practically impossible to perform a study with sufficient significance for analyzing this rare event as a primary end point in one center. Therefore, in this study we used a surrogate endpoint based on the NICE Guidelines of clinical indicators of possible early-onset neonatal infection aiming to detect the “compromised neonate” who are at risk for sepsis [20]. Our inability to report on the GBS status of the neonate is a weakness of this study, however since IAP masks the presence of GBS, it is reasonable to assess these cases of “probable early GBS infection in a neonate” (defined as symptoms and signs of infection in a neonate born to a GBS positive mother, and bacterial cultures from obtained from the neonate that were negative for GBS) as an end-point [23, 24].

Unfortunately the assessment of the neonate could not be blinded to the GBS status of the mother, as the information was included in the medical records. On the other hand, we used a formal structured and validated scale to assess the neonatal outcomes [20].

Although this is the largest study to date following GBS-Positive women after membrane stripping, sample size was not sufficient to establish formal non-inferiority. Furthermore, prior to performing a non-inferiority trial, obstetricians would need to form a consensus regarding a negligible clinical difference in the relevant study outcomes. Differences in population and practice may limit generalizability of our results in other countries. Thus, a large multicenter randomized controlled trial in a heavy colonized population would be useful to confirm our conclusions.


Membrane stripping is a common and simple obstetrical procedure that may positively influence the transition from pregnancy to the onset of labor, increasing the chance of spontaneous labor and reducing the need for formal induction of labor and the incidence of post-term pregnancy. Its utility is emphasized especially in situations in which our ability to hasten delivery is limited (i.e. grandmultiparity or previous cesarean delivery). Our study and the literature to date shows that antepartum membrane stripping in GBS carriers is not associated with increased maternal or neonatal adverse outcomes as compared to non-carriers. We believe that this study is of clinical relevance with wide application for obstetricians in the community as well as in hospital facilities, providing reassurance and expanding treatment options for GBS carriers.


We are gratefully acknowledge the obstetrical staff, physicians and midwives of the delivery room at Hadassah, who contributed to this work. Presented at the 34th Annual Meeting of the Society for Maternal-Fetal Medicine, New Orleans, LA, February, 2014, and received the Award for Research Excellence.

Ethical approval: The study was approved by the Hadassah Medical Organization institutional review board (IRB N°: 0204-11-HMO)

Author Contributions

Conceived and designed the experiments: DK YH SEF OP RNP YE. Analyzed the data: TRY ES WEA. Wrote the paper: DK YH SEF OP RNP YE.


