birth position, accoucheur, and perineal outcomes: informing women about choices for vaginal birth

Birth Position, Accoucheur, and PerinealOutcomes: Informing Women About Choices

for Vaginal Birth

Allison Shorten, RN, RM, MSc, Jacki Donsante, RN, RM, MSc,and Brett Shorten, BA, MCom

ABSTRACT: Background: The literature is tentative in establishing links between birthposition and perineal outcome. Evidence is inconclusive about risks and bene®ts of women'soptions for birth position. The objective of this study was to gain further evidence to informperinatal caregivers about the e�ect of birth position on perineal outcome, and to assist birthattendants in providing women with information and opportunities for minimizing perinealtrauma.Methods: Data from 2891 normal vaginal births were analyzed. Descriptive statisticswere obtained for variables of interest, and cross-tabulations were generated to explorepossible relationships between perineal outcomes, birth positions, and accoucheur type.Logistic regression models were used to examine potential confounding and interaction e�ectsof relevant variables. Results: Multiple regression analysis revealed a statistically signi®cantassociation between birth position and perineal outcome. Overall, the lateral position wasassociated with the highest rate of intact perineum (66.6%) and the most favorable perinealoutcome pro®le. The squatting position was associated with the least favorable perinealoutcomes (intact rate 42%), especially for primiparas. A statistically signi®cant associationwas demonstrated between perineal outcome and accoucheur type. The obstetrician groupgenerated an episiotomy rate of 26 percent, which was more than ®ve times higher thanepisiotomy rates for all midwife categories. The rate for tear requiring suture of 42.1 percentfor the obstetric category was 5 to7 percentage points higher than that for midwives. Intactperineum was achieved for 31.9 percent of women delivered by obstetricians compared with56 to 61 percent for three midwifery categories. Conclusion: Findings contribute to growingevidence that birth position may a�ect perineal outcome. Women's childbirth experiencesshould re¯ect decisions made in partnership with midwives and obstetricians who are equippedwith knowledge of risks and bene®ts of birthing options and skills to implement women'schoices for birth. Further identi®cation and recognition of the strategies used by midwives toachieve favorable perineal outcomes is warranted. (BIRTH 29:1 March 2002)

Injury to the genital tract during vaginal birth, eitherthrough tearing or episiotomy, can lead to adversehealth outcomes for women, ranging from minortemporary discomforts to severe pain, bleeding,dyspareunia (painful intercourse), infection, urinaryincontinence, fecal incontinence, and interferencewith establishment of breastfeeding (1,2). Short-and long-term health problems resulting from peri-neal trauma can signi®cantly a�ect women's qualityof life (3), and they are less likely to experience theseproblems when their perineum is intact (3,4). Theliterature supports restrictive use of episiotomy and

Allison Shorten is a Lecturer (Midwifery) in the Faculty of Healthand Behavioural Sciences, University of Wollongong; Jacki Donsanteis a Midwife/Clinical Care Co-ordinator in the Birthing Unit ofWollongong Hospital; and Brett Shorten is a tutor in quantitativeanalysis at The University of Wollongong, Wollongong, New SouthWales, Australia.

Address correspondence to Allison Shorten, RN, RM, MSc,Department of Nursing, University of Wollongong, North®eldsAve, Wollongong, NSW, Australia, 2522.

Ó 2002 Blackwell Science, Inc.

18 BIRTH 29:1 March 2002

emphasizes the need for further evidence to supportlabor practices that minimize perineal trauma (5,6).Positioning of women at the time of birth andaccoucheur type are two potentially importantclinical factors.

Prevention of trauma to the genital tract duringbirth is a clinical challenge. Limited guidance isprovided by research-based literature about e�ectiveclinical practices that minimize perineal traumaduring normal vaginal birth. Evidence about therelationship between birth position and perinealtrauma remains tentative, with recommendationsfocusing on women making ``informed choices''based largely on positions they ®nd most comfort-able at the time of birth (7). Perinatal caregivers areleft with limited evidence on which to inform womenabout the most favorable positions for minimizingperineal trauma during birth.

A systematic review of the literature aboutwomen's position and second stage of labor (7)highlighted potential advantages for upright posturesover supine and lithotomy positions. Bene®ts such asreduced duration of second stage of labor, reducedrates of assisted birth, fewer episiotomies, and fewerabnormal fetal heart rate patterns were noted (7).Risks included a small increase in the likelihood ofsecond degree tear and blood loss greater than 500 ml(7). Problems with methodological quality and con-sistency between studies (due to di�erences in de®ni-tions of ``upright'' position, measurement of length ofsecond stage, de®nitions of perineal trauma, andpolicies for restrictive versus routine episiotomy use)made comparison of research results di�cult andreduced the robustness of conclusions. Therefore,clinical recommendations about positioning duringbirth focus on women's choice as an importantdeterminant (6,7). Given the responsibility of inform-ing women about their choices, birth attendants needclear evidence about the merits of various positionsfor vaginal birth.

The relationship between accoucheur type andepisiotomy has been examined, with evidence tosuggest that midwives perform fewer episiotomiesthan obstetric physicians, even when clinical factorsare controlled for (8±11). Low et al recently reportedthat women whose accoucheurs had high rates ofperforming episiotomy experienced an increased riskof major perineal trauma when compared withwomen whose accoucheurs had lower episiotomyrates (11). In fact, perineal outcomes for the formergroup were inferior even when an episiotomy was notperformed. The clear di�erences in perineal manage-ment that occur among various accoucheur groupsneed to be further identi®ed to improve perinealoutcomes for women. The relationship between birth

position, accoucheur type, and perineal outcome ispotentially important in highlighting ``best practice''in perineal management.

