Saturday 31 August
No 8766
1991
ORIGINAL ARTICLES
Randomised trial of laxatives in treatment of childhood encopresis
Primary faecal incontinence (encopresis) in children is usually treated with laxative medication and a modification
programme aimed at promoting regular toileting, but the effectiveness of laxatives has never been adequately investigated. 169 children with encopresis and evidence of stool on plain abdominal radiograph were randomly allocated to receive multimodal (MM) therapy (laxatives plus behaviour modification; n=83) or behaviour modification alone (BM; n=86). Mean (SD) follow-up was 55·1 (27·0) weeks and 56·7 (32·0) weeks, respectively. By 12 months follow-up 42 (51%) of the MM group and 31 (36%) of the BM group (p=0·079) had achieved remission (at least one 4 week period with no soiling episodes) and 52 (63%) vs 37 (43%) (p=0·016) had achieved at least partial remission (soiling no more than once a week). MM subjects achieved remission significantly sooner than BM subjects, and the difference in the Kaplan-Meier remission curves was most striking in the first 30 weeks of follow-up (p=0·012). The patterns of compliance with toileting in the treatment groups were almost identical, although about 1 in 8 children overall did not comply with the sitting programme. After exclusion of the 24 poor compliers, there was no significant difference between BM and MM groups. This study shows a clear advantage overall for the use of laxative medication, although the benefit may not be as great for children who are able to maintain regular
behaviour
toileting.
Introduction Persistent faecal incontinence without associated anatomical abnormality (encopresis) is a common chronic and disabling disorder in childhood. Bellman1 defmed encopresis as repeated, involuntary defaecation into clothing, occurring in children over 4 years of age, for a period of at least 1 month. The reported prevalence among children of 7-8 years is 2-3% for boys and 1-3% for girls1 and among 10-12-year-olds 1-3% and 0-3%, respectively.2 The management of encopresis has developed without adequate controlled evaluation; it is usually based on laxative treatment with a varying degree of behaviour modification aimed at promoting regular toileting.3-5 There have been just two comparative studies of the additional benefit conferred by laxatives, and both were inconclusive. One was a randomised controlled trial6 and the other a nonrandomised retrospective comparison.7 Both had such small sample sizes that quite a large true difference in efficacy between the treatment groups could easily have gone undetected by chance. On the other hand, it has been claimed that no laxative therapy is required to treat encopresis successfully.s There have been three reports of behaviour modification alone 6,7resulting in remission rates of 70%, 40%, and 38%, respectively. We have tested the hypothesis that multimodal therapy (laxative treatment together with behaviour modification) for encopresis is superior to behaviour modification alone in terms of both remission and relapse rates. ADDRESSES University of Melbourne Department of Paediatrics (T. Nolan, FRACP, C. Coffey, BSc) and Department of Ambulatory Paediatrics (G. Debelle, FRACP, F. Oberklaid, FRACP), Australia. Children’s Melbourne, Hospital, Royal Correspondence to Dr T. Nolan, Department of Paediatrics, Royal Children’s Hospital, Parkville 3052, Australia.
524
Patients and methods A two-group, parallel, randomised trial was carried out in subjects referred to a clinic specifically for the treatment of children with faecal incontinence. The primary study endpoint was time to remission from encopresis. The study was approved by the Royal Children’s Hospital ethics in human research committee. Randomisation took place only after informed written consent was obtained from parents or guardians of eligible patients. Subject eligibility criteria were: meeting the definition of encopresis; evidence of stool on plain abdominal radiograph; the absence of very severe and/or prolonged constipation necessitating previous hospital admissions for enemas and other treatments; age 4-16 years; attendance at normal school; absence of neuromuscular disorder (eg, spina bifida, cerebral palsy, muscular dystrophy); absence of Hirschprung’s disease (based on barium enema, rectal biopsy if clinically indicated); and no use of purgatives for at least 2
TABLE I-BASELINE CHARACTERISTICS OF STUDY GROUPS
weeks before baseline assessment. 222 new patients attended the clinic between July, 1987, and July, 1989. Of these, 181 were eligible. The reasons for exclusion were no
encopresis (17), severe constipation requiring disimpaction (6), already receiving purgative medication (3), too young (3), moving interstate or living beyond 100 km from the hospital (5), not attending normal school (4), and need for immediate psychiatric referral (3). The parents of a further 12 eligible subjects refused or were unable to provide written informed consent. The remaining 169 eligible children were randomly allocated to the multimodal therapy group (MM), or the behaviour modification only group (BM). By the logrank test and Freedman’s tabulated sample size estimates a total sample size of 155 would be adequate to show 50% superiority (60% vs 40%) of multimodal therapy over behaviour modification alone at the one-tailed alpha level of 0-05 with 80% certainty. Referrals from general practitioners accounted for 44% of randomised children, 35% were referred by school doctors or nurses, and only 8% were referred by consultant paediatricians. Randomisation was carried out by means of a stratified (primary vs secondary encopresis), blocked schedule (variable block size of 4, 6, and 8) by an individual not connected with the study. Each treatment
allocation
was
recorded
on
a
card in
an
opaque,
numbered, and sealed envelope and stored sequentially in four separate boxes corresponding to the assignment strata. A child was classified as having primary encopresis when he or she had never achieved faecal continence for longer than 1 month, and secondary encopresis if faecal continence had ever been achieved for longer than 1 month. Remission was defmed as at least one continuous 4-week period with no soiling episode and partial remission as soiling episodes occurring no more often than once per week. Relapse occurred when there was resumption of soiling more often than twice per month after a period of full remission. Treatment failure was defmed as no substantial progress towards remission by at least 6 weeks from study entry (BM group); it occurred in 32 cases (37%). These subjects were deemed to have reached a study endpoint (and were censored) at the time treatment failure was declared. 3 patients started laxative medication before 6 weeks-1 protocol violation (4 weeks), 1 child with severe symptomatic constipation whose parents refused to proceed (4 weeks), and 1 patient with severe emotional problems in whom no clinically satisfactory improvement occurred (5 weeks). Both groups received a standard paediatric behaviour modification intervention.3-S This consisted firstly of clarification during a joint parent-child interview of the postulated underlying physiological basis for encopresis. The bowel training programme used positive reinforcement for successful defaecation in the toilet, and additional reinforcement for each 24 h without soiling. Reinforcement consisted of parental praise, and the use of a star-chart diary (fitness training card) to indicate soiling-free days. A regular sitting programme of 5-10 min toilet-time within 30 min of each meal was the basis of the programme. Dietary advice, general counselling, and support were provided by the paediatrician. Psychiatric assessment or treatment was initiated when indicated clinically. In addition, the protocol for the MM group used laxative therapy in two phases. The initial disimpaction phase consisted of 3-day
tConstlpated *Green
on
abdominal
radiograph.
