British Journal of Anaesthesia 89 (3): 466-72 (2002) analgesia and backache: a randomized controlled comparison with intramuscular meperidine for analgesia during labour B. A. Loughnan *, F. Carli 2, M. Romney, C. J. Dore 3 and H. Gordon Departments of Anaesthesia and Obstetrics and Gynaecology, rthwick Park Hospital, Watford Road, Harrow, Middlesex HA 3UJ, UK 2 Department of Anesthesia, McGill University, 68 Pine Avenue West, Montreal, Canada H3A A. 3 Medical Research Council Clinical Trials Unit, 222 Euston Road, London NW 2DA, UK ^Corresponding author Background. Concern has been expressed that epidural analgesia for labour may be associated with a higher incidence of backache. Methods. A prospective randomized trial investigating the effect of epidural analgesia on the outcome of labour in nulliparae, mothers were randomized to receive either epidural analgesia or meperidine. A questionnaire on postnatal symptoms was sent to them 6 months after delivery. Results. n all, 6 mothers were studied; 30 were randomly allocated to receive i.m. meperidine up to 300 mg and 30 to receive epidural bupivacaine. The response rate to our questionnaire was 83%. ntention-to-treat analysis showed similar prevalence rates of postpartum backache in the epidural (48%) and meperidine groups (50%), with an observed difference (epidural-meperidine) of-2% (95% Cl, - to +6%). After excluding mothers with backache before delivery, there were also similar incidence rates of postpartum backache in the epidural (29%) and meperidine groups (28%), observed difference % (95% Cl, -8 to +0%). Conclusions. analgesia in labour was not associated with an increase in the prevalence or incidence of backache. BrJAnaesth 2002; 88: 466-72 Keywords: anaesthetic techniques, epidural; analgesia, obstetric; analgesics opioid, meperidine; complications, backache Accepted for publication: February 26, 2002 Mac Arthur and colleagues reported an association between coexistence of other symptoms in association with backthe development of new postpartum back pain and the ache, we also enquired about the presence of other administration of epidural analgesia in labour. However, symptoms, their study was retrospective and relied on the mothers' recollections of events. Back pain has since been noted to be a common occurrence in pregnancy. 2 A more recent study Methods suggested that epidural analgesia and the development of The study was approved by the local research ethics postpartum backache may be unrelated phenomena, 3 but committee and was designed as a part of a pragmatic there have been few large prospective randomized con- randomized controlled trial comparing the outcomes of trolled trials to date that have compared the prevalence of labour between mothers randomized to receive i.m. backache in mothers who have received epidural analgesia meperidine with those randomized to receive epidural with those who received meperidine for labour analgesia. analgesia for pain relief in labour. 5 Only nulliparous The present study is a randomized prospective compari- women with an open mind about their choice of analgesia son of the prevalence of backache in mothers who received were recruited to the study. They were approached by an epidural analgesia compared with mothers who had experienced midwife in the later stages of pregnancy and meperidine. Because other investigators 3 4 have noted the agreed to be randomized with respect to their first analgesia. The Board of Management and Trustees of the British Journal of Anaesthesia 2002
analgesia and backache Randomized to Requested further analgesia n=508 Randomized treatment Randomized to epidural n=249 Randomized to meperidine n=259 Analgesia received Received n=29 n=29 + epidural n=8 n=20 Refused n=30 Entonox n=2 n=0 Received n=94 + epidural n=74 n=63 Refused n=65 Entonox n=2 Backache n=08 n= n=3 n=5 n=6 n=4 n=2 n=0 n=52 n=68 n=44 n=30 n=33 n=30 n= n= Fig Flow diagram showing prevalence of back pain 6 months postpartum by analgesia received in labour by mothers who requested further analgesia and were randomized (n=6), returned the questionnaire (n=509) and answered the backache question («=508). On admission to the labour ward, the diagnosis of labour was confirmed. When requesting further analgesia, the envelope identifying their randomized treatment was opened and mothers were offered analgesia according to their randomization. They either received meperidine 00 mg by repeated i.m. injection (up to three doses), or a lumbar epidural was inserted and, after an initial dose of 0.25% bupivacaine 0 ml, an infusion of % bupivacaine was administered until the second stage of labour at the rate of 0-5 ml h" with top-ups of 0.25% bupivacaine 5 ml as required (this was the standard epidural analgesia administered on the labour ward at the time of the study commencement). For ethical reasons, the mothers were reassured at recruitment that they could opt out from the trial at any stage. Twenty-four hours after delivery, the mothers were visited on the ward and asked about back pain and headache before and during pregnancy. For example, 'Did you suffer from backache?' and if 'yes', 'Was this during pregnancy, before pregnancy, or both?'