Our study aimed to provide further evidence toinform midwives and obstetricians about the e�ect ofbirth position on perineal outcome and to assist birthattendants in providing women with information andopportunities for minimizing perineal trauma. It alsoaimed to build on previous research relating to thein¯uence that midwifery and obstetric practices haveon perineal outcomes and ultimately the quality oflife for women who experience preventable perinealtrauma.

The study addressed two research questions: First,what is the e�ect of maternal birth position onperineal outcome? Second, what is the e�ect ofaccoucheur type on perineal outcome?

Methods

Retrospective analysis of 2891 consecutive ``normalvaginal births'' was conducted. The births occurred ina regional teaching hospital in New South Wales,Australia, between 1st April 1998 and 1st April 2000.The study was approved by the Human ResearchEthics Committee for the Illawarra Area HealthService, Wollongong, New South Wales.

The study site is a public hospital, where mostwomen receive public insurance. In cases of ``normallabor'' for ``publicly insured women,'' the birth isconducted by midwives, student midwives, or obstet-ric registrars. Specialist obstetricians attended normalvaginal birth for ``privately insured women'' only. Aprevious investigation in this hospital suggested thatfew clinically relevant di�erences exist between thetwo groups, that is, no signi®cant di�erences inmaternal weight, parity, primiparity, gestationaldiabetes, or pregnancy-induced hypertension (9).Although statistically signi®cant di�erences werefound in maternal height and newborn birthweight,the magnitude of these di�erences was small (1 cm[0.4 in] and 0.09 kg [about 3 oz], respectively).

The present dataset included information onpregnancy history, labor history, birth position,accoucheur type, perineal outcome, and baby Apgarscore at 5 minutes. For this study, midwives (or otherattendant midwife) were asked to note the mother'sposition at the time of birth. To facilitate consistentrecording of birth position, a chart was located nearthe ``birth register,'' depicting six birth positions,assigned letters A through F, where position A wassemi-recumbent (semi-sitting), B was lateral (side-lying), C was all fours (hands and knees), D waskneeling (upright), E was standing, and F wassquatting.

BIRTH 29:1 March 2002 19

Statistical Analysis

In the logistic regression analysis presented below,birth position was coded as a series of categoricalvariables, with the number one (1) indicating that abirth position had been used and a zero (0) otherwise.Estimation of these models required that one of thesevariables served as the reference category, in this casethe semi-recumbent position. Perineal outcomes wererecorded as intact perineum, graze not sutured, ®rstdegree tear sutured, second degree tear sutured, thirddegree tear sutured, episiotomy, and labial tearsutured. The logistic regression models used categor-ical variables for whether the women experienced anepisiotomy, a tear requiring suture, or an intactperineum as dependant variables. In each case thevariable was coded as one (1) if the condition waspresent and zero (0) otherwise. Accoucheur groupsincluded specialist obstetricians, other doctors(obstetric registrars and general practitioners), ``core''midwives (specialist labor/birthing unit midwives),``other'' midwives (those who rotate into the labor/birthing unit), and student midwives. This factor alsowas coded as a series of categorical variables equal toone (1) if the accoucheur was of that type and zero (0)otherwise. The reference category was specialistobstetricians.

As mentioned, both previous research and clinicalfactors suggest that perineal outcomes may be in¯u-enced by accoucheur type, birth position, or both,and these possible associations are the focus of thisstudy. Other potential explanators of perineal out-come and/or confounders of these associations(birthweight, maternal age, parity, use of epiduralanesthesia, length of second stage of labor, Apgarscore) were also selected on the basis of clinicalplausibility, conditioned by availability of data.Hence, our model building and estimation strategywere based on including variables available within thedataset that are theoretically plausible explanators ofperineal outcome, rather than relying on statisticalprocedures such as stepwise regression to determinewhich variables to include in the model.

The SPSS software (12) was used for all analysesof this dataset. Descriptive statistics were generatedfor all variables, based on the pooled dataset. Cross-tabulations explored possible relationships betweenperineal outcomes and the variables of interest, birthposition and accoucheur type. The potential e�ects ofconfounding and interaction e�ects of other variableswere examined by using logistic regression modelswith perineal outcome (episiotomy, perineal trauma,intact perineum) as the dependant variables. Aseries of multiplicative interaction variables wereconstructed to examine the possibility of interaction

e�ects between accoucheur type and birth position.We point out that no signi®cant interaction e�ectswere found in any of the models, hence interactionvariables are not included in the results presented.The logistic regression models were completed sepa-rately for nulliparous (primiparous) and multiparouswomen. Statistical signi®cance is assessed by SPSSby means of the Wald statistic with one degree offreedom (1 df ).

