Trends in Pediatricians’ Developmental Screening: 2002–2016

The Periodic Survey collects data from a national, random sample of nonretired AAP members. The 2016 survey explored pediatricians’ knowledge, attitudes, and practices regarding screening and managing children for developmental delays or problems. The survey instrument was developed collaboratively by AAP research staff, the AAP’s Council on Children With Disabilities and Section on Developmental and Behavioral Pediatrics, and the Autism Intervention Research Network on Physical Health at Massachusetts General Hospital. Many questions were replicated from a 2002 survey, and the use of formal screening tools (ie, standardized and validated measures) was also asked about in 2009. The AAP Institutional Review Board approved all surveys.

In 2016, an 8-page, self-administered questionnaire was sent to a random sample of 1638 active, US-based members of the AAP. It included pediatric residents and excluded those subspecialty subboarded. Survey participants were sent up to 7 paper mailings between November 2015 and March 2016, and electronic surveys were sent to nonrespondents after the second and fourth paper mailings. The response rate in 2016 was 47%. The 2009 survey was sent to 1620 AAP members with a response rate of 57%. The 2002 survey, which was sent to 1617 members, resulted in a response rate of 58%. The analytic sample was restricted to postresidency pediatricians who provided primary care and reported assessing for developmental risk in patients <36 months of age (2016, N = 469; 2009, N = 532; 2002, N = 562).

The primary dependent variable (used in logistic regression analysis) was pediatrician use of a standardized general developmental screening tool. Respondents were asked, “How often do you or your staff use the following methods or tools to identify children birth to 35 months of age at risk for A) developmental delay/problems and B) autism?” Respondents were only asked about autism screening in 2016, and thus, the primary focus of our analysis is on developmental screening (for which we have responses across all 3 surveys); we are in the process of preparing an additional article for publication that will provide an in-depth analysis of our results on ASD screening. Additional findings on autism screening will be reported in another article. A list of standardized instruments is provided (Table 1). Responses were dichotomized to indicate those who usually use at least 1 standardized screening tool (always or almost always use a formal screening tool = 1; sometimes, rarely, or never use a formal screening tool = 0) for all 3 surveys.

Other variables of interest (available only in 2002 and 2016) include who in the practice administered screens and reviewed results as well as perceived barriers to screening for developmental delays. We also examined several variables related to referral, including the estimated percentage of the respondents’ patients with possible developmental problems who were referred to state EI or specialists, how likely respondents were to refer patients to EI or specialists on the basis of certain conditions, and perceived barriers to EI referral.

The primary independent variable was survey year. To assess the independent effect of survey year on reporting standardized screening tool use, we also included a set of pediatrician demographics and practice characteristics that were available across the 3 surveys in the multivariable logistic regression model as controls: practice area (suburban, rural, urban and inner city, or urban and not inner city), practice setting (solo or 2-physician practice; group practice or health maintenance organization [HMO]; medical school, hospital, clinic, or other), hours worked per week (continuous), sex (female or male), and age (continuous). Our choice of independent variables to include in the multivariable model, although constrained by the variables that we had available across all 3 surveys, was informed by previous research on this topic using the Periodic Survey.⁴  Although the surveys collected information on patient insurance status, the measure suffered from substantial missing data in 2016 and was not significantly associated with our dependent variable in the multivariable model; thus, it was excluded from the analysis.

We assessed nonresponse bias by comparing the analytic samples to their respective target samples using key demographic variables available in the AAP member database (age, sex, and US geographical region). For each survey year, one-sample proportion tests compared the analytic and target samples on the basis of sex and region, and a t test compared the samples on the basis of age.

χ² tests compared rates of developmental screening tool use, administration of developmental screenings, and barriers to screening across survey years. A multivariable logistic regression model examined changes in standardized screening tool use over time while controlling for the other variables listed above. Additionally, χ² tests compared rates of referral, developmental conditions leading to referral, and barriers to EI referral across survey years. Data were analyzed by using IBM SPSS Statistics 25 (IBM SPSS Statistics, IBM Corporation). P ≤.05 was considered significant.

