Antireflux Procedures in Children With Neurologic Impairment: A National Survey of Physician Perspectives

We conducted a cross-sectional electronic survey study of physicians who practice at tertiary-care pediatric hospitals in Canada between November 2021 and February 2022. The design and administration of the survey followed guidelines developed by Burns et al.¹⁷  This study was approved by the Conjoint Health Research Ethics Board at our institution.

The survey was codeveloped by a pediatric surgeon, a pediatric gastroenterologist, a pediatric emergency medicine/palliative care physician, and complex care/hospitalist physicians based on clinical experience and review of existing literature. This multispecialty team determined the survey topics, question stems (including clinical scenarios), and the response options for each question. The survey contained 3 sections: (1) participant demographics; (2) general questions about ARPs (diagnostic tests, potential risks and harms, perceived barriers); and (3) case-based questions designed to imitate “real world” clinical practice (Fig 1). Following a clinical vignette, respondents were asked to report their likelihood to consider an ARP and what treatments or interventions they would attempt before an ARP (full questionnaire in Supplemental Fig 5). The vignettes corresponded broadly to the following clinical scenarios: (1) severe aspiration events (“Severe Asp”); (2) moderate aspiration events (“Mod Asp”); (3) failure to thrive because of vomiting or GERD (“FTT”); and (4) pain possibly secondary to GERD (“Pain”). A 4-point Likert scale was used with space for written comments after each question.

This instrument was pilot tested in 5 cognitive interviews with relevant pediatric specialists and subspecialists for content validity, question clarity, item interpretation, and item redundancy. Revisions were made after each interview. After the fifth interview, no further modifications were required. This survey was also reviewed by an expert in survey design. A final survey was created consisting of 8 demographic questions, 7 general ARP questions, and 4 clinical scenarios with 6 questions each. The survey was translated into French by a member of our research team with both French and English language proficiency.

Qualtrics software was used to administer the survey, collect, and store data. Invitations to participate were distributed by e-mail to all members of the relevant divisions at the 14 pediatric hospitals across Canada. Pediatric surgeons were recruited by e-mail invitation through the Canadian Association of Pediatric Surgeons. Potential participants were contacted using Dillman’s tailored design method for mixed-mode surveys.¹⁸  The introductory e-mail contained a link where the recipient could indicate their eligibility and express interest. A survey link was sent directly to each individual respondent to allow for tracking of response rates. A total of 3 reminders were sent at 1-week intervals. An implied consent form was presented before the survey.

Respondents were physicians who practice at tertiary-care pediatric hospitals in Canada and represent 1 of the following subspecialty groups: hospital or general pediatricians, complex care pediatricians, pediatric gastroenterologists, and pediatric surgeons. These subspecialties were chosen as those most likely to be involved in decision-making regarding the management of GERD in this complex population of children. The study is limited to tertiary care hospitals as these sites typically have access to ARPs. Additional eligibility criteria included the respondent having treated a child with GERD, NI, and enteral feeding in the past year and having the ability to respond in either English or French.

Descriptive statistics were calculated, including means with SD and interquartile ranges for continuous variables, and count with proportions for categorical variables. The primary outcome variable of likelihood to consider an ARP was coded as 2 categories (consider and do not consider) and physician characteristics were coded as years in practice (<10, 11–20, ≥30), subspecialty (hospital pediatrics or complex care, gastroenterology, surgery) and location of practice (province).

Bivariable association between potential predictor variables and likelihood to consider an ARP in each clinical scenario were determined using χ-square tests. In bivariable analyses, a significant level of 0.3 was used for variable selections. Any variable with P value < .3 was planned to be included in the multivariable models (although no associations reached this threshold). Multivariable logistic regression modeling was, therefore, conducted including all variables because of their clinical importance to determine if there were any independent effects on the primary outcome of physician consideration of ARP. Multiple imputation technique was used in a sensitivity analysis to account for any missing or incomplete information obtained from the surveys.

A total of 119 physicians indicated eligibility via the invitation link. Of those, 83 provided at least partial responses giving a response rate of 70%. Full responses were available for 76. Table 1 presents respondent characteristics. The majority of respondents represented hospital pediatrics, complex care, and/or general pediatrics, whereas 15% (n = 15) of respondents identified as gastroenterologists and 5% (n = 4) as pediatric surgeons. There was representation in this sample from every province with a tertiary care pediatric hospital and a total of 12 institutions.