Results

Tables 1 and 2 report summary statistics for allvariables used in the study, separately for categoricaland continuous variables. Registered midwives per-formed more than 80 percent of all normal vaginalbirths, obstetricians and other doctors accounted fornearly 10 percent of such births, and studentmidwives a further 7.5 percent. Given that all birthswere classi®ed as ``normal'' vaginal, it is not surpri-sing that 99 percent of babies in this study had5-minute Apgar scores of 7 or more, with an overallmean of 9.16. Mean parity was just over one, butparity showed great variability, with a maximumvalue of 10 being recorded (Table 2); approximately60 percent of women were multiparous. Women were13 to 45 years of age (mean 27 yr), and the meannewborn birthweight was 3.39 kg (approximately 7.5lb). Duration of second stage of labor was highlyvariable (Table 2), although the mean duration wasslightly more than 30 minutes. In terms of overallperineal outcomes, 55.4 percent of women had anintact perineum, 38.1 percent received a tear requi-ring suture, 6.5 percent an episiotomy, and 0.9percent a third degree tear (Table 1).

Table 3 summarizes information about eachaccoucheur group. Although ``other'' midwivesaccounted for just over one-half (1447/2891) of allvaginal births during the study period, as expectedboth ``core'' midwives and obstetricians deliveredmore babies on average. Student midwives accountedfor 216 births, and other physicians (primarilyobstetric registrars) accounted for only 53.

Birth Position and Perineal Outcomes

Over one-half (56%) of all women delivered in thesemi-recumbent position. Women also used the all-fours (18.9%), lateral (12.2%), and standing (9.5%)positions, mostly when attended by midwives; a fewdelivered in the squatting (2.1%) or upright kneeling(1.3%) positions.

Since the relationship between birth position andperineal outcome was the focus of the study, Table 4presents a cross-tabular analysis of these factors. The


``traditional'' semi-recumbent birth position wasassociated with higher episiotomy rates than otherpositions. This ®nding was at least partly due to theoverwhelming popularity of episiotomy with obste-tricians and other physicians (i.e., of the 288 birthscarried out by physicians, 243 [84.4%] occurred inthis position), who performed much higher rates ofepisiotomy in general. In fact, in most cases the lowerrates of episiotomy observed for alternative birthpositions appeared to have come at the expense ofhigher rates of tear requiring suture, since intactperineum rates for semi-recumbent, all fours, kneel-ing, and standing were all similar (54%±55%). In thisstudy an ``intact'' perineum was de®ned to includegrazes and tears not sutured as well as fully intactperineum outcomes (Table 1). The lateral birthposition performed well, resulting in below-averagerates of episiotomy and, especially, tear requiringsuture, and hence, achieving by far the highest intactrate at 66.6 percent. Conversely, the squatting posi-tion had the highest tear requiring suture rate(> 53%) and the lowest intact rate (just under 42%).It is worth noting that, although not shown inTable 4, the squatting position was also associatedwith the highest rate of the most severe trauma (thirddegree tear) (3.2% compared with 0.9% overall,Table 1).

The lateral position was associated with the mostfavorable perineal outcome pro®le and the squattingposition with the worst (Table 4). There was littleevidence that the other more ``alternative'' birthpositions (all fours, kneeling, standing) performedbetter than the semi-recumbent position, apparentlyindicating the substitution of tear requiring suture forepisiotomy.

It also should be pointed out that small numbers ofwomen delivering in the kneeling and squattingpositions contributed to high standard errors forthese variables in the logistic regression models, sothat little can be added to the discussion of Table 4with respect to these positions. These results do implythat the squatting position may have the lowest rate ofintact perineum, however, other things being equal.Table 5 con®rmed that episiotomy rates were higherfor the semi-recumbent position than for the lateral,standing, and especially, all-fours positions, even aftercontrolling for accoucheur type and other possibleconfounding factors. However, this occurred only inthe case of nulliparous women. In addition, it shouldbe noted that the overlap in 95% con®dence intervalsfor odds ratios in Table 5 suggests that we cannotreject the hypothesis that other birth positions lead tosimilar episiotomy outcomes. Table 6 suggests thatthe lateral position also demonstrated an advantageover the semi-recumbent in terms of lower tear

requiring suture rates, but the latter rates are as highor even slightly higher (although not statisticallysigni®cantly di�erent) for all other birth positionscompared with the semi-recumbent position. Again,all 95% con®dence intervals reported in Table 6overlap, suggesting no signi®cant di�erences in tearrequiring suture rates among these birth positions.The net result of these e�ects on episiotomy and tearrequiring suture is that the lateral position had thehighest probability of leading to an intact perinealoutcome, in particular signi®cantly outperforming thesemi-recumbent, all-fours, and standing positions inthe case of multiparas. Other positions demonstratedsimilar intact perineum rates, however, except fornulliparas who birthed in the squatting position; theyexperienced a lower rate of intact perineum.

Table 1. Summary Information for All Categorical Vari-ables in the Study

VariableNumberof Births (%)

Perineal outcomeIntact 1122 (38.8)Graze/tear not sutured 479 (16.6)Total intact 1601 (55.4)

1st degree tear 332 (11.5)2nd degree tear 633 (21.9)3rd degree tear 27 (0.9)Labial tear 109 (3.8)Total TRS 1101 (38.1)

Episiotomy 189a (6.5)

Total 2891 (100.0)

Accoucheur typeObstetrician 235 (8.1)Other physician 53b (1.8)Core midwife 940 (32.5)Other midwife 1447 (50.1)Student midwife 216 (7.5)Total 2891 (100.0)

Birth positionSemi-recumbent 1619 (56.0)Lateral 353 (12.2)All fours 545 (18.9)Kneeling 38 (1.3)Standing 274 (9.5)Squatting 62 (2.1)Total 2891 (100.0)

Epidural 256 (8.9)Apgar < 7 14 (0.5c)Nulliparas 1117 (38.8d)Multiparas 1763 (61.2d)

a Three women experienced both episiotomy and 3rd degreetearÐthese appear in the 3rd degree tear category in this table;b comprising 47 deliveries by obstetric registrars and six by generalpractitioners; c based on 2881 valid responses; d based on 2880 validresponses.TRS� tear requiring suture.


