Indications for caesarean sections at ≥34 weeks among nulliparous women and differential composite maternal and neonatal morbidity

Authors


Abstract

Objective

To compare composite maternal and neonatal morbidities (CMM, CNM) among nulliparous women with primary indications for caesarean section (CS) as acute clinical emergency (group I; ACE), non-reassuring fetal heart rate (group II) and arrest disorder (group III).

Design

A multicentre prospective study.

Setting

Nineteen academic centres in the USA, with deliveries in 1999–2002.

Population

Nulliparous women (n = 9829) that had CS.

Methods

Nulliparous women undergoing CS for three categories of indications were compared using logistic regression model, adjusted for five variables.

Main outcome measures

CMM was defined as the presence of any of the following: intrapartum or postpartum transfusion, uterine rupture, hysterectomy, cystotomy, ureteral or bowel injury or death; CNM was defined as the presence of any of the following: umbilical arterial pH <7.00, neonatal seizure, cardiac, hepatic, renal dysfunction, hypoxic ischaemic encephalopathy or neonatal death.

Results

The primary reasons for CS were ACE in 1% (group I, n = 114) non-reassuring FHR in 29% (group II; n = 2822) and failed induction/dystocia in the remaining 70% (group III; n = 6893). The overall risks of CMM and CNM were 2.5% (95% confidence intervals, CI, 2.2–2.8%) and 1.9% (95% CI 1.7–2.2), respectively. The risk of CMM was higher in group I than in group II (RR 4.1, 95% CI 3.1, 5.3), and group III (RR 3.2, 95% CI 2.7, 3.7). The risk of CNM was also higher in group I than in group II (RR 2.8, 95% CI 2.3, 3.4) and group III (RR 14.1, 95% CI 10.7, 18.7).

Conclusions

Nulliparous women who have acute clinically emergent caesarean sections are at the highest risks of both composite maternal and neonatal morbidity and mortality.

Introduction

In the USA, the rate of caesarean section among nulliparous women has increased from 21% in 1996 to 27% in 2003, reflecting a 28% increase over an 8-year period.[1] A multicentre study (2005–2007) involving 19 hospitals noted that one in three nulliparous women underwent caesarean delivery.[2] The risk factors and purported reasons for the increase are advanced maternal age,[3, 4] ethnicity with increased likelihood among black women,[5] gestational age (lower at 39 weeks versus beyond),[6] overweight and obesity,[7, 8] high-risk pregnancy,[1] tocophobia,[9, 10] labour induction,[5, 11] high fetal station in active phase of labour,[12] suspected macrosomia,[13] practice patterns within a hospital,[14] and liability concerns.[15, 16]

The disquieting aspects of the temporal increase in the caesarean section rate are the appreciably higher likelihood of intra- and postpartum complications associated with caesarean section[17-19] and in subsequent pregnancies an increased rate of placental complications, small for gestational age, preeclampsia, uterine rupture and hysterectomy.[18, 20, 21] The World Health Organization's recommendation that the caesarean section rate should be 15%[22] is understandable and concerted attempts to decrease the rates of caesarean section are underway.[23, 24]

The increasing burden of maternal morbidity and mortality in the USA[25, 26] has led to a recent reemphasis on, and acute need for, putting the ‘M’ back in Maternal-fetal Medicine.[8, 27, 28] Several previous studies have dealt with maternal and neonatal morbidities after caesarean section but have failed to take into consideration the indication for surgery.[8, 12, 18] Indeed, one of the issues that remains unknown is the relationship, as well as quantification, between the primary indication for caesarean section and maternal and neonatal morbidity. This provided the impetus for our research, where we sought to examine this relationship and also to quantify the risk for maternal and newborn complications of caesarean among nulliparous women according to the primary indication for caesarean section. Although it may be intuitive that the maternal and neonatal morbidities with caesarean vary with the indication, the risks for morbidities need to be quantified in order to counsel women, assure that appropriate clinicians are available at delivery for resuscitation, and design future studies whose aims are to lower these morbidities.

