An Alarm Watch for Daytime Urinary Incontinence: A Randomized Controlled Trial

This trial was registered (ANZCTR; 12611000828921) and was reported by using the Consolidated Standards of Reporting Trials checklist. Approval for the study was granted by the Sydney Children’s Hospitals Network Human Research Ethics Committee (HREC 11/CHW/12).

Children aged 5 to 13 years, with daytime urinary incontinence, who were referred to the tertiary urinary continence service at the Children’s Hospital at Westmead in Sydney, Australia, and who were prescribed timed-voiding as part of urotherapy, were invited to participate in the study. Referrals to the service were from the child’s family doctor or pediatrician. Participants continued to receive other treatments (including anticholinergic medications, treatment of constipation, behavioral therapy and psychological support) as deemed appropriate by their treating doctor during the study.

Children with daytime urinary incontinence of at least twice per week for at least 2 weeks before enrolment were eligible to participate. This included children with attention-deficit/hyperactivity disorder, intellectual disability, and other conditions. Exclusion criteria were children with organic causes for their incontinence who would be unable to toilet train (including defects of the central nervous system or underlying urologic abnormalities as defined by the physician) or where a timed-voiding program was contraindicated (for example, where the family would find the timed-voiding program overburdensome because of family stress).

Participants were randomly assigned, in an equal ratio, to receive either a personalized alarm watch or a standard watch, which was otherwise identical, for a 3-month treatment period. Both watches resembled a typical wristwatch worn by children, with a choice of colors for the watch band chosen by each child. Participants were stratified by age, sex, and severity of daytime urinary incontinence. Randomization was performed centrally, and participants were allocated treatment through an independent telephone randomization service. The investigators and data analysts were blinded to the allocated treatment; however, blinding of parents (who assessed the treatment outcome) and children was not possible.

In the intervention arm, the watch was set by the trial coordinator to vibrate at ∼2 hourly intervals during the day, at times defined by the parents to fit in with the child’s daily routine, and then were “locked” to prevent tampering. Participants were encouraged to void when the watch vibrated. In the control arm, identical watches (with time and date set by the trial coordinator but with the alarm disabled and locked) were given to participants who were encouraged to void regularly at ∼2 hourly intervals. Other aspects of urotherapy, including education about the importance of regular voiding, use of correct posture for voiding and lifestyle advice such as adequate hydration, avoidance of caffeinated drinks, encouragement of a healthy diet that has adequate fiber intake, and assessment and management of constipation, were provided to all participants equally.

Each child was followed up for a total of 9 months from enrolment, inclusive of the 3-month treatment period. Data collected at baseline included demographic data, a 14-day diary (recording severity and frequency of urinary incontinence), previous and current interventions used, quality of life data (by using the Paediatric Incontinence Quality of Life Questionnaire [PinQ], a modified bladder specific quality of life questionnaire^4,15 ) and a baseline time and volume chart (measuring fluid intake and urine output for two 24-hour periods). Participants also underwent uroflometry and a postvoid bladder scan to assess for bladder function and abnormal postvoid residual urine volumes. The uroflow was conducted when the child felt an urge to void. The postvoid bladder ultrasound was conducted immediately after the child voided, by the same researcher. Postvoid residual volumes were defined as abnormal if >30 ml in children aged <7 years or >20 ml in children aged ≥7 years, according to the age specific International Children’s Continence Society (ICCS) definition.⁷ 

During the 3-month study period, parents were asked to maintain a paper-based diary, where they recorded their child’s treatment response daily, including whether they had wet, the amount of urinary leakage, whether the child voided regularly during the day, and whether the child used the watch. At 3 months, data were collected about treatment outcomes and quality of life, and the uroflow and bladder scan assessments were repeated. Once the 3-month treatment period was completed, participants were allowed to keep their watch, which was unlocked by the trial coordinator. Participants were then followed up every 2 months for a total of 9 months from enrolment to assess for relapse after the treatment period.