In a nonresponse analysis comparing survey respondents to nonrespondents, there were no differences in sex or region in 2016, but the mean age of respondents was slightly higher than that of nonrespondents, with the respondents’ mean age being 47.8 years compared with 44.7 years for nonrespondents (P < .001). No differences were seen in 2002, whereas similar differences were seen in age in 2009 and 2016.

Across survey years (Table 2), the samples became increasingly female (54% in 2002 and 66% in 2016; P < .001), and the average age increased to 49 years (P < .001). The practice area distribution of pediatricians remained relatively stable over time, but in 2016, pediatricians were more likely to work in group-practice settings relative to previous survey years (P = .006).

Overall, the reported use of any standardized screening tool among pediatricians has increased from 21% in 2002 to 63% in 2016 (Table 1; P < .001). The standardized screening tool with the largest increase in reported use is the ASQ,⁹  rising from 9% in 2002 to 48% in 2016 (P < .001). In 2016, 14% of pediatricians reported using the Denver II, and 19% report using the PEDS instrument. Use of ASQ and PEDS tools increased in each survey year, whereas reported use of the Denver II remained stable across surveys. In 2016, 72% of pediatricians reported using 1 or more standardized autism-specific screening tools (Modified Checklist for Autism in Toddlers, Pervasive Developmental Disorders Screening Test II, or Screening Tool for Autism in Toddlers and Young Children) to identify patients at risk for ASD.

Survey results indicate that screenings shifted from being administered primarily by the pediatrician (86%) in 2002 to other staff (60%) in 2016 (P < .001). However, review and interpretation of the screenings were conducted by 96% of pediatricians in 2016, indicating that even as other practice staff became more involved in the administration of screenings over time, pediatricians remained heavily involved in the review of screening results.

In 2016, the most commonly perceived barrier to screening children for risk of developmental delay or ASD was time limitations, with 57% of pediatricians agreeing or strongly agreeing (Table 3). There was no difference in reporting of time as a barrier between pediatricians who administered the screen and those who did not. Time limitations, inadequate reimbursement for conducting a standardized screening, lack of office staff to perform screenings, belief that screening is not an appropriate role for pediatricians, and lack of confidence in screening ability declined as reported barriers to screening from 2002 to 2016. The only barrier that increased over time was lack of treatment options for positive screen results, which increased from 9% to 21% (P < .001).

At a bivariate level, the results in Table 1 indicated a substantial increase in reported standardized screening tool use across surveys. To assess whether this trend of increased standardized tool use remains statistically significant even when taking into account the changing demographics and practice characteristics of pediatricians over time, we employed a multivariable logistic regression model estimating the use of any standardized tool for developmental screening (Table 4). Even after adjusting for other available characteristics, pediatricians in 2016 were much more likely to report using a standardized screening tool relative to pediatricians in 2002 (adjusted odds ratio = 7.3; 95% confidence interval = 5.4–9.9). Among our control variables, pediatricians practicing in urban, inner-city areas and rural areas were more likely than those in suburban areas to report using a standardized screening tool, whereas women (adjusted odds ratio = 1.47; 95% confidence interval = 1.1–1.8) were more likely than men to report standardized tool use.

On average, pediatricians in 2002 self-reported referring 41% (SD = 40.9; median = 20%) of their patients who were identified as at risk for developmental problems to EI. That number increased to 59% (SD = 38.2; median = 75%) in 2016 (2-tailed test, P < .001; Wilcoxon rank test, P < .001), with pediatricians on average referring 22% (SD = 25.8; median = 10%) of patients who were identified with developmental problems to specialists before referring them to EI (given the skew present in these figures, we have provided means, medians, and results for both parametric and nonparametric tests). Although these figures may be subject to recall bias, they provide evidence of increased referral activity among pediatricians. Referral rates did not differ by screening tool used.