Tables 5 to 7 enable a comparison of the e�ect ofvarious birth positions on perineal outcome. Arelated question is whether birth position, per se,had any signi®cant association with perineal outcome(i.e., is the birth position variable as a wholeassociated with perineal outcome?). To answer thisquestion, we employed a variation of the likelihood

ratio test for subsets of co-e�cients outlined byAldrich and Nelson (13, pp 59±60), based on the v2

distribution with 6 df. These tests suggested that birthposition did a�ect the risk of episiotomy, for bothnulliparas (v2� 20.36, p < 0.005) and multiparas(v2� 30.91, p < 0.001). Similarly, signi®cant e�ectswere found for birth position on the risk of tears

Table 3. Summary Characteristics of Accoucheur Groups

AccoucheurType

No. ofPractitioners

No. ofBirths

Mean No.of Births

Median No.of Births

Minimum±MaximumNo. of Births

Obstetrician 7 235 33.6 20 3±100Other physician 16 53 3.3 2 1±17Core midwife 15 940 62.7 61 15±155Other midwife 59 1447 24.5 18 1±141Student midwife 14 216 15.4 14.5 2±35

Table 4. Perineal Outcomes by Birth Position and Accoucheur Type (%)

Birth Position

PerinealOutcome/Accoucheur Type

Position ASemi-recumbent

(n� 1619)

Position BLateral

(n� 353)

Position CAll Fours(n� 545)

Position DKneeling(n� 38)

Position EStanding(n� 274)

Position FSquatting(n� 62)

Total(n� 2891)

EpisiotomyObstetrician 26.0 * * * * * 26.0Other physician 22.9 * * * * * 15.1Core midwife 6.6 6.4 2.1 * 4.1 * 5.4Other midwife 7.4 1.5 1.3 * 1.9 * 4.5Student midwife 3.9 * 0.0 * * * 3.2Total 9.6 5.4 1.5 0.0 2.6 4.8 6.6

TRSObstetrician 42.3 * * * * * 42.1Other physician 34.3 * * * * * 39.6Core midwife 35.4 24.5 39.0 * 41.2 * 35.4Other midwife 36.6 30.7 46.6 * 43.6 * 39.4Student midwife 32.6 * 47.4 * * * 36.1Total 36.6 28.3 44.4 44.7 42.7 53.2 38.1

IntactObstetrician 31.7 * * * * * 31.9Other physician 45.7 * * * * * 47.2Core midwife 58.2 69.1 58.8 * 54.6 * 59.3Other midwife 56.0 67.8 52.1 * 54.5 * 56.1Student midwife 63.6 * 52.6 * * * 61.1Total 54.0 66.6 54.1 55.3 54.7 41.9 55.4

* < 30 Births recorded for that combination of position and accoucheur type.TRS� tear requiring suture.

Table 2. Summary Information for All Quantitative Variables Used in the Study

VariableNo. of ValidResponses Mean

StandardDeviation Range

Birthweight (kg) 2831 3.39 0.53 1.29±5.31Mother's age (yr) 2861 27.33 5.45 13±45Mother's parity (No.) 2880 1.05 1.16 0±10Length of second stage (hr) 2889 0.56 0.63 0.02±5.92Apgar score (5 min) 2881 9.16 0.63 4±10


requiring suture, for both nulliparas (v2� 11.32,p < 0.10) and multiparas (v2� 38.73, p < 0.001).As for the odds of achieving an intact perineum, birthposition was signi®cant in the case of multiparas only(v2� 39.06, p < 0.001).

Accoucheur Type and Perineal Outcomes

The in¯uence of accoucheur type on perineal out-come (Table 4) demonstrated the poor perinealoutcomes for the obstetrician group, and to a slightlylesser extent, for ``other'' physicians. In broad terms,for normal vaginal childbirth, episiotomy rates forobstetricians were ®ve times higher than those formidwives. The common justi®cation is that a liberaluse of episiotomy prevents tear requiring suture.However, this is not the case here, since obstetricians'tear requiring suture rates (42.1%) were 5 to 7percentage points higher than those of the variousmidwife groups. Interestingly, the third degree tearrate for the obstetrician group was 1.3 percent, alsoabove the overall average of 0.9 percent. Thus, liberaluse of episiotomy did not seem to have prevented``serious'' tears. The net e�ect of high episiotomy andtear requiring suture rates among obstetricians wasthat their intact perineum rate (31.9%) was 25 to 30

percentage points below those of the various midwifetypes (56%±61%).

One potentially important caveat here is thatperineal outcomes by accoucheur type could beconfounded by di�erences in maternal parity. Nosuch justi®cation was found for the outcomes ofobstetricians, however, since only 36.2 percent of theirbirths were for nulliparas, below the overall rate of38.8 percent. Hence, controlling for parity would belikely to make obstetricians' outcomes appear to beeven less favorable than is the case in Table 4.

