An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Kathy Goggin, PhD, Emily A. Hurley, PhD, MPH, Andrea Bradley-Ewing, MPA, MA, Brian R. Lee, PhD, Carey Bickford, BA, Kimberly Pina, MPH, Evelyn Donis de Miranda, BHS, Alexander Mackenzie, BS, Kirsten B. Delay, Sarah Schlachter, MEd, Areli Ramphal, LMSW, David Yu, MD, Kirsten Weltmer, MD, Sebastian Linnemayr, PhD, MPhil, MA, Christopher C. Butler, MD, Jason G. Newland, MEd, MD, and Angela L. Myers, MD, MPH.
Author Information and AffiliationsIn the United States, children with acute respiratory tract infections (ARTIs) are prescribed at least 11.4 million unnecessary antibiotic prescriptions annually. One chief contributor to overprescribing is inadequate parent–provider communication, born out of providers' misconceptions about parents' desire for antibiotics. However, efforts to reduce overprescribing have only indirectly targeted communication or been impractical to implement in US settings because of the time required for providers to explain why antibiotics are not necessary and are potentially harmful.
The aim of this study was to compare 2 feasible interventions for enhancing parent–provider communication (lower-intensity communication vs higher-intensity communication) to reduce the rate of inappropriate antibiotic prescribing among children with ARTIs.
This mixed-methods study consisted of a cluster-randomized comparative effectiveness trial at an academic children's hospital and private practice as well as qualitative assessment methods. Using a 1:1 randomization, we assigned 41 providers (pediatricians and nurse practitioners) to receive a higher-intensity or lower-intensity intervention. All providers received a 20-minute, in-person general education training on the pros and cons of antibiotics, the impact of inappropriate use, antibiotic-prescribing guidelines, and common reasons for antibiotic misuse. Providers in the higher-intensity arm received an additional 50-minute, in-person training to enhance their confidence in the use of parent-centered communication strategies (eg, open-ended questions, affirming and eliciting parents' thoughts/concerns) and the study educational trifold brochure during consultations. We then enrolled 1600 parent–child dyads (children aged 1-5 years; higher intensity, n = 696; lower intensity, n = 904) in which the children presented with ARTI symptoms to a provider trained to use 1 of the 2 interventions. Before their consultation, all parents completed a baseline survey and viewed a 90-second gain-framed antibiotic educational video. In addition, parent–child dyads consulting with providers trained in the higher-intensity intervention received a gain-framed educational trifold brochure promoting cautious use of antibiotics and rated their interest in receiving an antibiotic, which was shared with their provider before the visit.
Data for the primary outcome (rate of inappropriate antibiotic prescribing) were garnered from independent medical record reviews conducted by physicians according to diagnostic and treatment guidelines. Secondary outcomes (quality of parent–provider communication, shared decision-making, and visit satisfaction) were collected via parent surveys immediately following the visit as well as by a 2-week follow-up telephone call with parents to assess revisits (ie, additional clinic visits for the same illness, not including scheduled follow-ups) and adverse drug reactions. Due to the 2-stage nested design (parents nested within providers and clinics), we employed a generalized linear mixed-effect regression model (GLMM) to compare intervention outcomes. We also conducted and analyzed qualitative interviews with providers following the intervention period.
The overall rate of inappropriate prescriptions among all enrolled patients was similar among those who consulted with a higher-intensity (7.8%) vs a lower-intensity (9.4%) provider (difference = −1.6%; 95% CI, −4.4% to 1.1%). In the GLMM, we found that the odds of receiving inappropriate antibiotic treatment did not vary significantly between intervention arms. Secondary outcomes of revisits and adverse drug reactions did not vary between arms, and parent ratings of shared decision-making, satisfaction with quality of parent–provider communication, and visit satisfaction were similarly high in both arms. Provider satisfaction with the interventions was high, with all stating they would recommend the program to others.
The higher-intensity intervention (which consisted of additional patient-provider communication training and support materials) did not produce a meaningful difference in inappropriate antibiotic prescription compared with the lower-intensity arm. Qualitative and survey results suggest that implementing these 2 evidence-based interventions is feasible in pediatric outpatient settings.
As a comparative effectiveness trial, our study is limited in that it cannot determine the effectiveness of either intervention compared with baseline or usual care. Another limitation was the uneven distribution of patients across sites and providers, possibly reducing the power to detect a difference between intervention arms.
Note for the reader: The Background section is reproduced from: Goggin K, Bradley-Ewing A, Myers AL, et al. Protocol for a randomised trial of higher versus lower intensity patient-provider communication interventions to reduce antibiotic misuse in two paediatric ambulatory clinics in the USA. BMJ Open. 2018;8(5):e020981. doi:10.1136/bmjopen-2017-020981 [PMC free article: PMC5942422] [PubMed: 29743330] [CrossRef]
Antibiotic overuse and misuse contribute to the development of antibiotic-resistant infections that if left unchecked are estimated to cause 10 million deaths yearly worldwide by 2050.1 In the USA, antibiotic-resistant infections are responsible for at least 23 000 deaths and an additional 2 million infections annually.2 Inappropriate antibiotic use also increases incidence of antibiotic-associated adverse drug reactions (e.g. rash, diarrhea, nausea and vomiting), which result in >140,000 emergency department visits every year.3
The majority of all antibiotic prescribing in the USA occurs in the outpatient setting4 where children receive 49 million prescriptions annually.5 Children with acute respiratory tract infections (ARTIs) receive >70% of these prescriptions of which at least 29% are unnecessary (i.e., either to treat a presumed viral illness or an unnecessary broad-spectrum antibiotic).5 Despite some improvements, the most recent estimates suggest that antibiotics are prescribed for approximately 50% of ARTIs while it is estimated that only 27% of ARTIs are caused by bacterial infection.6 As a result, children are receiving up to 11.4 million unnecessary antibiotic prescriptions annually.6 Strikingly, an almost identical number was noted in a similar study conducted 16 years earlier (11.1 million unnecessary antibiotic prescriptions), suggesting there are considerable gains still to be made in reducing inappropriate use.7
Inappropriate antibiotic prescribing in the ambulatory setting has many causes, but the interaction between parents/legal guardians (hereafter referred to as parents) and providers is central. For their part, some parents still harbor misconceptions that make them think antibiotics are necessary when they are not.8 Nevertheless, parents generally desire antibiotics for their children only when absolutely necessary9 and do not expect antibiotics for common colds.10 Instead, parents become dissatisfied when providers minimize children's symptoms, fail to acknowledge parents' appropriate concerns and/or do not offer a contingency plan if symptoms fail to resolve.11,12
Despite evidence to the contrary, providers perceive significant parental pressure for antibiotics and fear damaging the parent–provider relationship if they withhold prescriptions.13,14 Combined with the ever-increasing time constraints and focus on parent satisfaction ratings inherent in modern clinical practice, these beliefs greatly contribute to ineffective parent–provider communication about antibiotics. When providers perceive that a parent expects or hopes for an antibiotic, they are more likely to prescribe one.15,16 In a study of children with viral ARTIs where no prescription should have been given, providers gave a prescription to 52% of parents they believed were expecting an antibiotic compared with only 9% of parents who they believed were not expecting an antibiotic.17 Adding to this problem is providers' mistaken belief that they can accurately predict parents' desires. In fact, providers' ability to accurately predict parents' expectation for an antibiotic is significantly worse than chance at 24%–41% concordance.13,17 Even though parents rarely state a desire for antibiotics (1% of the time in clinical recordings), providers report frequent parent demands for antibiotics.18 Providers also mistakenly believe that meeting perceived parental expectations for antibiotics is necessary for parent satisfaction.14 Parental satisfaction, however, is not related so much to whether or not they receive an antibiotic but more to the quality of communication with their provider.11,14 In fact, a recent observational study demonstrated that the use of what they termed ‘positive treatment recommendations’ (i.e., comfort care) plus ‘negative treatment recommendations’ (i.e., antibiotics will not help) was associated with the highest parent satisfaction.19 Current efforts to improve appropriate antibiotic use have only indirectly targeted parent–provider communication20,21 or have been found to be impractical.22 As described in several meta-analytic and systematic reviews, interventions have typically focused on education about antibiotics for providers or patients.20,21 While many have been successful in increasing knowledge about antibiotics and nationally antibiotic prescribing has evidenced modest reductions,9,23 more effective strategies that go beyond attempts to educate are needed to reduce overprescribing rates to levels that will have a significant impact. A limited number of studies have targeted parent–provider communication or shared decision-making, and they have already produced greater reductions in antibiotic prescribing than studies that focused only on increasing knowledge.21 Of the communication interventions tested, only one has directly targeted the provider's perceptions of parental expectations alongside antibiotic education and shared decision-making.24 This past study, which employed intensive provider training and a multipage patient–provider interactive educational booklet, resulted in a significant decrease in antibiotic use in a UK sample. The intervention, however, was viewed as burdensome by providers and impractical for most real-world settings.22 Effective, practical interventions are needed that address provider misconceptions about parent expectations, facilitate shared-decision making and improve aspects of communication that are most likely to increase parental satisfaction.
Specific Aim 1: Conduct a multisite, parallel group, cluster-randomized comparative effectiveness trial to compare the effect of two feasible interventions for enhancing parent–provider communication on the rate of inappropriate antibiotic prescribing (primary outcome).
Specific Aim 2: Compare the impact of each intervention on the following secondary outcomes: rate of revisits (ie, additional clinic visits for the same illness, not including scheduled follow-ups), adverse drug reactions, and parent–provider communication and visit satisfaction.
