British Journal of Obstetrics and Gynaecology January 2001, Vol. 108, pp. 34±40

The Ipswich childbirth study: one year follow up of alternative methods used in perineal repair Adrian Grant b,c, Beverley Gordon a,*, Carolyn Mackrodat a,e, Elizabeth Fern a, Ann Truesdale c,d, Sarah Ayers c Objective To assess the long term implications of four alternative approaches to postpartum perineal repair. Design A strati®ed randomised controlled trial using a 2x2 factorial design. Setting Original recruitment at the Maternity Unit at Ipswich Hospital NHS Trust, a district general hospital, between 1992 and 1994. Sample Seven hundred and ninety three women who had participated in the Ipswich Childbirth Study ± a trial among women who had required repair of episiotomy or ®rst or second degree tears ± at least one year previously. Methods Self-completed postal follow up at least one year after recruitment to trial comparing 1. two-stage repair leaving the skin unsutured with standard three-stage repair, and 2. polyglactin 910 with chromic catgut as suture material for the repair. Main outcome measures Failure to resume pain-free intercourse; persistent perineal pain; perineum feeling different; resuturing; time to resume pain-free intercourse; and dyspareunia. Results Fewer women allocated two-stage repair reported that the perineum felt different (30% versus 40%; RR 0.75; 95% CI 0.61 to 0.91; 2P , 0.01); otherwise there were no clear differences between the two methods. Women allocated polyglactin 910 were less likely to have dyspareunia (8% versus 13%; RR 0.59, 95% CI 0.39 to 0.91; 2P ˆ 0.02) and less likely to fail to resume pain-free intercourse (8% versus 14%; RR 0.57, 95% CI 0.38 to 0.87; 2P , 0.01). Conclusion Two-stage repair of perineal trauma leaving the skin unsutured appears to reduce the likelihood of the perineum feeling different from before delivery, in addition to less pain and dyspareunia initially; there were no apparent disadvantages. Polyglactin 910 reduces dyspareunia long term, indicating that the short term bene®ts of this material over chromic catgut persist.

INTRODUCTION Many women sustain perineal trauma during childbirth that requires surgical repair. For most, the associated pain and discomfort is temporary but in a minority it persists as chronic pain and discomfort 1. Two factors which may be linked with morbidity are the suturing technique and the choice of suturing material. The usual technique for repair is to close the perineal skin edges using interrupted or continuous subcuticular suturing after repair of the vagina and deeper perineal a

Ipswich Hospital NHS Trust, Ipswich, UK Health Services Research Unit, Aberdeen, UK c Perinatal Trials Service, National Perinatal Epidemiology Unit, Oxford, UK d Medical Statistics Unit, London School of Hygiene and Tropical Medicine, UK e Colchester General Hospital, Colchester, UK b

* Correspondence: Ms B. Gordon, Midwifery Of®ce, Ipswich Hospital NHS Trust, Heath Road, Ipswich, IP4 5PD UK.

q RCOG 2001 British Journal of Obstetrics and Gynaecology PII: S03 06-5456(00)0001 6-4

tissues. Because of the possibility that the skin sutures themselves may cause some of the morbidity experienced by women we evaluated a novel two-stage technique of repair that included leaving the perineal skin apposed, but unsutured. In a randomised comparison with the standard three-stage repair, which includes interrupted or subcuticular skin sutures, we found that the new technique appeared to reduce pain and dyspareunia, at least up to three months 2. There were no apparent disadvantages such as breakdown requiring resuturing. Nevertheless, we were concerned that these bene®ts might not be sustained and we therefore surveyed a sub-sample of participants 12 months after trial entry. In the original study we also took the opportunity to compare the two absorbable suture materials, polyglycolic acid and chromic catgut. The ®ndings in the ®rst three months that the use of polyglycolic acid sutures was associated with less early pain and fewer wounds requiring resuturing 3 were consistent with comparable trials 4. Longer term bene®ts were less certain, however, and so we also assessed these in the one-year follow up study reported here.