Material and methods

We carried out a secondary analysis of data from the Maternal-Fetal Medicine Units (MFMU) network's ‘Caesarean registry’ cohort.[18] An Institutional Review Board approval from East Virginia Medical School, VA, was obtained for an analysis of the caesarean section registry data. This multicentre prospective cohort study (1999–2002) was conducted in 19 centres and the data were abstracted daily by trained nurses based on standardised protocols and data coding instructions.[18, 29] The primary goal of the registry was to study contemporary issues associated with caesarean section. All women undergoing a primary caesarean section with an infant of at least 500 g or 20 weeks’ gestation at a participating centre were prospectively ascertained.[30]

Primary exposure groups

Acute clinical emergency caesarean (group I; ACE) sections were those with primary indication of at least one of the following : umbilical cord prolapse, placental abruption, placenta praevia with haemorrhage, or uterine rupture.[31] Group II included women who underwent a caesarean for fetal distress or non-reassuring fetal heart rate tracing as the primary indication. Women with primary indication for caesarean section that included cephalopelvic disproportion, failure to progress or failed induction were assigned to group III.

Maternal and neonatal outcomes

Composite maternal morbidity (CMM) was defined as the presence of any of the following: intrapartum or postpartum transfusion, uterine rupture, hysterectomy, cystotomy, ureteral or bowel injury or death. As previously reported,[18] cystotomy may have been elective or accidental, and these were not distinguished in the data. Based on clinical impression the surgeon reported whether there was a bowel or ureteral injury, although the specific type of bowel injury was not ascertained. Composite neonatal morbidity (CNM) included one or more of the following: umbilical arterial pH <7.00, neonatal seizure, cardiac, hepatic, renal dysfunction, hypoxic ischaemic encephalopathy, or neonatal death (within the first month).

Statistical analysis

Maternal sociodemographic characteristics, intrapartum course and neonatal outcomes were compared across the three indications for caesarean section. Multivariable regression models were constructed to examine the indications for caesarean as a predictor of selected maternal/neonatal morbidity and mortality outcomes. These models were adjusted for maternal age, ethnicity, marital status, education and cigarette use.

To ensure that these associations were unaffected by observed confounding bias, we examined the association based on propensity scoring methods. Three logistic regression models were developed comparing groups I and II, I and III, and groups II and III. In these models, we adjusted the risks for maternal age, ethnicity, marital status, education and cigarette use. From each of the models were estimated the probability of being in each of the groups (propensity scores). In the second stage, we fit a log-binomial regression model for each of the maternal and neonatal outcome based on the three groups, with the models weighted by the inverse probability of treatment weights.[32] Statistical analyses were performed using SAS version 9.3 (SAS Institute, Cary, NC, USA).

Cohort composition

The caesarean section registry included a total of 73 257 women; among these women, 40% (n = 29 669) had repeat caesarean, 25% (n = 18 261) had trial of labour after caesarean, and 35% (n = 25 327) had primary caesarean section. We restricted the study to nulliparous women who delivered non-anomalous singleton births at 34 weeks or more with vertex presentation. Thus, among the 25 327 who had primary caesarean, we excluded 61% (n = 15 498) women for the following reasons: parous women (n = 6723), multiple gestation (n = 3458), non-vertex presentation (n = 2266), gestational age <34 weeks (n = 1492), other indications for caesarean (n = 1208; based on indications other than the ones listed to define the three groups), anomalous fetus (n = 338) and stillbirths (n = 13). After all (sequential) exclusions, 9829 women remained in the study.

Results

Among the 9829 nulliparous women that met the eligibility criteria, the primary reasons for caesarean section were ACE in 1% (n = 114), non-reassuring FHR in 29% (n = 2822) and failed induction/dystocia in the remaining 70% (n = 6893). Among the 114 women that had an ACE caesarean section, the two most common indications were cord prolapse (37%) and abruption (32%).

Table 1 provides the characteristics of the cohorts across the three groups. In each group about 20% had less than a high school education and over 70% had a body mass index ≥30 kg/m2 at caesarean section. The peripartum findings of the groups are described in Table 2. The prevalence of chorioamnionitis, as well as the decision as to incision time interval, and utilisation of general anesthesia varied among the groups. The cervical dilation at the time of caesarean was 0–4 cm in over 90% of the nulliparous women in all three groups (Figure 1).

Table 1. Maternal characteristics of nulliparous women
CharacteristicsCaesarean among nulliparous P
Gr I: ACE (= 114)Gr II: NR FHR (= 2822)Gr III: Arrest (= 6893)
  1. Gr, group; ACE, acute clinical emergency; NR-FHR, non-reassuring fetal heart rate; BMI, body mass index; HELLP, haemolysis elevated liver function, and low platelets.

