Objective. To examine complaints for which children were referred to a pediatric rheumatology service and to determine whether there are specific complaints that are more likely to indicate the presence of chronic arthritis or chronic, systemic inflammatory disease.
Methods. A retrospective chart review of 414 children referred to the pediatric rheumatology service at the Children’s Hospital of Oklahoma from April 1998 to July 2001.
Results. Musculoskeletal pain was the most common complaint for which children were referred (n = 226). Of these, 111 had musculoskeletal pain as an isolated complaint. One of these children had a chronic inflammatory disease. Another 115 children had pain as 1 of several reasons for seeking a rheumatology consultation, including positive results on laboratory tests (antinuclear antibody, erythrocyte sedimentation rate, and rheumatoid factor). Nineteen of these children had a chronic inflammatory disease, including 12 with juvenile rheumatoid arthritis (JRA). Thus, musculoskeletal pain as a presenting complaint had a strong negative predictive value for the presence of either JRA (0.95) or any other chronic inflammatory disease that might be characterized by arthritis. Children who were referred, in part, because of positive antinuclear antibody and/or rheumatoid factor tests were no more likely to have a chronic inflammatory disease than children who did not include such results as a reason for referral. Joint swelling, in contrast, was the most likely complaint to be associated with a diagnosis of JRA.
Conclusions. Musculoskeletal pain was the most common reason for referral to our pediatric rheumatology clinic. However, isolated musculoskeletal pain, in the absence of other signs or symptoms, is almost never a presenting complaint of children with chronic forms of arthritis. Children with arthritis more commonly present with complaints of joint swelling and/or gait disturbance. Neither ANA nor rheumatoid factor evaluations were useful in evaluating children with musculoskeletal complaints.
Pediatric rheumatology is a relatively new subspecialty of pediatrics, with an American Board of Pediatrics certification available only since 1992. Understandably, then, the number of physicians certified in this specialty remains relatively small; only 179 physicians currently hold board certification in pediatric rheumatology (American Board of Pediatrics). Furthermore, in a survey published in 1997, 45 of the 125 medical schools in the United States reported not having a pediatric rheumatologist on staff.1 Thus, many practicing physicians who care for children have had little or only limited exposure to pediatric rheumatology as practiced by formally trained pediatricians.
These considerations make pediatric rheumatology a particularly interesting subspecialty from the standpoint of illuminating how and when primary care physicians utilize pediatric subspecialty services. We have previously shown, for example, that primary care physicians frequently refer children with pauciarticular juvenile rheumatoid arthritis (JRA), the most common form of chronic arthritis in white children,2 to orthopedic surgeons, even when such children present with classic signs and symptoms of this disease.3 We have speculated that primary care physicians may have a preference for using specific consultants depending on the presenting complaint of the child. In the case of our previous study, children with pauciarticular JRA most frequently presented with joint swelling and gait disturbance. We speculated that primary care physicians are less likely to consider arthritis when confronted with this clinical situation and more likely to consider the panoply of conditions for which orthopedic surgeons have considerable expertise.4
Although there have been several studies that have examined diagnoses made by pediatric rheumatologists5–8 documenting the spectrum of rheumatic disease in children, limited medical literature exists about the complaints for which pediatric rheumatologists are consulted. The latter is of particular interest, because children are frequently referred to specialists or subspecialists for evaluation of complaints for which the diagnosis has yet to be made. We, therefore, undertook this study to answer 2 specific questions: 1) Are there particular complaints for which pediatric rheumatologists are consulted by primary care physicians? 2) Are there specific complaints or clusters of complaints in children that are likely to predict the presence of chronic inflammatory diseases for which pediatric rheumatologists have particular expertise?
MATERIALS AND METHODS
Patients and Patient Records
The study consisted of a retrospective chart review of all children seen for their initial pediatric rheumatology consultation at the Children’s Hospital of Oklahoma rheumatology clinic between April 1998 and July 2001. Institutional review board approval was obtained for this project. Records of children previously seen by another rheumatologist (eg, for follow-up of a previously diagnosed condition) were excluded. Patient records from these dates all document a chief complaint or chief complaints articulated by the patient (or parent) as described by Weed.9,10 In certain cases, identification of the problem for which the patient was referred was affirmed or clarified by direct contact with the referring physician at the time of consultation.
