Rapid Maxillary Expansion in Children with Nocturnal Enuresis: A Randomized Placebo-controlled trial

Rapid maxillary expansion (RME) has previously been reported to have a potentially curative effect on nocturnal enuresis (NE) or bedwetting.^1,,,,–6  The term nocturnal enuresis is used for involuntary voiding of urine at night in children aged 5 years or older.⁷  After asthma and allergies, NE is the most common health problem in school children, affecting about 10% of 7-year-olds.^8,–10  The pathogenesis is multifactorial and complex but has, over the past century, become somewhat clearer. While enuresis was previously considered a primarily psychiatric disorder, the understanding now is that the following three causes are crucial: (1) nocturnal polyuria, (2) detrusor overactivity, and (3) impaired arousal response during sleep. Furthermore, for a subgroup of patients, enuresis can be caused by upper airway obstruction.¹¹ 

NE is highly hereditary, and the risk is 5 to 7 times higher if one parent has a history of the condition.⁸  Several “enuresis genes” have been discovered, although genotype and phenotype are not tightly linked, and the genes found have not led to further pathogenetic insights.¹² 

Treatment is usually recommended from 6 years of age, and the two first-line therapies are either the enuresis alarm, which conditions the child to wake up before wetting the bed, and desmopressin, an analogue of the pituitary hormone vasopressin, which reduces nocturnal urine output.^13,14  Unfortunately, these therapies are successful in only approximately half of the patients. The agreed second-line therapies, namely, anticholinergics or antidepressants, are limited by risks and side effects. On the other hand, approximately half of the subgroup of enuretic children with heavy snoring or sleep apnea gain resolution after adenotonsillectomy.^11,15 

Persisting NE often causes low self-esteem and negatively affects quality of life.^16,17  Therefore, the search for alternative antienuretic therapies is of great importance. Timms,¹  in his pioneering article from 1990, found that 10 enuretic children with upper airway obstruction became dry after RME. Since then, several other nonrandomized studies have been published indicating that approximately half of the children become dry or have their enuresis frequency much reduced.^2,3,6 

The aims of this prospective, randomized, placebo-controlled study were to investigate (1) whether RME was a useful treatment method for NE, (2) whether the treatment effect was due to placebo, and (3) whether any prognostic variables could be identified.

The study protocol was approved by the regional ethical review board in Uppsala, Sweden, registration number 2013/375, and adhered to the Declaration of Helsinki. The trial was registered at ClinicalTrials.gov (Identifier: NCT02178826), and the protocol was published at that site before recruitment started.

Participants were recruited at the tertiary enuresis center at Uppsala University Children's Hospital, Uppsala, Sweden. Children younger than 14 years with primary NE were invited to participate in the study. A consecutive recruitment process was carried out between the years 2013 and 2018.

The inclusion criterion was primary NE with at least 7 wet nights out of 14. Exclusion criteria were daytime incontinence, ongoing antienuretic treatment, or concomitant urological, endocrinological, nephrological, or psychiatric disorders.

Well-functioning patients with treated attention-deficit hyperactivity disorder were not excluded. Most subjects were nonresponders to first-line treatment and were therefore classified as “therapy resistant.”

All children were examined by a pediatrician. A medical history focused on micturition habits, previous treatment, heredity, and sleep disorders was taken. All families registered enuresis episodes on a voiding chart over 2 weeks and daytime voided volumes and nocturnal urine production (via the weighing of diapers or sheet covers) during one weekend. These data were expressed as percentages of expected bladder capacity for the child's age, according to the Hjälmås formula.¹⁸ 

The study protocol is described in Figure 1. Patients were randomly allocated to either the intervention group or the placebo group. The patients in the intervention group carried out RME for 10–14 days (T1), while the placebo group were treated with a sham appliance for 14 days (T1). This appliance was identical to the one used in the intervention group, except that the expansion screw did not generate any expansion. When the randomization was revealed, the sham appliance was replaced with an active RME appliance. For retention, the appliance was kept in situ for 6 months, after which it was removed and a final registration was carried out (T6_mon). Both groups continued documentation of wet and dry nights throughout the study period.

The orthodontic treatment was carried out at the Department of Orthodontics, Public Dental Service, Uppsala, Sweden. The treatment consisted of the transverse expansion of the maxilla, using an RME appliance with a Hyrax screw soldered to orthodontic bands on the permanent first molars (Figure 2). Parents were instructed to activate the Hyrax screw twice daily, which generated an expansion rate of approximately 0.5 mm per day. The endpoint was defined as when the occlusal surface of the palatal cusp of the upper first permanent molar came into contact with the occlusal surface of the buccal cusp of the lower first permanent molar.

Dental study casts were taken at baseline (T0) and after 6 months (T6_mon). The baseline casts were used for classification of the occlusion, according to Angle.¹⁹  The intermolar, interpremolar, and intercanine distances were evaluated at T0 and T6_mon. The intermolar distances were measured at the shortest intermolar linear distance at the gingival margins and at the mesiobuccal cusp tips of the teeth. The interpremolar and intercanine distances were measured at the buccal cusp tip. To evaluate the outcome reliability for the study cast measurements, 20 randomly selected cases were measured twice. The random sequence was generated with a random integer set generator at www.random.org. A digital caliper was used for all measurements (Pluradent Art.-Nr: 12842, Offenbach, Germany).