. Six months after delivery, a postal questionnaire was sent to the mothers. They were asked about a number of symptoms: frequent backache, frequent headaches or migraines, shoulder symptoms, neck symptoms, urinary incontinence and pain, tingling or weakness in the arms or legs. For each symptom they were asked whether it was present, when it had started and how it affected their daily functioning. For example, they were asked, 'Do you suffer from frequent backache?' and if 'yes', 'When did this first start? Was it before you became pregnant, during your pregnancy, or since you gave birth?'. The two randomized groups were compared using Mann-Whitney U tests and Fisher's exact tests where appropriate. Analysis was performed comparing responders and non-responders to the 6-month questionnaire. We compared the groups on the basis of their treatment and also compared characteristics of the women who did and did not have new symptoms at 6 months postpartum. Stepwise logistic regression was performed to investigate factors influencing the presence of each symptom, either as a persisting symptom or a new symptom 6 months postpartum. Possible predictor variables considered were randomized treatment, maternal age, height, weight, social class, marital status, ethnic group, induced labour, durations of first and second stages, oxytocin administration, mode of delivery, gestational age and foetal weight. This study was part of a trial investigating the effect of epidural analgesia on Caesarean section rate. 5 The prospective sample size of 68 for the trial had been calculated to detect a reduction in the Caesarean section rate from 7.8% in the epidural group (the observed rate in the unit in 988) to a rate of 3.8% in the non-epidural group. We calculated that this sample size would also allow us to detect a difference in rates of chronic backache such as that reported by MacArthur and colleagues for epidural (9%) and non-epidural (%) with a power >99% and using a significance level of 5%. Recruitment to the trial was slower than anticipated and, of the planned 68 mothers, 6 were randomized to receive either epidural analgesia (n=30) or i.m. meperidine (n=30) during labour. Results A total of 508 women returned the questionnaires 6 months after delivery, 249 of which had received the epidural and 259 had received meperidine. The overall response rate was 467
Loughnan et al. Table Obstetric and patient characteristics of responders and non-responders to 6 months postpartum questionnaire. Summary statistics are median (interquartile range (QR)) for continuous variables or frequencies (%) for categorical variables. P-values are from Mann-Whitney U test or Fisher's exact test Variable n-responders Responders f-value w/total n or median % or QR /(/total /( or median % or QR Randomized treatment Treatment received +epidural Neither Age (yr) Final weight (kg) Height (cm) Married Caucasian Social class // nduced First stage duration (h) Second stage duration (h) Syntocinon Mode of delivery rmal vaginal nstrumental Caesarean Gestational age (weeks) Foetal weight (kg) 5/02 7/02 50/02 34/02 /02 25 73 63 58/02 49/0 67/00 32/0 8.9 55/02 67/02 24/02 /02 3.3 50 7 49 33 22-29 65-8 58-67 57 49 67 32 6.2-.5-2.0 54 66 24 39-4 2.9-3.6 250/509 82/509 282/509 4/509 4/509 75 62 358/508 39/505 379/496 53/504 8.5.2 305/507 300/509 44/509 65/509 3.3 49 6 55 28 24-30 68-84 58-66 70 63 76 30 6.-.5-2.3 60 59 28 3 39-4 3.0-3.7 0.02 0.08 0.4 0.0 0.007 0.06 0.2 Table 2 Reported prevalence of backache 24 h and 6 months post delivery 24 h postpartum: 'Did you suffer from backache before delivery?' 6 months postpartum: 'Do you suffer from frequent backache?' f 'yes', 'When did it first start?' Before delivery Since delivery frequent backache Total Total 09 25 34 58 43 0 43 254 8 2 489 83%. There was no significant difference between the two randomized groups in terms of response rates, with 83 and 84% returned in the epidural and meperidine groups, respectively. Of the 249 women who had been randomized to the epidural group, 29 received their randomized treatment (88%), while of the 259 women randomized to the meperidine group, 94 received their randomized treatment (75%). An intention-to-treat approach was used for the analysis, comparing women on the basis of their randomized treatment allocation. Figure shows the prevalence of backache 6 months postpartum in the two randomized groups subdivided by the mode of analgesia actually received during labour, for mothers who requested further analgesia and were randomized (n=6), returned the questionnaire (n=509) and answered the backache question (n=508). Table compares the characteristics of the responders and non-responders to the 6-month postpartum questionnaire. There was no significant difference between patient characteristics of the groups in terms of treatment randomized, treatment received, labour characteristics or