Survey results indicate that pediatricians in 2016 considered a wider variety of developmental conditions when making the decision to provide an EI or specialist referral (Table 5). Pediatricians were more likely in 2016 than in 2002 to report being “very likely” to refer a patient to either EI or a specialist for the following developmental conditions: global developmental delays, loss of developmental milestones, delayed speech, sensory impairment, abnormal muscle tone and/or motor delay, family concern, newborn failing a hearing screen, and risk of neglect or abuse.

In 2016, the most commonly perceived barrier to EI referral was lack of feedback from the EI program about the child’s progress and outcomes, which was reported by 38% of respondents (Table 6). From 2002 to 2016, lack of understanding of the EI program’s processes (46% to 30%) and lack of information about EI and its services (42% to 29%) declined as reported barriers (P < .001). Inconsistent quality of services increased as a perceived barrier from 2002 to 2016 (24% to 30%; P = .02).

In 2006, in part as a response to the 2002 Periodic Survey showing low rates of standardized developmental screening, the AAP convened an expert panel of affiliated professionals to develop new guidelines for pediatricians to improve the early identification of children with developmental disorders through surveillance and screening.⁶  The result was a new policy statement and clinical algorithm for developmental surveillance and screening, implementation guidance for the primary care clinician, and substantial AAP educational efforts to promote screening. In a follow-up survey, reported screening rates increased from 21% in 2002 to 48% in 2009.⁴  In 2016, 63% of pediatricians reported use of standardized developmental screening tools. This represents a tripling in screening over 15 years and a doubling since the 2006 guidelines.

Similar improvements in screening were found by using parent self-report data,¹³  in which a much lower screening rate of 30% was reported, but some states’ rates approximated those in this pediatrician-report. Medical record–based investigations in large clinical programs reported that >80% of children were screened.^14,16  Although these studies reflect rates at individual health centers, they also provide a more accurate accounting of screening rooted in visit documentation. Although gaps remain, a consistent increase in screening has occurred since the 2006 AAP guidelines trending toward universal screening of children for developmental problems.

The current survey suggests that several factors contribute to increased developmental screening. One is a shift toward tool administration by other office staff (60%), which may reflect a response to the earlier concerns about time constraints.^5,6  Although administration seems to have changed hands, nearly all pediatricians (96%) report maintaining responsibility for result interpretation. With the rise of electronic medical record systems, additional changes in screening administration may be occurring that are not reflected in our findings, such as using other staff or automation or the use of nonstandardized milestone lists for screening. Future surveys will benefit from exploring the contributions and limitations of these new modalities.

Improvements may also be attributable to a perceived decline in barriers. Our findings indicate that pediatricians are now less likely to report barriers to screening. Other research shows a similar decrease in barriers when quality-improvement processes were put in place to expand the practice.^{14,16,18–20}  Despite the significant decrease from 80% in 2002, 57% of the pediatricians in 2016 indicated that time was a barrier, calling for continued attention to this issue. Additionally, 21% of pediatricians in 2016 indicated lack of treatment options for positive screen results as a barrier, increasing from 9% since 2002. Improved access to evaluation and treatment options for children with concerning screen results is needed. Although this study did not investigate specific successful screening practices, others have reported benefits from use of a computerized decision aid, office- or home-based online screening, and practice education.^18–21 

Screening children for developmental delays is the first step in a process of early identification and treatment of developmental disorders. As we move toward universal developmental screening, attention must be directed to the subsequent recommended steps of referral, evaluation, and initiation of treatment. Pediatricians now report higher rates of referral to EI for children who are identified as at risk for a developmental problem, with referral being more likely across a wide set of developmental issues. The referral increases may be due to better understanding of EI and a decrease in perceived barriers. Improvements in EI referral were also reported in practice-improvement projects using centralized electronic referral systems, patient navigators, tracking systems, and EI partnerships.^13,19  However, inconsistent quality of EI programs is more commonly reported in this survey as a barrier to referral (24% in 2002 and 30% in 2016). Additionally, the highest-reported barrier to referral in 2016 was the lack of feedback from programs (38%), suggesting a need for improved communication between pediatricians and local EI professionals, as has been implemented in another project.²² 