The study design was continuously guided by input from a 15-member community advisory board (CAB), inclusive of parents, providers, and community stakeholders (47% Hispanic, 20% African American, 20% exclusively Spanish-speaking). The CAB was majority female (80%), closely reflecting the proportion of parents attending outpatient pediatric visits with their children in our study setting (our final study sample of parents was 84% female). CAB members were recruited from patient and family advisory councils and local community service organizations serving Latinx and other ethnically diverse communities. CAB members participated in 12 meetings over a 3-year period that spanned project development, implementation, and analysis and were cofacilitated by the parent research associate (Ms Carey Bickford) and a study team member (Ms Andrea Bradley-Ewing, MPA, MA). CAB meetings were held on a monthly, then quarterly basis and alternately conducted in English and Spanish with live translation provided. All meetings were conducted in person, with optional teleconference capabilities available. CAB members chose the location, day, and time of each meeting to best accommodate their schedules. In addition to the CAB, the parent research associate was an integral member of the project team, attending biweekly project meetings to ensure parents' perspectives were integrated into all aspects of the project. Additional stakeholders (eg, physicians, nurses, health literacy teams, administrators, parents, and children) were also consulted to assist in the development of study materials and maximize the extent to which the study design and procedures could be integrated into normal clinic flow.
During the project development phase, CAB members worked collaboratively with the research team to develop and refine study procedures and materials, including recruitment, consent, and data collection materials. Members of the research team drafted initial versions of the study procedures and materials and shared them with CAB members who provided feedback. CAB members repeatedly emphasized the importance of taking the time to fully explain the research project and providing materials with terms that patients and parents could easily understand. Additionally, we consulted with a health literacy team consisting of physicians, nurses, and social workers who provided recommendations to ensure study materials were appropriate for parents with various reading, numeracy, and comprehension abilities. As a result, all study materials, including a 1-page study information flyer, 1-page consent form, and educational materials, were designed to be patient- and family-centered as well as culturally appropriate, and to include information that CAB members believed would be most important to parents. CAB members provided input on data collection procedures and instruments, highlighting the need to ensure survey questions assessed parents' primary concerns, were brief and comprehensive, and did not delay health care for the patient. This input reflects the selection of CollaboRATE and the 9-item Shared Decision-Making Questionnaire (SDM-Q-9).
CAB members also collaboratively developed parent and provider educational tools. Provider tools included a video message from parents describing what they value most during their consultations (eg, respect, reassurance, a plan of action to treat their child's symptoms), as well as a communication training video simulating an ideal consultation from a parent's perspective. Parent-focused educational tools included a 90-second, gain-framed animated video (Appendix A) featuring a young child with an upper respiratory infection and information on appropriate use of antibiotics. Gain framing presents a situation or outcome in terms of the gains or positive aspects (in this case, the gains a child and family will enjoy if they are not given an antibiotic when it is not needed). Parents in the higher-intensity intervention arm also received a gain-framed25 educational trifold brochure detailing when antibiotics are indicated and when they are not, as well as specific information about their child's diagnosis, including at-home care to improve symptoms and a contingency plan if their child's symptoms did not improve. All parent-focused materials were translated by bilingual/bicultural research staff and CAB members using a modified version of the International Society of Pharmacoeconomics and Outcomes Research Principles of Good Practice for Translation and Cultural Adaptation.67 Our 9-step translation process included preparation, forward translation, reconciliation, back translation and review, cognitive debriefing, finalization, proofreading, and creation of a final report (Appendix B).
During the implementation phase, CAB members provided input on strategies to strengthen participant recruitment and retention as well as modifications to study materials and procedures. They gave feedback on barriers that the study was experiencing (eg, reduced enrollment) and received timely updates on study progress. Input provided by parents and stakeholders helped improve study logistics, including maximizing participant recruitment and retention efforts by introducing the study in private examination rooms and providing text message reminders for follow-up surveys to improve retention among parents who were unreachable by phone.
Our engagement process added indispensable value to the design and implementation of our study. Further, engaging both English- and Spanish-speaking parents and community stakeholders allowed the development of novel, patient-centered, and culturally appropriate study materials and resources. In addition to this study, input from CAB members impacted polices within our institution, including translation (recognizing the 9-step translation process implemented in this study as the gold standard for the institution) and research participant payment policies (no longer requiring participants to provide a Social Security number in order to receive study incentive payments of <$600 annually, which included all participants in our study). We will plan to continue engaging CABs in our research. Additionally, this project has fomented interest in community and patient engagement among researchers and care providers across our institution.
Other positive impacts of stakeholder and family engagement included identifying nontraditional dissemination channels to share the study with parents and community members. These channels included the International Children's Advisory Network, Inc (iCAN) Summit presentation, a nonacademic conference that allowed us the opportunity to share the study engagement process with representatives of patient and parent advisory groups from around the world. Additional presentations included meetings with our institutional patient and family advisory councils (PFACs), including a PFAC for Spanish-speaking families and a regional PFAC of parents of medically complex children.
Some challenging aspects of patient/stakeholder engagement included engaging both English- and Spanish-speaking parents on the CAB. Additional staff time and resources were needed to develop CAB meeting materials (ie, agenda/presentation/mail reminders) and study materials in both English and Spanish. Additional staff and CAB member time was required to conduct meetings in English and Spanish while allowing time for translation. While partnering with parents and stakeholders who speak different languages was challenging, the results of this engagement led to the development of high-quality, parent-centered, culturally appropriate educational resources and promoted co-learning among CAB members as well as between the study team and CAB members.
Note for the reader: Certain subsections (Study Setting, Participants, Interventions, Study Outcomes, Sample Size and Power, Data Collection and Sources, Analytical and Statistical Approaches) are adapted from the study's published protocol: Goggin K, Bradley-Ewing A, Myers AL, et al. Protocol for a randomised trial of higher versus lower intensity patient-provider communication interventions to reduce antibiotic misuse in two paediatric ambulatory clinics in the USA. BMJ Open. 2018;8(5):e020981. doi:10.1136/bmjopen-2017-020981 [PMC free article: PMC5942422] [PubMed: 29743330] [CrossRef]
The aim of this study was to compare 2 feasible interventions for enhancing parent–provider communication to reduce the rate of inappropriate antibiotic prescribing. To achieve this aim, we conducted a multisite, parallel-arm, cluster-randomized comparative effectiveness trial in 2 pediatric outpatient clinics, with a balanced randomization of providers assigned to the higher-intensity intervention (prescribing education and communication skills training) and the lower-intensity intervention (prescribing education only). We hypothesized that compared with parent–child dyads managed by providers trained in the lower-intensity intervention, parent–child dyads managed by providers in the higher-intensity intervention would have lower rates of inappropriate antibiotic prescribed, as well as superior secondary outcomes of revisits, adverse drug reactions, and parent–provider communication and satisfaction. Parent–child dyads enrolled in the study were exposed to either the higher or lower-intensity intervention according to the provider conducting their clinic visit (see Figure 1 for flow of study participants). Data for the primary outcome (rate of inappropriate antibiotic prescribing) were garnered via medical record reviews. Secondary outcomes (quality of parent–provider communication, shared decision-making, and satisfaction) were collected via parent surveys immediately following the visit as well as by a 2-week follow-up telephone call with parents (revisits and adverse drug reactions). Parents' and providers' views of the acceptability and feasibility of the interventions were also assessed.
Schematic Diagram of Parent–Child Dyad Participant Flow.
We selected 2 ambulatory pediatric clinics as study sites: (1) an academic medical facility (Children's Mercy Hospital Primary Care Clinics [CMH PCC]) and (2) both locations of a private practice (Heartland Primary Care [HPC]). CMH PCC serves patients within the Kansas City metropolitan area, of which 41% identify as African American/Black and 29% identify as Hispanic. Of our patient sample at CMH PCC, 68% had public insurance, 23% had private insurance, 7% were self-paying, and 2% had military/other insurance. HPC provides pediatric care in 2 settings in suburban Kansas City, with a patient population that is 14% African American/Black and 16% Hispanic. Of our patient sample at HPC, 39% had public insurance, 55% had private insurance, and 6% had military insurance. At both study sites, approximately 20% of parents were Spanish speaking. CMH PCC treats approximately 2100 children per year with ARTIs, and HPC providers care for roughly 2000 children per year with ARTIs. We chose these sites based on patient volume (clinic records revealed an adequate yearly rate of appointments with children meeting our study eligibility criteria), as well as the opportunity to compare differential effects of the interventions in an academic vs private practice with highly similar patient populations.
We chose to randomly assign providers rather than clinics, as providers tend to have more consistent beliefs and behaviors within their practice than with other providers in other practices.26 In addition, the risk of contamination was small, as intervention components were designed to not be easily transferred between providers.
We screened all providers at CMH PCC and HPC for eligibility. Inclusion criteria required providers be a pediatrician (medical doctor [MD] or doctor of osteopathic medicine [DO]) or nurse practitioner (certified pediatric nurse practitioner [CPNP] or advanced practice registered nurse [APRN]) who would be actively and independently conducting outpatient consultations with our target patient population during the study period. Providers who primarily saw patients at urgent care clinics or specialty clinics or who were assigned to administration during the study period were excluded.