Another consideration arises from the fact that allnormal vaginal births were included in the study,including small (preterm) infants, who are more likelyto be delivered by an obstetrician and to requireexpedited birth. This could be a confounding issuethat helps to explain the pattern of perineal outcomesattributable to obstetricians, but it had little signi®-cance in this study. Although the percentage of small(< 2500 g) babies (n � 142) delivered by obstetricians(9.9%) was slightly higher than the percentage of allbabies delivered by obstetricians (8.1%), the di�erencewas not statistically signi®cant, according to a chi-square goodness of ®t test (v2� 2.03, v20.05,3 df � 7.61).A similar result was found when babies of less than37 weeks' gestation (n� 190) were considered.

Table 5. Risk Factors for Episiotomya from Logistic Regression Modelb

Nulliparas (n� 1087) Multiparas (n� 1696)

ExplanatoryVariable Coe�cient

WaldStatistic

OddsRatio (OR)

95% CIfor OR Coe�cient

WaldStatistic

OddsRatio (OR)

95% CIfor OR

Accoucheurc

Other physician )0.84 2.06 0.43 0.14±1.36 )0.57 0.65 0.57 0.14±2.25Core midwife )1.39 18.34* 0.25 0.13±0.47 )1.46 18.13* 0.23 0.12±0.46Other midwife )1.68 27.97* 0.19 0.10±0.35 )1.78 28.54* 0.17 0.09±0.32Student midwife )2.37 13.06* 0.09 0.03±0.34 )2.06 10.09* 0.13 0.04±0.45

Birth positiond,e

Lateral )0.54 2.63 0.58 0.30±1.12 )0.38 0.80 0.68 0.30±1.58All fours )1.16 8.66* 0.31 0.15±0.68KneelingStanding )1.77 5.84 0.17 0.04±0.72 )0.52 1.06 0.60 0.22±1.60Squatting )0.05 0.01 0.96 0.27±3.41

Other variablesBirthweight 0.17 0.60 1.18 0.77±1.80 0.37 2.24 1.45 0.89±2.36Mother's age 0.03 2.20 1.03 0.99±1.08 0.07 6.38 1.07 1.02±1.13Parity na na na na )0.39 5.52 0.68 0.49±0.94Epidural )0.34 1.16 0.71 0.38±1.32 )0.59 1.23 0.56 0.20±1.57Apgar score(5 min)

)0.36 7.06* 0.70 0.54±0.91 )0.02 0.01 0.98 0.65±1.55

Length 2nd stage 0.41 10.46* 1.51 1.18±1.94 1.01 17.95* 2.74 1.72±4.37

* Signi®cant at the 0.01 level. Signi®cant at the 0.05 level.a Dichotomous variable� 1 if episiotomy experienced, 0 otherwise; b a total of 108 cases were not included in the model due to missing data;c omitted category is specialist obstetric physicians; d omitted birth category is birth position A (semi-recumbent); e blank cells indicate that nowomen both gave birth in this position and experienced episiotomy.


Table 4 documents a particularly good perinealoutcome pro®le for student midwives, who achievedthe highest intact perineum rates (61.1%), with little

use of episiotomy (3.2%) and below average inci-dence of tear requiring suture (36.1%). This wasapparently not due to a more favorable caseload in

Table 6. Risk Factors for TRSa from Logistic Regression Modelb



WaldStatistic

OddsRatio (OR)


WaldStatistic

OddsRatio (OR)

95% CIfor OR

Accoucheurc

Other physician 0.65 1.92 1.91 0.76±4.77 )0.51 0.90 0.60 0.21±1.71Core midwife 0.16 0.39 1.18 0.71±1.96 )0.18 0.73 0.83 0.55±1.26Other midwife 0.20 0.60 1.22 0.74±2.01 0.07 0.12 1.07 0.72±1.61Student midwife 0.09 0.07 1.09 0.57±2.10 )0.31 1.25 0.73 0.42±1.26

Birth positiond

Lateral )0.15 0.60 0.87 0.60±1.25 )1.24 29.04* 0.29 0.18±0.45All fours 0.28 2.66 1.32 0.95±1.84 0.06 0.15 1.06 0.79±1.42Kneeling 0.37 0.41 1.45 0.47±4.49 0.18 0.16 1.20 0.49±2.96Standing 0.44 3.90 1.55 1.01±2.40 )0.23 1.25 0.80 0.54±1.18Squatting 0.88 4.06 2.42 1.02±5.71 0.07 0.04 1.07 0.50±2.29

Other variablesBirthweight 0.53 17.23* 1.70 1.32±2.18 0.98 71.78* 2.65 2.12±3.32Mother's age 0.05 12.51* 1.05 1.02±1.07 0.09 49.98* 1.09 1.07±1.12Parity na na na na )0.70 81.84* 0.50 0.49±0.94Epidural )0.04 0.03 0.97 0.67±1.40 )0.13 0.23 0.88 0.53±1.47Length 2nd stage )0.01 0.02 0.99 0.83±1.18 0.35 5.77 1.42 1.07±1.90

* Signi®cant at the 0.01 level. Signi®cant at the 0.05 level.a Dichotomous variable� 1 if TRS experienced, 0 otherwise; ba total of 100 cases were not included in the model due to missing data; c omittedcategory is specialist obstetric physicians; d omitted birth category is birth position A (semi-recumbent).TRS� tear requiring suture.