Referrals to specialists can be used to evaluate for etiology and co-occurring conditions that may impact treatment and family understanding. Rates of such referrals are high and have also increased for children who are at risk for developmental delay, with significantly increased referrals to both EI and specialists for global delays, milestone loss, and sensory impairment being observed. Despite these improvements, only 46% of providers reported referring for delays in speech and language (although these delays are common in children with ASD), hearing loss, and neurogenetic conditions. Referral rates to specialists remain low for abuse and neglect or family concerns, but these are generally beyond the scope of these tools. Although rates of referral for specialty evaluation are high, only 21% of providers are referring for concerns before referral for EI, which is in line with common reports from pediatricians about the difficulties in children’s access to specialty services.^23,24 

There are some limitations to this study. Because this survey is based on respondent self-report, there is potential for inaccurate estimations of screening or referrals rates. Additionally, social desirability bias may have caused some pediatricians to overestimate their rates of certain developmental practices. However, the previous surveys had the same methodology, suggesting that these figures represent true change. Furthermore, the sample was drawn from the AAP member database and may not be representative of all pediatricians and other child medical professionals. Response rates for the Periodic Survey declined across survey years, as they have for surveys more broadly.²⁵  However, response rates for the Periodic Survey compare favorably with other health care provider surveys, and previous research indicates that there is no conclusive link between a survey’s response rate and response bias.^26–28  Although it is possible that there are unmeasured attributes (eg, interest in developmental screening) that are influencing survey response, our nonresponse bias analysis finds that there are no substantial differences in age, sex, or regional characteristics between respondents and nonrespondents, and sex and age were included as controls in the multivariable logistic regression model. Lastly, these data measured responses to questions about developmental screening but not ASD screening. Given the overlap between ASD and other developmental disorders, respondents may have included their ASD screening practices in the reporting on general developmental screening, causing overreporting or underreporting of developmental screening.

The encouraging results of this study show increased use of standardized developmental screening tools since 2002, with nearly two-thirds of pediatricians now performing standardized screening of children for developmental problems. This improvement has occurred with declines in perceived barriers to screening, although time limitations remain a strong concern and with staffing shifts in standardized screening administration in pediatric practices. However, one-third of pediatricians are not completing standardized screening, with barriers remaining, leaving many children unscreened for developmental disorders and potentially resulting in delayed identification and treatment of problems that may affect children and families for a lifetime. With more screening, pediatricians are now also referring a higher percentage of patients with concerns identified by developmental screening to both EI and to developmental and medical specialists, with referral occurring for a wider set of conditions. To further increase the rates of both developmental screening and referral of patients with concerns for treatment, increased attention should be focused on improving screening and referral systems. This may include the incorporation of screening tests and referral and tracking systems into electronic medical record systems, allowing for better tracking of child outcomes and feedback to the medical home. Improved communication systems between practices and community intervention, therapy, and education programs will similarly aid in tracking. Pediatricians must also continue advocating for effective and evidence-based interventions across EI and treatment programs to ensure that each child will reach his or her potential. A revision of the 2006 guidelines on developmental surveillance and screening⁷  will soon be published, incorporating many of the findings and conclusions from this survey.

This research was supported by the American Academy of Pediatrics. The 2016 survey and development of this article was supported by the Health Resources and Services Administration of the US Department of Health and Human Services (cooperative agreement UA3 MC11054: Autism Intervention Research Network on Physical Health). This information or content and conclusions are those of the authors and should not be construed as the official position or policy of, nor should any endorsements be inferred by, the Health Resources and Services Administration, Department of Health and Human Services, or US Government. This work was conducted through the Autism Speaks Autism Treatment Network, serving as the Autism Intervention Research Network on Physical Health. The 2002 survey was supported by the US Department of Education Office of Special Education Programs (grant H02 MC00073) and the National Center of Medical Home Initiatives for Children With Special Needs.

AAP

American Academy of Pediatrics

ASD

autism spectrum disorder

ASQ

Ages and Stages Questionnaire

EI

early intervention

HMO

health maintenance organization

PEDS

Parents’ Evaluation of Developmental Status