Charts were reviewed for the chief complaint as well as demographic information that included: the patient’s age, sex, referral source, duration of symptoms, and ultimate diagnosis. Diagnoses of JRA and specific subtype designations,11 systemic lupus erythematosus (SLE), 12 juvenile dermatomyositis (JDMS),13 ankylosing spondylitis,14 localized scleroderma, 15 Wegener’s granulomatosus,16 and acute rheumatic fever17 were made based on standard criteria. Noninflammatory forms of musculoskeletal pain, overuse syndromes, and mechanical forms of musculoskeletal pain were diagnosed based on history, physical findings, and, where appropriate, laboratory and/or radiographic findings as described by Sherry and Malleson.18 Viral arthritis was diagnosed in children with transient synovitis that resolved spontaneously within 2 months of onset.
Results were entered into a computer database and transferred to a commercially available statistical software package (GraphPad Prism, San Diego, CA). Comparisons between groups (eg, JRA versus non-JRA) were undertaken using 2-tailed independent t tests. Proportions of different patients in specific subgroups were analyzed from contingency tables using the Fisher exact test. Statistical significance was assumed for P values < .05.
The positive or negative predictive values (NPVs) for a particular laboratory test, symptom, sign, or finding (eg, musculoskeletal pain) for chronic arthritis in children were determined conventionally:
Charts of 482 children seen for an initial rheumatology consultation between April 1998 and July 2001 were reviewed. Sixty-eight children were excluded because they were referred by another rheumatologist and, thus, might represent a source of ascertainment bias, leaving 414 charts available for analysis. Demographics of the study population are shown in Table 1. The chief complaints for which children were referred are noted in Table 2
Musculoskeletal pain was by far the most common complaint for which children were referred to the clinic for the first time (n = 226). Of these children, 111 presented with musculoskeletal pain alone. The remaining 115 children with pain as part of their chief complaint presented with additional symptoms that included joint swelling (n = 24), gait disturbance (n = 11), fever (n = 14), morning stiffness (n = 5), fatigue (n = 16), rash (n = 7), and/or positive laboratory tests (n = 35).
Isolated musculoskeletal pain, in the absence of other signs or symptoms, was, as noted above, the chief complaint in 111 patients. Only 1 of these patients had a rheumatic disease (ankylosing spondylitis), and none had JRA. Indeed, of the 76 children diagnosed with JRA during the study period, only 12 (16%) included pain as part of their chief complaint, whereas the remaining 64 (84%) did not include pain at all in the complaints articulated by either the parents or patients. Thus, musculoskeletal pain, with or without other symptoms, had a strong NPV for JRA (NPV = 0.95).
Children with polyarticular JRA (n = 34) were no more likely to include pain as their chief complaint than children with pauciarticular JRA (n = 37). Six of the 34 children with polyarticular and 5 of the 37 with pauciarticular JRA complained of pain (P = .760). Even when children with JRA were subgrouped based on age, children with polyarticular disease were no more likely to articulate pain as part of their chief complaint than were children with pauciarticular disease, as shown in Table 3. Children 6 years of age or older with polyarticular JRA were the children most likely to articulate pain as part of their presenting complaint (6 [32%] of 19). However, this was not a statistically significant proportion when compared with children with polyarticular disease who were under 6 years old (P = .067). Boys with JRA were no more or less likely to complain of pain than girls. Four (15%) of 27 boys and 8 (16%) of 49 girls with JRA were referred with pain as part of their chief complaint.
Five children with the systemic form of JRA were diagnosed during this study period. Predictably, all 5 of these children presented with fever as the primary reason for seeking a rheumatology evaluation. Musculoskeletal pain was included as an indication for rheumatology evaluation in 1 of these 5 children.