www.bjog-elsevier.com

FOLLOW UP OF IPSWICH CHILDBIRTH STUDY 35

METHODS The original trial was carried out at Ipswich Hospital NHS Trust from 1992 to 1994. In a ®rst phase, women who had sustained an episiotomy or laceration (®rst or second degree) during a normal spontaneous delivery and who had formally given consent were eligible. From 1993 the trial was extended to include women delivered by simple (non-rotational forceps or vacuum extraction) instrumental delivery. Before the study the standard technique used in Ipswich Hospital was a three-stage perineal repair. Most of these were performed by midwives; some operators used a continuous subcuticular suture for skin closure while most used interrupted sutures. During the trial, however, operators were encouraged to use continuous subcuticular suturing for skin closure in light of the available evidence 5. The three-stage approach was compared with a two-stage approach. In this, the vagina was repaired in the usual way. During repair of the perineal muscles, the aim was to leave the skin edges no more than half a centimetre apart when the woman was in the lithotomy position, with the intention of leaving the skin unsutured. Nevertheless, the repair actually performed was left to the operator's discretion; skin sutures could be used for women allocated two-stage repair if this was judged necessary. The study was therefore a pragmatic comparison of two suturing policies as they might be used in practice. Within a 2x2 factorial design, the study also compared the two suturing materials, polyglactin 910 (Vicryl, Ethicon Ltd, Edinburgh, UK) and chromic catgut. Instruction and training were given for the two experimental aspects ± two±stage repair and the use of polyglactin 910 ± in preparation for the trial. Randomisation occurred after completion of the third stage of labour. The randomisation schedule used

balanced blocks varying in size between four and twelve and was strati®ed by type of delivery. Serially numbered, sealed opaque envelopes containing the allocation details, suture material, and a data sheet to be completed by the operator immediately after the repair, were prepared and sealed in advance by an independent research assistant. Trial entry was signalled by opening the next envelope in sequence. All envelopes were accounted for. Outcomes at 24 to 48 hours, 10 days and three months have already been reported 2,3. We describe here a follow up to one year postpartum. Because this was not part of the original plan for the study, this was limited to women recruited after June 1993. It included all women who had an instrumental delivery together with women recruited after a spontaneous delivery over 15 months until August 1994. A total of 988 women were therefore eligible. No attempt was made to contact the 69 women who had not responded to the three-month questionnaire. A total of 919 women were therefore sent questionnaires at least 12 months after trial entry. The single page form asked about: persistent pain; whether the perineum felt different from before delivery; the need for resuturing; and return to, and pain during, sexual intercourse. A total of 793 forms were returned (86% of those sent a form and 80% of all eligible): 247 from women who had had an instrumental delivery and 546 from women who had had a spontaneous birth. Simple descriptive statistics were used to relate those in the one-year follow up with those in the original trial. The primary analyses for one-year comparisons was by intention-to-treat, and the statistical tests were two-sided; the Yates correction was used for x 2 statistics. Relative risks (RR) with 95% con®dence intervals were calculated where appropriate. Secondary analyses were strati®ed by the method of delivery, the operator's principal skin

Table 1. Description of groups at trial entry. Values are given as n (%) unless otherwise shown.

Maternal age: mean [SD] Previous vaginal delivery Previous perineal suturing Not known Mode of this delivery Spontaneous Instrumental Birthweight (g): mean [SD] Perineal injury Episiotomy Laceration 2 nd degree 1 st degree 3 rd degree Not known

Women in original study

Women followed up to one year

(n ˆ 1780)

(n ˆ 793)

2±stage (n ˆ 396)

3±stage (n ˆ 397)

Polyglactin 910 (n ˆ 395)

Chromic catgut (n ˆ 398)

28.3 [4.9] 693 (39) 643 (36) 5 (0)

28.9 [4.8] 282 (36) 264 (33) 3 (0)

29.1 [4.7] 140 (35) 134 (34) 1 (0)

28.6 [4.9] 142 (36) 130 (33) 2 (1)

28.5 [5.0] 137 (35) 131 (33) 0 (0)

29.2 [4.6] 145 (36) 133 (33) 3 (1)

1469 (83) 311 (17) 3505 [491]

546 (69) 247 (31) 3530 [508]

273 (69) 123 (31) 3556 [528]

273 (69) 124 (31) 3504 [487]

269 (68) 126 (32) 3556 [504]

277 (70) 121 (30) 3504 [512]

662 (37) 1082 (61) 26 (1) 6 (0) 4 (0)

359 (45) 415 (52) 14 (2) 3 (0) 2 (0)

172 (43) 216 (55) 5 (1) 3 (1) 0 (0)

187 (47) 199 (50) 9 (2) 0 (0) 2 (1)

189 (48) 198 (50) 5 (1) 2 (1) 1 (0)

