By automating the process of screening and alerting the physician to those who screened positive with our decision support system, we have significantly decreased the burden of identifying relevant guidelines and screening of patient families in our clinics.
This study, like all research, has limitations. The most important limitation is that CHICA selects questions for the families to answer based on patient’s age and what is known about the patient at the time of the encounter. Therefore, every question is not asked of every patient, which could lead to some underestimates of prevalence in our population for some risks.
By contrast, because the system retains information from 1 visit to the next, CHICA may ask a question more than once over the course of several visits. For instance, if CHICA learns there is a smoker at home in 1 visit, in the next it may ask if that person has quit. Another example might be that CHICA learns that a patient has asthma in 1 visit, it may ask whether symptoms occur during the day or night or how often symptoms occur. Finally, CHICA will repeat a question after a sufficient period time (months or years) have passed. This may cause some risk rates to overestimate the population prevalence. Regardless, the rates of positive screening are informative as an indication of the rates at which pediatricians might find these problems if they had a process to ask the questions.
It should be noted that we have an urban patient population with high minority and Medicaid rates. Therefore, these risk rates may not apply to other populations. It could also be argued that adolescent patients may elect not to answer more sensitive questions honestly. In fact, our results reveal low rates of some risky behaviors in the adolescent group, lower than previously reported,27
for example, rates of illicit drug use or alcohol. The rates reported here are very low when compared with the national statistic from the most recent Youth Risk Behavior Surveillance survey (http://www.cdc.gov/HealthyYouth/yrbs
) in 2009. On the other hand, reported rates of unprotected sex are close to national statistics, and identifying patients who think they might be pregnant (5.6%) is extremely valuable clinically. We have observed that when the PSF is handed to teenagers, they are often accompanied by adults. This undoubtedly limits a teenager’s willingness to disclose some behaviors. Interestingly, paper surveys typically achieve higher rates of disclosure of sensitive topics than face to face or telephone surveys.28
However, the logistics in our setting may not promote some disclosures.
Alternative approaches such as private kiosks with audio computer assisted structure interviews may be more promising in the adolescent population.29
However, in general, we found that all questions were answered ~90% of the times they were asked. Moreover, we think that the rates at which risk factors are found (over 2 per patient per encounter) can help physicians focus their efforts. It is clear that the PSF provides an efficient way to capture risk data in the waiting room. It promises to increase both the sensitivity and the efficiency of routine screening for health risks in the primary care pediatric setting. Moreover, once the risk data are in the system, they can be tracked for future reminders and support can be generated for the physician at future visits.
This study provides useful information about how commonly certain risk factors are found in the clinical environment. Often, research about the prevalence of these risk factors is conducted outside of clinical care. These results represent the rates at which patients and families report risks when they are present in a clinical setting. A deeper understanding of the under-recognized risks can help formulate a rational strategy for approaching preventive care in pediatrics.