Low-level evidence for behavioral intervention to treat drooling in children with neurodisability

Q What are the effects of behavioral interventions for the treatment of drooling at the level of body functions and structures, activity and participation (all International Classification of Functioning, Disability and Health constructs), and quality of life in children with neurodisability?

METHOD

Design: Systematic review. This systematic review was designed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses protocols (Hutton et al., 2015). The review protocol was registered online with the International Prospective Register of Systematic Reviews Protocols known as PROSPERO.

Data sources: Databases MEDLINE, CINAHL, AMED, CENTRAL, Embase, PsycINFO, Scopus, and Web of Science were searched for studies up to 2 June 2017, using subject terms: “child,” “neurodisability,” “drooling,” “behavioral,” “intervention,” and associated relevant subject headings. In addition, a manual hand search of reference lists of full-text articles in English was performed by the first author and a limited search of unpublished sources in the gray literature was conducted using Trove, The Agency for Healthcare Research and Quality, Australian Institute of Health and Welfare, National Health Medical Research Council, and Registry of Open Access Repositories. In sum, 763 papers were screened, from which six studies were selected while another two studies were added from an additional hand search of full-text references, resulting in eight studies that were included in this SR. Four of them were published between 1977 and 1989 and four between 2006 and 2011.

Study selection and assessment: Studies were included if they: (1) included participants between the ages of 0 and 18 years with a diagnosis of neurodisability and drooling; (2) provided behavioral interventions that targeted drooling or a drooling-related behavior; and (3) used experimental study designs including randomized controlled trials (RCTs), clinical controlled trials, and single-case experimental design (SCED, i.e. designs with control) studies. After removal of duplicates, review articles, conference proceedings, editorials, poster presentations, and articles in languages other than English, two reviewers independently applied selection criteria to title, abstract, and full-text articles. Most differences were resolved through consensus. If not, a third party was consulted until a final decision was reached. After study selection, two reviewers independently extracted data on the following variables: study design, participant number and diagnostic information, behavioral intervention, target behavior definition, outcomes of interest, outcome measure used, main findings, and level of evidence. To determine the level of evidence of studies, the American Academy of Cerebral Palsy and Developmental Medicine Methodology to Develop Systematic Reviews of Treatment Interventions (Darrah et al., 2008) was used.

 Because of heterogeneity in population diagnoses, in data collection methods and measures used in the included studies, and in reported study outcomes, a narrative approach was used to synthesize the data on key study variables, study quality, and risk of bias. Risk of bias assessment was performed using the Risk of Bias assessment in N-of-1 Trials (RoBiNT) Scale (Tate et al., 2015) for single-case experimental designs and the Cochrane Collaboration Tool for RCTs (Higgins, 2008) by two independent reviewers. Cases, where there were different ratings, were resolved by consensus.

Outcomes: Outcomes assessed included target behavior related to either drooling frequency (e.g. drooling drops per minute) and/or severity (e.g. percentage of chin wetness, percentage of wet time samples, pools of saliva, or drooling status). In addition, “activity-level outcomes” were considered such as mouth drying as a response and improved head control. Authors also include the outcomes “eye contact” and “imitation of vocalization” as one study – in addition to reduce drooling – was also aimed at increasing the beginning language. Finally, outcomes related to participation or quality of life were considered. No review-generated outcomes such as effect size were reported.

MAIN RESULTS

Positive outcomes were reported in all eight studies that included a total of 35 participants with different clinical presentations. Seven studies used SCEDs (four studies a withdrawal/reversal design (level of evidence IV), three studies a multiple baseline design (level of evidence II)) and one was a RCT (level of evidence II) with 25 participants. Behavioral intervention in these studies included the use of a combination of procedures, such as reinforcement (including token economy), prompting, self-management, instruction, extinction, overcorrection, and fading. Each study assessed body functions or structures’ outcomes (drooling frequency and severity); three included activity outcomes (mouth drying, head control) and none assessed participation or quality of life. No validated measures were used to assess outcomes of interest in the included studies. No adverse effects were reported. In six studies with SCED, only visual inspection of graphical data was reported and none of these studies reported a structured visual analysis of the data. While each study reported positive effects of intervention, risk-of-bias was high. Significant quality issues related to recruitment, risk-of-bias, and external validity were present in all eight studies, greatly limiting the interpretation as to how effective behavioral interventions are, in reducing drooling in children with neurodisability.

