To our knowledge, this is the first study investigating the hospitalisation rates among newly diagnosed patients with PMR and the reasons for hospitalisation. In our study cohort, one in five newly diagnosed PMR patients was referred for hospitalisation at disease presentation before establishing the FTC. Our results demonstrated a significant difference (< 0.0001) on the initial CRP level between hospitalised and non-hospitalised patients (mean ± standard deviation-SD) 99.53 ± 59.36 vs 45.82 ± 36.96 mg/lt, without differences in the PMR-related symptoms. The high initial CRP levels and the consequent fear of a potentially severe underlying disease may be the main reason for admission to the hospital (Fig. 2).
In a Swedish study from 1995, which investigated the mortality among 220, temporal biopsy negative PMR patients [11] 72% of all PMR patients were hospitalised. However, in this study, the patients with initial GCA-related symptoms were not excluded, leading to a potential selection bias. Furthermore, inpatient days were not presented, and it is generally unclear if the hospitalisation included patients referred from the rheumatology outpatient clinic. A Norwegian study from 1997 analysed the incidence of GCA/PMR in the county of Aust Agder (catchment area of 98,000 inhabitants) over 8 years [7]. They encouraged GPs to refer patients with GCA/PMR suspicion to the local Department of Rheumatology. During the study period, only three patients were referred to the department of internal medicine with a tentative PMR diagnosis, while in the previous 8 years, 31 cases of PMR & GCA were diagnosed by other than Rheumatology departments. However, it is unclear if these patients were hospitalised or diagnosed in the outpatient clinic.
In a short communication a research group from Turkey reported, that about 30% of the PMR patients were hospitalised for a mean period of 7 ± 3 days before referral to rheumatology unit [12]. Somehow, in this study, initial GCA related symptoms as headache and sight loss were reported in 43 and 7.5% of the patients. Furthermore, no information is available about hospitalisations reasons and when during the disease patients were hospitalised. Another significant difference was the very high duration (13.2 ± 12.4 months) between the first symptom onset and admission or referral to a rheumatology clinic, compared to our 2013–2017 cohort (mean 12.30 ± 12.30 weeks). An explanation can be that the physician might not be so aware and experienced on PMR diagnosis in countries with low PMR incidences [2].
In our study, the main reason for hospitalisation was musculoskeletal symptoms in association with significantly elevated levels of inflammatory markers. Non-rheumatologists initially evaluated hospitalised patients in the emergency medical department, which could be the reason for the frequent initiation of treatment with antibiotics (Supplementary S1), a high number of inpatient days of care (Table 3), higher initial prednisolone dose (Table 1), and finally increased use of imaging modalities (Table 2) to exclude other conditions, compared to the non-hospitalised group. In the study by Schönau V et al. [19], patients with fever or inflammation of unknown origin (FUO and IUO) were referred to the Immunology and Infectious Disease Clinic for further examinations. With the help of a PET-CT scanning and later on confirmed by a rheumatologist, the diagnosis of PMR was made in 6% of patients with FUO and 18% with IUO. These results are in line with our observations that patients with PMR symptoms can be referred to a non-rheumatological department or hospitalised, leading to a diagnostic delay and excessive, expensive examinations.
Hospitalised patients had a significantly (p = 0.0018) shorter duration of symptoms before diagnosis compared to the non-hospitalised group. In 2018, using the FTC approach, the duration of symptoms before diagnosis decreased significantly and was similar to hospitalised patients. The difference in symptoms’ duration can be explained by the time from referral to evaluation in the rheumatological outpatient clinic (approximately 4–6 weeks) before establishing the FTC.
Twelve patients with a PMR diagnosis made by a rheumatologist did not fulfil the classification criteria for PMR. In 10 of the cases, the reason was initially normal inflammatory markers. ESR and CRP usually support PMR diagnosis, and increased inflammatory markers are mandatory for satisfying the PMR classification criteria algorithm [16]. Studies, however, show that 1.5–22.2% of patients with PMR present with normal acute-phase reactants [20, 21]. In our study, none of the 12 excluded patients was hospitalised.
The fact that PMR is the second most common rheumatic disease in the elderly [22, 23] and the number of people aged over 60 years expected to increase dramatically in all countries with a high prevalence of PMR [24] emphasises the importance of our findings. The rapid evaluation of patients with PMR by using the FTC approach led to a significant decrease in hospitalisation rates and the inpatients days of care, highlighting another advantage of the FTC as a potentially more cost-effective approach for diagnosing PMR.
One of this study’s strengths is that it performed in one endemic region for PMR [2, 3], where GPs are used to managing patients with these conditions [3]. The electronic journal system (Cambio COSMIC) ensured all medical records’ availability, hence collecting all necessary clinical and paraclinical data. Besides, the emergency department in SVS is the only available department for admissions in our catchment area, securing the registration of hospitalised PMR patients.
Our study has limitations. The study’s design is retrospective and included only patients referred to the hospital, thereby excluding patients diagnosed and treated in primary care. Nevertheless, it is a common international practice, GPs treating PMR patients and referring challenging cases to specialist healthcare for advanced diagnostics and treatment. Consequently, our study reflects the real population of PMR patients referred to specialist healthcare. The groups of hospitalised vs non-hospitalised patients were comparable in PMR related symptoms, but we could not compare the severity of symptoms, disability, or comorbidities, which may be one of the reasons for hospitalisation. Somehow, diagnostic imaging modalities were performed in the majority of hospital patient (Table 2) and other causes for PMR symptoms as malignancy or infections were excluded. Furthermore, none of the patients in the 2018 cohort developed malignancy during the first year of follow-up. The association between PMR and cancer is unclear, with previous studies yielding mixed results [11, 25,26,27,28].
Another limitation is that we did not routinely examine the 2013–2017 PMR cohort for subclinical GCA. It is reported in 16–21% of PMR patients, and the possibility of coexistent GCA arises in PMR patients with constitutional symptoms and markedly elevated acute phase reactants ([29]). In our cohort, the hospitalised patients underwent a CT scan of chest/abdomen with contrast or PET-CT in almost 40% of cases (Table 2) without evidence of large-vessel vasculitis. Furthermore, none of the hospitalised patients has was diagnosed with GCA during the first-year follow-up after the treatment initiation. Another limitation of our study is that we included patients with an established PMR diagnosis but not patients with PMR suspicion ending with other final diagnoses. We identified our patient population using an electronic database, where patients are registered according to their final diagnosis and not the tentative/referred diagnosis. More studies evaluating the impact of FTC on patients with PMR suspicion are warranted in the future.