score.
cycles of 5 ml ’Microlax’ enemas (Pharmacia: sodium citrate 90 mg, sodium lauryl sulphoacetate 9 mg, sorbic acid 5 mg, glycerol, sorbitol, distilled water) on day 1, one 5 mg bisacodyl suppository after school and one in the evening on day 2, and a 5 mg bisacodyl tablet after school and one in the evening on day 3. Up to four cycles (12 days) were undertaken. Further cycles were prescribed if there was later evidence of stool reaccumulation. The subsequent maintenance phase consisted of’Agarol’ (Knoll AG: liquid paraffin, phenolphthalein, benzoic acid, sorbic acid) 5-30 ml once or twice each day, senna granules, and/or bisacodyl tablets. Doses were adjusted to maintain at least daily defaecation and were increased if there was persistent or recurrent stool retention (large-calibre stools, less frequent defaecation, abdominal pain, or reappearance of soiling). Constipation was assessed with the stool retention rating record" based on a plain abdominal radiograph at baseline and 2 weeks after randomisation. A score of 10 or more classified the patient as constipated. Radiographs were scored blindly by an experienced paediatric radiologist not connected with the study. Intraobserver repeatability was high (random sample of n 20, intraclass r 0-93, The parent form of the child behaviour checklist p < 0001). (CBCL)12 was used as a measure of behaviour disorder at baseline and at 12 months. A summary behaviour problems standardised T-score of more than 63 classified a child as maladjusted.12 Items 6 (bowel movements outside toilet), 49 (constipated, doesn’t move bowel), 56f (stomachaches or cramps), 107 (wets self during the =
=
525
Fig 2-Kaplan-Meier curves for remission from incontinence).
treatment groups
(time
to
analysis by the Kaplan-Meier product limit estimator (with logrank and Wilcoxon tests), and a Cox proportional hazards model were used with remission and relapse as failure-times. Zero time was the time of random assignment for remission, and the time of remission for relapse. The proportionality assumption was tested with an interaction between the outcome group and log time. Comparison of proportions of subjects in remission and partial remission was carried out by means of the X2 test for 3, 6, and 12 months of follow-up by intention to treat. t tests and nonparametric tests were used as appropriate. Scores from the behaviour measures were assessed by analysis of covariance with the gain score (score at baseline minus score at yearly review) as the dependent variable, with adjustment for pretreatment score and other covariates.14 Although sample size estimates were based on a one-tailed fx, all tests and confidence intervals are reported as Survival
Fig 1-Patient follow-up and study endpoints. *Of the 32 BM subjects who reached an end-point, 17 later achieved remission
with the
help of laxatives.
two-tailed. 108 (wets the bed) on the CBCL behaviour problems scale relate to behaviours or symptoms which could be directly related to soiling. Therefore, the CBCL was also scored with these items set to 0 to avoid confounding of symptomatic improvement directly attributable to resolution of soiling with behaviour changes (QCBCL score). The socioeconomic status measure was the two-factor Green score which is a weighted aggregate of scores for years of maternal education and occupation of the head of the household.13 The fitness training card was used to monitor clinical progress, compliance with the bowel training programme (excellent = 2-3 sits/day; good = 1-2 sits/day; poor = less than 1 sit/day), toilet defaecation frequency, medication compliance (excellent = more than 75% taken; good = 25-75 %; poor=less than 25%), and adverse effects of laxatives. Clinic follow-up was usually done at 2, 6, 14, 30, and 52 weeks and thereafter as required. There was, however, variation in the frequency and subsequent duration of follow-up depending on clinical progress.
day),
TABLE II-PRESCRIBED TREATMENT IN MM GROUP
Results The baseline characteristics of the 169 randomised patients are given in table I; more girls, more older children, and more children who had had previous laxative treatment and toileting programmes were allocated to the MM group. In the BM group, twice as many children were passing stool into the toilet at least daily, and there were more who had problems with urinary continence. There were similarly high rates of faecal retention (on the stool retention record) and maladjustment (on the CBCL). For other baseline
features, randomisation produced equivalent distributions. The average duration of follow-up was 55(SD 27-0) weeks for the MM group, and 56-7 (32-0) weeks for the BM group. Over this time the mean numbers of clinic attendances were 7-2 (35) and 7-1 (3-3), respectively (p=073). All but 7 patients were reviewed after about 12 months’ follow-up. Over the first 12 months of follow-up, 4 patients in the BM group, and 3 patients in the MM group
lost to follow-up (fig 1). The times at which BM patients reached study endpoints are also shown in fig 1. Patients who were discharged in remission after 12 months’ follow-up were assumed to have remained in remission. Details of prescribed treatment in the MM group are given in table II. 8 children (10%) were not prescribed any rectally administered medication. 79% received 9 or more days of rectally administered disimpaction therapy. Over subsequent months, 25% had additional disimpaction cycles. In the maintenance phase, the median duration of laxative use was 6 months, with more than 33% being prescribed laxative medication for more than 9 months. Only 3 patients were admitted to hospital for disimpaction. 29 families were referred for psychiatric evaluation and were
*Median duration of laxative use=6 6 mo.