Table 7. Contributing Factors for Intact Perineuma from Logistic Regression Model (n� 2791)b



WaldStatistic

OddsRatio (OR)


WaldStatistic

OddsRatio (OR)

95% CIfor OR

Accoucheurc

Other physician 0.36 0.42 1.43 0.48±4.25 0.77 2.41 2.16 0.82±5.71Core midwife 1.16 11.78* 3.17 1.64±6.14 0.83 15.04* 2.28 1.50±3.46Other midwife 1.21 13.23* 3.34 1.74±6.38 0.62 9.32* 1.87 1.25±2.79Student midwife 1.51 14.47* 4.50 2.07±9.78 1.03 13.75* 2.81 1.63±4.84

Birth positiond

Lateral 0.36 3.54à 1.44 0.98±2.09 1.26 33.42* 3.52 2.30±5.38All fours 0.04 0.05 1.04 0.74±1.46 0.19 1.60 1.21 0.90±1.62Kneeling 0.08 0.02 1.09 0.34±3.42 0.08 0.03 1.08 0.43±2.68Standing )0.08 0.14 0.92 0.59±1.70 0.31 2.34 1.36 0.92±2.01Squatting )1.03 3.96 0.36 0.13±0.99 0.21 0.30 1.24 0.57±2.68

Other variablesBirthweight )0.61 21.01* 0.55 0.42±0.71 )1.01 78.28* 0.37 0.29±0.46Mother's age )0.06 18.35* 0.94 0.92±0.97 )0.10 60.77* 0.91 0.89±0.93Parity na na na na 0.74 95.76* 2.09 1.81±2.42Epidural 0.19 0.87 1.20 0.82±1.77 0.27 1.03 1.31 0.64±2.18Length 2nd stage )0.18 3.23à 0.84 0.69±1.02 )0.71 20.18* 0.49 0.36±0.67

* Signi®cant at the 0.01 level. Signi®cant at the 0.05 level.à Signi®cant at the 0.10 level.a Dichotomous variable� 1 if intact perineum experienced, 0 otherwise; b a total of 100 cases were not included in the model due to missing data;c omitted category is specialist obstetric physicians; d omitted birth category is birth position A (semi-recumbent).


terms of parity. Both percent of nulliparas andaverage maternal parity were found to be almostidentical for all three midwife groups. A cautionshould be noted, however, that student midwives hadhigher third degree tear rates (2.3%) than otheraccoucheurs (0.8%), leading to a signi®cantly higherrelative risk ratio of 2.81.

With respect to the issue of core versus othermidwives, few consistent di�erences are found inTable 4, suggesting that individual variations inoutcomes, preferences for di�erent birth positions,and other management practices are more importantthan any systematic di�erences between the twogroups (in, for example, experience or amount ofrecent practice). However, it is perhaps not surprisingthat core midwives should demonstrate slightlyhigher intact perineum rates overall (59.3% vs56.1%), which is entirely due to lower tear requiringsuture rates (35.4%).

The results in Table 4 controlled only for accouch-eur type and birth position in seeking to explainvariations in perineal outcome, and the relationshipssuggested could be confounded by other importantfactors. Therefore, to make the ®ndings more robust,Tables 5 to 7 report the results of logistic regressionmodels designed to identify risk factors for episiot-omy, tear requiring suture, and intact perineumseparately for nulliparas and multiparas using anexpanded set of potential explanators in addition toaccoucheur type and birth position. These includedmaternal age and parity, use of epidural anesthesia,length of second stage of labor, newborn birthweight,and, in case of episiotomy only, 5-minute Apgar score(designed to act as an indicator of possible need toexpedite birth due to symptoms of fetal distress).

In general, the logistic regression results presentedin Tables 5 to 7 con®rmed inferences from Table 4.Given that the omitted accoucheur category isspecialist obstetricians, the signi®cant negative coef-®cients in Table 5 mean that, after controlling forother relevant variables, the odds of a womanexperiencing an episiotomy were signi®cantly reducedif the accoucheur was a midwife rather than anobstetrician. Conversely, no signi®cant di�erence wasdemonstrated in the probability of episiotomy acrossthe three midwife categories. The lower incidencewith midwife accoucheurs was not at the expense ofelevated probability of women experiencing tearrequiring suture, as shown in Table 6. Rates of thelatter did not di�er signi®cantly across accoucheurtypes (odds ratios close to 1 and an insigni®cant v2

statistic for the accoucheur variable as a whole). Thenet e�ect, shown in Table 7, was that midwifeaccoucheurs were far more likely to achieve an intactperineal outcome, evidenced by odds ratios in the

range of 2 to 4. Again, there was no evidence of anysigni®cant di�erence in intact perineal outcomesacross midwife types.

According to chi-square statistics based on likeli-hood ratio tests (5 df), discussed earlier, the accouch-eur variable as a whole had a signi®cant associationwith perineal outcome, especially with respect toepisiotomy and intact perineum. Accoucheur typewas signi®cantly associated with episiotomy for bothnulliparas (v2� 32.18, p < 0.001) and multiparas(v2� 33.64, p < 0.001). As noted above, accoucheurtype had no signi®cant association with the odds oftear requiring suture, which is not consistent with theclaim that midwife accoucheurs achieve lower episi-otomy rates at the expense of higher tear rates.Overall, the odds of intact perineum were associatedwith accoucheur type for both nulliparas (v2� 21.30,p < 0.001) and multiparas (v2� 19.04, p < 0.005).

Other Factors

Tables 5 to 7 also estimated the e�ect of several otherfactors on the probability of episiotomy, tear requi-ring suture, and intact perineum. For both paritygroups, newborn birthweight was found to elevate theodds of, especially, tear requiring suture, but to havea much more modest e�ect on the probability ofepisiotomy. Nevertheless, an increase of 1 kg inbirthweight was found to signi®cantly reduce the logodds of an intact perineum (Table 7).