AUTHOR’S CONCLUSION

Despite the prevalence of drooling in children with neurodisability and the likely common use of behavioral interventions, very few experimental studies have been undertaken to determine whether these interventions work. Low-level evidence suggests behavioral interventions may be useful for the treatment of drooling in children with neurodisability. Well-designed intervention studies are urgently needed to determine effectiveness.

COMMENTARY

Saliva control is a behavioral response that is learned through a process of perception-action loops in combination with social contingencies during the first 4 years of life for most typically developing children (Van Hulst et al., 2018). A subgroup of youth with developmental disabilities need support to acquire saliva control in case this does not develop through maturation and natural contingencies. In most reviews in the last thirty years, oral-motor training and behavioral interventions are appraised as the first step before more invasive medical interventions to reduce drooling are offered (see for instance, Harris & Purdy, 1987). Nevertheless, there is still a very limited number of studies on the effect of behavioral interventions that implement experimental designs with control, as this systematic review rightfully points out. However, this shortage of well-controlled studies for behavioral interventions for drooling is also a recurrent conclusion in past reviews (Van der Burg et al., 2007a, 2007b). One might ask why this did not change during all those years. Perhaps, the organization of a study for (groups of) participants who suffer from drooling and may profit from a specific and uniform behavior intervention program is challenging because sufficient numbers of participants for a RCT or multiple baseline across participants design may not be available or not available at the same time, withdrawal/reversal designs may not be acceptable or meaningful (e.g. after a successful self-management program) and/or multiple baseline across participants, situations, or behaviors designs may transcend the capacity for longitudinal measurements in treatment settings. As no adverse events were reported in the eight studies in this systematic review, one might suppose that if these events do occur, behavioral interventions are likely stopped, leading to incomplete data collection and resulting in a study that is not publishable.

This systematic review was conducted according to well-established methods and guidelines (e.g., Schlosser et al., 2007). The general methodology in the review process was sound and well described in detail (with additional information online), as well as the results of data extraction and risk of bias analysis. However, there may have been some selection bias. It is not clear from the text nor from Table SII “Search strategy in MEDLINE” if search terms were complete to retrieve all behavioral studies. For instance, targets for behavioral intervention like “mouth drying” or “posture control” are missing. Because of the chosen age range from 0 to 18 years, six interesting SCEDs studies by Lancioni and colleagues on behavioral interventions that make use of microswitch technology to reduce drooling in individuals with multiple disabilities and severe to profound learning disabilities with participants >18y are not included in this systematic review (G. E. Lancioni et al., 2008a, 2008b, 2009c, 2011a, 2011b, 2013). One might argue that developmental age in this population is more relevant than chronological age, and as a consequence, chronological age is no valid reason to exclude these studies in this systematic review. Studies with a quasi-experimental design, such as the non-concurrent multiple baseline design (see Van der Burg et al., 2009, 2016) were also not included in this review. However, these studies may be valuable “to identify and synthesize the evidence for behavioral interventions to treat drooling,” as was the purpose of this systematic review. In case, children above the age of 6 did not develop adequate saliva control and still suffer from chronic and severe drooling, baseline drooling must have been high before intervention, although this is not controlled by concurrent data collection in a multiple baseline design. Finally, there might also have been linguistic bias as only papers in English were included and some kind of publication bias because of the removal of conference proceedings and poster presentations, which might have contained unsuccessful and therefore unpublished studies on behavioral intervention to treat drooling. There was no report on the adherence to the review protocol and no quantification of agreement of selection of papers between assessors. During quality assessment, the assessors were not blinded for the results of the study and no interrater agreement on the coding of the studies was reported.