526
TABLE m—REM!SStON AND IMPROVEMENT RATES
By 6 months, the MM group remission rate was 49%, and was substantial improvement in a further 11 % (table Both remission and improvement rates were only III). slightly higher by 1 year. BM group rates were significantly lower than MM group rates at 3 and 6 months, and the improvement rate was significantly lower at 12 months. The remission rates at 12 months were 51 % and 36% (p = 0-079, 95% CI on difference, upper limit 29%). Soiling frequency was higher in the BM than in the MM group at 3 and 6 months (rank sum test p 0- 11 and 0-004), but at 12 months the BM subjects who had not reached a study endpoint had significantly fewer soiling episodes than subjects in the MM group (p < 0-001). Of the 32 BM patients who reached a study endpoint, 17 (53%) later achieved remission with the help of laxative medication. Remission was achieved within 6 months by 12 (38%) patients, and by 12 months in 16 (50%). The influence of compliance with the toileting programme on remission is shown in fig 3. The BM children who complied poorly (n = 11) did significantly worse than all others overall (Wilcoxon test p=0016, logrank test p=0-06). After removal of the poor compliers in both treatment groups from the data set (n=24), proportional hazards analysis was repeated for phases I and II of follow-up with the same covariates as in the principal evaluation. There was no significant difference between BM and MM groups in phase I (hazard ratio 1 -49,95% CI 0-90, 2-47; p=012) or in phase II (hazard ratio 0-73, 95% CI 0-32,1-70; p =046). Children with secondary encopresis who received MM therapy did significantly better (Wilcoxon test p=0005, logrank test p 0-01) than all other groups. There were no significant interactions between treatment group and sex, maladjustment (on the CBCL), extent of previous treatment, coexistence of enuresis (noctural or diurnal), socioeconomic status, presence of single parent family, or constipation at baseline assessment. There was no significant difference in outcome for patients by their treating clinicians. More MM than BM subjects experienced at least one relapse (33/55 [51%] vs 17/46 [37%]), but the 14% difference was not significant (p=021, 95% CI for difference - 5%, 33%). The proportional hazards analysis resulted in a hazard ratio of 1-56 (95% CI 0-85, 2-86; p 0-16) for MM compared with BM subjects. Only 1 BM subject had more than one relapse, compared with 6 MM subjects. Of MM children who accepted medication, 28 (35%) reported at least one episode of transient abdominal pain (61%), watery stool (54%), nausea (7%), or rectal pain (7%). No subject had symptoms severe enough to require hospital admission. At 12 months’ follow-up, the adjusted mean improvement in QCBCL summary behaviour problem T-scores for the 76 available MM group subjects was 3-10 (p=0005), compared with 1-81 (p=0-21) for the 43 BM subjects still in the study. However, the difference in these adjusted gain scores was not significant (p = 0-48). Of the 17 BM and 17 MM subjects with nocturnal enuresis at baseline for whom follow-up data were available, remission from enuresis was recorded in 7 (41%) and 5 (29%), respectively. 7 children in each group had diurnal enuresis, and 4 in each group recovered during follow-up.
there
=
*p < 0 05, tp < 0 005.
co-management
(BM
group 17
[20%],
MM group 12
[15%]; p=036). The patterns of compliance with toileting were similar in the BM and MM groups (excellent in 50 [60%] and 46 [58%]; good in 22 [27%] and 21 [26%]; poor in 11 [13%] and 13 [16%], respectively). 1 in 8 children overall did not comply with the sitting programme. Compliance with prescribed medication in the MM group was excellent in 51 (65%) and good in 22 (28%); only 6 (8%) were poorly
compliant. Non-compliant subjects were significantly more likely to be the children of single parents (38% vs 14%; p=0006) and to be maladjusted on CBCL summary behaviour problem T-scores (71% vs 38%; p=0-003). There was no significant difference in sex, age, difficulty with initial toilet training, socioeconomic status, or presence of stool retention at baseline between compliant and non-compliant children. Reduction in stool retention was measured over the first 2 weeks of treatment by change in the radiograph stool retention rating scores. There was a mean decrease of 1-84 (SD 6-08) in BM group subjects (p 0-012 by signrank test) and of 3-17 (6-31) in the MM group (p =0-0001). The =
between-group difference, however, significant (p=015).
was not
statistically
The MM group came into remission significantly sooner than the BM group (fig 2), and the difference in the remission curves is most striking in the first 30 weeks of follow-up (logrank test p=0-12, Wilcoxon test p=0-012). The proportionality assumption was not satisfied for the full follow-up period, and separate hazard functions were modelled for follow-up before and after 26 weeks (phase I and phase II, respectively). After adjustment for age, sex, stratification group, and socioeconomic status, the phase I hazard ratio was 1-91 (MM vs BM, 95% confidence interval [CI] 1-18, 3-09; p = 0-008). However, the hazard ratio was not significant for phase II (102, 95% CI 046, 2-26;
p=095).
Fig 3-Kaplan-Meier curves showing effect of compliance on remission.
treatment
=
=
8 children in the BM group and 4 in the MM group showed development of enuresis.