As expected, higher maternal parity among mul-tiparas was found to reduce the probability of bothepisiotomy and tear requiring suture signi®cantly,and thus to increase the probability of intactperineum. This con®rmed earlier research suggestingthat the actual parity of multiparas should be takeninto account in studies of perineal outcomes, ratherthan simply classifying each woman as nulliparous ormultiparous (9). After controlling for other factors,maternal age was found to be a signi®cant predictorof perineal outcome. Older mothers were more likelyto experience tear requiring suture, and to havereduced probability of intact perineum if they weremultiparas.

Length of second stage of labor was a highlysigni®cant explanator of the use of episiotomy (longersecond stage increased the odds of episiotomy). Thisfactor was also positively associated with tear requir-ing suture outcomes in the case of multiparas only,whereas essentially no association was observed inthe case of nulliparas. It is not clear why this was so,or whether this ®nding has any clinical signi®cance,but it may re¯ect the reluctance of accoucheurs(particularly midwives) to perform episiotomy onmultiparas, given that episiotomy and tear requiring


suture may, in some cases, be viewed as ``substitute''outcomes. The net e�ect of a longer second stagelabor was found to reduce the odds of achievingintact perineum, other factors held constant. The useof epidural anesthesia was not found to be asigni®cant predictor of any of the three perinealoutcomes studied.

In the episiotomy model only (Table 5), 5-minuteApgar scores were included as a possible indicator offetal distress necessitating episiotomy. This e�ect wascon®rmed in the case of nulliparas only, with lowerApgar scores being associated with elevated odds ofepisiotomy. However, the e�ect was not strongenough to explain variations in the episiotomyexperience for more than a fraction of the sample.

As is usually the case with retrospective studiessuch as this, it was possible to think of severaladditional variables not available within the datasetthat may have helped to further explain variations inperineal outcomes. It should be pointed out, how-ever, that in an overlapping study at the samehospital site during approximately the same timeperiod, we could include a much larger number ofpossible risk factors for episiotomy (9). None of theadditional risk factors unavailable to this dataset wassigni®cant in that study.

In the logistic regression models presented inTables 5 to 7, the three dependent variables (episiot-omy, tear requiring suture, intact perineum) in factrepresented an exhaustive set of perineal outcomes.Hence, it should be noted that in reality only two ofthe models were independent, with the third beingessentially deterministic. To guard against any ana-lytical problems that could have occurred as a result,the three models were also estimated by means of adiscriminant analysis with three groups. Results,available on request from the authors, con®rmedthat the same explanatory variables found to besigni®cant in Tables 5 to 7 appeared to be associatedwith perineal outcome.

Discussion and Conclusions

In drawing conclusions about the e�ect of birthposition on perineal outcome, the evidence from thisstudy highlights the potential bene®t of using thelateral (side-lying) position, which appears to beassociated with an increase in likelihood of intactperineum. ``Alternative'' birth positions (all fours,kneeling, standing) did not perform better than themore ``traditional'' semi-recumbent position, withsimilar perineal outcomes. The squatting positiondemonstrated least favorable results, although fewwomen birthed in this position. In further explainingperineal outcomes, even when birth position, parity

and other relevant variables were taken into account,midwife accoucheurs were far more likely to achieveintact perineum. If the accoucheur was an obstetri-cian rather than a midwife, the odds of an episiotomyor tear requiring suture were greatly elevated.

These ®ndings are consistent with earlier work(8,9) in which women with private health insurancein Australia were less likely to experience intactperineum than those who had public insurance.Compared with these investigations, our study pro-vided a larger sample of women with more speci®caccoucheur information, so that where insurancestatus previously served as a proxy for accoucheurtype, our study more clearly identi®ed practitionertype, birthing style (positions used), and perinealoutcomes. This association between accoucheur typeand perineal outcome is consistent with ®ndings byLow et al (11), in that the most favorable perinealoutcomes were achieved by clinicians who performedfewer episiotomies and were midwifery rather thanobstetric clinicians.

Although a systematic review of the literature (7)highlighted the possible advantages of upright orlateral positions for the second stage of labor, thisstudy identi®ed the potential bene®ts of a lateralposition over other upright positions, even whenparity was accounted for. Given the small number ofwomen using squatting and kneeling positions, wecould not compare perineal outcomes with otherstudies where these positions demonstrated a bene®t(14).

Although the extent to which birthweight relatesto perineal outcome is unclear (14), our study foundthat a birthweight increase of 1 kg approximatelyhalved the log odds of an intact perineum fornulliparas, with a greater e�ect for multiparas.Consistent with Murphy and Feinland (14), anincrease in length of second stage of labor reducedthe odds of experiencing intact perineum, otherfactors held constant. Again consistent with otherstudies (9,14), an increase in parity signi®cantlyincreased the probability of intact perineum. Despitethis, older multiparas had a reduced probability ofintact perineum when other factors were controlledfor. In light of the results of Murphy and Feinland(14), who found that maternal age more than 40 yearsincreased the risk for perineal trauma in multiparas,this ®nding may add evidence to the range of possiblee�ects of maternal age on birth outcomes.