Patient visit data from the 6 months before randomization were used to assign each provider to a large- or small-patient-volume arm. The study statistician then stratified the randomization of providers to ensure each study arm was balanced across large and small volume providers and across clinics. This randomization was conducted before provider recruitment and consent. Eligible providers were recruited during brief study orientation sessions held between January and April 2017. During the orientation, interested providers completed informed consent and were given a sealed envelope prepared by the study statistician that contained their intervention arm assignment.
Recruitment and enrollment of parent–child dyads began in March 2017 and continued through March 2019. During this period, study staff prescreened potentially eligible parent–child dyads using the clinics' electronic medical record (EMR) system each morning and periodically throughout the day. Dyads were eligible if the patient was between ages 1 and 5 years (ie, before sixth birthday), presented with any ARTI symptoms (eg, cough, congestion, difficulty breathing, sore throat, ear ache), and his/her parent was able to give informed consent and perform study tasks in either English or Spanish.26 Children were not eligible if they had received an antibiotic in the last 30 days; had a concurrent probable bacterial infection (eg, urinary tract infection, soft tissue infection); had known immunocompromising conditions (eg, HIV, malignancy, solid-organ transplant, chronic corticosteroid use); or had factors that make shared decision-making around prescribing an antibiotic extremely complex, such as children with complex chronic care conditions (eg, cystic fibrosis)27 or who require hospitalization during the visit. We included patients with penicillin allergy, because shared decision-making with this group is especially important given more limited treatment options. Parents or children who had previously participated in the study were not eligible to participate again.
Potentially eligible dyads were identified by prescreening all appointments, and parents were given a study flyer during check-in. Clinic staff greeted potentially eligible dyads in the examination room before the provider arrived, gave a short synopsis of the study, and offered eligibility screening. If >1 caregiver was with the child, staff asked them to designate 1 person who would independently complete the informed consent and all assessments. Providers played no role in identifying, screening, or consenting potentially eligible dyads, or in collecting data. Parent–child dyads were unaware which arm their providers had been enrolled and trained in. Eligible parent–child dyads were enrolled and consented before completing baseline assessments and proceeding to their consultation. According to normal clinic procedures, Spanish-speaking parents either saw providers with an interpreter or saw providers who were fluent in Spanish. Parents were offered a $10 gift card for completion of each of the 3 surveys (baseline, postconsultation satisfaction, 2-week follow-up) for a total maximum compensation of $30.
The intervention was informed by a series of evidence-based interventions conducted in the United Kingdom and Europe: Enhancing the Quality of Information-sharing in Primary care (EQUIP),24,28 Improving the Management of Patients with Acute Cough Trial (IMPACT),29,30 Stemming the Tide of Antibiotic Resistance (STAR), and Genomics to combat Resistance against Antibiotics in Community-acquired lower respiratory tract infection in Europe (GRACE).31
Providers randomly assigned to this arm received 2 trainings. The first was a 20-minute, in-person general education training provided by study physicians (A.M. and J.N.) which covered common reasons for antibiotic misuse;5,32 pros and cons of antibiotics33; the impact of inappropriate use of antibiotics6; common respiratory conditions for which antibiotics are prescribed, including sinusitis, acute otitis media, community-acquired pneumonia, and streptococcal pharyngitis (strep throat); and viewing/discussing the 90-second antibiotic educational video (described below and in Appendix A). Didactic and interactive learning strategies were employed to review the national published guidelines, which included the appropriate diagnostic criteria to help distinguish a viral ARTI from a bacterial ARTI, as well as the recommended narrow-spectrum antibiotic for bacterial ARTI.34-39
Second, providers randomly assigned to the intensive intervention received a 50-minute, in-person training on parent-centered communication skills provided by a behavioral psychologist (K.G.). The training used a variety of educational strategies, including lectures, group discussions, and viewing/discussion of motivational and role model videos. The goal was to enhance providers' confidence in the use of parent-centered communication strategies (eg, asking open-ended questions, affirming and eliciting parents' thoughts/concerns) and the study gain-framed educational trifold brochure to conduct key aspects of the EQUIP/IMPACT/STAR/GRACE interventions during consultations. These providers learned to (1) elicit parents' expectations, (2) affirm parents' concerns, (3) provide an evidence-based estimate of likely illness duration, (4) provide gain-framed (ie, highlighting the positive aspects of a decision in more detail before presenting the downsides) antibiotic information, (5) recommend options for symptom relief, (6) identify triggers for reconsultation and contingency plans, and (7) elicit parents' thoughts on the plan. These target communication behaviors were meant to be integrated into the regular consultation and not to add significant time to the visit (which we assessed via audio recordings in our fidelity monitoring). Providers also learned to use the study trifold brochure to ensure that they completed all necessary aspects of the intervention and to provide written notes for parents to refer to after the visit. The study trifold brochure provided gain-framed information about when antibiotics are and are not necessary and what risks are involved in taking antibiotics. Research has shown that people react to the same trade-off in different ways depending on whether the possible outcomes are presented as losses or gains.40 In this study, we trained providers and tailored our parent materials to highlight the gains of not using antibiotics (eg, staying safe from adverse effects, making sure that effective cures are available in the future, knowing that their child's body will fight off most ARTIs on its own) in order to increase parents' comfort with not receiving an antibiotic prescription for their child. Drawing from the EQUIP study,28 the outside of the brochure includes a place to write the child's first name; check boxes to indicate the diagnosis, recommended home care treatments, and reasons for consultation; expected recovery time, if antibiotics are needed; and tips for communicating with providers.
To reduce their reliance on guessing what parents want, providers were trained to rely on parents' antibiotic desire ratings, which were taken after parents viewed the video and were provided at the start of each visit via a sticky note that study staff placed on the door of the consultation room. To assess fidelity to the communication skills, audio recordings were made for a randomly selected subsample of visits (10%) in both the higher- and lower-intensity arms. These audio recordings were objectively coded by raters blind to the intervention arm for use of key communication strategies using established methods that we have successfully employed in other studies.41,42 We delivered in-person provider training, because studies have shown the value of an active approach over more passive web-based versions,43 and we also conducted yearly refresher trainings during the study period.
In examination rooms before the consultation, parents completed the baseline survey, viewed the video with accompanying brochure, and rated their desire for antibiotics via a tablet computer. The video used gain-framed messages40,44 to explain when antibiotics are and are not indicated while emphasizing the risk of adverse effects and the creation of resistant organisms. It also highlighted what information should be provided during the consultation (eg, an estimate of illness duration, recommendations for system relief and triggers for reconsultations and contingency plans). Parents in this arm also viewed the educational portion of the brochure (ie, the inside) on the computer tablet.
This intervention was modeled on proven parent- and provider-focused educational interventions used in previous studies.20,36,43,45-52 Providers completed the same 20 minute, in-person general education training described previously. Parents received the same parent training described previously except that they did not receive a hard copy of the study brochure and their antibiotic desire ratings were not shared with providers.
Several measures were taken to reduce the likelihood of contamination between arms. Specifically, (1) we trained study team members to ensure that all of their communications (written or in person) with providers in the lower-intensity arm did not reveal any of the strategies from the higher-intensity training; (2) as part of our training for providers, we reviewed the importance of keeping intervention arms distinct in randomized controlled trial designs; (3) we directly asked higher-intensity providers to pledge not to share any details of the additional communication skills training with their colleagues randomly assigned to the lower-intensity arm; (4) we controlled the dissemination of the brochure to ensure that only parents who were consulted by providers in the higher-intensity arm received it; and (5) we offered communication strategies for dealing with requests from lower-intensity-arm colleagues who ask for more information.
Our primary outcome was rate of inappropriate antibiotic prescribing. We hypothesized that the rate of inappropriate antibiotics prescribed by providers in the higher-intensity arm would be lower than the rate produced by providers in the lower-intensity arm. Inappropriate prescribing was assessed by study physicians, blinded to study arm, who reviewed the medical record documentation for each enrolled patient's visit to determine whether inappropriate antibiotic prescribing had occurred. Consistent with the norm in the antibiotic stewardship literature, we excluded topical antibiotics in our overall count of antibiotics administered and our assessment of what was an inappropriate antibiotic. Prescriptions were considered inappropriate if they met any of the following criteria: (1) antibiotic prescribed for an apparently viral ARTI, (2) antibiotic prescribed for a presumed bacterial ARTI that does not meet Table 1 criteria, (3) broad-spectrum antibiotic prescribed for a bacterial ARTI in a child without a penicillin allergy, or (4) nonrecommended alternative antibiotic prescribed for a bacterial ARTI (see Table 1) in a child with a penicillin allergy. We determined penicillin allergy status by reading the patient's EMR.
Appropriate Antibiotic Selection (Primary Antibiotic: Amoxicillin).
Instead of relying on diagnostic codes for the index illness as has been done in previous studies,53,55 the study physicians assessed the appropriateness of the patient's diagnosis by reviewing detailed symptoms, physical examination findings, and diagnostic tests in the EMR. This guarded against the potential bias of relying on diagnostic codes alone, as clinicians sometimes assign diagnostic codes to support their antibiotic prescribing.12 Children thought/suspected to have a bacterial infection needed documentation of the specific diagnoses and the clinical criteria confirming that diagnosis (listed in Table 2). Ten percent of all EMR reviews were verified by the other study physician blinded to the initial coding and study arm. Study physicians' initial coding of appropriate vs inappropriate antibiotic treatment disagreed in 8.9% of cases. Nevertheless, all disagreements were discussed and reconciled throughout the study to ensure consistency in coding across coders. All EMRs with a diagnosis of acute otitis media were reviewed by both study physicians.
Diagnostic Criteria for ARTIs.