527
Discussion These results clearly show that the acquisition or resumption of faecal continence is promoted by the addition of laxative therapy to the structured toileting programme, and that successful treatment is accompanied by a significant reduction in emotional maladjustment. This finding contrasts with that of the only other published randomised evaluation of laxative therapy6 but the negative outcome reported in that study can be attributed to several methodological shortcomings. The randomised comparison applied only to 3 months of laxative therapy, no rectal medication was used, and the sample size was so small that the observed lack of difference in outcomes could easily have occurred by chance. In our study, most of the patients had had substantial previous treatment, which usually consisted of oral laxative therapy without any serious attempt to modify toileting behaviour. Nevertheless, the study group represented a spectrum of severity and history of incontinence. There is no reason to believe that the results cannot be generalised to the usual paediatric patient population in community or
hospital practice. Randomisation
produced treatment groups which were largely comparable, with the possible important exception of the amount of previous laxative therapy, and the extent to which the child regularly defaecated into the toilet. In both cases, this difference in distribution would probably have made remission less likely in the MM group. Although laxative medication improved outcomes overall, the benefit was slight for children who were able to comply with an effective toileting programme. There were nearly identical patterns of compliance with toileting in both treatment groups, although about 1 in 8 children overall was poorly compliant. Behavioural and family factors identified children who were likely to be non-compliant. However, we believe that these factors should be used only as a guide rather than as criteria on which to judge the need for laxative medication. Children with secondary encopresis who were treated with MM therapy did significantly better than all other groups. This is not a surprising result; secondary encopresis is likely to be related to simple faecal accumulation, whereas in primary encopresis more complex pathophysiological and behavioural factors may operate. We considered the use of laxative placebo for this study but thought that rectal administration of placebo was unlikely to be acceptable to either ethics committees or parents, and there is the possibility of a therapeutic effect of even rectally administered water. It seems unlikely that a placebo effect could account for the observed difference in outcomes.
The use of study endpoints to provide an escape for those BM subjects who made inadequate clinical progress was a necessary prerequisite to mounting this trial. The study clinicians adhered well to the criteria for BM treatment failure. It seems reasonable to expect that prolonged continuation of therapy without laxatives would not have produced outcomes superior to the MM group. The rate of remission in BM patients who were switched to MM therapy after leaving the study was very similar to that in the MM group. The mechanism of action of laxatives in faecal incontinence is not completely understood and is probably not simply related to the debulking effect on chronically retained stool. It is possible that the stimulation of colonic
smooth muscle not only increases colonic motility itself,15 but also increases awareness of motility and rectal filling. The remission and improvement rates achieved by MM-treated patients in this study are similar to those reported in previous follow-up studies.3.4,16 Although such studies have repeatedly shown that good outcomes are achievable, the same studies have also shown that up to a third of patients remain troubled by at least some degree of incontinence. The usefulness of biofeedback training for these treatment-resistant patients also remains to be properly assessed. 17-19 There has been no systematic attempt to develop primary prevention strategies. Until more effective treatments are found, we believe that the use of MM therapy remains the best overall approach. However, since compliant children may in some cases do just as well without laxatives, a preliminary trial of behaviour modification may occasionally obviate the need for any medication. We thank Dr Jill Sewell, Dr Alf Nicholson, Mr Ted Byrt, Ms Janet Bishop, Ms Dorothy Jones, and Ms Judy Wright. This study was funded by the National Health and Medical Research Council, and by the Royal Children’s Hospital Research Foundation.
REFERENCES 1. Bellman M. Studies (suppl): 1-154.
on
encopresis. Acta
Paediatr Scand 1966; 170
2. Rutter M, Tizard J, Whitmore K. Education, health and behaviour. Psychological and medical study of childhood development. New York: Wiley, 1970: 219-20. 3. Levine MD, Bakow H. Children with encopresis. A study of treatment outcome. Pediatrics 1976; 845: 852. 4. Lowery SP, Srour JW, Whitehead WE, Schuster MS. Habit training as treatment of encopresis secondary to chronic constipation. J Pediatr Gastroenterol Nutr 1985; 4: 397-401. 5. Wright L. Outcome of a standardized program for treating psychogenic encopresis. Prof Psychol 1975; 6: 453-56. 6. Berg I, Forsythe I, Holt P, Watts J. A controlled trial of ’Senokot’ in faecal soiling treated by behavioural methods. J Child Psychol Psychiatry 1983; 24: 543-49. 7. Halpern WI. The treatment of encopretic children. Am Acad Child Psychiatry 1977; 16: 478-99. 8. Jolley H. A paediatrician’s view on the management of encopresis. Proc R Soc Med 1976; 69: 21-22. 9. Pinkerton P. Psychogenic megacolon in children: the implications of bowel negativism. Arch Dis Child 1958; 33: 371-80. 10. Freedman LS. Tables of the number of patients required in clinical trials using the logrank test. Stats Med 1982; 1: 121-29. 11. Barr RG, Levine MD, Wilkinson RH, Mulvihill D. Chronic and occult stool retention. A clinical tool for its evaluation in school-aged children. Clin Pediatr 1979; 18: 674-86. 12. Achenbach TM, Edelbrock E. Manual for the child behaviour checklist and revised child behaviour profile. Burlington, Vermont: University of Vermont, 1983. 13. Green LW. Manual for scoring socioeconomic status for research on health behaviour. Publ Health Rep 1970; 85: 815. 14. Laird N. Further comparative analyses of pretest-posttest research design. Am Statist 1983; 37: 329-30. 15. Loening-Baucke VA, Younoszai MK. Effect of treatment on rectal and sigmoid motility and chronically constipated children. Pediatrics 1984; 73: 199-205. 16. Debelle G, Nolan T, Coffey C. Treatment outcome for encopresis managed by behaviour modification and cathartic therapy. Aust Paediatr J 1989; 10: 334 (abstr). 17. Leoning-Baucke V. Modulation of abnormal defecation dynamics by biofeedback treatment in chronically constipated children with encopresis. J Pediatr 1990; 116: 214-22. 18. Shepherd K, Hickstein R, Rose V, Nasser C, Cleghorn GJ, Shepherd RW. Faecal incontinence in childhood: a multidisciplinary approach including boifeedback. Aust Paediatr J 1989; 25: 351-55. 19. Wald A, Chandra R, Gabel S, Chiponis D. Evaluation of biofeedback in childhood encopresis. J Pediatr Gastroenterol Nutr 1987; 6: 554-58.