  2. Data presented as % (n).

  3. Arrest disorder included caesarean for failed induction, cephalopelvic disorder, failure to progress.

Maternal age (years)
<2028.9 (33)24.2 (683)18.4 (1269)<0.001
≥356.1 (7)10.1 (286)9.6 (664)
Ethnicity
Afro-American32.5 (37)41.9 (1183)26.5 (1829)<0.001
Caucasian43.8 (50)33.3 (940)41.9 (2892)
Hispanic19.3 (22)19.2 (541)25.6 (1765)
Other4.4 (5)5.6 (158)5.9 (407)
Marital status
Married42.1 (48)42.3 (1193)51.8 (3570)<0.001
Single56.1 (64)56.2 (1587)47.3 (3257)
Unknown1.8 (2)1.5 (42)0.9 (66)
Education
Less than high school21.9 (25)22.8 (643)19.1 (1313)<0.001
High school28.1 (32)24.8 (701)25.4 (1751)
Beyond high school24.6 (28)25.7 (726)29.7 (2048)
Missing25.4 (29)26.7 (752)25.8 (1781)
Gestational age (weeks)
34–3631.6 (36)10.8 (304)4.6 (318)<0.001
37–4057.9 (66)66.3 (1871)69.5 (4791)
≥4110.5 (12)22.9 (647)25.9 (1784)
Smoker14.1 (16)13.3 (374)10.8 (747)0.002
Alcohol2.6 (3)3.4 (97)2.8 (198)0.320
Drugs5.3 (6)4.1 (116)2.3 (161) 
Pre-pregnancy BMI (kg/m 2 )
15–18.93.5 (4)4 (118)4 (264)<0.001
19–24.928 (33)32.5 (917)34 (2274)
25–29.924.5 (29)19 (505)20 (1325)
30–39.95 (6)13 (366)15 (980)
>400 (0)3.5 (104)4 (271)
Missing39 (45)28 (793)23 (1517)
BMI at delivery (kg/m 2 )
18.5–24.99.7 (11)5.6 (157)4.3 (299)<0.001
25.0–29.910.5 (12)9.3 (263)5.9 (413)
30.0–39.935.9 (41)28.2 (797)27.7 (1907)
≥40.040.4 (46)43.3 (1223)47.0 (1033)
Missing3.5 (4)13.5 (382)14.9 (1033)
Maternal hypertension
Gestational hypertension4.4 (5)4.6 (129)6.6 (457)<0.001
Preeclampsia16.7 (19)13.0 (367)12.7 (877)
HELLP/Eclampsia10.5 (12)0.5 (15)0.3 (17)
Maternal diabetes7.0 (8)7.7 (218)9.4 (645)0.029
Table 2. Peripartum characteristics of nulliparous women
CharacteristicsCaesarean among nulliparous P
Gr I: ACE (= 114)Gr II: NR FHR (= 2822)Gr III: Arresta (= 6893)
  1. Gr, group; ACE, acute clinical emergency; NR-FHR, non-reassuring fetal heart rate.

  2. Data presented as % (n) or mean ± SD.

  3. a

    Arrest disorder included caesarean for failed induction, cephalopelvic disorder, failure to progress.

PROM before 37 weeks 2.6 (3)1.7 (49)0.9 (68)0.036
Labour
Failed induction2.6 (3)1.2 (35)1.0 (70)<0.001
Induction29.8 (34)14.9 (1185)43.4 (2989)
Spontaneous26.3 (30)19.8 (560)6.5 (450)
Augmentation13.2 (15)32.5 (916)46.3 (3194)
Missing28.1 (32)4.5 (126)2.8 (190)
Chorioamnionitis4.4 (5)11.8 (332)22.6 (1560)<0.001
Decision to incision interval (min)22 ± 3636 ± 4062 ± 56<0.001
General anaesthesia52.6 (60)17.9 (506)3.5 (243)<0.001
Figure 1.

Maximum cervical dilation before caesarean section.