A summary of the clinical scenarios and likelihood to consider ARP is shown in Fig 1. Two of the scenarios (#1 – Severe Asp and #4 - Pain) were equally split with approximately half of respondents indicating that they were very or somewhat likely to consider ARP (n = 45, 56% for scenario 1 and n = 44, 57% for scenario 4), whereas the other half were somewhat or very unlikely to consider. Respondents were less likely to consider ARPs for scenario 2 (Mod Asp), wherein only 19% (n = 15) of them would consider this intervention. Scenario 3 (FTT) was most favored for ARP with 78% (n = 60) of respondents indicating they were very or somewhat likely to consider. When asked which ARP they are likely to present as an option (fundoplication or GJT), most respondents would offer GJT only (22% to 49%), whereas very few would offer only fundoplication (4% to 8%) (Supplemental Table 3). On bivariable analysis, none of the potential predictor variables studied were significantly associated with the respondents’ likelihood to consider an ARP in any of the clinical scenarios (Supplemental Table 4). Multivariable logistic regression was performed including all variables (because of suspected clinical importance) with a cutoff of statistical significance at P < .05. This likewise did not reveal any independent predictors of physician likelihood to consider ARP in any clinical scenario. There were missing responses in 7 surveys (8%), but sensitivity analysis suggested this does not significantly impact our results (Supplemental Information).

There were no respondents who indicated they would proceed immediately to ARP in any scenario, instead suggesting a wide range of diagnostic investigations or treatments to be considered before (Table 2). The most common intervention was feeding changes (including changing rate or volumes or switching to continuous feeding) with 49% to 90% of respondents choosing this option. Prokinetics were also commonly trialed, particularly in scenarios 1, 3, and 4 (73%, 86%, and 78% respectively). Some respondents would trial baclofen (13% to 32%), NJ tube feeds (23% to 52%), formula changes (38% to 76%), or blenderized diets (24% to 54%). Depending on the scenario, up to half of respondents (19% to 54%) also indicated they would attempt to treat a comorbid condition, most commonly sialorrhea.

In terms of diagnostic testing before ARP, over half of respondents would order an upper gastrointestinal series for either GJT (n = 53, 64%) or fundoplication (n = 57, 69%). About one third would complete a gastric emptying scan (n = 30, 35% for GJT; n = 37, 45% for fundoplication) and progressively smaller numbers would request impedance studies (n = 16, 19% for GJT; n = 35, 42% for fundoplication), fluoroscopic evaluation of feeding and swallowing or videofluoroscopic feeding study (n = 30, 36% for GJT; n = 21, 25% for fundoplication) or pH probes (n = 12, 14% for GJT; n = 24, 29% for fundoplication) (Supplemental Fig 4). Aside from feeding and swallowing studies, all other diagnostic tests were more commonly ordered before fundoplication compared with GJT.

Respondents were asked about the perceived risks associated with fundoplication. Most were concerned the procedure would be ineffective (n = 60, 72% moderately or very concerned) or cause worsening GERD symptoms (n = 67, 81% moderately or very concerned) (Fig 2). For GJT, the highest levels of concern were regarding the need for continuous feeding (n = 62, 75% moderately or very concerned), need for emergent GJT changes (n = 60, 72% moderately or very concerned), and major late complications such as intussusception (n = 52, 63% moderately or very concerned) (Fig 3).

The majority of respondents indicated that it was very easy or somewhat easy (n = 57, 77%) to access GJT although one-quarter had difficulty accessing this procedure. Only half reported that it was very or somewhat easy (n = 39, 51%) to access fundoplication at their institution. Most respondents (70%) were concerned about the ability of patients or families to access tertiary care sites for either routine or emergent GJT care.

This study provided a broad overview of the opinions and reported-practice of tertiary-care based physicians across Canada with respect to the use of ARPs among children with NI and gastrostomy tubes. A substantial majority of respondents were pediatricians practicing in hospital medicine or complex care, corresponding to the higher numbers of these specialists in Canada. Our results, thereby, predominantly reflect the perspectives of this group. We discovered considerable practice variability with no clear relationship to subspecialty, experience, or location of practice, although we may not have achieved sufficient power to detect these differences. Some responses reflected issues with access to ARPs because of a variety of geographic barriers (distribution of population) and resource limitations (surgeon or radiologist availability), which likely also impact physician practice. The scenario considered most amenable to ARP was scenario 3 (FTT), which is consistent with available guidelines recommending ARP for refractory GERD symptoms and failure of intragastric feeds.^3,16  Even still, nearly one-quarter of respondents would not offer an ARP in this scenario, raising questions about guideline awareness and adherence. The scenario that had the lowest likelihood of ARP was scenario 2 (Mod Asp), which presented a child with chronic lung disease and infrequent episodes of aspiration pneumonia. Interestingly, scenarios 1 (Severe Asp) and 4 (Pain) were evenly split in in terms of likelihood to consider ARP. To some degree, this variability may reflect the lack of evidence available to guide clinical decision-making, which is an established contributor to practice variability.¹⁷  No randomized controlled trials have established the effectiveness of GJT or fundoplication compared with nonsurgical management in children with NI and gastrostomy feeds. Observational studies have mainly conducted head-to-head comparison of GJT and fundoplication,^14,18  including a meta-analysis (2015) finding no difference in mortality or frequency of pneumonia.¹⁹  Interestingly, most physicians did not consider ARPs to be second-line treatment, instead suggesting many other interventions they would attempt before ARP. For instance, many would trial prokinetics before ARP even though existing guidelines do not recommend prokinetic agents for treatment of GERD (except in severe refractory cases).^3,9 