Unfortunately, the setting in which outcomes were measured and interventions took place was not described in the results of the systematic review. This is relevant because drooling frequency and severity may vary within a participant, depending on (amongst other factors) the activities/tasks that are required and the social situations in which they are performed (Van der Burg et al., 2006). In addition, generalization of drooling-related behavior to other relevant social contexts is not a matter of course and demands careful planning and/or additional intervention (De Bruijn et al., 2017; Van der Burg et al., 2016). Although the information on follow-up assessment was described (if available) on the level of the individual studies, the authors did not address the maintenance of behavioral interventions, and neither the assessment of generalization across settings. As Jones (1982) pointed out, maintenance of effects of an intervention for drooling may also decrease after a study ends (referring to the study of Rapp & Bowers, 1979) that implemented auditory cueing for swallowing and wiping responses, using an experimental control group design with follow-up assessment over a period of 9 months (Rapp, 1980), when everyone goes “back to business,” and both the focus of the participant on saliva control and the social attention and reinforcement for non-drooling and adequate saliva control fades out. Both maintenance and generalization of effect are important considerations for clinical decision-making and establishing the contribution of behavioral interventions in relation to various medical and non-medical intervention options for drooling.

Participants from the included studies in this review were treated in very different settings such as a public school (Kay et al., 2006), a special class in a regular school (Dunn et al., 1987), a nonresidential center for individuals with learning disabilities (Drabman et al., 1979), a research/treatment facility (Sethy et al., 2011), and a residential setting for individuals with multiple or severe/profound learning disabilities (Lancioni et al., 1989, 2009a, 2009b; Richman & Kozlowski, 1977). It should be recognized that the learning abilities of participants in combination with setting characteristics determine which behavioral procedures and training intensity are appropriate and feasible. As a consequence, no general conclusions on the effect of “behavioral interventions” as a group of interventions can be drawn. The evidence for specific behavioral intervention programs for drooling must be judged for differentiated subgroups in relation to personal characteristics and contextual factors. Moreover, in systematic reviews and meta-analyses of pharmacologic (medication or injections) and surgical interventions for drooling, evidence is also evaluated and rated for each treatment option separately.

Despite the low level of evidence, the authors of the systematic review still recommend behavioral interventions because they “may be useful for treatment of drooling in children with neurodisability.” This conclusion applies to behavioral intervention for anterior drooling (i.e. spillage of saliva from the mouth that is clearly visible). In case of posterior drooling (i.e. spillage of saliva through the oropharynx and into the hypopharynx, with the risk of aspiration resulting in (recurrent) infections and progressive lung disease), which is also related to swallowing problems in this population and must be differentiated from anterior drooling through careful assessment, behavior interventions aimed at increasing swallowing frequency are contra-indicated.

From a clinical perspective, the selection of a behavioral goal for an intervention for drooling asks for a functional analysis to determine the behavioral deficit that leads to drooling, such as maintaining head and posture control, not efficient or not frequently enough swallowing (and/or wiping), insufficient divided attention during the execution of dual tasks, etc. If swallowing efficiency is unlikely to improve (consult a speech-language therapist), wiping off spilled saliva from the mouth and chin area may be added or chosen as the primary goal of intervention. After this, dependent on personal and setting characteristics, appropriate antecedent and/or consequent behavioral techniques have to be selected and combined in an intervention plan that is manageable by a therapist, teacher, trainer, and/or parents/caregivers. Also, do plan for generalization and maintenance of the effect from the beginning. Be aware that if external cueing by electronic devices or prompts, or reinforcement with token economies remain necessary for the maintenance of the effect, it must be verified that this is practically and socially acceptable for the participant and the people around him/her on the long term. For further considerations and clinical guidelines, see Van der Burg et al. (2007a) and the American Academy of Cerebral Palsy and Developmental Medicine Sialorrhea Care Pathway (https://www.aacpdm.org/publications/care-pathways/sialorrhea). Finally, as longitudinal data collection within an experimental design with control is mandatory for building up evidence, those clinical who have access to the target group and the opportunity to document changes in detail are urged to seek cooperation with researchers to evaluate their programs. The call from the authors of the current systematic review for well-designed behavioral intervention studies is now upgraded to “urgent” by them.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the author(s).