The Institute of Medicine’s consensus report on building safer health systems advocates the development of systems “to make it harder for people to do something wrong and easier for them to do it right” [6]. Progressive efforts in health information technology (HIT) have worked to customize the electronic health record (EHR) to enhance clinical practice. One way of doing this has been through the integration of best practice alerts (BPA) into the EHR to support clinical decision making. Our study is unique in that the BPA was evaluated in an outpatient pediatric urology setting, which, to our knowledge, has not been studied before.
In our study we implemented a BPA as a guide for practitioners to increase awareness of the VUR guidelines with the goal of increasing adherence. While the vast majority of clinicians (> 87%) were screening patients with weight and ultrasound initially and at their one-year follow-up visits both before and after BPA implementation, adherence to other recommended measures were less consistent. In our practice, weight measurements are built into the intake process and therefore part of existing clinic flow. We suspect that the high rate of obtaining ultrasound is related to agreement with the AUA guidelines in the perceived beneficial utility in decision making and clinical discussions with families.
It is unclear as to why some parts of the guidelines were followed more closely than others in this cohort. Part of this observation may be related to when the BPA appeared within the EHR workflow (immediately upon opening a patient chart) and ease of incorporating practices into existing clinic flow. Height measurements significantly increased following implementation. Obtaining height at check-in may fit more easily into the existing practice (along with weight) and medical assistants may be more receptive to changes in practice. There is a potential that billing may also play a role in obtaining vital signs, as at least two vital signs were previously needed for higher outpatient billing codes. Interestingly, blood pressure measurements were largely unchanged pre- and post-BPA. As BP is usually part of the intake process, it seems counterintuitive that other intake measurements increase while this did not. Studies have reported a low percentage of children have their BP measured, even in the emergency room setting, partially due to difficulty in obtaining measurements in children and low perceived clinical utility [7]. Therefore, the lack of change in obtaining BP measurements may be related to the higher barriers of incorporation into intake flow. Studies in other specialties on guideline-based BPAs have not demonstrated significant improvement in adherence to the guidelines, though issues may lie in the development of the BPA and where it appears within the EHR [8, 9]. Reasons cited as barriers to BPA success in the EHR are concerns about formatting of the alert, alert fatigue, time pressures in clinic, and poor fit within clinic workflow [10, 11].
There was a high loss to follow-up rate with roughly half of patients following up at 1 year pre-BPA as opposed to 40% post-BPA. Patients were significantly more likely to follow-up if they were younger (roughly 2.5 years), being treated with antibiotic prophylaxis, or had a family history of reflux. This is consistent with findings in the literature that increasing patient age and non-chronic conditions are associated with higher missed pediatric urology appointment rates [12]. We did find higher levels of follow-up in patients treated with antibiotics and this may be due to antibiotic prescription expiration prompting patient families to return for evaluation and medication. We also found higher levels of follow-up in patients with a family history of reflux, which may be due to increased awareness of the need for follow-up in family members with similar conditions. Interestingly, there was no association between grade of reflux and rate of follow-up. While this study focused on guideline adherence by providers, future studies could examine the effect of a similar intervention aimed at parents and caregivers, alerting them to schedule an annual follow-up visit with the goal of reducing missed appointment rates.
There were no changes in patients on antibiotic prophylaxis or decisions to pursue surgery indicating that the BPA did not interfere with practice patterns for interventions and decision may be made on individual clinical information. There was also no significant change in obtaining urinalysis. Variability in adherence to other guidelines has been attributed in part to providers questioning the applicability of all parts of guidelines, drive for individualized care, and possible disagreement with parts so it is possible that this may have some basis here [13–16].
This study is limited by the nature of study design in that data was prospectively collected through standardized data fields within the clinical note and later extracted and manually reviewed retrospectively. We are unable to assess beyond chart review the reasoning behind clinical decision making and with standardized fields within the templated note; this may limit some of the detailed description of clinical decision making. Additionally, the control group participants were seen more than 1 year before those in the intervention group. Clinic staff and responsibilities may have changed during this time, affecting measures. Another limitation may be the practice setting within which this BPA was implemented. Brenner et al performed a systematic review of 69 studies focusing on the effects of health IT on patient safety outcomes [17]. They concluded that most studies showed mixed results, but those that did demonstrate benefit were more likely to be in an inpatient setting [17–19]. Given that the BPA appeared in an outpatient setting at a tertiary referral center, our findings may not be generalizable to other practice settings. Finally, we did not survey the providers to determine exactly what challenges or barriers they faced with adopting the recommendations contained within the BPA, thereby limiting our ability to determine the reason for the lack of effectiveness of this intervention.
Although the BPA for VUR did not provide the desired effect of changing practice patterns across the board, it does provide interesting insights. Many studies looking at drug-drug interaction (DDI) alerts cite “alert fatigue” as a reason for overriding the alerts [20–22]. The EHR now has many built-in alerts and often clinicians are clicking through the alert without fully acknowledging the content. Additionally, clinicians become more desensitized to alerts if there are multiple within a single patient encounter, with acceptance of reminders or alerts dropping by 30% for each additional reminder received per encounter [11, 22]. When looking to improve the effectiveness of BPAs and clinical decision support systems, clinicians look for integration into the current workflow, the timing of such alerts, and aim for reduced complexity of the system [23, 24]. Although the BPA we implemented was built to be actionable (placing orders upon seeing the BPA) and fit into workflow, this still did not translate to changes in practice [21]. Future studies are needed to better understand exactly why providers are not adhering to the guidelines and to determine the best way to effect successful implementation of evidence-based guidelines into clinical practice.