As with our previous study, a significant percentage of children with JRA (n = 20; 26%) were referred from an orthopedic surgeon. This percentage was slightly higher for children under the age of 12 years. Sixty of the children with JRA were under the age of 12 years, and 19 of these children (32%) were seen by orthopedic surgeons before being referred to the pediatric rheumatology service. Not surprisingly, orthopedic surgeons were significantly more likely to refer children with JRA than primary care physicians (P = .008). However, orthopedic surgeons were no less likely to refer children for evaluation of musculoskeletal pain than were primary care physicians or other medical specialists. Of the 52 children referred from orthopedic surgeons, 28 (52%) were referred for evaluation of musculoskeletal pain, and 11 of these patients (21%) were referred for evaluation of musculoskeletal pain as an isolated symptom. The percentage of children referred by orthopedic surgeons for evaluation of musculoskeletal pain as an isolated complaint was not statistically significant compared with percentages referred by either primary care physicians or other medical specialists (P > .34).
JRA is not the only chronic inflammatory condition that may have arthritis as a prominent feature. When data were analyzed to include all chronic inflammatory diseases that may have arthritis as a component (spondyloarthopathy, SLE, JDMS, mixed connective tissue disease, and localized scleroderma), pain was still found to have a negative correlation with disease. Of the 226 children who were referred because of musculoskeletal pain (with or without other indicators for consultation), 20 had a chronic inflammatory disease (including the 12 children with JRA). Diagnoses associated with musculoskeletal pain in this group included 3 children with SLE, 4 children with spondyloarthopathy (3 with ankylosing spondylitis and 1 with psoriatic arthritis), and 1 child with Raynaud’s phenomenon. Thus, even when children with a broad spectrum of rheumatic diseases were included in the analysis, musculoskeletal pain was found to have a strong NPV (NPV = 0.91) in discerning which patients may have 1 or another of these diseases. These results are summarized in Table 4.
The vast majority of the patients with musculoskeletal pain as an isolated complaint (90 [81%] of the 111) had mechanical musculoskeletal or overuse syndromes as the explanation for their pain. Diagnoses in this category included hypermobility syndrome, plantar fasciitis, patello-femoral syndrome, and Osgood-Schlatter’s Disease. The remaining 11 had fibromyalgia / psychogenic pain syndromes19,20 (n = 4), toxic synovitis of the hip (n = 2), reflex sympathetic dystrophy (n = 1), serum sickness (n = 1), and viral arthritis (n = 3). Children with musculoskeletal pain as part of their chief complaint (eg, musculoskeletal pain and a positive antinuclear antibody [ANA] test) had a broader spectrum of diagnoses, including chronic inflammatory diseases such as JRA and SLE, as noted above. The largest number of these children (n = 47; 41%), however, had mechanical musculoskeletal pain/overuse syndromes.
Among the different complaints articulated by patients and/or parents, joint swelling was most strongly correlated with JRA. Of the 76 patients with JRA, 55 (72%) reported joint swelling as part of their chief complaint. In contrast, 52 of the remaining 338 patients (15%) reported joint swelling as part of their chief complaint. This difference was statistically significant (P < .0001). Among patients or parents reporting joint swelling as a chief complaint and/or reason for referral, the positive predictive value of joint swelling for JRA was 0.51.
Joint swelling, representing proliferative synovium and/or effusion, was detected by physical examination in 13 additional patients with JRA in our clinic without either parents or the referring physician identifying such swelling as a reason for the referral. In addition, synovitis was detected in 1 patient each with SLE, JDMS, and an undifferentiated connective tissue disease in instances where joint swelling was not indicated as a reason for referral. Other diagnoses associated with mild joint swelling that were not mentioned by the parent, child, or referring physician were: viral arthritis (n = 1), acute rheumatic fever (n = 1), acute lymphocytic leukemia (n = 1), and Henoch-Schönlein purpura (n = 1).
Gait disturbance was also highly predictive in identifying children with JRA. Twenty-three of the 76 children (30%) diagnosed with JRA presented with this complaint. In contrast, gait disturbance was present in only 14 of 338 children with diagnoses other than JRA. This difference was highly significant (P < .0001). However, the positive predictive value of gait disturbance for JRA was still relatively low (0.59). In children with JRA, gait disturbance, usually limping, was most prominent after periods of rest (eg, with arising in the morning or after naps) and improved with activity. Other diagnoses associated with gait disturbance included toxic synovitis of the hip (n = 5), mechanical musculoskeletal pain (n = 4), viral arthritis (n = 2), Legg Calvè Perthes disease (n = 1), infectious (tuberculous) arthritis (n = 1), and Kawasaki disease (n = 1).