Respondents rated high levels of comfort with ARPs even though their reported practice varied considerably. We hypothesized that practices were often concordant within an institution and/or subspecialty but varied between them. Internally consistent practices could result in higher levels of physician comfort even with a lack of evidence. However, this association was not confirmed in our results. Other research suggests a dissociation between physician-reported comfort and clinical decision-making.^20,21  It would thereby be possible to have comfort even if decision-making was not optimal or evidence-based. We cannot exclude the possibility that some variation could be because of different interpretations of the clinical scenarios. However, our results are supported by prior studies that report similar degrees of practice variation in the treatment of GERD. A large cross-sectional study in the United States revealed that the odds of undergoing an ARP in the southwest were twice that of the northeast, even after adjusting for age, case-mix, and case-volume.²²  Another study by Goldin et al reported changing rates of ARPs in relation to common procedures, such as appendectomies, at 36 children’s hospitals in the United States between 2001 and 2006. The overall decline in ARPs was unequally distributed with institutional rates that ranged from 40 fewer per year to 10 greater per year in comparison with 2001.¹¹  A national survey of pediatric surgeons also found that about 80% of them required an upper gastrointestinal series before ARP and only 13% required a pH probe, similar to the proportions in our study.²³ 

Existing evidence cannot determine who would benefit from an ARP, particularly in children with NI and other complex needs. There is a need for future research including large prospective, comparative studies that can determine crucial outcomes, such as impact on respiratory morbidity and GERD symptoms. Another area that is under-explored is the impact of ARP on child and family quality of life. One small prospective study suggested improvement with fundoplication in the short term.²⁴  Another interview study with parents whose child had undergone ARP better reflected the complexity of these scenarios, highlighting how feeding tolerance, caregiving requirements, continuous feeding, and changes in health care utilization could also mediate the impact of these procedures on quality of life.²⁵  It is also concerning that physicians may be under-appreciating and, thereby, under-treating GERD symptoms, particularly in scenario 4 (Pain). Existing evidence suggests that pain and other distressing symptoms are often overlooked in children with NI and other complex conditions,^26–28  whereas physicians may be more comfortable making decisions in the face of “objective” evidence, such as episodes of aspiration pneumonia or growth failure (as in scenarios 1 and 3).

Even as we strive to build a stronger evidence base informing the use of ARPs, the clinical issues and dilemmas persist. The complexity of these children’s care and the nonspecific nature of these symptoms alongside social and logistical variables complicates the development of guidelines and standardization on the use of ARPs. Intentional shared decision-making (SDM) is a valuable approach that could acknowledge the limitations of current evidence and allow for patient and parent values, goals, and preferences to be emphasized and incorporated.²⁹  One framework suggests that for medical decisions where there is more than 1 reasonable option without a clearly favorable benefit to burden ratio, an in-depth exploration of family preferences can ensure that child and parent values align with the chosen option.³⁰  Children with medical complexity (including NI) are less likely to be engaged in SDM,^31,32  potentially because of the difficulty predicting outcomes in the setting of multiple comorbidities. However, these parents appreciate the complexity of these decisions and highly value their own involvement.³²  Uncertainties in the decision for ARP and the child’s course should be acknowledged, which can further facilitate parents’ participation and comfort.³² 

Although 70% of individuals indicating eligibility were included, this study enrolled a relatively small sample skewed largely toward pediatric hospital medicine and complex care physicians, limiting generalizability to other subspecialties. These nuanced and complex clinical scenarios are difficult to describe in a brief vignette. The diagnosis of GERD may not have been obvious, but vignettes were intentionally designed to reflect “real world” scenarios where formal diagnostic testing is seldom used. The survey underwent a structured, iterative process of development, including cognitive interviewing but not formal clinical sensibility testing, which would have strengthened our findings. The style of the questions may have inhibited respondents from communicating more details about their perspectives, although there was space for written comments after each question. The distribution of responses, with some scenarios having more positive responses for ARP, suggests that the questions could discriminate different practice styles. However, it remains possible that some of the variation in responses is related to different interpretations of the clinical scenarios. Other treatment options for GERD were not addressed in this study (such as surgical jejunostomy tube) as they are infrequently used in Canada. No concerns about this omission were raised in the cognitive interviews.

With the lack of evidence regarding the impact of ARPs on clinical and patient-centered outcomes, considerable variability exists in physician perspectives and reported practice. Strengthening the evidence base with prospective studies should be prioritized. As with many complex medical decisions, SDM would acknowledge the limitations in our knowledge and best incorporate patient and family preferences, values, and goals when approaching decisions about ARP.

We thank Dr Brent Hagel, Dr Tamara Simon, Dr Marialena Mouzaki, and Dr Sarah Lai for their assistance in developing and testing the questionnaire; and the contributions of our survey respondents.