The primary outcome of interest was complete response by 3 months, as defined by ICCS, as an indication of initial success.⁷  According to ICCS, a complete response is defined as a 100% reduction, a partial response is defined as a 50% to 99% reduction, and a nonresponse is defined as a <50% reduction in frequency of wetting. Secondary outcomes include treatment adherence (defined as a child who wears the watch for at least 75% of the time and always or mostly voiding at the appropriate times when they did so, as reported by the parents on the treatment diary); time to achieve complete response (defined as the period from trial commencement until 14 consecutive dry days were achieved); relapse after complete response (defined as recommencement of daytime wetting); change in frequency and severity of wetting from baseline; abnormal postvoid residual urine volumes; adverse effects; treatment satisfaction; and change in bladder specific quality of life score. We chose a chose a multimodal approach to assess adherence. with a cutoff point between 60% to 80%, which is commonly accepted as an indication of treatment adherent in the literature.¹⁶ 

A sample size of 360 children was required to detect a 20% difference in the rate of complete response between the alarm watch and control watch groups (considered a clinically significant difference), with 80% power and a significance of 0.05. Data analysis was by intention to treat. Descriptive results are presented as percentages and counts for discrete variables and as mean and SD or median and interquartile range for quantitative variables.

The primary analysis was the comparison of intervention and control arms with respect to proportion of subjects with a complete response by 3 months by using a χ² test. Mann–Whitney U test was used for comparison of number of wet days per week at relapse, diary data, and treatment adherence, because of the expected skewness in these variables. The comparison of time to complete response between the arms used a Cox-proportional hazards model, adjusting for baseline variables.

The study was conducted over 6 years (2012–2017). Of 380 children who were approached for participation in the trial, 243 (63.9%) consented, with 120 randomly assigned to the alarm watch and 123 to the control watch. A total of 4 participants in the intervention group and 13 in the control group were lost to follow-up during the treatment period. A total of 6 participants discontinued treatment: 1 from the intervention group and 5 from the control group (Fig 1). Recruitment was stopped after 6 years because ongoing enrolment was no longer feasible because potential participants increasingly preferred to purchase their own alarm watch, which had become commercially available in Australia shortly after commencement of the trial.

At baseline, 62% were girls, the average age was 8.0 years, the degree of wetting was moderate to severe, and the baseline modified PinQ score was 2.8 for the control group and 3.0 for the alarm group (out of a possible 6). Sex, age, frequency and severity of wetting, and mean quality of life scores were similar between groups, although the alarm group had more participants with abnormal postvoid residual urine volumes (21% vs 12%; Table 1).

At the end of the three-month treatment period, 22% of the alarm watch group and 17% of the control group (difference: 5%; 95% confidence interval [CI]: −5% to 16%; P = .4) achieved a complete response. In the alarm group, the recorded frequency of wetting was slightly lower (65% vs 49%; difference: 16%; 95% CI: 3% to 29%; P = .05; see Table 2 for details). Fewer participants had an abnormal postvoid residual urine volume (15% vs 10%; difference: 6%; 95% CI: −3% to 14%; P = .04), with more having a resolution of their abnormal postvoid residual urine volume if it was initially present. There was no difference in the change in PinQ score before and after treatment between the 2 groups (difference: 0.09; 95% CI: 0.08 to −0.29; P = .1; Table 2).

Participants assigned to the alarm watch were more treatment adherent than those assigned to the control watch (40% were adherent, compared with 10% of the control watch group; difference: 30%; 95% CI: 20% to 40%; P < .001; Table 2). The median time to achieve a complete response was similar between groups, with the 42 and 39 days (P = .8) for the alarm and control groups, respectively (Table 2).

Six months after completion of treatment, the relapse rates were similar between the 2 groups and were 42% and 33% (difference: 9%; 95% CI: −38% to 19%; P = .8) for the alarm and control groups, respectively (Table 3).

More parents from the alarm watch group reported that the treatment was helpful in reducing the child’s frequency and severity of wetting. More parents in the control watch group reported that their child did not respond to the watch to prompt timed-voiding. There was no difference in the frequency of side effects, the child’s attitude to the watch, or family acceptance of the watch between the groups (see Table 4).