All patients underwent cardiorespiratory polygraphy to assess respiration during sleep and arousals. This was performed with a portable sleep device (NOX T3, Nox Medical, Reykjavík, Iceland) at baseline, after RME, and at 6 months. The results from these measurements will be reported separately.

All children who failed to obtain a satisfactory level of dryness were re-referred to the enuresis center.

The sample size calculation was based on previous studies by Nevéus et al. on similar populations.²⁰  The number of wet nights out of 14 (±1) was assumed to be 11.5 ± 2.7 without treatment. The calculation indicated that 18 participants in each group were required in order to have a 90% chance of detecting a decrease of 3 wet nights in the intervention group compared with the placebo group. The significance level was set at 5%.

Informed written and verbal consent was acquired from the families before the patients were randomized for treatment allocation. The randomization was based on a computer-generated sequence using SPSS software (version 17, IBM SPSS Statistics, Chicago, Ill). To ensure homogeneity between the groups, the random allocation was carried out in blocks of different sizes using a concealed principle in a 1:1 ratio. The sequence was held by a member of the staff at the orthodontic clinic, who administered the group allocation. The recruitment and randomization process were in this way separated.

The patients were blinded to the allocation sequence, and the group allocation was not revealed until T1. Thereafter, it was not possible to blind either patients or care providers to the interventions.

To evaluate the primary outcome, the number of wet nights during 14 nights' follow-up after the intervention/placebo treatment, a random intercept linear mixed model was used. Study groups (intervention/placebo), time (T0, T1), and the interaction variable groups × time were fixed factors, and the model's estimated marginal mean differences of outcome within and between study groups was reported with 95% confidence intervals (CIs). The mixed model is similar to the analysis of variance for repeated measurement but has the advantage that patients with missing data were evaluated with the assumption of missing at random. The model residuals were tested with Shapiro-Wilk normality distribution test, and no violation was present.

A linear mixed model was also used to compare the number of wet nights during 14 nights' follow-up at the end of the intervention (T1) when all study subjects had had the intervention and after 6 months (T6_mon) compared with T0. At T6_mon, the study subjects were classified as responders/nonresponders and analyzed with logistic regression to try to identify potential prognostic variables. The logistic regression was adjusted for the number of wet nights reported at T0. A responder was defined as having reported a >50% reduction in the number of wet nights at T6_mon compared with T0. Linear regression was used to identify potential prognostic variables for the continuous outcome, change of number of wet nights T0 to T6_mon, adjusted for number of wet nights at T0.

To evaluate the reliability of measurements on study casts (intermolar width, interpremolar width, and intercanine width), 17 study patients were measured twice on examination at T0, and the intraclass correlation (ICC) was estimated. The ICC was estimated for each outcome by linear mixed model with patients as a random factor. As the ICC is the ratio of the outcome variance between patients divided by total variance (between and within patients), the standard deviation (SD) between and within patients from the mixed model was also presented.

A P value less than .05 was considered statistically significant, and the analyses were performed with IBM SPSS Statistics 22 (Armonk, NY) and STATA release 14 (STATACorp, College Station, Tex).

The patient flow throughout the study is demonstrated in Figure 1. Of the 38 patients enrolled, 18 patients were allocated to the intervention group (18 boys, mean age 10.3 years, SD 1.8 years) and 20 to the placebo group (17 boys and three girls, mean age 10.2 years, SD 1.8 years). This is summarized in Table 1. Two children from the placebo group dropped out shortly after allocation, as one patient was unable to take dental impressions and another refused to have the RME appliance fitted.

The upper arch was somewhat expanded with good occlusion, and no other side effects on the dentition were apparent. The mean transverse maxillary expansion for the whole group was 4.8 mm (SD 2.2) at the intermolar distance, 5.7 mm (SD 1.7) at the interpremolar distance, and 2.2 mm (SD 1.3) at the intercanine distance.

Thirteen children (35%) presented with a Class II malocclusion (Table 2). Five children had a unilateral crossbite. Most of the children had normal occlusions. Twenty-one patients (60%) presented with a positive family history of NE.

Table 3 shows the reduction of wet nights during 2 weeks at T0 and T1 and the difference between the two groups. The intervention group showed a significant reduction of wet nights from T0 to T1, mean −2.2 (95% CI, −3.7 to −0.8; P = .003), and the placebo group showed no significant reduction of wet nights, mean −0.6 (95% CI, −2.0 to 0.8; P = .40). The intervention group showed a larger reduction in the number of wet nights compared with the placebo group; however, the difference between the study groups was not statistically significant, mean −1.6 (95% CI, −3.6 to 0.4), P = .11.

The long-term effect for the whole group 6 months after the intervention is presented in Table 4. The number of wet nights decreased significantly from a mean of 11.9 to a mean of 8.5, mean difference –3.2 (95% CI, −4.5 to −1.8) from T0 to T6_mon. The mean reduction 6 months after expansion was also significant, P = .002.