We determined whether children seen by providers in the 2 study arms differed in terms of revisits and/or adverse drug reactions. Data on these clinical outcomes were collected via follow-up phone calls with parents 2 weeks after the visit. Parents were asked whether any additional health care visits and/or treatment occurred and whether antibiotics were given to the child. They were also asked whether any adverse effects or adverse drug reactions occurred. Parents were asked to report on when their child's symptoms improved, whether contingency prescriptions were filled, use of home care treatment suggested by the provider, assessment of the video and brochure, and satisfaction with study participation.
We assessed parent ratings of shared decision-making using an adapted version of the 3-item CollaboRATE questionnaire.56 This very brief (<30-second) scale was developed with the input of end users and assesses the effort that providers put forth to initiate shared decision-making. Members of our CAB and participants in several studies have strongly preferred the CollaboRATE scale to other measures of shared decision-making, especially for more routine health care issues.57 Items are “How much effort was made to: (1) help you understand your child's health issue? (2) listen to the things that matter most to you about your child's health issues? and (3) include what matters most to you in choosing what to do next?” Each item is scored on a 10-point response scale ranging from 0 = “no effort was made” to 9 “every effort was made.” In a simulation study, the CollaboRATE scale demonstrated discriminative validity between 6 standardized patient–provider encounters that included varied amounts of shared decision-making, concurrent validity with other measures of shared decision-making, excellent test-retest reliability, and sensitivity to change.58
We used a single item: “How satisfied were you with the communication between you and your child's health care provider?” with a 5-point Likert-type response format ranging from 1 = “very dissatisfied” to 5 = “very satisfied.”
We used a single item: “Overall, how satisfied were you with the visit?” with a 5-point Likert-type response format ranging from 1 = “very dissatisfied” to 5 = “very satisfied.”
All eligible providers at the study sites were invited to participate. A total of 1600 parent–child dyads were recruited based on inclusion criteria. Prior research examining our primary outcome showed 30% of the antibiotics prescribed in the outpatient ARTI visits were inappropriate.6 Prior behavioral intervention studies have produced 20% to 81% reductions in inappropriate prescribing,53,54 with statistically significant differences between intervention and control arms.54 Based on the intraclass correlation coefficient (ICC) observed in the Meeker et al study,54 which is most similar to our study, we assume an ICC of 0.04. By assuming 30% inappropriate prescribing at baseline, a 20% decrease in the lower-intensity arm, and a conservative 50% decrease in the higher-intensity arm following intervention with 40 providers (clusters), as well as an α of .05 and 80% power, we needed a sample size of 760 per arm to detect a 9% difference between arms (inappropriate antibiotic prescribing 24% in the lower-intensity arm vs 15% in the higher-intensity arm after intervention). Consistent with our historical retention rates in similar studies in the same setting, we chose to protect against an attrition rate of 5% and aimed to recruit 1600 participants to ensure adequate power to assess our primary and secondary outcomes.
As stated previously, all providers received an initial 20-minute intervention training. Providers randomly assigned to the higher-intensity arm received an additional 50-minute training. Both interventions began with parent–child dyad enrollment in March 2017 and concluded in March 2019. We also conducted yearly refresher trainings for providers in each study arm.
Parents completed the baseline survey (5 minutes) before the consultation and then a postconsultation survey after (3 minutes; see Figure 1 for flow of study participants). A follow-up survey (3 minutes) was conducted 2 weeks after the visit because this is typically the end of the expected duration for an ARTI. This duration also allowed us to assess resolution of the child's illness, additional health care visits or revisits, additional treatments if contingency prescriptions were filled, and the presence of any adverse drug reactions.
At baseline, providers completed a brief survey collecting demographic data and their views on parent interaction regarding antibiotics for viral illness, their comfort with telling parents that antibiotics are not necessary, and their concern about parents' responses. Once parent–child dyad recruitment was complete, we conducted a brief survey mirroring the baseline provider assessment and a brief (~15-minute) semistructured individual with providers to learn about their experience of being in the study, as well as to solicit suggestions for improvement and ideas about disseminating the results to other settings. Providers did not receive incentives for study participation.
Before their scheduled visit with a provider, all enrolled parents completed a baseline survey on a computer tablet. Items included demographics, questions about their antibiotic knowledge, and interest in antibiotics for their child's current condition. Parents then viewed the video, again indicated their interest in antibiotics for their child's current condition, and rated their perceived likelihood of actually receiving antibiotics during their visit. Immediately after the visit, parents were asked to complete a second tablet survey, including their rating of shared decision-making during the visit, satisfaction with parent–provider communication, and overall satisfaction with the visit. Two weeks after seeing the provider, parents were contacted by study staff via phone or email to complete the follow-up survey, which collected data for the secondary outcomes of revisits, adverse effects, and illness duration. EMR data were abstracted by the study physicians for each enrolled parent–child dyad, using a standardized data collection form, and evaluated to determine the appropriateness of antibiotic prescribing. All data were collected and stored using Research Electronic Data Capture (REDCap).59 We collected reasons for exclusion, withdraw, or refusal on a separate questionnaire form during the eligibility screening process described previously.
All analyses were conducted using an intention-to-treat (ITT) strategy. Initial analyses examined the underlying distributions of the primary and secondary outcomes. We constructed an analytic model to assess the impact of intervention type on our primary outcome of inappropriate antibiotic prescribing. This is a 2-stage nested design, with parents nested within providers (level 1 units) and study site (level 2 units). We employed a generalized linear mixed-effect regression model (GLMM) using Stata,60 which allows for easy specification of both fixed and random effects, including accommodating ≥1 cluster variables. Alternative covariance structures were investigated; as hypothesized, the exchangeable (or compound symmetry) structure sufficed. We employed robust standard errors to help minimize misspecification and to examine time as a potential random effect. The data were analyzed using a post-test only approach, as the study was not designed to look at changes in inappropriate antibiotic prescribing and no preintervention data were collected. This meant that we could not examine changes from baseline or prebaseline rates of inappropriate antibiotic prescribing by intervention arm.
Next, we examined the effects of the potential covariates (eg, parent's/patient's gender, insurance type, parent's self-reported race and ethnicity, parent's educational attainment, and provider's years of clinical experience) on the primary and secondary outcomes. Our goal was to identify parsimonious final models with the fewest covariates that most accurately described the outcomes. Additionally, we explored the heterogeneity of treatment effect (HTE), or the possibility that 1 or both interventions work better for specific groups of patients. Variables considered included clinic, provider type (physician vs nurse practitioner), provider years of experience, patient age, patient gender, parent education level, parent gender, parent race/ethnicity, parent preferred language, parent relationship to patient (mother, father, other), and parent health insurance type. We created a binary indicator for each variable and included each as an interaction term in the regression models. We examined these interaction terms across intervention arms and explored within-arm differential trends in our primary and secondary outcomes over time.
The study surveys and forms were designed with required responses and automatic skip patterns, reducing missing data within a study survey. Study staff were present when individuals were taking surveys to answer questions and to ensure that all data were provided appropriately. To prevent loss to follow-up, study staff collected multiple modes of contact from individuals enrolled in the study, and a monetary incentive was provided for the third survey that was collected by phone. Study staff scripts reiterated the number of surveys, monetary incentives, and outline of the study in clear language to help participants understand study expectations and how they would be reimbursed.
All but 1 of the 1600 patients enrolled were included in the ITT analysis of the primary outcome (the 1 excluded had missing clinical medical record data, unrelated to study procedures). Analyses for the secondary outcomes were conducted with available data from the 84% of participants who completed the 2-week follow-up survey where data for these secondary outcomes were collected. We did not impute values for participants who did not complete the 2-week follow-up (complete-case analysis). Missing survey items were rare (<1% for all variables with relevant denominator), so we did not use multiple imputation techniques. The few missing data did not exhibit any systematic patterns.
As discussed under “Patient and Stakeholder Engagement,” we conducted focus groups and interviews with clinic, parent, and community stakeholders to help adapt the interventions and study design for our targeted population. Further, we collected qualitative data for both parent and provider study participants during the follow-up periods. All parents who completed the 2-week follow-up phone interview were invited to provide qualitative, free-response data.
After patient enrollment was complete, study staff conducted in-person follow-up interviews with all active study providers. Interviews aimed to assess provider perceptions of factors that influence parents' desire for an antibiotic for their child and inappropriate antibiotic prescribing in their practices. Providers were questioned about the study design, materials, and trainings they received. The lower-intensity-arm providers were introduced to the higher-intensity materials and asked to provide feedback. All providers were asked about the feasibility of and recommendations for implementing the intervention in other practices.
All parent free responses during the 2-week follow-up calls were transcribed and later analyzed by creating a REDCap output of all parent comments organized by each standardized question. Bilingual/bicultural study staff reviewed comments and identified common keywords and salient thematic content. Comments were reviewed separately by each study staff member to classify positive and/or negative comments and themes. Provider interviews were audio-recorded, transcribed by a professional transcription service, and uploaded to Dedoose software for coding.61 We first applied structural codes to the interview transcripts based on the original topics of the interview guide.62 In the next round of coding, study staff examined all excerpts coded in each structural code and created subcodes according to emergent themes or evaluation content (eg, positive, neutral, or negative comments). Each structural code and its corresponding subcode were analyzed by 2 independent members of the study staff, who used the Dedoose Analytic Tool to compare data between participant groups of interest (eg, intervention arm, study site, type of provider). Analysis of each code was documented with an analytic memo. Senior study staff periodically examined an audit trail of the analysis (eg, codebook development, memos, analytic outputs).