Duration of symptoms also had some predictive value in identifying children with JRA who were referred to our rheumatology service. The duration of symptoms (mean + standard deviation) for children with JRA was 5.3 + 8.8 months (range: 1 week to 4 years), whereas children diagnosed with other entities was 18.3 + 27.5 months (range: 1 week-13 years). This difference was highly significant (P < .0001). Children with other rheumatic diseases (eg, SLE, JDMS, spondyloarthopathy) also had a longer duration of symptoms (14.2 months + 22.3) than children with JRA (P = .003).
In 135 children (both with and without pain as a chief complaint), laboratory test results were indicated as an additional reason for the referral. Test results included positive ANA (n = 90) or IgM rheumatoid factor (IgM-RF) determinations (n = 16), and/or elevated erythrocyte sedimentation rate (ESR; n = 47). Diagnoses made on the basis of the history, physical examination, and laboratory evaluation in rheumatology clinic are shown in Table 5.
Positive rheumatoid factor test results were cited as a reason for referral in 16 patients. Of these 16 patients, 3 (19%) had JRA. In all 3 patients, prominent synovial proliferation, morning stiffness, and fatigue would have established the diagnosis without a positive test. Children referred because of a positive rheumatoid factor test were no more likely to have JRA than children in whom IgM-RF test results were not cited as a reason for referral (P = 1.00).
Positive ANA tests were cited as a reason for referral in 90 patients. Of these, 14 children (16%) had JRA. Children in whom a positive ANA test was cited as a reason for referral were no more likely to have JRA than children for whom a positive test result was not indicated (P = .65). Among the children for whom a positive ANA test was cited as a reason for referral, there were an additional 10 children with other rheumatic diseases. These included 7 children with systemic lupus erythematosus, and 1 each with undifferentiated connective tissue disease, autoimmune thrombocytopenia, and Raynaud’s phenomenon. The latter had no clinical or laboratory evidence of systemic disease and had negative assays for antibody to DNA and the Ro and La antigens. Even when these children are included in the analysis, only 24 (27%) of the 90 patients referred for positive ANA tests had a chronic inflammatory disease that could be discerned on the basis of the history, physical examination, and other supportive laboratory studies. The constellation of musculoskeletal pain and a positive ANA test was particularly unlikely to identify children with rheumatic disease. Only 1 (4%) of 24 patients referred with these complaints had a chronic inflammatory disease, SLE. Thus, children referred for musculoskeletal pain and a positive ANA test were even less likely than the general study population to have a chronic inflammatory disease such as JRA, SLE, JDMS, or spondyloarthopathy (P = .038).
In 47 children, an elevated ESR was included as a reason for referral. Of these, 18 (38%) were found to have JRA, and an additional 3 were found to have other chronic inflammatory diseases (2 with systemic lupus and 1 with uveitis without arthritis). Children who were referred because of an elevated ESR were significantly more likely to have JRA (P = .009) than children in whom ESR results were not cited as a reason for referral. However, the positive predictive value for JRA of an elevated ESR was small (0.38) and only minimally higher for any chronic inflammatory disease (0.45).
Musculoskeletal pain is a common symptom of arthritis in adults and is an important measure of response to treatment in both rheumatoid arthritis21 and osteoarthritis22, 23 in this population. Our study presents the surprising finding that musculoskeletal pain is a rare presenting complaint for children with JRA. Indeed, the articulation of musculoskeletal pain had a strong NPV for either JRA or other chronic inflammatory diseases that are frequently associated with arthritis. They support a previous study by Inocencio,24 who did not report a single case of JRA among 61 children presenting with musculoskeletal pain in a primary care setting. As with our previous study,3 gait disturbance and joint swelling were the most common reasons for referring children who were eventually diagnosed with JRA.