170 (43) 217 (55) 9 (2) 1 (0) 1 (0)

q RCOG 2001 Br J Obstet Gynaecol 108, pp. 34±40

One-year methods comparison

One-year materials comparison

36 A. GRANT ET AL.

suturing technique (subcuticular, interrupted or mixed), suturing material (for the methods comparison), and method of repair (for the materials comparison). x 2 statistics were used to test for possible interactions. With about 400 women in each comparison, the follow up study had more than 80% power (2P , 0.05) to detect a difference of 6% in the rate of failure to resume pain-free intercourse (12% vs 6%). RESULTS The proportion of women who had instrumental deliveries was higher in the follow up than in the original study (31% vs 17%; Table 1) because the timing and arrangement of the follow up were designed to maximise information about perineal repair after instrumental delivery (these women were only eligible in the latter part of the trial, and follow up was extended amongst them). For this reason women followed up at one year, in comparison with the total trial group, were less likely to have had a spontaneous delivery (69% vs 83%), more likely to have had an episiotomy (45% vs 37%) (Table 1) and more likely to have been repaired by a doctor (38% vs 24%; Table 2). The randomised groups making up the two follow up trial comparisons were, however, balanced in these respects (Tables 1 and 2). Three women reported that they had had another baby since trial entry. Women completed the follow up questionnaires on average 400 days after trial entry and the randomised groups were similar (reported as mean (SD)) in this respect: two-stage 404(70) days; three-stage 396(65) days; polyglactin 910 400(69) days; chromic catgut 400(66) days. Table 3 summarises the ®ndings of the

one-year follow up, ®rstly for the comparison of the two methods of repair and secondly for the comparison of the two suture materials. In the comparison of two±stage with three±stage repair, 7% of women reported `pain or general discomfort from where they had been stitched when the baby was born', and this was similar in the two trial groups. Fewer women allocated the two-stage policy said that the area cut or torn during childbirth felt different (30% versus 40%; RR 0.75, 95%CI 0.61 to 0.91; 2P , 0.01). The descriptions that explained this difference were: `uneven', `smaller', and `more rigid'. Seven respondents de®nitely had resuturing: one allocated two-stage and six allocated three-stage repair. Five further women who only responded to the three-month questionnaire are also known to have been resutured (two versus three), giving totals of three allocated two-stage and nine allocated three-stage repair. The time to resumption of sexual intercourse, the number of women reporting dyspareunia at ®rst, and current dyspareunia were similar in the two groups; 10% versus 11% (RR ˆ 0.89; 95%CI 0.59 to 1.33) had failed to resume pain-free intercourse. The results of the secondary, strati®ed analyses are shown in Fig. 1. These were statistically consistent with the overall result for all women, and none of the tests for interaction were signi®cant. In the comparison of Polyglactin 910 with chromic catgut, there was no difference between the groups in the number of women who still had pain or general discomfort where they had been stitched. Fewer women allocated polyglactin 910 reported that the cut or torn area felt different (31% versus 38%; RR ˆ 0.83; 95%CI 0.69 to 1.01; 2P ˆ 0.07), the most common reason for the difference being a `ridge' or `scar tissue'.

Table 2. Management after trial entry. Values are given as n (%) unless otherwise shown.

Technique of skin closure Subcuticular Subcuticular and interrupted Interrupted Two-stage only None Material used Polyglactin 910 only Chromic catgut only Polyglactin and catgut None Status of operator Midwife SHO Registrar/consultant Student/other

Women in original study

Women followed up to one year

One-year methods comparison

One-year materials comparison

(n ˆ 1780)

(n ˆ 793)

2±stage (n ˆ 396)

3±stage (n ˆ 397)

Polyglactin 910 (n ˆ 395)

Chromic catgut (n ˆ 398)

247 (14) 5 (0) 731 (41) 784 (44) 13 (1)

144 (18) 2 (0) 286 (36) 357 (45) 4 (1)

9 (2) 1 (0) 33 (8) 351 (89) 2 (1)

135 (34) 1 (0) 253 (64) 6 (2) 2 (1)

83 (21) 1 (0) 133 (34) 177 (45) 1 (0)

61 (15) 1 (0) 153 (38) 180 (45) 3 (1)

881 (49) 888 (50) 10 (1) 1 (0)

392 (49) 396 (50) 5 (1) 0 (0)

195 (49) 199 (50) 2 (1)

197 (50) 197 (50) 3 (1)

390 (99) 3 (1) 2 (1)

2 (1) 393 (99) 3 (1)