No 8766
1991
ORIGINAL ARTICLES
Randomised trial of laxatives in treatment of childhood encopresis
Primary faecal incontinence (encopresis) in children is usually treated with laxative medication and a modification
programme aimed at promoting regular toileting, but the effectiveness of laxatives has never been adequately investigated. 169 children with encopresis and evidence of stool on plain abdominal radiograph were randomly allocated to receive multimodal (MM) therapy (laxatives plus behaviour modification; n=83) or behaviour modification alone (BM; n=86). Mean (SD) follow-up was 55·1 (27·0) weeks and 56·7 (32·0) weeks, respectively. By 12 months follow-up 42 (51%) of the MM group and 31 (36%) of the BM group (p=0·079) had achieved remission (at least one 4 week period with no soiling episodes) and 52 (63%) vs 37 (43%) (p=0·016) had achieved at least partial remission (soiling no more than once a week). MM subjects achieved remission significantly sooner than BM subjects, and the difference in the Kaplan-Meier remission curves was most striking in the first 30 weeks of follow-up (p=0·012). The patterns of compliance with toileting in the treatment groups were almost identical, although about 1 in 8 children overall did not comply with the sitting programme. After exclusion of the 24 poor compliers, there was no significant difference between BM and MM groups. This study shows a clear advantage overall for the use of laxative medication, although the benefit may not be as great for children who are able to maintain regular
behaviour
toileting.
Introduction Persistent faecal incontinence without associated anatomical abnormality (encopresis) is a common chronic and disabling disorder in childhood. Bellman1 defmed encopresis as repeated, involuntary defaecation into clothing, occurring in children over 4 years of age, for a period of at least 1 month. The reported prevalence among children of 7-8 years is 2-3% for boys and 1-3% for girls1 and among 10-12-year-olds 1-3% and 0-3%, respectively.2 The management of encopresis has developed without adequate controlled evaluation; it is usually based on laxative treatment with a varying degree of behaviour modification aimed at promoting regular toileting.3-5 There have been just two comparative studies of the additional benefit conferred by laxatives, and both were inconclusive. One was a randomised controlled trial6 and the other a nonrandomised retrospective comparison.7 Both had such small sample sizes that quite a large true difference in efficacy between the treatment groups could easily have gone undetected by chance. On the other hand, it has been claimed that no laxative therapy is required to treat encopresis successfully.s There have been three reports of behaviour modification alone 6,7resulting in remission rates of 70%, 40%, and 38%, respectively. We have tested the hypothesis that multimodal therapy (laxative treatment together with behaviour modification) for encopresis is superior to behaviour modification alone in terms of both remission and relapse rates. ADDRESSES University of Melbourne Department of Paediatrics (T. Nolan, FRACP, C. Coffey, BSc) and Department of Ambulatory Paediatrics (G. Debelle, FRACP, F. Oberklaid, FRACP), Australia. Children’s Melbourne, Hospital, Royal Correspondence to Dr T. Nolan, Department of Paediatrics, Royal Children’s Hospital, Parkville 3052, Australia.
524
Patients and methods A two-group, parallel, randomised trial was carried out in subjects referred to a clinic specifically for the treatment of children with faecal incontinence. The primary study endpoint was time to remission from encopresis. The study was approved by the Royal Children’s Hospital ethics in human research committee. Randomisation took place only after informed written consent was obtained from parents or guardians of eligible patients. Subject eligibility criteria were: meeting the definition of encopresis; evidence of stool on plain abdominal radiograph; the absence of very severe and/or prolonged constipation necessitating previous hospital admissions for enemas and other treatments; age 4-16 years; attendance at normal school; absence of neuromuscular disorder (eg, spina bifida, cerebral palsy, muscular dystrophy); absence of Hirschprung’s disease (based on barium enema, rectal biopsy if clinically indicated); and no use of purgatives for at least 2
TABLE I-BASELINE CHARACTERISTICS OF STUDY GROUPS
weeks before baseline assessment. 222 new patients attended the clinic between July, 1987, and July, 1989. Of these, 181 were eligible. The reasons for exclusion were no
encopresis (17), severe constipation requiring disimpaction (6), already receiving purgative medication (3), too young (3), moving interstate or living beyond 100 km from the hospital (5), not attending normal school (4), and need for immediate psychiatric referral (3). The parents of a further 12 eligible subjects refused or were unable to provide written informed consent. The remaining 169 eligible children were randomly allocated to the multimodal therapy group (MM), or the behaviour modification only group (BM). By the logrank test and Freedman’s tabulated sample size estimates a total sample size of 155 would be adequate to show 50% superiority (60% vs 40%) of multimodal therapy over behaviour modification alone at the one-tailed alpha level of 0-05 with 80% certainty. Referrals from general practitioners accounted for 44% of randomised children, 35% were referred by school doctors or nurses, and only 8% were referred by consultant paediatricians. Randomisation was carried out by means of a stratified (primary vs secondary encopresis), blocked schedule (variable block size of 4, 6, and 8) by an individual not connected with the study. Each treatment
allocation
was
recorded
on
a
card in
an
opaque,
numbered, and sealed envelope and stored sequentially in four separate boxes corresponding to the assignment strata. A child was classified as having primary encopresis when he or she had never achieved faecal continence for longer than 1 month, and secondary encopresis if faecal continence had ever been achieved for longer than 1 month. Remission was defmed as at least one continuous 4-week period with no soiling episode and partial remission as soiling episodes occurring no more often than once per week. Relapse occurred when there was resumption of soiling more often than twice per month after a period of full remission. Treatment failure was defmed as no substantial progress towards remission by at least 6 weeks from study entry (BM group); it occurred in 32 cases (37%). These subjects were deemed to have reached a study endpoint (and were censored) at the time treatment failure was declared. 3 patients started laxative medication before 6 weeks-1 protocol violation (4 weeks), 1 child with severe symptomatic constipation whose parents refused to proceed (4 weeks), and 1 patient with severe emotional problems in whom no clinically satisfactory improvement occurred (5 weeks). Both groups received a standard paediatric behaviour modification intervention.3-S This consisted firstly of clarification during a joint parent-child interview of the postulated underlying physiological basis for encopresis. The bowel training programme used positive reinforcement for successful defaecation in the toilet, and additional reinforcement for each 24 h without soiling. Reinforcement consisted of parental praise, and the use of a star-chart diary (fitness training card) to indicate soiling-free days. A regular sitting programme of 5-10 min toilet-time within 30 min of each meal was the basis of the programme. Dietary advice, general counselling, and support were provided by the paediatrician. Psychiatric assessment or treatment was initiated when indicated clinically. In addition, the protocol for the MM group used laxative therapy in two phases. The initial disimpaction phase consisted of 3-day
tConstlpated *Green
on
abdominal
radiograph.