Previous authors have noted that children with JRA may not verbalize pain even when their disease is active.25 The failure to articulate pain may be more common in younger children and reflect differences in how sensations from joints are interpreted.26 However, our study is the first, to our knowledge, to indicate that verbal articulation of pain is extraordinarily rare in children who present with JRA. It seems unlikely that this finding is attributable to children’s inability to localize or verbalize extremity pain. Children with acute rheumatic fever or self-limited forms of arthritis (such as viral arthritis) often presented with acute, severe joint pain. Furthermore, we did not find that older children with JRA were significantly more likely to express pain as part of their chief complaint than were younger children. We therefore speculate that the indolent processes of leukocyte infiltration into the synovium and proliferation of the synovial tissue lead to a subacute type of inflammation to which many children acclimatize and eventually accept as “normal.” This is not to conclude that children with JRA do not experience pain. Rather, we can only conclude that pain is not the most conspicuous feature of the illness at presentation.
Results from abnormal laboratory tests were indicated as a reason for referral in 33% of our patients. Neither ANA nor rheumatoid factor tests had any discriminatory ability to identify patients with JRA, as has been previously reported.27–30 Indeed, based on our data and on these previously published studies, it seems prudent to suggest that ANA tests not be performed on children with musculoskeletal pain and otherwise normal physical examinations, as the diagnostic yield is likely to be extraordinarily low, whereas the potential for triggering needless anxiety quite high. Similarly, it is doubtful that IgM rheumatoid factor tests have any utility in pediatrics, and they should not be part of the “ routine” evaluation of children with musculoskeletal complaints. Of the screening tests frequently used as part of an “arthritis work-up, ” ESR seemed to be more useful in identifying children with JRA than either IgM-RF or ANA tests. However, even when analyzed as part of a highly selected population (ie, patients referred to a rheumatology clinic with an elevated ESR identified as a reason for referral), the positive predictive value for JRA was low. Similarly, in a clinical setting highly suggestive of JRA (eg, a preschool-aged child presenting with gait disturbance and a swollen knee), a normal ESR should not be used as a criterion for excluding JRA in the differential diagnosis. In the final analysis, the diagnosis of JRA and related illnesses will continue to require a careful history and physical examination, and the laboratory will continue to have only a supportive role allowing the exclusion of other disease entities (eg, acute lymphocytic leukemia).
There is nothing in our data that allows us to be sure that physicians who referred children to the rheumatology service did so because they believed that those children may have a chronic form of inflammatory arthritis or a systemic inflammatory disease. The number of children referred because of positive results on ANA, IgM-RF, and/or ESR testing (33% of the patients in this study), however, suggests that physicians frequently consider arthritis or rheumatic disease in children presenting with musculoskeletal complaints. The frequency with which pain was identified as a chief complaint in our patients and the common use of ANA and/or IgM-RF results as a screening test suggest that primary care physicians may approach musculoskeletal complaints in children using models that are better-suited to adult rheumatology. This is hardly surprising, as it is likely that most physicians’ exposure to the specialty of rheumatology has been predominantly in the study of rheumatic diseases in adults with adult rheumatologists as teachers and mentors.
As we had noted in our earlier study, a significant number of children with JRA (26%) were referred initially to orthopedic surgeons. It was not surprising, then, that patients referred from orthopedic surgeons were much more likely to have JRA than children referred from primary care physicians or medical specialists, as such children were the beneficiaries of another level of scrutiny that in all likelihood assisted in narrowing and defining the diagnosis. More puzzling is why primary care physicians choose orthopedic surgeons as the primary referral choice for such a large percentage of children with JRA. It is interesting to speculate that this pattern may reflect the possibility that primary care physicians, using adult rheumatology models, do not consider JRA in the differential diagnosis of painless joint swelling and gait disturbance in young children. Equally likely is the possibility that orthopedic surgeons are chosen in this clinical setting because they are perceived to have expertise in caring for a broader range of musculoskeletal problems that might be heralded by painless joint swelling in children. We have considered the possibility that primary care physicians prefer to refer to orthopedic surgeons because of the relative inavailability of pediatric rheumatologists. As with our previous study,4 we believe that is an unlikely explanation. The wait for a new patient to be scheduled for an appointment in our clinic is usually 1 to 2 weeks, and children are seen within 24 to 48 hours if a primary care physician makes direct phone contact with the rheumatologist.
