Results
According to the original cohort based on registry data, 87 patients below 15 years were diagnosed with a spinal tumor in Denmark from January 1st, 1996 to December 31st, 2018.
17 patients were excluded due to misclassification, as they were not classified as classical spinal tumors, being brain tumors, Langerhans cell histiocytosis, lymphoma, hemangioma, hamartoma, sarcoma, and bone sarcomas with location in the spinal cord. Further, 12 patients were excluded due to missing records or insufficient data, (Fig. 2).
The final cohort of the present study included 58 patients aged 0-14 years diagnosed with a spinal tumor. 64% (37/58) had a prior musculoskeletal diagnosis, and 89% (33) of these were a misdiagnosis. Thereby a total of 57% (33/58) had a musculoskeletal misdiagnosis. The majority (64%, 21/33) of the misdiagnoses were non-specific including musculoskeletal pain (46%, 16/33), and accidental musculoskeletal lesions, often evaluated at the emergency room (15%, 5/33) such as torticollis, muscle strains, and sprains. A specific rheumatic misdiagnosis occurred in 36% (12/33) and included arthritis, arthropathy, osteomyelitis, discitis, inflammatory spondylopathy, reactive arthritis, and scoliosis.
In Table 1 we compare the clinical characteristics: age, gender, tumor type, metastases, treatment, and comorbidities of the group with versus without a musculoskeletal misdiagnosis. There was no difference in gender distribution between the two groups. Patients with a prior misdiagnosis were older (median age 10.5 vs. 5.9, p=0.04). Most of the tumors were low-grade tumors, present in 82% vs 68% (p=0.35). Ependymoma was the most common type of tumor in both groups (found in 27% vs. 20%, p=0.56). Low-grade astrocytoma was slightly more prevalent in the group with a misdiagnosis (24% versus 16%, p = 0.53), whereas high-grade astrocytoma, glioblastoma, medulloblastoma and primitive neuroectodermal tumor (PNET) occurred more often in patients without a misdiagnosis (32% vs. 9%, p=0.002), Table 1. Metastases to the brain were less frequent in the group with a misdiagnosis (9 % vs. 24%, p=0.15). Additional treatment was less prevalent in patients with a misdiagnosis: 21% received steroids or pressure relieving operations compared to 40% of patients without a misdiagnosis (p=0.15), and only 15% received chemotherapy compared to 40% of the patients without misdiagnosis, (p=0.04). Comorbidity occurred less frequent in patients with a misdiagnosis compared to those without (39% versus 60%, p=0.18), mainly due to a lower frequency of CNS comorbidity, including epilepsy and mental retardation, present in 6% with misdiagnosis and 24% without (p=0.15), Table 1.
In Table 2, we present a comparison of the clinical presentation for the group with versus without a misdiagnosis, with pain and paresis as the most common presentation in both groups. Musculoskeletal symptoms were present in 100% of patients with a misdiagnosis versus 56% of patients without (p= < 0.001). In the patients with a misdiagnosis the most common presenting symptom was localized pain in the lower limb, neck, and/or back, occurring in 81% compared to 28% of the patients without a misdiagnosis (p < 0.001), Table 2.
Neurological symptoms were less common in patients with a misdiagnosis (63%) compared to those without (96%; p=0.004), (Table 2). Furthermore, general symptoms such as fatigue, fever, and weight loss were less frequent in patients with a misdiagnosis (21% vs. 52%, p=0.02). Physical findings were highly prevalent and did not differ between the two groups, with abnormal neurological findings, particularly paresis, sensory deprivation, and disordered reflexes, being the most frequent (Table 2).
Mistreatment in the form of physiotherapy, chiropractor or painkiller occurred only among the patients with a misdiagnosis (36%, p < 0.001). Almost a third (27%) of the misdiagnosed children had received prior physiotherapy treatment. The time from first symptom until evaluated at the hospital (parental and primary care interval) did not differ compared to the cases not receiving physiotherapy, being respectively 60 days (IQR 22; 540, range 0;788), compared to 59 days (IQR 13; 100, range 3;730), p=0.34. Though, when comparing to the patients without a misdiagnosis they have a shorter interval with 27 days (IQR 4;91, range 0-880), p=0.10.