1329 (75) 76 (4) 361 (20) 14 (1)

484 (61) 50 (6) 255 (32) 4 (1)

238 (60) 26 (7) 129 (33) 3 (1)

246 (62) 24 (6) 126 (32) 1 (1)

236 (60) 23 (6) 134 (34) 2 (1)

248 (62) 27 (7) 121 (30) 2 (1)

q RCOG 2001 Br J Obstet Gynaecol 108, pp. 34±40

FOLLOW UP OF IPSWICH CHILDBIRTH STUDY 37 Table 3. Follow up at least 12 months after entry to Ipswich childbirth study. Values are given as n (%). Methods comparison 2-stage (n ˆ 396) Still pain or general discomfort where stitched? None Mild Moderate Severity not stated Not known Area cut or torn feels different Not known Resutured Not known Timing of resumption of sexual intercourse Tried, but too painful By 6 months By 3 months Could not remember when No partner Not known Dyspareunia at ®rst, if resumed Dyspareunia now, if resumed Failure to resume pain-free intercourse a b c d e

x y2 x y2 x y2 x y2 x y2

ˆ ˆ ˆ ˆ ˆ

Materials comparison

3-stage (n ˆ 397)

Polyglactin 910 (n ˆ 395)

Chromic catgut (n ˆ 398)

368 18 9 1 0 117 a 1 1 2

(93) (5) (2) (0) (0) (30) (0) (0) (1)

371 22 3 0 1 157 a 1 6 0

(93) (6) (1) (0) (0) (40) (0) (2) (0)

370 18 7 0 0 124 b 1 2 2

(94) (5) (2) (0) (0) (31) (0) (1) (1)

369 22 5 2 1 150 b 1 5 0

(93) (6) (1) (1) (0) (38) (0) (1) (0)

1 56 328 6 4 1 142 39 40

(0) (14) (83) (2) (1) (0) (36) (10) (10)

3 41 338 9 4 2 148 42 45

(1) (10) (85) (2) (1) (1) (37) (11) (11)

1 50 334 7 3 0 126 c 30 d 31 e

(0) (13) (85) (2) (1) (0) (32) (8) (8)

3 47 332 8 5 3 164 c 51 d 54 e

(1) (12) (83) (2) (1) (1) (41) (13) (14)

8.33 (1df) 2P , 0.01. 3.20 (1df) 2P ˆ 0.07. 7.01 (1df) 2P , 0.01. 5.33 (1df) 2P ˆ 0.02. 6.19 (1df) 2P ˆ 0.01.

Resuturing was less common after repair with polyglactin 910. The numbers were: two versus ®ve amongst respondents at 12 months and none versus ®ve amongst non respondents, giving totals of two versus ten. There was no clear difference in time to resumption of intercourse. Women allocated polyglactin 910, however, were significantly less likely to report dyspareunia at ®rst (32% versus 41%; RR ˆ 0.77; 95%CI 0.64 to 0.93; 2P , 0.01), and any dyspareunia at the time of follow up (8% versus 13%; RR ˆ 0.59; 95%CI 0.39 to 0.91; 2P ˆ 0.02). Signi®cantly fewer women in this group failed to resume pain-free intercourse by one year (8% versus 14%; RR ˆ 0.57; 95%CI 0.38 to 0.87; 2P ˆ 0.01). The results of the strati®ed analyses are shown in Fig. 2. Again, none of the tests for interaction were signi®cant. DISCUSSION We decided to extend follow up to one year when it became obvious that some women were still having signi®cant problems with perineal healing at and beyond three months. We also particularly wanted to assess long term outcome after instrumental delivery. These were the reasons why the follow up only recruited women in the later period of the trial. Because of local logistics, however, it was not possible to do this for women q RCOG 2001 Br J Obstet Gynaecol 108, pp. 34±40

who had been randomised in the last two months of recruitment other than for those who had an instrumental delivery. The eligibility for the trial had been extended to include instrumental deliveries only after the trial had become established and this is the other reason why there is a relatively large proportion of women delivered in this way in this follow up study (Table 1). The trial groups compared were, however, well matched in this and other respects. Response rates were satisfactory (80% of those eligible) and there was no suggestion of any biased response within the randomised comparisons. The apparent bene®ts of the two-stage technique in terms of pain and dyspareunia three months postpartum were not seen at longer term follow up. Women in the two-stage group, however, were less likely to report that the perineum felt different from before the delivery, with fewer reports of unevenness and constriction or rigidity. Although this difference was not observed among women whose repair was performed by an operator who used the subcuticular method (Fig. 1), the test for interaction was non-signi®cant, and the data are statistically compatible with the overall result. There was no clear difference between the two policies in other outcomes, particularly in respect of the timing of resumption of intercourse and dyspareunia. Follow up to three months in the comparison of the suture materials identi®ed bene®ts favouring polyglactin 910 in terms of short term pain and use of analgesia, with