Our data support the previously published work of Bowyer and colleagues, 6 demonstrating the frequency with which nonrheumatic musculoskeletal conditions are referred to pediatric rheumatology clinics. In that study, examining the diagnoses made in pediatric rheumatology clinics in the United States over a 3-year period, “Arthralgias,” “Positive ANA,” “ Hypermobility,” and “Limb Pain” were the second through fifth most common diagnoses made after JRA and cumulatively represented a greater number of diagnoses (n = 3216) than JRA (n = 2017). These nonrheumatic musculoskeletal pain syndromes seem to be common, and if identified, respond well to symptomatic treatment.18 Osgood-Schlatter’s Disease may be considered an archetype for these entities, which are characterized by isolated musculoskeletal pain that is worse with activity and better with rest.
Our study is the first, however, to attempt to link specific complaints or reasons for referral with specific diagnoses. The current medical economic environment places a premium on judicious use of expensive subspecialty medical care. We have already demonstrated3 that the large majority of children with pauciarticular JRA, the most common rheumatic disease in white children, see another medical specialist (usually an orthopedic surgeon) before they see a pediatric rheumatologist. This was true even when children presented with the classic symptoms of the disease: joint swelling and gait disturbance. Our current study demonstrates that the most common reason for referral to pediatric rheumatologists, musculoskeletal pain, seldom leads to a diagnosis of JRA or other chronic illnesses associated with arthritis. Better delineation of which children may have chronic arthritis based on their presenting complaints, therefore, may allow more efficient use of subspecialty services such as pediatric orthopedics and rheumatology. Meanwhile, both academic orthopedists and pediatric rheumatologists need to be aware of the frequency with which children present to primary care physicians with musculoskeletal complaints and be prepared to train residents in the details of the history and physical examination that are critical to diagnosis.
This study was supported through a summer fellowship from the University of Oklahoma Health Sciences Center’s Native American Center of Excellence (Ms McGhee) and through a summer fellowship from the American College of Rheumatology and the University of Oklahoma summer honors research program (Mr Burks). Dr. Jarvis is the recipient of research awards from the National Arthritis Foundation and the Oklahoma Chapter of the Arthritis Foundation.
- Received October 1, 2001.
- Accepted February 12, 2002.
- Reprint requests to (J.N.J.) Department of Pediatrics, Rheumatology Research, Basic Sciences Education Bldg #235A, OU Medical Center, 940 Stanton L. Blvd, Oklahoma City, OK 73104. E-mail:
- ↵Bohan A, Peter JB. Polymyositis and dermatomyositis. N Engl J Med.1975;344– 347, 403–407
- ↵Dajani AS, Ayoub EM, Bierman FZ, et al. Guidelines for the diagnosis of rheumatic fever: Jones Criteria, updated 1992. JAMA.1992;87 :302– 307
- ↵Sherry DD, Malleson PN. Nonrheumatic musculoskeletal pain. In: Cassidy JT, ed. Textbook of Pediatric Rheumatology. Philadelphia, PA: WB Saunders Co;2001:362– 380
- ↵Sherry DD, McGuire T, Mellins E, Salmonson K, Wallace CA, Nepom B. Psychosomatic musculoskeletal pain in childhood: clinical and psychologic analyses in 100 children. Pediatrics.1991;88 :1093– 1099
- ↵Inocencio J. Musculoskeletal pain in primary pediatric care: analysis of 1000 consecutive general pediatric clinic visits. Pediatrics.1998;102 :1468
- ↵Cabral DA, Petty RE, Fung M, Malleson P. Persistent antinuclear antibodies in children without identifiable rheumatic or autoimmune disease. Pediatrics.1992;89 :441– 444
- Deane PMG, Liard G, Siegel DM, Baum J. The outcome of children referred to a pediatric rheumatology clinic with a positive antinuclear antibody test but without an autoimmune disease. Pediatrics.1995;95 :892– 895
- Eichenfield AH, Athreya BH, Doughty RA, Cebul RD. Utility of rheumatoid factor in the diagnosis of juvenile rheumatoid arthritis. Pediatrics.1985;78 :480– 484
- ↵Malleson PN, Sailer M, Mackinnon MJ. Usefulness of antinuclear antibody testing to screen for rheumatic diseases. Arch Dis Child.1997;77 :299– 304
- Copyright © 2002 by the American Academy of Pediatrics