score.
cycles of 5 ml ’Microlax’ enemas (Pharmacia: sodium citrate 90 mg, sodium lauryl sulphoacetate 9 mg, sorbic acid 5 mg, glycerol, sorbitol, distilled water) on day 1, one 5 mg bisacodyl suppository after school and one in the evening on day 2, and a 5 mg bisacodyl tablet after school and one in the evening on day 3. Up to four cycles (12 days) were undertaken. Further cycles were prescribed if there was later evidence of stool reaccumulation. The subsequent maintenance phase consisted of’Agarol’ (Knoll AG: liquid paraffin, phenolphthalein, benzoic acid, sorbic acid) 5-30 ml once or twice each day, senna granules, and/or bisacodyl tablets. Doses were adjusted to maintain at least daily defaecation and were increased if there was persistent or recurrent stool retention (large-calibre stools, less frequent defaecation, abdominal pain, or reappearance of soiling). Constipation was assessed with the stool retention rating record" based on a plain abdominal radiograph at baseline and 2 weeks after randomisation. A score of 10 or more classified the patient as constipated. Radiographs were scored blindly by an experienced paediatric radiologist not connected with the study. Intraobserver repeatability was high (random sample of n 20, intraclass r 0-93, The parent form of the child behaviour checklist p < 0001). (CBCL)12 was used as a measure of behaviour disorder at baseline and at 12 months. A summary behaviour problems standardised T-score of more than 63 classified a child as maladjusted.12 Items 6 (bowel movements outside toilet), 49 (constipated, doesn’t move bowel), 56f (stomachaches or cramps), 107 (wets self during the =
=
525
Fig 2-Kaplan-Meier curves for remission from incontinence).
treatment groups
(time
to
analysis by the Kaplan-Meier product limit estimator (with logrank and Wilcoxon tests), and a Cox proportional hazards model were used with remission and relapse as failure-times. Zero time was the time of random assignment for remission, and the time of remission for relapse. The proportionality assumption was tested with an interaction between the outcome group and log time. Comparison of proportions of subjects in remission and partial remission was carried out by means of the X2 test for 3, 6, and 12 months of follow-up by intention to treat. t tests and nonparametric tests were used as appropriate. Scores from the behaviour measures were assessed by analysis of covariance with the gain score (score at baseline minus score at yearly review) as the dependent variable, with adjustment for pretreatment score and other covariates.14 Although sample size estimates were based on a one-tailed fx, all tests and confidence intervals are reported as Survival
Fig 1-Patient follow-up and study endpoints. *Of the 32 BM subjects who reached an end-point, 17 later achieved remission
with the
help of laxatives.
two-tailed. 108 (wets the bed) on the CBCL behaviour problems scale relate to behaviours or symptoms which could be directly related to soiling. Therefore, the CBCL was also scored with these items set to 0 to avoid confounding of symptomatic improvement directly attributable to resolution of soiling with behaviour changes (QCBCL score). The socioeconomic status measure was the two-factor Green score which is a weighted aggregate of scores for years of maternal education and occupation of the head of the household.13 The fitness training card was used to monitor clinical progress, compliance with the bowel training programme (excellent = 2-3 sits/day; good = 1-2 sits/day; poor = less than 1 sit/day), toilet defaecation frequency, medication compliance (excellent = more than 75% taken; good = 25-75 %; poor=less than 25%), and adverse effects of laxatives. Clinic follow-up was usually done at 2, 6, 14, 30, and 52 weeks and thereafter as required. There was, however, variation in the frequency and subsequent duration of follow-up depending on clinical progress.
day),
TABLE II-PRESCRIBED TREATMENT IN MM GROUP
Results The baseline characteristics of the 169 randomised patients are given in table I; more girls, more older children, and more children who had had previous laxative treatment and toileting programmes were allocated to the MM group. In the BM group, twice as many children were passing stool into the toilet at least daily, and there were more who had problems with urinary continence. There were similarly high rates of faecal retention (on the stool retention record) and maladjustment (on the CBCL). For other baseline
features, randomisation produced equivalent distributions. The average duration of follow-up was 55(SD 27-0) weeks for the MM group, and 56-7 (32-0) weeks for the BM group. Over this time the mean numbers of clinic attendances were 7-2 (35) and 7-1 (3-3), respectively (p=073). All but 7 patients were reviewed after about 12 months’ follow-up. Over the first 12 months of follow-up, 4 patients in the BM group, and 3 patients in the MM group
lost to follow-up (fig 1). The times at which BM patients reached study endpoints are also shown in fig 1. Patients who were discharged in remission after 12 months’ follow-up were assumed to have remained in remission. Details of prescribed treatment in the MM group are given in table II. 8 children (10%) were not prescribed any rectally administered medication. 79% received 9 or more days of rectally administered disimpaction therapy. Over subsequent months, 25% had additional disimpaction cycles. In the maintenance phase, the median duration of laxative use was 6 months, with more than 33% being prescribed laxative medication for more than 9 months. Only 3 patients were admitted to hospital for disimpaction. 29 families were referred for psychiatric evaluation and were
*Median duration of laxative use=6 6 mo.
