While national and international registries of pediatric rheumatology patients have grown over the last 10–20 years, analysis of individual center populations has been lacking in the literature. Although viewing the field of pediatric rheumatology through the lens of a single-center experience has limitations with respect to the advancement of treatment and diagnosis of rare diseases, it can shed a unique light on the supply-demand challenges facing the field today. Analyses by Denardo et al., Bowyer et al., and Rosenberg have previously looked into pediatric rheumatology diagnoses at the individual clinic and health system level, but there has been little published in the last 20 years to compare to our current study [3,4,5]. It is hard to equate clinic volumes given multiple obscured factors like the number of providers, catchment area, etc., but compared to our median monthly new patient rate of 52–137 patients, these previously reported population numbers equate to an average of 71–172 new patients per clinic per year, demonstrating a substantial difference in patient load. The proportions of rheumatic disease diagnoses within the Denardo et al. and Bowyer et al. cohorts were reported to be 38% and 40.5%, respectively [3, 4]. In the Rosenberg cohort, out of 3268 patient referrals, a diagnosis was reached in only 2098 patients (64.2%), and of those diagnosed, 50.9% had rheumatic disease. Therefore, if we assume that all undiagnosed patients did not have a rheumatic disease (likely not correct), the rheumatic disease diagnosis rate of all referred patients would be 32.6%, with the true proportion likely higher, as some amount of the undiagnosed patients likely did have a yet-to-be-diagnosed rheumatic condition [5]. Again, the comparison to our clinic’s 23.1% rheumatic disease diagnosis rate is difficult given our policy of offering appointments to all referred pediatric patients, but all previously reported cohorts had notably higher rates of rheumatic diagnoses. Juvenile rheumatoid arthritis/JIA was the most common rheumatic diagnosis in all three studies at 53%, 39.4%, and 31.6%, comparable to our JIA prevalence of 45.6% [3,4,5]. Of the remaining non-rheumatic diagnoses, musculoskeletal conditions (56%, 36.1%) were most common, but at a smaller proportion than our 61.7% [3, 5]. Therefore, despite the previously reported populations having lower total patient volume and less rheumatic disease overall, the proportions of specific rheumatic conditions within the total rheumatic diagnosis cohort seemed to be similar to our current population, with our clinic having a higher rate of non-rheumatic disease.

The pediatric rheumatology workforce supply in the United States is projected to significantly lag demand over the next few decades. As of 2018, 42 out of 50 states were noted to have less than one pediatric rheumatologist per 100,000 children and 30% of practicing pediatric rheumatologists self-reported as likely to retire in the following 10 years [2]. And although there may be almost 400 pediatric rheumatologists practicing in the US and it’s likely that adult rheumatologists may see pediatric patients in various settings, the total clinical full-time equivalents (FTEs) devoted specifically to pediatric rheumatic care was reported to be 287 FTEs in 2015, even when including nurse practitioners (NPs) and physician assistants (PAs) [2]. Demand for pediatric rheumatology care was estimated at 382 FTEs in 2015, already a shortfall of 95 FTEs with the 2015 workforce, and this gap is only expected to worsen by 2030 with the projected supply of 231 FTEs insufficient for the projected demand of 461 FTEs [2]. Strategies have been recommended to increase the supply of pediatric rheumatology providers, including increasing exposure to the field during medical school and residency, decreasing fellowship training from 3- to 2-year commitments, increasing NP and PA utilization, and financial incentive programs [2, 6].