The overall risks of CMM and CNM were 2.5% (95% CI 2.2–2.8)and 1.9% (95% CI 1.7–2.2), respectively. The CMM and CNM for each group are depicted in Figure 2. Group I had the highest CMM (Table 3) and CNM (Table 4) of the three groups. The most common CMM was transfusion, ranging from 2.1% (group II) to 7.9% (group I). The propensity score-adjusted risk ratio for CMM was four-fold higher in group I than II (RR 4.1, 95% CI 3.1, 5.3) and over three-fold higher in group I than III (RR 3.2, 95% CI 2.7, 3.7); however, the risks were similar between groups II and III (Table 3). Renal dysfunction was the most common neonatal end-organ damage. The risk of CNM varied across the three groups (Table 4). The risk of CNM was over 14-fold higher (RR 14.1, 95% CI 10.7, 18.7) among infants delivered by caesarean section for ACE indications in comparison with those born by caesarean for failed induction/dystocia.

Table 3. Risks of maternal complications among nulliparous women in relation to indications for caesarean delivery
 Gr I: ACE (= 114)Gr II: NR FHR (= 2822)Gr III: Arresta (= 6893)RR (95% CI)b I versus IIRR (95% CI)b II versus IIIRR (95% CI)b I versus III
  1. Gr, group; ACE, acute clinical emergency; NR-FHR, non-reassuring fetal heart rate.

  2. Data presented as % (n).

  3. a

    Arrest disorder included caesarean for failed induction, cephalopelvic disorder, failure to progress.

  4. b

    Adjusted for maternal age, ethnicity, marital status, education and cigarette use by propensity score (weighting).

  5. c

    Injury to bladder, bowel or ureter.

  6. d

    The maternal death in group I was due to cardiopulmonary arrest; in groups II and III the cause was hypertensive encephalopathy.

  7. e

    Composite maternal morbidity consisted of any of the following: transfusion, uterine rupture, hysterectomy, maternal injury or death.

Transfusion7.9 (9)2.1 (58)2.5 (169)3.5 (2.6, 4.6)0.8 (0.7, 1.0)2.5 (2.1, 3.0)
Uterine rupture00.04 (1)0
Hysterectomy0.9 (1)0.3 (7)0.2 (14)1.4 (0.6, 3.1)33.1 (18.53, 59.2)69.2 (39.0, 122.4)
Maternal injuryc1.8 (2)0.3 (7)0.1 (8)7.3 (3.4, 15.8)2.0 (1.0, 4.1)14.6 (7.2, 29.7)
Maternal deathd11.1 (1)0.9 (1)0.5 (1)11.9 (1.7, 83.0)1.6 (0.2, 16.6)37.1 (5.3, 260.3)
Composite MMe9.7 (11)2.2 (63)2.6 (176)4.1 (3.1, 5.3)0.8 (0.7, 1.0)3.2 (2.7, 3.7)
Table 4. Risks of neonatal complications among nulliparous women in relation to indications for caesarean delivery
 Gr I: ACE (= 114)Gr II: NR FHR (= 2822)Gr III: Arresta (= 6893)RR (95% CI)b I versus IIRR (95% CI)b II versus IIIRR (95% CI)b I versus III
  1. Gr, group; ACE, acute clinical emergency; NR-FHR, non-reassuring fetal heart rate; UA, umbilical arterial; NC, not calculable (due to the small sample size and the model would not converge).

  2. Data presented as % (n).

  3. a

    Arrest disorder included caesarean for failed induction, cephalopelvic disorder, failure to progress.

  4. b

    Adjusted for maternal age, ethnicity, marital status, education and cigarette use by propensity score (weighting).

  5. c

    One of the following: hypoxic ischaemic encephalopathy, neonatal seizure, renal, cardiac or hepatic dysfunction.

  6. d

    Composite neonatal morbidity included any of the following: UA pH <7.00, end-organ damage or neonatal death.

UA pH <7. 0014.1 (12)3.8 (76)0.7 (21)4.1 (3.3, 5.22)5.6 (3.9, 8.0)20.4 (13.3, 31.4)
Renal dysfunction0.0 (0)1.4 (39)0.3 (19)NC5.0 (3.3, 7.6)NC
Cardiac dysfunction1.8 (2)1.2 (35)0.1 (7)NC11.6 (6.1, 22.0)14.7 (6.9, 31.5)
Hepatic dysfunction1.8 (2)0.6 (17)0.09 (6)NC7.2 (3.6, 14.5)15.0 (6.6, 34.1)
Neonatal seizures1.8 (2)0.9 (26)0.2 (14)NC4.6 (2.8, 7.5)5.6 (3.2, 9.8
Hypoxic ischaemic encephalopathy1.8 (2)0.8 (23)0.03 (2)NC30.7 (8.9, 106.4)51.6 (12.9, 206.8)
End-organ damagec3.5 (4)2.6 (72)0.5 (34)1.3 (1.0, 1.7)5.0 (3.7, 6.8)5.7 (4.0, 8.2)
Neonatal death0.9 (1)0.3 (8)0.03 (2)NC10.5 (2.9, 37.9)32.3 (8.0, 130.5)
Composite neonatal morbidityd11.4 (13)4.4 (125)0.7 (51)2.8 (2.3, 3.4)6.0 (4.7, 7.7)14.1 (10.7, 18.7)
Figure 2.