526
TABLE m—REM!SStON AND IMPROVEMENT RATES
By 6 months, the MM group remission rate was 49%, and was substantial improvement in a further 11 % (table Both remission and improvement rates were only III). slightly higher by 1 year. BM group rates were significantly lower than MM group rates at 3 and 6 months, and the improvement rate was significantly lower at 12 months. The remission rates at 12 months were 51 % and 36% (p = 0-079, 95% CI on difference, upper limit 29%). Soiling frequency was higher in the BM than in the MM group at 3 and 6 months (rank sum test p 0- 11 and 0-004), but at 12 months the BM subjects who had not reached a study endpoint had significantly fewer soiling episodes than subjects in the MM group (p < 0-001). Of the 32 BM patients who reached a study endpoint, 17 (53%) later achieved remission with the help of laxative medication. Remission was achieved within 6 months by 12 (38%) patients, and by 12 months in 16 (50%). The influence of compliance with the toileting programme on remission is shown in fig 3. The BM children who complied poorly (n = 11) did significantly worse than all others overall (Wilcoxon test p=0016, logrank test p=0-06). After removal of the poor compliers in both treatment groups from the data set (n=24), proportional hazards analysis was repeated for phases I and II of follow-up with the same covariates as in the principal evaluation. There was no significant difference between BM and MM groups in phase I (hazard ratio 1 -49,95% CI 0-90, 2-47; p=012) or in phase II (hazard ratio 0-73, 95% CI 0-32,1-70; p =046). Children with secondary encopresis who received MM therapy did significantly better (Wilcoxon test p=0005, logrank test p 0-01) than all other groups. There were no significant interactions between treatment group and sex, maladjustment (on the CBCL), extent of previous treatment, coexistence of enuresis (noctural or diurnal), socioeconomic status, presence of single parent family, or constipation at baseline assessment. There was no significant difference in outcome for patients by their treating clinicians. More MM than BM subjects experienced at least one relapse (33/55 [51%] vs 17/46 [37%]), but the 14% difference was not significant (p=021, 95% CI for difference - 5%, 33%). The proportional hazards analysis resulted in a hazard ratio of 1-56 (95% CI 0-85, 2-86; p 0-16) for MM compared with BM subjects. Only 1 BM subject had more than one relapse, compared with 6 MM subjects. Of MM children who accepted medication, 28 (35%) reported at least one episode of transient abdominal pain (61%), watery stool (54%), nausea (7%), or rectal pain (7%). No subject had symptoms severe enough to require hospital admission. At 12 months’ follow-up, the adjusted mean improvement in QCBCL summary behaviour problem T-scores for the 76 available MM group subjects was 3-10 (p=0005), compared with 1-81 (p=0-21) for the 43 BM subjects still in the study. However, the difference in these adjusted gain scores was not significant (p = 0-48). Of the 17 BM and 17 MM subjects with nocturnal enuresis at baseline for whom follow-up data were available, remission from enuresis was recorded in 7 (41%) and 5 (29%), respectively. 7 children in each group had diurnal enuresis, and 4 in each group recovered during follow-up.
there
=
*p < 0 05, tp < 0 005.
co-management
(BM
group 17
[20%],
MM group 12
[15%]; p=036). The patterns of compliance with toileting were similar in the BM and MM groups (excellent in 50 [60%] and 46 [58%]; good in 22 [27%] and 21 [26%]; poor in 11 [13%] and 13 [16%], respectively). 1 in 8 children overall did not comply with the sitting programme. Compliance with prescribed medication in the MM group was excellent in 51 (65%) and good in 22 (28%); only 6 (8%) were poorly
compliant. Non-compliant subjects were significantly more likely to be the children of single parents (38% vs 14%; p=0006) and to be maladjusted on CBCL summary behaviour problem T-scores (71% vs 38%; p=0-003). There was no significant difference in sex, age, difficulty with initial toilet training, socioeconomic status, or presence of stool retention at baseline between compliant and non-compliant children. Reduction in stool retention was measured over the first 2 weeks of treatment by change in the radiograph stool retention rating scores. There was a mean decrease of 1-84 (SD 6-08) in BM group subjects (p 0-012 by signrank test) and of 3-17 (6-31) in the MM group (p =0-0001). The =
between-group difference, however, significant (p=015).
was not
statistically
The MM group came into remission significantly sooner than the BM group (fig 2), and the difference in the remission curves is most striking in the first 30 weeks of follow-up (logrank test p=0-12, Wilcoxon test p=0-012). The proportionality assumption was not satisfied for the full follow-up period, and separate hazard functions were modelled for follow-up before and after 26 weeks (phase I and phase II, respectively). After adjustment for age, sex, stratification group, and socioeconomic status, the phase I hazard ratio was 1-91 (MM vs BM, 95% confidence interval [CI] 1-18, 3-09; p = 0-008). However, the hazard ratio was not significant for phase II (102, 95% CI 046, 2-26;
p=095).
Fig 3-Kaplan-Meier curves showing effect of compliance on remission.
treatment
=
=
8 children in the BM group and 4 in the MM group showed development of enuresis.