The demand side of the supply-demand shortfall may be a more complicated issue to address. Despite the 4–6 attending physicians, 3–4 nurse practitioners, and 1–3 pediatric rheumatology fellows that saw patients throughout our study period, it was and continues to be a struggle to see our large patient load without long appointment wait times. Moreover, even though there are limited studies focused on wait times for rheumatology evaluation, this is not a problem unique to our division. One study of adult patients referred to Ontario rheumatologists from 2000 to 2013 noted a median wait time from referral to rheumatologist consultation of 74 days, decreasing to 66 days for patients with systemic inflammatory rheumatic disease [7]. In pediatric rheumatology, organizations in the United Kingdom and in Canada have set benchmark times for rheumatology evaluation at 4 weeks from referral for non-systemic JIA, but there is limited data on whether United States pediatric rheumatology centers can or do meet these guidelines [7, 8]. During the study period, the median time between referral and appointment (wait time) for all patients was found to be 42.0 days, outside the recommended 4 weeks for rheumatology appointment wait times. However, for those patients eventually diagnosed with a rheumatic condition, the median wait time was found to be much lower at 13.8 days, well within the recommended timeframe. Wait times for individual rheumatic diagnoses were found to vary, but patients with Raynaud phenomenon were the only ones with wait times outside of 28 days. In those patients diagnosed with a non-rheumatic condition, median wait time was 49.0 days, with infection-related diagnoses (reactive arthritis, serum sickness, transient synovitis) the only category inside of 28 days. These findings seem to suggest that our providers are proficient at triaging referrals based on likelihood of rheumatic disease, recommending earlier appointments for those deemed high-risk and those at low risk receiving later appointments.

It might be prudent in our case, and in pediatric rheumatology as a whole, to focus on strategies to decrease demand for non-essential referrals, targeting those 76.9% of new patient referrals that do not have a rheumatic disease. One potential way to reduce referrals for non-rheumatic disease is to target primary care provider education. Previous studies have reported on the inappropriate ordering of laboratory testing by primary care providers, including antinuclear antibody (ANA) levels and rheumatoid factor, and the improper interpretation of musculoskeletal pain as a symptom of rheumatic disease in the pediatric population [9, 10]. The Choosing Wisely campaign has also previously highlighted unnecessary autoantibody panels and repeat ANA testing in its “Top 5” practices that add to the cost of care without improving quality [11]. In our cohort, benign musculoskeletal pain made up 61.8% of our non-rheumatic disease diagnoses and 47.5% of all new patients seen during the study period. In the 1223 patients (46.4% of the cohort) who had musculoskeletal pain listed in the reasoning for referral to pediatric rheumatology, only 11.6% were diagnosed with a rheumatic condition. Similarly, of the 546 patients with “positive ANA” in their referral reason, either as the sole reason or in conjunction with other symptomology, 7.1% were diagnosed with rheumatic disease. By improving the ability of primary care providers to conduct musculoskeletal examinations and correctly order and interpret rheumatology laboratory testing, we may be able to limit referrals for non-rheumatic ailments.

An additional focus on the correct identification of benign musculoskeletal pain as a somatic symptom of depression and anxiety may also be helpful in reducing non-rheumatic referrals. In the last decade, numerous studies have shown a decline in the overall mental health of pediatric and adolescent patients, with significant increases in rates of depression, anxiety, and mental-health-related emergency department visits [12, 13]. There is a high prevalence of somatic symptoms in patients with depression and anxiety, and these patients may report only somatic symptoms at their initial primary care provider evaluation [14]. Such a presentation may lead to a pediatric rheumatology referral for evaluation of potential inflammatory causes of pain. In our population, somatic disorders like AMPS and chronic fatigue syndrome were diagnosed in 227 patients from 2019 to 2021, making up 8.6% of all new patients seen during that period. We may be able to reduce the amount of unnecessary pediatric rheumatology referrals by targeting these few simple topics for primary care education, especially in under-resourced communities.

This study is limited by its single-center population, which makes generalizability difficult to assess, especially given our practice of offering all patients appointments regardless of the likelihood of true disease during the referral process. The COVID-19 pandemic appearing during the study period may have also altered rheumatology referral quantity and quality. In-person clinic appointments were drastically limited between February 2020 and May 2020 leading to a signficiant drop in new patient appointments. However, the limited dip in referral numbers with a rapid return to baseline levels makes this source of error unlikely. March 2021 was an outlier in terms of referral quantity that does not have such an easy explanation. Patients with non-rheumatic diagnoses doubled from just the month before while rheumatic diagnoses stayed constant. The one potential explanation that has been discussed is that Alabama saw its largest peak in COVID-19 cases in December 2020 - January 2021, so it is possible that the increase in non-rheumatic diagnoses was related to non-specific post-viral symptoms. Finally, “no-shows” of scheduled referrals and those patients diagnosed initially while inpatient were not counted in our analysis, and it is unclear how this affected the overall rates of diagnosed rheumatic disease.