mode of delivery. However, married women were more likely to respond to the questionnaire (P= 0.0), as were older women (P=0.02) and Caucasians (P=0.007). There were no significant differences between the randomized groups in terms of these characteristics (results not shown). There were 489 women who completed questionnaires at both 24 h and 6 months postpartum. They were asked about backache on both occasions. Table 2 shows the responses at 24 h and 6 months postpartum. Many of the women gave inconsistent responses. There were 58 mothers who had stated 24 h postpartum that they had suffered from backache before delivery but stated at the 6-month follow-up that their backache had only commenced after delivery. There were also 25 mothers who said at 24 h that they had not had antepartum backache but at 6 months stated that their backache had commenced before delivery. 468
analgesia and backache Table 3 Prevalence of symptoms 6 months postpartum. Summary statistics are frequencies (%). P-values are from Fisher's exact test Symptom P-value /(/total n % /(/total n % Backache Headache Neck ache Urinary incontinence Arm symptoms Leg symptoms 9/249 56/248 60/247 46/246 36/244 5/243 48 23 24 9 5 2 30/259 68/256 70/256 5/258 49/255 46/246 50 20 9 9 0.2 Table 4 ncidence of new symptoms 6 months postpartum. For each symptom, the table shows its occurrence in patients not suffering the particular problem antepartum. Hence the variable numbers of subjects in the epidural and meperidine groups within the table. Summary statistics are frequencies (%). P-values are from Fisher's exact test Symptom P-value n/total n n/total n Backache Headache Neck ache Urinary incontinence Arm symptoms Leg symptoms 52/82 26/28 32/29 26/226 22/230 26/28 29 5 0 49/78 25/23 37/223 33/2 28/234 23/223 28 7 4 0.0 0.7 The prevalence of each postnatal symptom at 6 months in the two randomized groups is compared in Table 3. There was no significant difference between the two randomized groups in the frequency of any of the symptoms enquired about. There was also no significant difference in the prevalence of backache between the epidural and meperidine groups (48 vs 50%), with an observed difference in backache prevalence (-epidural-meperidine) of-2.4% (95% C,-. to+6.3%). n the stepwise logistic regression models to predict the presence of each symptom at 6 months, there was a positive relationship between the occurrence of backache and duration of the first stage (odds ratio (OR) for a h increase,.08; 95% C,.03-.4; P=0.002). With respect to headaches, there was a positive relationship with non- Caucasian ethnic groups (OR,.70; 95% C,.07-2.7; P=0.03) and a negative relationship with maternal age (OR, 4 for a yr increase in age; 95% C, 9-9; P=0.02). For urinary incontinence, there was a positive relationship with instrumental delivery (OR,.72; 95% C,.04-2.84; P=0.03). For pain, tingling or weakness in the arms there was a negative relationship with maternal height (OR, 4 for a cm increase in height; 95% C, 0-9; P=0.0), a positive relationship with length of first stage (OR,.07 for a h increase; 95% C,.0-.5; P=0.03) and a positive relationship with non-caucasian ethnic group (OR, 2.49; 95% C,.38-4.47; P=0.002). For pain, tingling or weakness in the legs there was also a positive relationship with non-caucasian ethnic group (OR, 2.82; 95% C,.70-4.69; P<0.00). Table 4 is a comparison between the two randomized analgesia groups after omitting women who stated 6 months postpartum that the symptoms began before or during pregnancy. There were no significant differences between the two randomized groups in the incidence of any new postpartum symptoms. The incidence of new backache in the epidural group was 29% and in the meperidine group 28%, with an observed difference in the incidence of backache (epidural-meperidine) of.0% (95% C, -8.2 to +%). Other new symptoms such as headache, neck ache, urinary incontinence, and arm or leg symptoms were reported by around % of each group. Stepwise logistic regression analysis performed to predict whether or not a woman had each new symptom showed relationships similar to those found when predicting the presence of any symptoms. There was a negative relationship between headache and maternal age (OR, 9 for a yr increase; 95% C, -7; P=0.007) and a positive relationship with being married (OR, 2.36; 95% C;.00-5.55; P=0.05). There was a positive relationship between urinary incontinence and oxytocin administration (OR, 2.36; 95% C,.7^.77; P=0.02) and a positive relationship with instrumental delivery (OR,.86; 95% C,.00-3.45; P=0.05). For arm symptoms, there was a negative relationship with maternal height (OR, 4; 95% C, 9-9; P=0.03) and a positive relationship with non- Caucasian ethnic group (OR, 3.85; 95% C,.84-8.07; P<0.00). For leg symptoms there was a positive relationship with non-caucasian ethnic group (OR, 2.24; 95% C,.8-4.; P=0.0). 469