527
Discussion These results clearly show that the acquisition or resumption of faecal continence is promoted by the addition of laxative therapy to the structured toileting programme, and that successful treatment is accompanied by a significant reduction in emotional maladjustment. This finding contrasts with that of the only other published randomised evaluation of laxative therapy6 but the negative outcome reported in that study can be attributed to several methodological shortcomings. The randomised comparison applied only to 3 months of laxative therapy, no rectal medication was used, and the sample size was so small that the observed lack of difference in outcomes could easily have occurred by chance. In our study, most of the patients had had substantial previous treatment, which usually consisted of oral laxative therapy without any serious attempt to modify toileting behaviour. Nevertheless, the study group represented a spectrum of severity and history of incontinence. There is no reason to believe that the results cannot be generalised to the usual paediatric patient population in community or
hospital practice. Randomisation
produced treatment groups which were largely comparable, with the possible important exception of the amount of previous laxative therapy, and the extent to which the child regularly defaecated into the toilet. In both cases, this difference in distribution would probably have made remission less likely in the MM group. Although laxative medication improved outcomes overall, the benefit was slight for children who were able to comply with an effective toileting programme. There were nearly identical patterns of compliance with toileting in both treatment groups, although about 1 in 8 children overall was poorly compliant. Behavioural and family factors identified children who were likely to be non-compliant. However, we believe that these factors should be used only as a guide rather than as criteria on which to judge the need for laxative medication. Children with secondary encopresis who were treated with MM therapy did significantly better than all other groups. This is not a surprising result; secondary encopresis is likely to be related to simple faecal accumulation, whereas in primary encopresis more complex pathophysiological and behavioural factors may operate. We considered the use of laxative placebo for this study but thought that rectal administration of placebo was unlikely to be acceptable to either ethics committees or parents, and there is the possibility of a therapeutic effect of even rectally administered water. It seems unlikely that a placebo effect could account for the observed difference in outcomes.
The use of study endpoints to provide an escape for those BM subjects who made inadequate clinical progress was a necessary prerequisite to mounting this trial. The study clinicians adhered well to the criteria for BM treatment failure. It seems reasonable to expect that prolonged continuation of therapy without laxatives would not have produced outcomes superior to the MM group. The rate of remission in BM patients who were switched to MM therapy after leaving the study was very similar to that in the MM group. The mechanism of action of laxatives in faecal incontinence is not completely understood and is probably not simply related to the debulking effect on chronically retained stool. It is possible that the stimulation of colonic
smooth muscle not only increases colonic motility itself,15 but also increases awareness of motility and rectal filling. The remission and improvement rates achieved by MM-treated patients in this study are similar to those reported in previous follow-up studies.3.4,16 Although such studies have repeatedly shown that good outcomes are achievable, the same studies have also shown that up to a third of patients remain troubled by at least some degree of incontinence. The usefulness of biofeedback training for these treatment-resistant patients also remains to be properly assessed. 17-19 There has been no systematic attempt to develop primary prevention strategies. Until more effective treatments are found, we believe that the use of MM therapy remains the best overall approach. However, since compliant children may in some cases do just as well without laxatives, a preliminary trial of behaviour modification may occasionally obviate the need for any medication. We thank Dr Jill Sewell, Dr Alf Nicholson, Mr Ted Byrt, Ms Janet Bishop, Ms Dorothy Jones, and Ms Judy Wright. This study was funded by the National Health and Medical Research Council, and by the Royal Children’s Hospital Research Foundation.
REFERENCES 1. Bellman M. Studies (suppl): 1-154.
on
encopresis. Acta
Paediatr Scand 1966; 170
2. Rutter M, Tizard J, Whitmore K. Education, health and behaviour. Psychological and medical study of childhood development. New York: Wiley, 1970: 219-20. 3. Levine MD, Bakow H. Children with encopresis. A study of treatment outcome. Pediatrics 1976; 845: 852. 4. Lowery SP, Srour JW, Whitehead WE, Schuster MS. Habit training as treatment of encopresis secondary to chronic constipation. J Pediatr Gastroenterol Nutr 1985; 4: 397-401. 5. Wright L. Outcome of a standardized program for treating psychogenic encopresis. Prof Psychol 1975; 6: 453-56. 6. Berg I, Forsythe I, Holt P, Watts J. A controlled trial of ’Senokot’ in faecal soiling treated by behavioural methods. J Child Psychol Psychiatry 1983; 24: 543-49. 7. Halpern WI. The treatment of encopretic children. Am Acad Child Psychiatry 1977; 16: 478-99. 8. Jolley H. A paediatrician’s view on the management of encopresis. Proc R Soc Med 1976; 69: 21-22. 9. Pinkerton P. Psychogenic megacolon in children: the implications of bowel negativism. Arch Dis Child 1958; 33: 371-80. 10. Freedman LS. Tables of the number of patients required in clinical trials using the logrank test. Stats Med 1982; 1: 121-29. 11. Barr RG, Levine MD, Wilkinson RH, Mulvihill D. Chronic and occult stool retention. A clinical tool for its evaluation in school-aged children. Clin Pediatr 1979; 18: 674-86. 12. Achenbach TM, Edelbrock E. Manual for the child behaviour checklist and revised child behaviour profile. Burlington, Vermont: University of Vermont, 1983. 13. Green LW. Manual for scoring socioeconomic status for research on health behaviour. Publ Health Rep 1970; 85: 815. 14. Laird N. Further comparative analyses of pretest-posttest research design. Am Statist 1983; 37: 329-30. 15. Loening-Baucke VA, Younoszai MK. Effect of treatment on rectal and sigmoid motility and chronically constipated children. Pediatrics 1984; 73: 199-205. 16. Debelle G, Nolan T, Coffey C. Treatment outcome for encopresis managed by behaviour modification and cathartic therapy. Aust Paediatr J 1989; 10: 334 (abstr). 17. Leoning-Baucke V. Modulation of abnormal defecation dynamics by biofeedback treatment in chronically constipated children with encopresis. J Pediatr 1990; 116: 214-22. 18. Shepherd K, Hickstein R, Rose V, Nasser C, Cleghorn GJ, Shepherd RW. Faecal incontinence in childhood: a multidisciplinary approach including boifeedback. Aust Paediatr J 1989; 25: 351-55. 19. Wald A, Chandra R, Gabel S, Chiponis D. Evaluation of biofeedback in childhood encopresis. J Pediatr Gastroenterol Nutr 1987; 6: 554-58.