Maternal and neonatal composite morbidity.

Discussion

Principal findings

Our first finding is that the CMM varied based on the indication for the caesarean section (Figure 2). Although ACE delivery is an uncommon indication for caesarean section (1%), it was done most expeditiously and was associated with the highest neonatal morbidity – even higher than the group that underwent caesarean section for non-reassuring FHR. This is an unexpected new finding, as the caesareans in Group I were not done for fetal distress alone. As the primary indication for caesarean section was used to categorise our groups, our hypothesis is that most caesareans in Group I may have been done to avoid fetal distress, as conditions such as prolapsed cord, abruptio placenta or placenta previa may not necessarily be associated with fetal distress. If the above clinical conditions were also associated with fetal distress, then these cases would have as a primary indication fetal distress and therefore would have been included in group II. With respect to maternal morbidity, blood transfusion, the most common morbidity, and CMM had a similar pattern: highest in the emergent caesarean section group and lowest in the non-assuring FHR group. Based on data from the MFMU caesarean registry, Rouse and colleagues[33] noted that among primary caesarean sections, general anaesthesia, placenta previa, and severe preoperative anaemia (haematocrit <25%) significantly increased the odds of transfusion. Although their study was not restricted to nulliparous women or to specific indications for caesarean, it is possible that these variables are associated with the risk of transfusion with ACE caesarean section. Uterine rupture and hysterectomy were uncommon events among nulliparous women. Maternal injury occurred significantly more often when caesarean section was done for acute clinical emergency than other indications (Table 3).

Our second finding is that the CNM differed between the three groups, which was highest for ACE caesarean section and lowest for arrest disorders. We analysed the risks of umbilical arterial pH <7.00 as well as end organ damage in those ≥34 weeks as proxies for acute events sufficient to cause cerebral palsy, as recommended by the American College of Obstetricians and Gynecologists.[34] Previously, it was reported that renal dysfunction was the most common end organ damage among hypoxic newborns,[35] and our study corroborated this finding. Although the risks of hypoxic ischaemic encephalopathy (HIE) and neonatal mortality were different between groups I and III, the wide confidence intervals for the risk ratio suggests some imprecision. Our findings corroborate those of previous publications[31, 36] that reported that the rate of neonatal acidosis is linked with indications for emergency delivery. Future studies should discern the factors that significantly ameliorate the risk of neonatal morbidity with emergent caesarean.

The third finding was that, regardless of the indications, over 90% of caesarean sections were done when the cervical dilation was 0–4 cm. Even when the caesarean section was done for arrest disorders, almost 90% of the women had cervical dilation of 4 cm or less and only 1% was noted to be 7–10 cm. Considering this was a prospective study at multiple academic centres, where clinicians were aware that the indications and outcomes of the caesareans sections were being reviewed, it was unexpected that the majority of women who had caesarean sections for arrest disorder were not in active phase. To a lesser extent, Zhang et al.[2], reporting for the Consortium on Safe Labor, noted that among nulliparous women who were induced and had caesarean section, 50% were not in the active phase of labour. Subsequent to the data collected for the MFMU caesarean registry, it has been established that labour may take more than 6 hours to progress safely from 4 to 5 cm of cervical dilation.[37]

Strengths and limitations of the study

The strengths of the study are noteworthy. Trained nurses collected the data for the caesarean sections registry based on a standardised protocol. The multicentre design of the study makes our findings generalisable to the obstetric population of nulliparous women. The analyses based on propensity scores with the inverse probability of treatment weighting scheme minimises the impact of bias due to selection in observational cohort studies.