Table 2 and Fig. 3, provide a comparison of diagnostic intervals for patients with and without a misdiagnosis. Patients with a misdiagnosis experienced a longer total interval (time from onset of symptoms until treatment) with a median time of 165 days (IQR 35;318), compared to 97 days (IQR 28; 176) for those without a misdiagnosis (p=0.07), mainly due to longer parental and primary care intervals. Half of the patients (52%) with a misdiagnosis had a total interval exceeding six months, compared to one third (32%) of patients without a misdiagnosis (p=0.18).
The first hospital doctor was a pediatrician in 60% of all cases. However, for patients with a misdiagnosis, orthopedic doctors or general doctors in the emergency room were the first hospital doctor in 21% of cases, compared to zero cases for patients without a misdiagnosis (p=0.02). A referral to a specialist occurred after the tumor diagnosis was established in 79% of cases, resulting in a median specialist interval of 0 days (IQR 0;1).
In order to further investigate any patterns or red flags in the group with a musculoskeletal misdiagnosis we performed an analysis comparing the 12 patients with a specific “rheumatic” misdiagnosis (including arthritis, arthropathy, osteomyelitis, discitis, inflammatory spondylopathy, reactive arthritis, and scoliosis) to the 21 patients with a non-specific musculoskeletal misdiagnosis (including musculoskeletal pain and accidental musculoskeletal lesions). Further the 12 patients with specific “rheumatic” misdiagnosis were compared to the 25 patients without a misdiagnosis, Supplemental Table S1. The patients with a specific rheumatic misdiagnosis predominantly had low-grade tumors (83%), and metastasis did not occur in this group. High-grade tumors occurred in respectively 17% and 19% of the children with misdiagnoses and in 32% of the children without musculoskeletal misdiagnoses (p = 0.45). Pain as the first presenting symptom occurred in 95% of the patients with non-specific misdiagnoses compared to 42% in the group with specific rheumatic misdiagnoses (p<0.001). The frequency of other symptoms and findings did not differ for the children with specific versus non-specific misdiagnoses. When comparing the diagnostic interval for these three subgroups we found an even longer secondary care interval in the subgroup with a specific misdiagnosis, including a long first hospital doctor interval of 128 days (IQR 13; 153), compared to 10 days (IQR 1; 74) for the subgroup with non-specific misdiagnoses (p=0.10), and 7 days (IQR 1; 45) for the group without a misdiagnosis (p=0.01).
The impact of a musculoskeletal misdiagnosis on overall survival is illustrated in a Kaplan-Meier curve (Fig. 4). The group with a prior misdiagnosis tended to have a higher 5-year survival of 83% (95% CI 63-92%), compared to 66% (95% CI 44-82%) for the patients without musculoskeletal misdiagnoses (p=0.15). For the total cohort, 55% achieved complete remission and 25% (15/58) died due to tumor or treatment. The median duration of follow-up was 10.6 years (IQR 5.2; 14.7), being 8.8 years (IQR 7.1; 21.1) for children with misdiagnoses and 11.3 (IQR 8.6; 12.6) years for the children without, p=0.45.
Sequelae were highly prevalent among all patients, reported in 63% with a misdiagnosis and 76% of patients without a misdiagnosis (p=0.37). Severe sequelae occurred in 42% of patients with a misdiagnosis and 56% of patients without (p=0.43). There was no significant difference in the frequency of musculoskeletal sequelae when comparing the two groups (24% vs. 20%, p=0.76). Similarly, there was no significant difference in the frequency of central nervous system (CNS) sequelae between the two groups (52% vs. 60%, p=0.60). The most frequent CNS sequelae were paraplegia, paresis, sensory impairment, incontinence, and decreased cognitive function.