Loughnan et al. Table 5 Obstetric and patient characteristics of those with new backache 6 months postpartum and those who had never had backache. Summary statistics are median (QR) for continuous variables or frequencies (%) for categorical variables. P-values are from Mann-Whitney U test or Fisher's exact test Variable backache Backache P-value /(/total n or median % or QR ///total n or median % or QR Randomized treatment Treatment received +epidural Neither Age (yr) Final weight (kg) Height (cm) Married Caucasian Social class // nduced First stage (h) Second stage (h) Syntocinon Mode of delivery rmal vaginal nstrumental Caesarean Gestational age (weeks) Foetal weight (kg) 30/259 35/259 4/259 82/259 /259 74 63 8/259 72/259 90/250 75/255 8.2 48/258 62/259 66/259 3/259 3.3 50 52/0 4 54 32 24-30 68-83 59-67 70 66 76 30 5.5-0.7-2. 57 63 25 39-4 3.0-3.7 20/0 62/0 8/0 /0 74 62 75/0 52/00 8/99 34/00 9.8. 66/0 54/0 3/0 6/0 3.2 5 20 6 8 24-30 66-84 57-66 74 52 82 34 7.5-.8-2.2 65 53 3 6 39^ 3.0-3.5 0.02.0 0.09 0.4 0.02 0.4 0.00.0 0.9 0. Table 5 compares the characteristics of women with and without new backache at 6 months postpartum, excluding those with backache before delivery. There were no significant differences between the groups in terms of randomized treatment allocated. However, there was a significant difference between groups in the type of analgesia actually received (P=0.02); women with new backache were more likely to have had epidural analgesia. The proportion of non-caucasian women was higher in those who developed new backache CP=0.02) and the duration of first stage was longer (P=0.00). Stepwise logistic regression was performed to predict whether a woman developed new backache. There was a positive relationship between new backache and non- Caucasian ethnic group (OR,.73; 95% C,.02-2.94; P=0.04), and a positive relationship with duration of first stage (OR,. for a h increase; 95% C,.04-.8; P=0.002), but no significant relationship with type of analgesia received. Discussion We have shown, like other investigators, that backache is an extremely common complaint postpartum and was present in 49% of mothers 6 months after delivery. When groups were compared on the basis of their randomized treatment allocation, there was no significant difference in the prevalence of backache between the epidural (48%) and meperidine (50%) groups. Our study was designed on a randomized prospective basis. n other non-randomized studies, the presence of backache or the choice of analgesia in labour may influence the mothers' reporting of back pain. Our comparison between the two analgesia groups was performed on an intention-to-treat basis as this is the least biased approach. 6 There was no significant difference between the groups when they were compared on this basis; the incidence of new backache in the epidural group was 29% and in the meperidine group 28%. When we compared those with new backache at 6 months with those who had never had backache (Table 5), the type of analgesia actually received was significantly different. administration was more common in those with new backache (82 out of 0 patients; 8%) than those without (76 out of 259; 68%). This is consistent with the finding of McQuay and colleagues 7 who suggest a small but slight increase in the development of new backache after epidural administration. However, this was based on a systematic review of results from non-randomized studies and would therefore incorporate any biases occurring in such studies. n our stepwise logistic regression model, the type of analgesia was no longer a significant predictor after including the duration of the first stage of labour and Caucasian ethnic group in the model. Our findings are similar to those of Ho well and colleagues. 8 n our study, the prevalence of backache in the meperidine group was 50% and in the epidural group was 48%. This differs considerably from the rate of new backache of % in the non-epidural group and rate of 9% in the epidural group in the retrospective study by MacArthur and colleagues, used in our prospective power 470
analgesia and backache calculation. When a post-hoc power calculation was performed for our study, it showed that we would have needed a sample size of 605 in each group to detect a 5% difference in the prevalence of backache (50 vs 45%) using the 5% significance level with a power of 80%. This is beyond the scope of most institutions and could only be performed in a large multi-centre trial. The response rate to our questionnaire was high (83%). This compares favourably with the % response rate of Mac Arthur and colleagues and the 63% response rate of Russell and colleagues 9 and is similar to the 88% response rate of Breen and colleagues. 