Originally, our primary outcome was parent-assessed shared decision-making, and rate of inappropriate antibiotic prescribing was among our secondary outcomes. In the first year of the study while we were preparing our protocol paper for submission,26 we garnered feedback from our expert consultants and CAB about our analytic plans. Their strong advice was that rate of inappropriate antibiotic prescribing was the more robust primary outcome and the most proximal to the public health problem our intervention sought to address. Although still of high importance, shared decision-making was seen as a more fitting mediating variable. Shared decision-making is now among our secondary outcomes, along with the originally proposed revisits, adverse drug reactions, and parent-rated quality of parent–provider communication and visit satisfaction. We sought feedback and approval from our PCORI program officer, who approved this change.
All other procedures in the original study protocol remained the same, though study physicians analyzing EMRs for appropriateness of antibiotic prescribing made iterative improvements to their assessment criteria to deal with unexpected challenges such as missing information.
A total of 51 providers were screened for eligibility from the 2 study sites. Five providers were ineligible because they either did not see patients without a resident or were or would be out on leave for a long period of time during the active recruitment period. One eligible provider declined to enroll in the study, and we were unable to reach 4 others after repeated attempts to contact them. A total of 41 providers enrolled (22 randomly assigned to the higher-intensity arm; 19 randomly assigned to the lower-intensity arm). All 41 providers conducted clinic visits with enrolled participants and all data were included. Three providers left their practice during the study (all in the higher-intensity arm), so end-of-study provider follow-up surveys and interviews were conducted with the 38 providers remaining in the study when enrollment ceased.
Figure 2 illustrates the overview of participant flow for parent–child dyads throughout study enrollment. Between March 2017 and March 2019, a total of 2245 parent–child dyads were screened for eligibility. This represents dyads that passed prescreening and were approached in waiting rooms by study staff. Of the total approached, 197 dyads were excluded during the eligibility screening. This resulted in data from 978 dyads within the higher-intensity arm and 1267 in the lower-intensity arm. Most children were ineligible because they had received an antibiotic within the past 30 days, had a disqualifying chronic condition, or their legal guardian was not present at the visit. An additional 447 potential parents decided not to participate. A total of 1601 dyads were enrolled in the study; however, 1 child in the higher-intensity arm became ineligible because he was hospitalized during the visit. Data for 1 child in the lower-intensity arm was excluded from the primary analysis because while she was seen and an antibiotic was prescribed, the EMR documentation was insufficient to determine whether the prescription was appropriate or inappropriate. This left us with data from 1599 children (696 higher intensity and 903 lower intensity) for the ITT analysis of the primary outcome. Two lower-intensity dyads were lost to follow up because they left the clinic before completing the post-visit survey. This left 1597 participants (901 in the lower-intensity arm and 696 in the higher-intensity arm) that completed both the baseline and postconsultation questionnaire. Two weeks after their visit with the provider, 84% of parent participants (n = 1337; 85% higher-intensity vs 83% lower-intensity parents) completed the follow-up phone survey. After multiple attempts, we were unable to contact all of the lost-to-follow-up dyads, although none actively communicated that they wanted to drop out of the study.
Parent–Child Dyad Enrollment.
Most providers were female (78%), with two-thirds being MD/DO physicians (66%) (Table 3). Providers' median years in practice after completion of training was 8, ranging between 1 and 40 years. Three-quarters of the providers practiced at CMH PCC. The demographic characteristics of providers randomly assigned to both intervention arms were similar. In a comparison of providers who participated with those who did not, there was no significant difference in gender, and the proportion of female to male providers mirrored national norms in pediatrics.
Provider Demographics.
Table 4 displays demographics for the 1599 dyads included in the primary analysis. Overall, 84.3% of parent participants were female. Most were reported as the mother (83.1%) or father (16.1%) of the pediatric patient. Roughly 19% of parents identified as Hispanic. Most participants identified their race as White/Caucasian (78.6%) followed by Black/African American (12.1%). Parents had a range of education levels, with post–high school training or some college accounting for approximately a third of the sample. Approximately 6% of the sample preferred to participate in Spanish. Demographic characteristics of enrolled pediatric participants were similar among those exposed to both interventions.
Parent and Patient Demographics.
When we compared patients who completed the 2-week follow-up with those who did not, the distributions of parent gender, parent race, parent ethnicity, patient age, and patient gender were comparable. Spanish-speaking parents and those who had obtained higher levels of education were more likely to complete the follow-up. For example, parents with a graduate degree (11.68% completed follow-up vs 6.46% lost to follow-up) or high school degree (19.31% completed follow-up vs 28.90% lost to follow-up) were more likely to complete the 2-week follow-up.
The aim of this study was to conduct a multisite, parallel-arm, cluster-randomized comparative effectiveness trial to compare 2 feasible interventions for enhancing parent–provider communication in order to reduce the rate of inappropriate antibiotic prescribing in children with ARTIs.
A total of 429 (26.8%) enrolled pediatric participants received an antibiotic prescription, and 139 (32.4%) of those prescriptions met the study criteria for being inappropriate (Table 5). Overall antibiotic prescribing did not vary significantly based on intervention arm, study site, or provider type (see Table 5). Prescribing occurred frequently among patients diagnosed with pneumonia (100%), acute otitis media (97.9%), and strep throat (91.7%).
Overall and Inappropriate Antibiotic Prescribing.
We hypothesized that the rate of inappropriate antibiotic prescribing would be lower among parent–child dyads managed by providers trained in the higher-intensity intervention than among dyads managed by providers trained in the lower-intensity intervention. Analyses of all enrolled patients revealed that 139 of the 1599 patients (8.7%) received an inappropriate antibiotic prescription. The most common reasons for an antibiotic to be considered inappropriate were (1) antibiotics prescribed for a presumed bacterial ARTI that did not meet criteria (n = 109 [78.4%]) and (2) broad-spectrum antibiotics prescribed for a bacterial ARTI in a child without a penicillin allergy (n = 24 [17.3%]). The rate of inappropriate prescriptions among all enrolled patients in each arm was similar among higher-intensity providers (54 of 696 [7.8%]) and lower-intensity providers (85 of 903 [9.4%]). However, the rate of inappropriate prescriptions among all patients who received an antibiotic was lower among the higher-intensity providers (54 of 196 [27.6%]) than among the lower-intensity providers (85 of 233 [36.5%]; difference in proportions = −8.9%; 95% CI, −17.7% to −0.1%).
As proposed in the study application, GLMMs were used to assess the primary outcome of rate of inappropriate prescribing in this 2-stage nested design [parents nested within providers (level 1 units) and study site (level 2 units)]. Providers (N = 41) were treated as a random effect. In the unadjusted GLMM examining the entire patient sample of 1599, we found that the odds of receiving inappropriate antibiotic treatment for the higher-intensity arm did not vary significantly compared with the lower-intensity arm (odds ratio [OR], 1.09; 95% CI, 0.56-2.10; Table 6).
ORs of Receiving Inappropriate Antibiotic Among All Enrolled Children (N = 1599).
After adjusting for provider type, clinic setting, and provider experience, a nonsignificant effect of intervention arm was still observed (OR, 0.99; 95% CI, 0.52-1.89). The ICC for the adjusted model was 0.093, suggesting a higher level of correlation between providers than we anticipated. The interaction of the higher-intensity arm and whether the provider was an MD/DO or not was significant in the adjusted GLMM (OR, 0.12; 95% CI, 0.04-0.37) of inappropriate treatment, suggesting the effect of the intervention arm was differentially distributed based on the provider type (ie, MD/DO vs CPNP/APRN). Specifically, the MD/DO providers in the higher-intensity intervention arm were less likely to prescribe an inappropriate antibiotic than the MD/DO providers in the lower-intensity intervention arm. The reverse was true for the CPNP/APRN providers, with those trained in the higher-intensity intervention being more likely to prescribe an inappropriate antibiotic than CPNP/APRN providers trained in the lower-intensity intervention. Rates of overall inappropriate treatment in the higher-intensity arm did not show significant differences when stratified by provider gender. After adjusting for location and provider years of experience, we found that male providers in the higher-intensity arm had significantly lower odds of prescribing inappropriate antibiotic treatment than did male providers in the lower-intensity arm (OR, 0.14; 95% CI, 0.05-0.37; P < .0001). For female providers, an opposite, though nonsignificant, association was noted whereby the female providers in the higher-intensity arm had increased odds of prescribing inappropriate antibiotic treatment (OR, 1.60; 95% CI, 0.85-2.98; P = .147) compared with female providers in the lower-intensity arm.
We conducted additional analyses to explore the rate of inappropriate antibiotic prescribing among the subsample of patients who had received a prescription for antibiotics. In an unadjusted GLMM examining only the patients who were prescribed an antibiotic, we found that the odds of receiving inappropriate antibiotic treatment for the higher-intensity arm did not vary significantly compared with the lower-intensity arm (OR, 0.77; 95% CI, 0.38-1.57).
After adjusting for provider type, clinic setting, and provider experience, the odds of inappropriate antibiotic prescribing among higher-intensity providers was still nonsignificant (OR, 0.61; 95% CI, 0.31-1.19) compared with lower-intensity providers (Table 7). The interaction term for the odds of receiving an inappropriate antibiotic based on intervention arm and provider type (ie, MD/DO vs CPNP/APRN) was significant (OR, 0.18; 95% CI, 0.07-0.44). Again, the MD/DO providers in the higher-intensity intervention arm were less likely to prescribe an inappropriate antibiotic than the CPNP/APRN providers in the higher-intensity intervention arm.