Table 5 contains results for the prognostic analyses. According to the International Children's Continence Society definitions, the trial resulted in one full responder, 10 intermediate responders, and 20 nonresponders at T6_mon. Since there was only one patient defined as a full responder, the outcome variable in the logistic regression was defined as full/intermediate responders vs nonresponders. The logistic regression, adjusted for number of wet nights at T0, showed no significant associations for any of the potential prognostic variables. The average daytime voided volume showed an odds ratio (OR) of 1.093 (95% CI, 0.998–1.198), indicating higher odds for responding with increasing bladder storage capacity, P = .055. The mean intermolar width at T0 showed an OR of 1.43 (95% CI, 0.94–2.16), indicating higher odds with increasing distance, P = .09.

Evaluation of outcome reliability for the study cast measurements showed high ICC, indicating excellent agreement between the test-retest measurements in all three areas. ICC ranged from lowest 0.989 (95% CI, 0.972–0.996) to highest 0.996 (95% CI, 0.988–0.998).

This was the first randomized placebo-controlled study of RME treatment of patients with therapy-resistant NE. A statistically significant reduction of wet nights was found after the RME treatment within the intervention group, but it was not large enough to show a significant difference when the placebo group and intervention groups were compared. The long-term mean reduction in number of wet nights for the whole group was 3.2 nights over 2 weeks. Although statistically significant, this reduction is, in most cases, not considered clinically relevant and none of the children in this study became completely dry during the trial. Thirty-five percent of the patients in this study had a reduction of wet nights >50% compared with baseline. This therapeutic effect was lower than in previous studies.^1,,,,–6  The study reporting the highest therapeutic benefit was the work of Al Taai et al.⁴  That study, however, reported the enuresis frequency in the number of enuresis episodes per night instead of number of wet nights per week, which makes it difficult to compare. A previous meta-analysis of six articles showed that the average rate of becoming completely dry 1 year after RME was 31%.²¹  Our previous study also indicated a higher response: 60%.⁵  The sample in the present study was composed of mainly therapy-resistant children with a high enuresis frequency. This may partly explain why there was a lower full response rate than in the earlier studies.^1,,,,–6  Another reason may be that there was a shorter follow-up time used in order to limit the effect of spontaneous resolution, which is approximately 15% per year.²²  Most previous studies had a longer follow-up time than the current study, and it is quite likely that some spontaneous resolution occurred during the observation time. The reduction of wet nights for the placebo group was minimal and, as in previous studies, the current study did not find a positive placebo effect of the appliance.^4,5 

A large average daytime voided volume and a wide maxilla may be associated with an increased chance of having the enuresis frequency reduced, and these two variables seemed to be of favorable prognostic value. The association between voided volume and a favorable outcome of the RME treatment is worth mentioning, even though it was not statistically significant. A larger study sample would possibly have strengthened the association. Larger voided volumes constitute an indirect sign of less detrusor overactivity. Detrusor overactivity is characterized by involuntary detrusor muscle contractions during the filling phase of the bladder. Consequently, it may be speculated that RME would work better in patients who have nocturnal polyuria, that is, an abnormally large nocturnal urine production, as a central pathogenetic factor behind their enuresis, instead of nocturnal detrusor overactivity.

Regarding the other possible prognostic factor, the nasal airway dimension is likely to be larger in patients with a wide maxilla. A significant correlation between reduction of NE and increase in nasal volume after RME has been previously reported.⁵  Responders have also been found to have larger nasal airway dimensions.²³  None of the children in the current study sample suffered from upper airway obstructions.

The strength of this study was that it was the first study of its kind with a randomized placebo-controlled design. The study group was representative of the group of patients who would benefit from an alternative treatment method to the first-line treatments available today. A thorough effort was also made to investigate mechanisms underlying the alleged treatment effect.

This study also had some weaknesses. First, the power analysis was based on the reduction of wet nights, and some other items were investigated as well. A larger number of participants would have been needed to carry out a fair predictive analysis. It is possible that some variables with a statistically significant prognostic value could have been found in a larger sample. Second, an untreated control group was not enrolled; it would not have been ethical to leave a group of children untreated for that long.

About one-third of the children in this study had their NE frequency reduced. They all would have wished to become completely dry, and although a statistically significant reduction in the number of wet nights was found after the intervention, it was, in most cases, not clinically satisfactory.

RME can in some cases be beneficial in the treatment of NE. The present study, however, was not able to prove that RME was effective in reducing NE for most of the participants.

RME may have a positive effect on NE in a subgroup of patients. A wide maxilla and large voided volumes at baseline could increase the chances of a successful treatment.

• Rapid maxillary expansion has a modest effect on children with therapy-resistant nocturnal enuresis.

• The treatment outcome does not seem to be due to a placebo effect of the appliance.

This study was funded by the Public Dental Service, Region Uppsala, Uppsala, Sweden, Uppsala-Örebro Regional Research Council (grant RFR-651101), Thuréus Foundation, and Gillberg Foundation.