38 A. GRANT ET AL.

Fig. 1. Strati®ed analyses for two-stage versus three-stage.

q RCOG 2001 Br J Obstet Gynaecol 108, pp. 34±40

FOLLOW UP OF IPSWICH CHILDBIRTH STUDY 39

Fig. 2. Strati®ed analyses for polyglactin 910 versus chromic catgut.

less clear differences at three months. However, the data now available describing longer term follow up show sustained important bene®ts with polyglactin 910. In particular, there were fewer reports of dyspareunia both q RCOG 2001 Br J Obstet Gynaecol 108, pp. 34±40

initially and at the time of follow up in the polyglactin 910 group, and a strong suggestion of less risk of resuturing. An estimated 6% fewer women in this group had failed to resume pain-free intercourse. This difference

40 A. GRANT ET AL.

was not seen in the sub-group repaired by an operator who used the subcuticular method. However, this ®nding was based on only eight women who reported this outcome, the data are compatible with the overall result (the con®dence interval overlaps the overall interval), and the test for interaction is not signi®cant. CONCLUSIONS On the basis of these new data, there still appear to be advantages of using a two-stage repair leaving the skin unsutured. There are also still no apparent disadvantages, particularly in respect of any increased risk of breakdown of the repair and need for resuturing. We recognise that the use of polyglactin 910 is associated with an increased risk of the need for material to be removed postnatally. Nevertheless, the bene®ts of this material, con®rmed and strengthened by this follow up, reinforce our view that polyglycolic acid based materials should replace chromic catgut in the repair of perineal trauma. We feel that there is now no need for further randomised comparisons of polyglactin 910 with chromic catgut. However, we do recognise that some people may wish to replicate our evaluation of the two-stage method of repair before its widespread implementation. Our trial may also raise questions about the need for any suturing of uncomplicated lacerations, and we think that a policy of leaving these unsutured should now be formally evaluated. Acknowledgements The authors would thank all the women who took part in

this study, as well as the midwives and obstetricians employed by the Ipswich Hospital NHS Trust, particularly: C. Colbourne, N. Cole. They would also thank: I. Chalmers, P. Brocklehurst, D. Elbourne, R. McCandlish, G. Chamberlain, J. Sleep, J. Gordon, K. Gallagher, L. Mallia, L. Nurse, C. Busby and K. McIntosh. The National Birthday Trust and East Anglia Region Locally Organised Research Scheme provided funding and Ethicon Ltd supplied much of the suture material used in the study. A. Grant receives salary support from the Chief Scientist Of®ce of the Scottish Executive Health Department, and the NPEU is supported by the Department of Health. The views expressed are those of the authors and not necessarily shared by the agencies that ®nancially supported the trial.

References 1. Grant A. Repair of perineal trauma after childbirth. In: Chalmers I, Enkin MW, Keirse MJNC, editors. Effective Care in Pregnancy and Childbirth, Oxford: Oxford University Press, 1989:1170±1181. 2. Gordon B, Mackrodt C, Fern E, Ayers S, Truesdale A, Grant A. The Ipswich Childbirth Study: (1) a randomised evaluation of two-stage postpartum perineal repair leaving the skin unsutured. Br J Obstet Gynaecol 1998;105:435±440. 3. Mackrodt C, Gordon B, Fern E, Ayers S, Truesdale A, Grant A. The Ipswich Childbirth Study: (2) a randomised comparison of polyglactin 910 (Vicryl) with chromic catgut for postpartum perineal repair. Br J Obstet Gynaecol 1998;105:441±445. 4. Kettle C, Johanson RB. Absorbable synthetic versus catgut suture material for perineal repair. The Cochrane Library, Issue 1. Oxford: Update Software, 2000. 5. Kettle C, Johanson RB. Continuous versus interrupted sutures for perineal repair. The Cochrane Library, Issue 1. Oxford: Update Software, 2000. Accepted 30 August 2000

q RCOG 2001 Br J Obstet Gynaecol 108, pp. 34±40