Odds of Receiving Inappropriate Antibiotic Among All Children Who Received an Antibiotic (N = 429).
We determined whether children seen by providers in the 2 study arms differed in terms of revisits and/or adverse drug reactions. Approximately 12% of all patients had a return visit to a medical provider within 2 weeks of their index visit. The rate did not vary between patients seen by higher-intensity or lower-intensity providers (Table 8). Adverse reactions to the prescribed antibiotic were relatively infrequent and similar among higher-intensity and lower-intensity arms (16.5% vs 12.8%; P = .27).
Secondary Outcomes.
We assessed parent ratings of shared decision-making using an adapted version of the 3-item CollaboRATE questionnaire.56 Parents ratings were extremely high overall, likely evidencing a ceiling effect. Both arms had similar nonparametric distributions for the CollaboRATE shared-decision-making scale (higher intensity: median, 9.0; interquartile range [IQR], 8.7-9.0; lower intensity: median, 9.0; IQR, 8.3-9.0).
Parent-rated satisfaction with their provider's communication during the visit and their overall visit satisfaction were also very high overall and similar between arms (see Table 8).
The secondary outcomes were further examined when stratified by provider type (ie, CPNP/APRN and MD/DO). The median CollaboRATE scores between lower-intensity and higher-intensity arms were similar for both CPNP/APRN providers (9.0 [IQR, 8.7-9.0] vs 9.0 [IQR, 8.3-9.0], respectively; P = .009) and for MD/DO providers (9.0 [IQR, 8.3-9.0] vs 9.0 [IQR, 9.0-9.0], respectively; P = .02). Among the CPNP/APRN group, a significantly different proportion of encounters with a satisfaction score of 5 was observed between the lower-intensity and higher-intensity arms, both for communication satisfaction (93.9% vs 87.8%, respectively; P = .006) and overall visit satisfaction (94.6% vs 88.7%, respectively; P = .005). Communication satisfaction between the low-intensity (93.1%) and high-intensity arms (94.0%) did not differ when examining the MD/DO group (P = .582); similar results were also observed for overall visit satisfaction among MD/DO providers (lower intensity: 91.1% vs higher intensity: 93.3%; P = .221).
We analyzed a total of 141 audio recordings, 62 from the higher-intensity providers and 79 from lower-intensity providers. Overall, we found no significant difference in consultation length when comparing visits from higher-intensity providers, whose visits lasted 12 minutes, with those from lower-intensity providers, whose visits lasted 10 minutes. Further, more visits from higher-intensity providers were characterized by ≤50% of provider talk time compared with visits conducted by lower-intensity providers (46.8% vs 35.4%, respectively).
All but 2 of the communication tasks were evident in a larger proportion of recordings from higher-intensity providers than in those of lower-intensity providers (all P values < .05). Higher-intensity providers were more likely than lower-intensity providers to have “fully explored parents' concerns” (16.1% vs 1.2%, respectively), “stated the child's illness duration” (79% vs 41.8%, respectively), “discussed appropriateness of antibiotics” (35.5% vs 15.2%, respectively), “used negative behavior framing to discuss antibiotics” (19.4% vs 1.3%, respectively), and “engaged parents in the decision for their child's treatment” (24.2% vs 11.4%, respectively). The 2 communication tasks that did not show a difference between arms were “recommended options for symptom relief” and “identified triggers for reconsultations and contingency plans,” which both demonstrated high rates in each arm.
At the conclusion of recruitment, all 38 enrolled providers completed a follow-up survey and interview. The majority of the higher-intensity providers (84.2%) rated the program as “very effective” compared with 57.9% of lower-intensity providers. Nearly two-thirds of lower-intensity providers rated the program as “very feasible” compared with 57.9% of higher-intensity providers. Nevertheless, all providers stated that they would recommend the program to other health care providers. When asked how satisfied they were with the study, 84.2% of all providers reported being “very satisfied.” Nearly all higher-intensity providers (94.7%) indicated that they were “very satisfied” with the program compared with just under three-quarters of lower-intensity providers (73.7%). Providers were also asked to estimate how often the enrolled parents were interested in receiving an antibiotic during their visit (scale, 0 to 100). Higher-intensity providers provided a significantly lower rating (47.8% of parents) than did lower-intensity providers (58.2% [difference = −10.4; 95% CI, −20.1 to −0.60]).
After reviewing the components of the Let's Talk About Antibiotics (LTAA) intervention, providers in each arm were asked which aspects they most liked. The majority of qualitative responses from providers in the higher-intensity arm included the brochure as a favorite component, while about half also included the video and a third included the communication training. Providers in the lower-intensity arm named the initial antibiotic-prescribing education and ongoing booster antibiotic education with similar frequency. When asked what they did not like about the program, most providers (63% higher intensity and 53% lower intensity) stated that they had no concerns or issues with their assigned intervention. Just over a quarter of providers (26% in both arms) mentioned some aspect of the research trial occasionally interfering with clinic flow. One provider in the higher-intensity arm voiced some concerns about the time associated with filling out the brochure. One provider in the higher-intensity arm and 1 in the lower-intensity arm mentioned that they did not like being recorded: “The only thing is that sometimes timing-wise they would just take 5 extra minutes or just having to get into the room” (provider in higher-intensity arm, MD, CMH PCC).
Across arms, the majority of providers gave positive reviews of the video, appreciating that it was brief and informative, and presented content in a way that both the child and parent could enjoy. Providers considered the video a valuable support tool that provided useful preparation for the clinical conversation: “I really liked it, too. I wish all parents could see that because sometimes they have no idea” (provider in higher-intensity arm, APRN, HPC). Providers thought the video was a useful tool for educating parents about antibiotics, particularly the risk of using antibiotics when they are not needed and the difference between infections caused by viruses and those caused by bacteria. Three providers (16%) in the higher-intensity arm and 1 (0.05%) in the lower-intensity arm responded that they were not sure how the video impacted the parents because they did not receive any feedback from them.
Providers in both interventions reported that the antibiotic education training was helpful, most often stating it reinforced their prescribing knowledge. “I think it solidified things that I knew and do and made me feel like, ok, I am doing that well… that it is appropriate for what they are looking for” (provider in the lower-intensity arm, DO, HPC). Providers also appreciated being able to reference the accessible booklets given as continuing education during the second cold and flu season: “It really allowed me to say that this is where your kid is, this is an antibiotic or this is a diagnosis that you absolutely need an antibiotic for” (provider in the higher-intensity arm, MD, CMH PCC).
During the interviews, study staff showed providers the information sheet on communication tips as a reference to the communication training. Of the providers in the higher-intensity arm who remembered the training, most thought the communication tips were helpful and reinforced prior knowledge: “Like, ‘What's the most important thing you're concerned about today?’ Just having those kind of conversation tactics was great” (provider in the higher-intensity arm, MD, CMH PCC).
When providers in the lower-intensity arm were shown the informational sheet and given some basic background about the communication training, most thought they already practiced good communication but that the training content might be valuable, especially for other providers.
Many providers in the higher-intensity arm considered the brochure a valuable support tool, appreciating that it was concise and contained both visual and written material to help families better understand their child's diagnosis and treatment, as well as what to expect in recovery and when to follow-up: “I like the brochure. And it's the tangible part of it and that you get to reiterate pieces you wanted to show them what to do and they know what to do at home” (provider in the higher-intensity arm, CPNP, CMH PCC). A few providers at CMH PCC had more tempered feelings about the brochure. Some felt that they were duplicating their paperwork, or that it was inconvenient and not a useful addition to the material already given to parents. When shown the brochure during the interview, most providers in the lower-intensity arm commented that they thought it would be a helpful communication aid for their visits. Like some of the providers in the higher-intensity arm, some in the lower-intensity arm were concerned that the brochure would be inconvenient to use because of its format and questioned its usefulness in addition to the discharge paperwork parents already receive.
Approximately half of the providers in the higher-intensity arm felt that knowing parents' level of antibiotic desire via sticky note on the examination room door was helpful preparation for the consultation. The other half of these providers felt that knowing parents' antibiotic desire did not impact their decision or overall consultation experience: “Because I was going to do what I thought professionally was the best for them, regardless of what they thought” (provider in the higher-intensity arm, CPNP, CMH).
Other providers thought knowing a parent's level of antibiotic desire via sticky note was good, but in practice they rarely experienced pushback from even parents who expressed high interest in receiving an antibiotic. Providers in the lower-intensity arm had similar comments about knowing parents' level of antibiotic desire as providers in the higher-intensity arm. Over half of the providers mentioned how knowing parents' desire for an antibiotic for their child would be helpful as a good support tool, particularly when providers have not yet built rapport with the family or when a parent was expecting an antibiotic. About a quarter of the providers in the lower-intensity arm felt that the sticky note would not be helpful because it would not make a difference or would make them simply anticipate conflict.
When asked about overall effectiveness in reducing inappropriate prescribing, 71% of providers who responded thought LTAA would be “very effective,” 29% responded “somewhat effective,” and none responded “not at all effective.” One provider who responded “very effective” elaborated, “particularly the training for the providers, communication, I think that would be very helpful” (provider in the higher-intensity arm, MD, CMH).
Another provider explained that their rating of “somewhat effective” was grounded in their beliefs of how difficult it is to change provider behavior. A few of the MD/DO providers in the lower-intensity arm at the academic medical center believed the program would be more helpful for other types of practitioners, including those in private practices, nurse practitioners, and providers who have been practicing for a long time.