Although timed-voiding is commonly prescribed for treating daytime urinary incontinence, treatment nonadherence remains a major barrier for children.^12,13  This trial demonstrated that compared with a standard (nonalarm) watch, an alarm watch improves treatment adherence, resulting in a reduction in episodes of incontinence and in normalization of postvoid residual urine volumes. However, this did not result in more children becoming completely dry, because no difference in complete response rates was found. It is possible that the alarm watch may be more effective in children who are expected to be less treatment adherent to timed-voiding. Both watches were equally acceptable and tolerable to children and their families, with no significant adverse effects reported.

We assessed the treatment response 6 months after completion of treatment to assess the sustainability of the treatment response. There was no difference between the two groups, with 42% and 33% of those who achieved a complete response relapsing 6 months after completion of treatment in the alarm and control groups, respectively. The relapse in those who were initially successful may be related to a reduction in motivation to continue timed-voiding. At 6 months after completion of the 3-month treatment period, more than half of all the participants reporting an improvement in the frequency of wetting and a third reporting an improvement in the severity of their wetting, but there was no difference between the groups. This similarity in response between the 2 watch groups may be because participants in the control group (who had the option of “turning on” the alarm mechanism on their watch after the 3-month treatment period) used their watch as an alarm watch during this period.

Compared with previous studies of timed-voiding for treating daytime urinary incontinence in children,^12,13  the complete response rate in the control arm of our study (who received a nonalarming watch and standard care) was higher than the rates reported in previous studies of timed-voiding in children without using any watch (17% in our study, compared with 7% in the retrospective study of timed-voiding and 0% in the control arm of the randomized controlled trial). The complete response rate in the intervention arm of our study was similar to the rate in the comparable arm of the previously published small trial, with relative effectiveness lower in our trial than the previous trial.¹³  Our study indicates that in children, wearing a watch that does not alarm to aid timed-voiding may increase the response rate, but to a lesser extent than wearing an alarm watch because the alarm watch improves treatment adherence to timed-voiding.

The findings of the normalization of postvoid residual urine volumes after treatment has implications for children with recurrent urinary tract infections because voiding postponement and abnormal postvoid residual urine volumes appear to be a causal factor.⁹  Our study was not powered to address the question of whether an alarm watch can reduce recurrent urinary tract infections in children with incomplete bladder emptying, but this is a hypothesis that is worthy of further exploration.

One of the limitations of our study was the inability to achieve the sample size anticipated, which may have impacted on the power to demonstrate whether the alarm watch can improve complete response rates. In the design of the study, we chose the complete response rate as the primary outcome because ICCS defined this as success. However, using the complete response rate as the primary outcome meant that this outcome, which requires a higher threshold to achieve than an outcome such as reduction in the frequency of urinary incontinence, may have compounded the problem of power. Despite extending our study to 6 years, we were unable to recruit sufficient numbers to the study, largely because of alarm watches being commercially available in Australia shortly after commencement of our study and the reluctance by families to participate in the Watch with Alarm for Timed-Voiding in Children study because many preferred to purchase their own alarm watch, rather than accepting the uncertainty of randomization and the potential of being randomly assigned to the control arm. This again highlights the difficulty of conducting trials among children where lack of personal equipoise can significantly impact on study success. Because our study was conducted among 5 to 13 year olds who were referred to a tertiary center, participants are more likely to be treatment resistant, and our results may not be generalizable to children seen in the primary care setting.

An alarm watch is more effective than a nonalarming watch in improving treatment adherence with timed-voiding and in normalization of postvoid residual urine volumes in children with daytime urinary incontinence. Although there were no differences in complete response and relapse rates and quality of life score changes between the two groups, the improvement in treatment adherence and normalization of postvoid residual urine volumes with the alarm watch may have implications for treating children with recurrent urinary tract infections who have incomplete bladder emptying and using alarming devices to improve treatment adherence in children in other clinical settings.

Preliminary data for this project were presented at the Continence Foundation of Australia/International Children’s Continence Society and Urogynaecological Society of Australia Conference in Cairns, Australia (2014), the Royal Australasian College of Physician Conference in Sydney (2018) and the International Children’s Continence Society Conference in Rome (2018).

CI

confidence interval

ICCS

International Children’s Continence Society

PinQ

Paediatric Incontinence Quality of Life Questionnaire