Several limitations of this study should be noted.A randomized controlled trial, although a�ording agreater reduction in bias than was possible in thisstudy, could not be conducted. In addition, it must benoted that it is di�cult to prove that causal relation-ships exist among accoucheur type, birth position,


and perineal outcome without conducting such atrial. Selection of birth position was either determinedby the practitioner, according to her or his individualpreference, or the woman for reasons of comfort. Inthe case of women with private insurance, obstetri-cians determine the position for birth in mostsituations. It was not possible to account for varia-tions in practices used by the accoucheurs in thestudy. Aspects of perineal management, such asperineal massage, use of oils, and perineal ``guard-ing,'' for example, were not recorded by accoucheursand have a potential impact on perineal outcome(3,6,15,16). These aspects will be highlighted inongoing research within the birthing unit.

One possible key issue a�ecting perineal outcomesfor di�erent accoucheur groups is the perceived valueof an intact perineum. Midwives may place a highervalue on achieving an intact perineum than otheraccoucheur groups. Although it was not within thescope of this study to examine these factors, ongoingresearch within the unit may reveal important di�er-ences in value systems that are re¯ected in accoucheurattitudes towards perineal outcomes.

Except in the case of student midwives, who have alimited level of experience, the status of ``core'' versus`òther'' midwives has served as a proxy for level ofmidwifery skill. The designation of obstetrician isalso assumed to carry a more highly specialized levelof skills and experience than the `òther'' physiciancategories. This may have lead to conclusions that aretentative and deserve closer analysis of the actuale�ect of speci®c perineal management skill levels onperineal outcomes.

The literature is clear in recommending restrictiveuse of episiotomy and the value of minimizingperineal trauma in terms of postpartum maternalhealth. Midwives in this study site achieved superiorperineal outcomes, irrespective of the range of birthpositions o�ered to women. Although the lateralposition was the most favorable, midwives achievedgood outcomes in the `àlternative'' positions, partic-ularly when compared with other accoucheur groups.Women who purchase private health insurance inAustralia, especially if multiparous, and choose tobirth with an obstetric specialist physician, shouldonly do so with the understanding that the perceivedbene®ts of private insurance must be weighed againstthe cost of an increased probability of experiencingperineal trauma.

Acknowledgments

We wish to thank Dr. Ken Russell, Director of theStatistical Consulting Unit, University of Wollon-

gong, for his advice concerning the analysis ofresults. We are grateful to the midwives of theBirthing Unit of Wollongong Hospital for theirongoing commitment to women and their work ingeneration of data for this study. We thank thereviewers for their helpful comments in the develop-ment of the paper. The cooperation of the IllawarraArea Health Service in conducting this research isacknowledged. However, the views expressed arethose of the authors and do not necessarily re¯ectthose of the health service.

References

1. Woolley R. Bene®ts and risks of episiotomy: A review of the

English literature since 1980. Part I. Obstet Gynecol Surv

1995a;50:806±820.

2. Woolley R. Bene®ts and risks of episiotomy: A review of the

English literature since 1980. Part II. Obstet Gynecol Surv

1995b;50:821±835.

3. Stamp G, Kruzins G, Crowther C. Perineal massage in labour

and prevention of perineal trauma: Randomised control trial.

BMJ 2001;322:1277±1280.

4. Klein MC, Gauthier RJ, Robbins JM, et al. Relationship of

episiotomy to perineal trauma and morbidity, sexual dys-

function and pelvic ¯oor relaxation. Am J Obstet Gynecol

1994;71:591±598.

5. Carroli G, Belizan J. Episiotomy for vaginal birth (Cochrane

Review). In: The Cochrane Library. Issue 2. Oxford: Update

Software, 2001.

6. Renfrew MJ, Hannah W, Albers L, et al. Practices that

minimize trauma to the genital tract in childbirth: A

systematic review of the literature. Birth 1998;25:143±160.

7. Gupta JK, Nikodem VC. Women's position during second

stage of labour (Cochrane review). In: The Cochrane Library.

Issue 1. Oxford: Update Software, 2000.

8. Shorten A, Shorten B. Episiotomy in NSW Hospitals 1993-

1996: Towards understanding variations between public and

private hospitals. Aust Health Rev 1999;21:18±32.

9. Shorten A, Shorten B. Women's choice? The impact of

private health insurance on episiotomy rates in Australian

hospitals. Midwifery 2000;16:204±212.

10. Nodine P, Roberts J. Factors associated with perineal

outcome during childbirth. J Nurse Midwifery 1987;32:123±

130.

11. Low LK, Seng JS, Murtland TL, Oakley D. Clinician-speci®c

episiotomy rates: Impact on perineal outcomes. J Midwifery

Women Health 2000;45:87±93.

12. SPSS, Inc. SPSS 10.0 for Windows 1999. Chicago: Author,

1999.

13. Aldrich JH, Nelson FD. Linear Probability, Logit, and Probit

Models. Newbury Park: Sage Publications, 1984.

14. Murphy PA, Feinland JB. Perineal outcomes in a home birth

setting. Birth 1998;25:226±234.

15. McCandlish R, Bowler U, Asten H, et al. A randomised

controlled trial of care of the perineum during second stage

of normal labour. Br J Obstet Gynaecol 1998;105:1262±

1272.

16. Lydon-Rochelle M, Albers L, Teaf D. Perineal outcomes and

nurse-midwifery management. J Nurse Midwifery 1995;40:

13±18.


birth position, accoucheur, and perineal outcomes: informing women about choices for vaginal birth

Documents