In direct contrast to comments about the usefulness of aspects of the LTAA program and how it might impact other providers' behavior, all but 3 of the providers in the higher-intensity arm who were asked “How, if at all, has the ‘Let's Talk About Antibiotics’ program influenced your own antibiotic-prescribing practices?” responded that they did not believe the program had a direct impact on their personal prescribing decisions. Only 1 provider reported that she thought the program had a “little bit” of influence on her own prescribing behavior. Two providers did not offer a response to this question.
Of the providers in the lower-intensity arm, 68% (13) said that the program did not influence their own prescribing practices, explaining for instance, that they already knew the prescribing guidelines. Five providers in this arm said that the program influenced their practice in some way. For example, some providers believed the program served as a good reminder: “Maybe just a good reminder that it's okay to take the time to explain and families can respond to that and will respond to that” (provider in the lower-intensity arm, MD, PCC).
One provider specifically mentioned the diagnostic tips booklets as being helpful, and a few others mentioned how the presence of study staff made them more aware of their own antibiotic-prescribing practices but did not necessarily change the way they prescribed.
When asked if their communication with parents had changed as a result of LTAA, most providers in the higher-intensity arm reported that it had changed for the better, whereas 2 thought their communication experience had not changed. Providers who thought their communication had changed elaborated that they spent more time explaining to parents about why their child did not need an antibiotic and were more empathetic toward parents' concerns:
I think [my communication is] just more open-ended. And then just also kind of re-emphasizing like, “Hey, antibiotics are not always good. We see these problems. It's good that little Johnny doesn't need antibiotics and here is why.” [provider in the high-intensity arm, MD, HPC]
When asked about barriers to implementing the higher-intensity LTAA program in the future, 45% of providers from both arms responded that they thought there would be no barriers. Others mentioned a few concerns that could become implementation barriers, including future program providers not being willing to change their practice. Other providers worried about time constraints, such as doubling paperwork when completing brochure and clinic records, as well as time in consultation.
When discussing ways to make the program easier to implement, many providers offered suggestions on how to make the video easily accessible without the assistance of study staff. Some suggested “showing the video on display, like on a TV” (provider in the higher-intensity arm, MD, HPC):
Maybe having it in a public area like in the waiting room or something so everybody is seeing it. So it's not set up for specific patients for specific concerns. It's just like everybody just sees it all the time. And it kind of sticks in their head. [provider in the high-intensity arm, MD, PCC]
When arranging trainings in new locations, providers suggested considering the setting and assessing “the clinic's practice and timing and clinic flow if they would actually be able to have the time for the education” (provider in the lower-intensity arm, MD, PCC). One provider felt that “some of the practices that I think are bigger and in a hospital-based setting, things like that, might be a little bit more challenging to get in there versus a smaller practice” (provider in the lower-intensity arm, MD, PCC). Others thought an online module would be helpful to make the training more accessible for providers' busy schedules, but as one noted, this approach might not be as engaging as in-person training.
Lastly, a small number of providers believed that the key to implementation success would be provider interest in the program: “[Providers] have to be really interested in being educated and it goes really well” (provider in the higher-intensity arm, MD, HPC). For locations where provider initial interest is low, some providers suggested emphasizing the benefit for the clinic, the families, and the fact that the program will “be providing a lot of education for the families, so hopefully, things… make their job easier” (provider in the lower-intensity arm, MD, PCC).
Parents who completed the 2-week follow-up survey (n = 1337 [84%]) were asked about their satisfaction with and recommendation for the program. The majority of all parents (92%) reported being “very satisfied” with the program, with an additional 7% indicating that they were “somewhat satisfied.” Only 2 parents (both in the higher-intensity arm) reported being “not satisfied” with their participation. Rates of reported satisfaction were similar between parents exposed to the higher-intensity (92.2% “very satisfied”) and lower-intensity (92.4% “very satisfied”) interventions. Qualitative comments from the 2-week follow-up suggested that overall, parents were satisfied with their participation in the study. Many parents commented that they learned new information about antibiotic use and home care treatments, which they valued for their children's present condition and for use in the future: “I thought the information was very helpful. I went back the following week for my daughter and was able to retain the antibiotic knowledge I learned from the study the week before” (English-speaking mother, aged 35 years).
Other parents said that they already knew much of the information provided in the program but mentioned how the materials would be useful for other parents who may not be as aware. About 60% of parents who received the lower-intensity intervention and almost 62% of parents who received the higher-intensity intervention reported being “extremely likely” to recommend the study to friends and family. About another third of parents in each arm indicated that they would “likely” recommend the study as well. Only 4 parents stated that they were “unlikely” to recommend the study, and all commented that that this was because their friends already knew the content.
Parents reported how helpful they believed the video was in helping them understand information about antibiotic use (Figure 3). Overall, more than half of parents thought the video was “very” or “mostly helpful.” Within the lower-intensity arm, 57% of parents thought the video was “very helpful,” and 22% thought it was “mostly helpful.” Within the higher-intensity arm, more than half of the parents found the video “very helpful,” with another 26% indicating that it was “mostly helpful.” When asked about the video, parents said they found it helpful, informative, easy to understand, and reassuring, and many appreciated the animated portrayal of antibiotics treating bacterial—not viral—infections: “I did not know some of the facts in the video and the sheet [brochure], so it was very helpful… it was eye opening” (English-speaking mother, aged 31 years).
Helpfulness of Educational Video and Brochure, as Rated by Parents.
A 28-year-old English-speaking father specified how he did not know what antibiotics were until he watched the video, while a 38-year-old Spanish-speaking father explained the video was “very good” because “a lot of times we think that the antibiotic will solve everything.” Some parents who were aware of the information before seeing the video felt it was a good “refresher” and/or would be helpful for other parents. Parents also liked that the video was brief, creative, and entertaining, and some gave feedback on how it was “nice to watch while waiting to be seen” (English-speaking mother, aged 33 years).
Parents exposed to the higher-intensity intervention were asked an additional question about the helpfulness of the brochure. Around 50% of participants found it “very helpful,” 25% “mostly helpful,” 18% “somewhat helpful,” and 6% “helpful.” Parents liked the brochure because it was well organized, easy to read, and promoted shared decision-making. These parents were also asked to specify which section of the brochure they found most helpful. Over half of the parents indicated that the information about antibiotics was the most helpful section. Others liked how the brochure provided an overview of their visit and treatment plan as well as a list of home care treatments they could refer to: “It was a good resource to have to reflect on when the child had symptoms again” (English-speaking mother, aged 43 years).
Participants also mentioned being able to share the brochure with spouses or friends. One preferred the brochure over the video because, “parents can take it home and actually have something to refer to and not have to call the doctor's office about 15 times to ask the same questions or get the same information regarding the doctor's recommendations” (English-speaking father, aged 30 years).
However, some parents in the higher-intensity arm did not remember receiving the brochure, and a few stated that they forgot about it and/or threw it away.
We conducted a multisite, parallel-arm, cluster-randomized comparative effectiveness trial to compare 2 feasible interventions for enhancing parent–provider communication in order to reduce the rate of inappropriate antibiotic prescribing in children with ARTIs. The overall rate of inappropriate prescriptions among all enrolled patients in each arm was similar among higher-intensity (7.8%) and lower-intensity (9.4%) providers. These rates are considerably lower than recently published estimates of inappropriate prescribing (14.3%) in the Midwest.32 However, they are consistent with recently published data documenting declines in inappropriate prescribing, especially among pediatricians.63 Nevertheless, they are still higher than findings from other successful intervention studies (rate of inappropriate antibiotic prescriptions ranging from 1.5% to 3.9%).64,65
In the unadjusted GLMM examining the entire sample, we found that the odds of receiving inappropriate antibiotic treatment in the higher-intensity arm did not vary significantly compared with the lower-intensity arm, even after adjusting for provider type, clinic setting, and provider experience. In the absence of a significant main effect, findings of this comparative effectiveness study indicate that the lower-intensity intervention should be recommended for further testing and dissemination.
The interaction of the higher-intensity arm and whether the provider was an MD/DO or CPNP/APRN was statistically significant in the adjusted GLMM of inappropriate treatment, suggesting that the MD/DO providers in the higher-intensity intervention were less likely to prescribe an inappropriate antibiotic than their CPNP/APRN colleagues. Results were similar for models exploring the rate of inappropriate prescribing in the subsample of patients who received an antibiotic in the study. These findings indicate that the higher-intensity intervention may have had a greater impact on the MD/DO providers, who subsequently provided fewer inappropriate antibiotic prescriptions. Why we observed this difference in this study is unclear, and we likely had too few CPNP/APRN providers to draw any definitive conclusions. Nevertheless, it might be because MD/DO providers are more likely than CPNP/APRN providers to go against guideline recommendations when they feel it is in the best interest of their patient. Providers in the higher-intensity arm were guided through an exploration of their thoughts and motivations throughout patient visits, during which they stated that they often feel pressure to inappropriately prescribe antibiotics. Such raised awareness might have had a larger impact on the MD/DO providers than on the CPNP/APRN providers in the higher-intensity arm. Future studies should continue to explore differences in response to interventions between different types of providers. Assessment of fidelity to the communication training content is consistent with this result, as it revealed that providers in the higher-intensity arm more consistently discussed key aspects known to improve shared decision-making and parents' satisfaction during their consultations than did providers in the lower-intensity arm. All but 2 of the providers in the higher-intensity arm reported that the program had improved their communication with parents, reporting that they were more empathetic about parents' concerns and spent more time explaining why antibiotics were not going to be helpful for their child. Given that the major difference between the higher and lower-intensity interventions was the addition of communication skills training for the higher-intensity arm, this result is consistent with other studies that have repeatedly demonstrated the strongest impact from studies that effectively target patient–provider communication.9,20,21,23,43
Secondary outcomes of revisits and adverse drug reactions did not vary between patients seen by providers in the higher-intensity or lower-intensity arm. Parent ratings of shared decision-making, satisfaction with quality of parent–provider communication, and visit satisfaction were all very high and similar between arms. Ceiling effects in the measures were apparent and likely reduced our ability to observe any true differences between arms. Although the P values are <.05 for the comparison of CollaboRATE scores between arms when stratified by provider type, we conclude they do not exhibit a meaningful difference. The data do suggest that for the CPNP/APRN providers there was higher satisfaction in the lower-intensity arm than in the higher-intensity arm; however, no differences were observed for the MD/DO providers. These findings support the recommendation of the lower-intensity intervention for future testing and dissemination, as it was not associated with a higher rate of any of the secondary outcomes that we explored.