  1. 1. Swann O. Induction of labor by stripping membranes. Obstetrics and gynecology. 1958;11(1):74–8. pmid:13504633
  2. 2. Thiery M BC, Keirse MJNC. The development of methods for inducing labour. Effective Care in Pregnancy and Childbirth.: Chalmers I, Enkin M, Keirse MJ, editors. Oxford; Oxford University Press; 2000.
  3. 3. McColgin SW, Bennett WA, Roach H, Cowan BD, Martin JN Jr., Morrison JC. Parturitional factors associated with membrane stripping. American journal of obstetrics and gynecology. 1993;169(1):71–7. pmid:8333480
  4. 4. Renfrew MJ, McFadden A, Bastos MH, Campbell J, Channon AA, Cheung NF, et al. Midwifery and quality care: findings from a new evidence-informed framework for maternal and newborn care. Lancet. 2014;384(9948):1129–45. pmid:24965816
  5. 5. de Miranda E, van der Bom JG, Bonsel GJ, Bleker OP, Rosendaal FR. Membrane sweeping and prevention of post-term pregnancy in low-risk pregnancies: a randomised controlled trial. BJOG: an international journal of obstetrics and gynaecology. 2006;113(4):402–8.
  6. 6. Tan PC, Jacob R, Omar SZ. Membrane sweeping at initiation of formal labor induction: a randomized controlled trial. Obstetrics and gynecology. 2006;107(3):569–77. pmid:16507926
  7. 7. Schuchat A. Group B streptococcus. Lancet. 1999;353(9146):51–6. pmid:10023965
  8. 8. Le Doare K, Heath PT. An overview of global GBS epidemiology. Vaccine. 2013;31 Suppl 4:D7–12. pmid:23973349
  9. 9. Baker CJ, Barrett FF. Transmission of group B streptococci among parturient women and their neonates. The Journal of pediatrics. 1973;83(6):919–25. pmid:4585831
  10. 10. Boulvain M, Stan C, Irion O. Membrane sweeping for induction of labour. The Cochrane database of systematic reviews. 2005(1):CD000451. pmid:15674873
  11. 11. American College of O, Gynecologists Committee on Obstetric P. ACOG Committee Opinion No. 485: Prevention of early-onset group B streptococcal disease in newborns. Obstetrics and gynecology. 2011;117(4):1019–27. pmid:21422882
  12. 12. Verani JR, McGee L, Schrag SJ, Division of Bacterial Diseases NCfI, Respiratory Diseases CfDC, Prevention. Prevention of perinatal group B streptococcal disease—revised guidelines from CDC, 2010. MMWR Recommendations and reports: Morbidity and mortality weekly report Recommendations and reports / Centers for Disease Control. 2010;59(RR-10):1–36.
  13. 13. Gabbe SG. Obstetrics: normal and problem pregnancies. Philadelphia: Elsevier/Saunders; 2012.
  14. 14. Van Dyke MK, Phares CR, Lynfield R, Thomas AR, Arnold KE, Craig AS, et al. Evaluation of universal antenatal screening for group B streptococcus. The New England journal of medicine. 2009;360(25):2626–36. pmid:19535801
  15. 15. Netta D VP, Bayliss P. Does cervical membrane stripping increase maternal colonization of group B streptococcus? Poster presented at: 23rd Annual Meeting of the Society for Maternal–Fetal Medicine; February 3–8, 2003; San Francisco, California POSTER SESSION V. American Journal of Obstetrics & Gynecology. 2002;187(6):S210–S38.
  16. 16. Crago MS, Gauer R, Frazier J. Clinical inquiry: Does cervical membrane stripping in women with group B Streptococcus put the fetus at risk? The Journal of family practice. 2012;61(1):60a–b. pmid:22220298
  17. 17. Arnold W. Cohen JG. Membrane sweeping and GBS: A litigious combination? OBG Management. September 2006.
  18. 18. Induction of Labour. NICE Clinical Guidelines, No. 70. National Collaborating Centre for Women's and Children's Health (UK). Jul 2008. Available from:
  19. 19. National Guideline C. Induction of labour Rockville MD: Agency for Healthcare Research and Quality (AHRQ); [12/1/2014]. Available from:
  20. 20. Antibiotics for early-onset neonatal infection: antibiotics for the prevention and treatment of early-onset neonatal infection. NICE clinical guidelines 149: The Royal College of Obstetricians and Gynaecologists.; [cited July 2014]. Available from:
  21. 21. Hill MJ, McWilliams GD, Garcia-Sur D, Chen B, Munroe M, Hoeldtke NJ. The effect of membrane sweeping on prelabor rupture of membranes: a randomized controlled trial. Obstetrics and gynecology. 2008;111(6):1313–9. pmid:18515514
  22. 22. Kabiri D, Hants Y, Raz-Yarkoni T, Shaulof E, Eventov-Friedman S, Nir-Paz R, et al. 303: Does membrane stripping in GBS carrier patients preclude adequate antibiotic treatment in labor? (STRIP-G study). American Journal of Obstetrics & Gynecology.212(1):S163.
  23. 23. Luck S, Torny M, d'Agapeyeff K, Pitt A, Heath P, Breathnach A, et al. Estimated early-onset group B streptococcal neonatal disease. Lancet. 2003;361(9373):1953–4. pmid:12801740
  24. 24. Carbonell-Estrany X, Figueras-Aloy J, Salcedo-Abizanda S, de la Rosa-Fraile M, Castrillo Study G. Probable early-onset group B streptococcal neonatal sepsis: a serious clinical condition related to intrauterine infection. Archives of disease in childhood Fetal and neonatal edition. 2008;93(2):F85–9. pmid:17704105
  25. 25. Laughon SK, Zhang J, Troendle J, Sun L, Reddy UM. Using a simplified Bishop score to predict vaginal delivery. Obstetrics and gynecology. 2011;117(4):805–11. pmid:21383643