The interpretation of our findings must be balanced with respect to a few limitations. In this cohort of nulliparous women who underwent caesarean sections, only about 3% had any maternal or neonatal morbidity/mortality. However, our definition of composite morbidity did not include some outcomes such as wound disruption or admission to neonatal intensive care unit, which contribute to prolonged hospitalisation and healthcare costs.[38, 39] It is very likely that the burden of maternal and newborn morbidities may have been underestimated. Another limitation is that we do not have data on the consensus among the centres regarding when to proceed with caesarean. It is unknown, for example, whether all women had intrauterine resuscitation before caesarean section for non-reassuring FHR.[40] The study was done in academic centres, and the rate of CMM and CNM may be different in community hospitals, especially when there is lack of 24/7 coverage by clinicians. These theoretical differences in peripartum complications between academic and other settings merit scrutiny. The data was collected prospectively; thus there is a potential for Hawthorne effect. Leung and colleagues,[41] for example, noted a significant reduction in birth trauma and asphyxia related to instrumental deliveries when clinicians were aware that the data was being collected prospectively. Likewise, it is possible that the rate of CMM and CNM was different when clinicians at various centres were aware of the data being collected compared with before the start or completion of the study. Our findings may not be applicable to deliveries before 34 weeks, women with multiple gestations, parous women or to women with prior caesareans. We acknowledge that we did not classify cesarean section as proposed by Lucas et al.[42] because this is not used in the USA. Future studies should quantify the CMM and CNM associated with the four-grade classification system.[42] Lastly, the data is over 10 years old (1999–2002) and the reported composite morbidity may not reflect current practices and sequelae. However, considering the sample size, the prospective nature of data collection and that there have been no substantial changes in surgical techniques involved with caesarean that decrease the CMM,[43] we think the findings are relevant to contemporary practice, and will serve as reliable historical data in the future.

Interpretation

Pregnancy and childbirth are not without complications.[44] The peripartum morbidity[45] and adverse outcomes that accompany subsequent deliveries[46] are appreciably higher with a caesarean section than in women who have a vaginal birth. As there is an increasing use of caesarean section,[1] there is a need to comprehend the reasons for the increase[2], as well as make concerted efforts to decrease its rate.[23, 24] We believe that an understanding of the risks of maternal and neonatal morbidity associated with various indications of caesarean section is important. Such a granular understanding will permit accurate counseling to women, development of predictive models and guide future investigations toward minimising the untoward sequelae. Hence, we undertook this secondary analysis of the three indications for caesarean section among nulliparous women. The relationship between the primary indication for caesarean section, as well as quantification of the associated CMM and CNM, has not been described previously.

Conclusions

Our study indicates that among nulliparous women at near term, the composite maternal and neonatal morbidity is significantly linked with the indication for caesarean section, with acute clinical emergency being done most expeditiously, and carrying the highest risks of maternal and neonatal morbidity and mortality. As there is a reliable online calculator for predicting the success of vaginal delivery following a caesarean section, as well as for the prediction of morbidity with a trial of labour after caesarean section,[47-49] the development of a model to predict the probability of caesarean section for ACE or non-reassuring FHR as indications among nulliparous women would potentially enhance resource allocation and diminish associated morbidity.

Acknowledgements

We acknowledge the assistance of the NICHD, the MFMU Network and the Protocol Subcommittee in making the database available to us. The contents of this report represent the views of the authors and do not represent the views of the NICHD MFMU Network.

Disclosure of interests

The authors declare they have no conflict of interests.

Contribution to authorship

S.P.C.: Designed the study and directed its implementation, including quality assurance and control. H.B.: Designed the study's analytic strategy and ran the statistical analysis.

I.A.H.: Helped conduct the literature review, prepared the Materials and Methods, and collected study data. Responsible of the IRB application and management. S.B.: Helped conduct the literature review and the Discussion sections of the text. J.B.H.: Helped in the design and Discussion section. M.M.: Helped with Results section of the study. M.E.d'A.: Helped to review the manuscript and provided enlightened critical points in the study. A.Z.A.: Helped review the manuscript and provided enlightened critical points in the study. A.M.V.: Helped review the manuscript and provided enlightened critical points in the study. C.V.A.: Senior statistician helped design the study and the statistical analysis, and provided critical review of the manuscript.

Details of ethics approval

The Eastern Virginia Medical School institutional review board approved the study on 7 May 2012, reference number 12-05-NH-0130.

Funding

The study had no funding.

Ancillary