4 Of some concern is the fact that the responders were more likely to be Caucasian and married than the non-responders were. Our study showed a positive relationship between non-caucasian status and the presence of back pain and other postnatal symptoms such as arm and leg symptoms. t is possible that this effect could be due to non-caucasian women with postnatal symptoms being more likely to respond than those without postnatal symptoms. However, in a study by Russell and colleagues, 9 European and West ndian women were more likely to respond than other groups. The prevalence of backache at 6 months in our study (49%) is similar to the prevalence of 44% found by Breen and colleagues 4 at 2 months postpartum, but considerably greater than the 30% found by Russell and colleagues 9 at months. The incidence of new backache at 6 months is somewhat less common (28%) in our study, while that of Russell and colleagues 9 was 5% at months. Russell and colleagues' study was retrospective but suggests that the prevalence of postpartum backache may be lower at than at 3 or 6 months after delivery. We chose a 6-month followup period as back pain is particularly relevant when the mother is weaning her child and possibly planning to return to employment. There was a higher proportion of non-caucasian women (35%) who developed new backache compared with Caucasian women (23%). n-caucasian women also developed other symptoms postpartum. Our local population, from which the study sample was taken, is comprised of 25% Gujerati Asians. Although theirs was a retrospective study, Mac Arthur and colleagues 0 found that backache, frequent headache, shoulder ache and pains and weakness in the arms and legs all occurred more commonly among Asian than among Caucasian women. Russell and colleagues 9 did not report the incidence of new backache with respect to other symptoms or to ethnic group, but did state that, of 36 women seen with new backache, 4 had possible psychological factors related to the backache. Our high prevalence of backache compared with other studies may be due to our study population. MacArthur and colleagues 0 postulate that dietary factors may play a role in the prevalence of postpartum problems among Asian women; the cause of backache in Asian women may be of dietary or sociological origin and warrants further investigation. The mothers gave inconsistent replies to the questionnaires 24 h and 6 months postpartum and this finding draws attention to possible problems associated with the interpretation of retrospectively collected data on postpartum backache. Russell and colleagues 9 found that memories even at 8 months postpartum were unreliable. This calls into question the findings of MacArthur and colleagues who found an association between epidural analgesia and backache but relied on the mothers' memories of events as long as yr previously. n a later prospective study MacArthur and colleagues found that although the incidence of low backache was significantly higher in the epidural group than in the non-epidural group on day, there was no significant difference at 7 days or 6 weeks. Our study, like that of Russell and colleagues 9, is restricted to nulliparous women. The studies of MacArthur and colleagues and Breen and colleagues 4 included multiparae. The disadvantage of including multiparae in a study on backache is that back pain may have occurred in relation to a previous pregnancy. n summary, we have shown that back pain is a common symptom 6 months postpartum and the incidence of new back pain was 28%. n a randomized prospective controlled trial, there was no significant difference in the prevalence between epidural and meperidine groups when an intention-to-treat analysis was used. Women can be reassured that the administration of an epidural per se is unlikely to cause backache. Acknowledgement The authors wish to acknowledge the financial support of the National Health Service Executive rth Thames (Previously the rth West Thames Regional Health Authority Locally Organised Research Scheme). References MacArthur C, Lewis M, Knox EG, Crawford JS. anaesthesia and long term backache after childbirth. Br Med J 990; 30: 9-2 MacEvilly M, Buggy D. Back pain and pregnancy: a review. Pain 996; 64:5-4 3 Russell R, Dundas R, Reynolds F. Longterm backache after childbirth: prospective search for causative factors. Br Med J 996; 3: 384-8 4 Breen TW, Ransil BJ, Groves PA, Oriol NE. Factors associated with back pain after childbirth. Anesthesiology 994; 8: 29-34 5 Loughnan BA, Carli F, Romney M, Dore CJ, Gordon H. Randomized controlled comparison of epidural bupivacaine versus pethidine for analgesia in labour. Br J Anaesth 2000; 84: 75-9 6 Newell DJ. ntention-to-treat analysis: implications for quantitative and qualitative research. nt J Epidemiol 992; 2: 837-4 7 McQuay H, Moore A. anaesthesia and low back pain after delivery. Br Med J 996; 3: 58 47
Loughnan et al. 8 Howell CJ, Kidd C, Roberts W, et al. A randomised controlled 0 MacArthur C, Lewis M, Knox EG. Comparison of long-term trial of epidural compared with non-epidural analgesia in labour. health problems following childbirth among Asian and Caucasian BrJ Obstet Gynaecol 200; 08: -33 mothers. 8r J Gen Pract 993; 43: 59-22 9 Russell R, Groves P, Taub N, O'Dowd J, Reynolds F. Assessing MacArthur A, MacArthur C, Weeks S. anaesthesia and long term backache after childbirth. Br Med J 993; 306: low back pain after delivery: a prospective cohort study. Br MedJ 99-303 995; 3: 336-9 472