Provider satisfaction with the interventions was high, with all providers stating they would recommend the program to others. Providers especially liked the video, antibiotic education, communication training, and brochure. Providers were mixed in their appreciation for the sticky note displaying parents' ratings of interest in receiving an antibiotic provided in the higher-intensity arm. Despite their positive comments about the program and reporting that other providers would benefit from it, only 1 female provider felt that the program had changed her own prescribing practices. There are several possible reasons for this finding, none the least that the overall rate of inappropriate prescribing observed in this sample was exceedingly low. Cognitive dissonance, or the well-documented difficulty that most providers have in admitting to themselves that they might not always perfectly follow prescribing guidelines, might also have contributed. Providers saw few barriers for implementation in other settings but noted concern that “other providers” are often unwilling to change their practice behaviors and that time constraints will be important considerations in any dissemination effort. Providers also offered useful advice about how to expose parents to the video in clinic waiting rooms and to fit the program into busy clinic schedules that will be critical for future dissemination efforts. In terms of scalability, it is important to consider that the brochure requires little training (it was covered in the last 15 minutes of the communication training), and the video could be downloaded and played even without internet connectivity. Parent satisfaction with LTAA was high, and the majority reported that they were likely to recommend the program to a friend. Parents liked the brochure's gain-framed antibiotic information, overview of their visit, and overview of their treatment plan, as well as the list of home care treatments they could refer to and share with others at home.
The lack of a significant main effect may indicate that shared decision-making and the other communication factors targeted by the higher-intensity intervention were not as strongly related to inappropriate prescribing as had been expected. This may indicate a cultural shift in parental expectations and/or provider comfort in withholding unnecessary antibiotics, challenging the relevance of early literature to the social and communication dynamics at play today. It may also be that the communication factors targeted in the higher-intensity intervention derived from established literature were not as strongly associated with inappropriate prescribing in this study context. Other recent studies have found success focusing on provider education about appropriate antibiotic prescribing and the effects of peer comparison.54,64 However, a follow-up study indicates that prescribing education, not peer comparison, showed the greatest long-term impact.65 Provider education interventions may be sufficient to yield long-term benefits, as parental expectations for antibiotics continue to decrease (from an overall cultural shift or from exposure to a high-quality video).
There are many possible explanations for why the higher-intensity intervention did not demonstrate a significant advantage over the lower-intensity intervention in our study. As a comparative effectiveness trial, our study is limited in that it cannot determine the effectiveness of either intervention compared with baseline or usual care. The overall low rate of inappropriate antibiotic prescribing across interventions and sites is encouraging, but our design does not allow us to draw any conclusions about the role of either intervention in these lower rates compared with usual care. It may be that the higher-intensity intervention is indeed preferable, but the unexpected, uneven distribution of patients among sites and providers reduced statistical power and our ability to detect a difference (type II error). We monitored the rate of accumulation of dyads by site and provider throughout the study and took steps to try to improve the balance across sites and providers. For example, we assigned more full-time staff to the academic site and focused our efforts on recruiting dyads seen by providers who had treated fewer enrolled dyads. This effort ensured that we recruited dyads from all enrolled providers, but the number of dyads per provider still ranged from 2 to 243. Although it is still possible that this uneven distribution masked a true difference between interventions, it is highly unlikely, as the main analysis controlled for clustering and the results exploring the HTE also revealed no difference after adjusting for clinic setting. Providers who received the higher-intensity intervention training may not have been given a sufficient “dose” of the communication training, because it was not feasible for providers to attend > 1 additional 60-minute training. While our fidelity monitoring demonstrated that overall, the higher-intensity providers practiced the targeted communication techniques more frequently than the lower-intensity providers, overall fidelity was modest, suggesting the 60-minute training plus clinic materials may not have been sufficient to shift behavior to the extent necessary to “move the needle” on prescribing rates.
In generalizing the findings from this study to other patients, providers/practices, and seasons during which ARTIs are prevalent, we offer the following considerations. In terms of pediatric patients, the statistical similarities between parent–patient dyads that chose to enroll and those that declined supports the notion that our sample closely represented the larger population pool from which it was drawn. We have confidence that findings at the parent–patient level could generalize to this population, but also note that the target population had some restrictions. We cannot generalize our findings to patients who had recent antibiotic use, chronic or immunocompromising medical conditions, or who were outside of the 1-year to 5-year age range. In terms of providers and their practices, having similar results at both an academic and private practice suggests potential generalizability to these contexts. However, the academic practice had implemented effective antibiotic stewardship training in its inpatient setting66 and has close ties with the private practice, so it is possible that providers at our study sites were already practicing better-than-average antibiotic prescribing. Although rates of antibiotic prescribing in our setting were lower than reported in previous literature, we do note that our methods for determining inappropriateness were more meticulous than other studies have used, making it difficult to make a direct comparison in these rates. We are unsure whether the findings related to interaction effects of the intervention and provider type can be generalized to the larger population. Finally, our results represent a 2-year data collection period, covering all seasons in which ARTI prevalence and etiology fluctuates. We do note that the flu season of 2017-2018 had a higher-than-usual burden on our community, which may have resulted in a higher proportion of pediatric patients with influenza and less pressure to prescribe antibiotics during that time.
Given the findings of this study, future research should focus on identifying and testing alternative dissemination strategies for the lower-intensity intervention. Because rates of inappropriate antibiotic prescribing were lower than anticipated in both arms, it may be useful to repeat the study in a setting of high rates of inappropriate prescribing. Future research may also consider adding a control/no-intervention arm to compare both the higher- and lower-intensity interventions with a usual care scenario. Further, future research may also benefit from estimating baseline rates of communication competence and inappropriate prescribing in the sample of providers. Our study also reveals some provider behaviors other than communication that are related to rates of inappropriate prescribing, such as inappropriate diagnosing or prescribing for strep throat, and may be worth exploring as potential areas for intervention.
Our study offers several useful findings. Overall, we demonstrated that implementing evidence-based provider and parent education on antibiotics can be feasible in academic and private pediatric outpatient practices. However, additional training and tools in parent–provider communication as delivered by the higher-intensity intervention did not significantly improve rates of inappropriate antibiotic prescribing beyond lower-intensity educational strategies. Our video demonstrated both feasibility and effectiveness in reducing parents' desire for antibiotics. Our community-engaged process for designing and implementing the intervention resulted in high satisfaction from both parents and providers. We also demonstrated that an intervention tailored for the fast-paced environment of US pediatric outpatient clinics can significantly increase targeted provider communication behaviors. Finally, our institution has established effective procedures for ensuring the Spanish-speaking population we serve are active, equal participants in CABs and patient-centered research.
We note a change from our original study proposal, which defined the primary outcome as shared decision-making.
Research reported in this publication was supported through a PCORI Program Award (CDR-1507-31759). All statements in this report, including its findings and conclusions, are solely those of the authors and do not necessarily represent the views of PCORI, its Board of Governors, or Methodology Committee. The authors wish to acknowledge the contributions of parent and child stakeholders, our CAB members, and clinical stakeholders at Children's Mercy Primary Care Clinics and Heartland Primary Care Clinics in designing this study. We also wish to thank all provider, parent/caregiver, and patient participants.
Research reported in this report was funded through a Patient-Centered Outcomes Research Institute® (PCORI®) Award (#CDR-1507-31759). Further information available at: https://www.pcori.org/research-results/2016/comparing-higher-and-lower-intensity-parent-clinician-communication-trainings
Stills From the 90-Second Animated Educational Video (PDF, 384K)
Details of 9-Step Translation Process (PDF, 106K)
Goggin K, Hurley EA, Bradley-Ewing A, et al. (2021). Comparing Higher and Lower Intensity Parent-Clinician Communication Trainings to Reduce Antibiotic Misuse in Children. Patient-Centered Outcomes Research Institute (PCORI). https://doi.org/10.25302/01.2021.CDR.150731759
The [views, statements, opinions] presented in this report are solely the responsibility of the author(s) and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute® (PCORI®), its Board of Governors or Methodology Committee.
This book is distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License which permits noncommercial use and distribution provided the original author(s) and source are credited. (See https://creativecommons.org/licenses/by-nc-nd/4.0/
Your browsing activity is empty.
Activity recording is turned off.
See more...
