Introduction

The time to diagnosis and thus the start of therapy has an impact on the progression of chronic inflammatory rheumatic diseases including rheumatoid arthritis (RA), psoriatic arthritis (PsA) and axial spondyloarthritis (axSpA). In Germany, approximately 1.2 million patients are affected by one of these three diseases.1 2 According to the European League Against Rheumatism (EULAR) recommendations for the management of early arthritis, symptomatic patients should be seen by a rheumatologist within 6 weeks after the onset of complaints and treatment with disease-modifying antirheumatic drugs (DMARDs) should be commenced within 12 weeks.3 4 Prompt diagnosis and initiation of therapy can prevent irreversible joint damage and guide patients into a long-lasting remission. Thus, economic benefits are generated due to the reduction of direct and indirect disease costs, such as incapacity to work costs.4–9 From a health economic point of view, initiating treatment within the first 12 weeks reduces the likelihood of cost-intensive therapy with biological drugs and targeted-synthetic DMARDS.10 Implementing these theoretical goals is challenged by the lack of rheumatologists in many countries.

This is also reflected in a 2020 EULAR survey in which 52% of 1873 patients and 59% of 1131 rheumatologists from 35 European countries described specialist consultation within 6 weeks as the biggest challenge.11 Therefore, EULAR calls for a feasible and valid approach to support especially general practitioners in the diagnosis and referral of patients with early RA since a standardised procedure for transferring patients with musculoskeletal problems from primary care physicians to rheumatologists does not exist.3

Approximately 100 million people in Europe are diagnosed with a rheumatic and musculoskeletal disease (RMD). It is speculated that at least 100 million more people live without a diagnosis. In 2017, the EULAR launched the Europe-wide campaign entitled ‘Don’t Delay, Connect Today’ to highlight RMDs as a public health concern of pandemic proportions and that early diagnosis and timely access to treatment can prevent further damage and burden on the individual and society (www.EULAR.org). While this campaign endeavoured to raise awareness, our study aimed to address a reason for the delay-the lack of rheumatologists prevents timely treatment of time-critical diseases.2 3 Additionally, the shortage of rheumatologists is exacerbated by geographic distribution. While the average distance in Germany to a rheumatologist for patients with RA in urban areas with >500 000 inhabitants is 12 km, patients from rural areas (<5000 inhabitants) have to travel an average of 32 km to a rheumatologist.12 Thus, the Rheuma-VOR study investigated a novel approach of ‘coordinated cooperation’ between the caregivers of patients with rheumatic diseases to improve the early diagnosis of inflammatory arthritis.

MethodsStudy design

Rheuma-VOR was a prospective non-randomised multicentre study aiming to establish a network to optimise rheumatological care and diagnostic processes.

From 1 July 2017 to 31 December 2020, patients were included in the study to pass a baseline data collection and a 1-year follow-up appointment. The inclusion criteria for the study were as follows:

Age at inclusion ≥18 years.

(Suspected) inflammatory rheumatic disease with International Classification of Diseases codes (ICD) M05, M06, M06.9, M13.0, M45, M46.1, M46.8, M07, M09.0 or L40.5.

Sufficient language skills and signed informed consent form.

Four university medical centres, three rheumatology centres, local rheumatological specialists, the associations of statutory health insurance physicians and primary care physicians, and the regional associations of the patient advocacy groups from four federal states with a population of about 14 million adults (approximately 20% of the German adult population), participated in the study. Two scientific institutes evaluated the results of the study with different focus. One evaluating institute was responsible for clinical effect evaluation and the other was responsible for health economic evaluation.

Primary care physicians such as general practitioners, internists, dermatologists and orthopaedists used screening questionnaires to document potential early cases of RA, PsA and axSpA, based on the characteristic symptoms of the classification criteria.

Additionally, the primary care physicians had the opportunity to join interactive training courses in basic rheumatology. The interactive training is based on a lecture on early symptoms, diagnostic criteria and examination procedures. During the study period, 20 open access courses were performed.

Primary care physicians sent the questionnaires by fax or a newly developed app for mobile devices to the federal state specific coordination centres. Multiprofessional teams in the specific coordinating centre assessed the likelihood of early arthritis. The teams consisted of a specialist in rheumatology, a clinical nurse specialist and a secretary.

The clinical nurse specialists processed, completed and prepared the screening questionnaires for the rheumatology specialists. The specialist evaluated and triaged the available data for the presence of one of these three conditions. If the criteria for referral were met (characteristic symptoms with elevated C reactive protein or erythrocyte sedimentation rate), the patient was assigned to obtain an appointment with a cooperating rheumatologist within 6 weeks. Appointments were coordinated by the secretaries with one of the 72 in the network participating rheumatologists. When the criteria for referral were not met, patients were returned to standard medical care without a rheumatological examination (figure 2).

When the participating rheumatologists confirmed a rheumatic disease, physicians and patients received questionnaires on sociodemographic data, diagnosis, disease activity, medication, health-related quality of life, well-being and activities of daily living. After 12 months, the questionnaires were issued a second time.

Additionally, from October 2018 to the end of the survey, a 15 min screening consultation was integrated at the Rhineland-Palatinate coordination office. All patients meeting the appropriate criteria but with a scheduled appointment later than 4 weeks were examined by rheumatologists within the coordination centre without further diagnostics (no lab, no X-ray) (figure 1).

Figure 1

Flow chart Rheuma-VOR. GP, general practitioner.

The intervention of the study is the implementation of the coordination centres in a way of a professional triage and schedule office, which can be used by all physicians in order to speed up time to diagnosis and a relief for rheumatologists. This is in contrast to the standard German care in which the general practitioner assesses the symptoms, and refers the patient, for example, one after another to a dermatologist, orthopaedist and rheumatologist until the diagnosis is made. Figure 2 shows the common referral process for rheumatic diseases in German standard care versus the Rheuma-VOR process with the focus on the coordination centre.

Figure 2

Common referral process for rheumatic diseases in German standard care versus Rheuma-VOR.

Data collection and quality control

Based on the study design, the screening questionnaires were derived from strict classification criteria: the disease-specific ACR classification criteria for RA, the slightly adapted CASPAR criteria combined with the ‘Psoriasis Epidemiology Screening Tool‘, and the ‘Early Arthritis for Psoriatic Patients‘ questionnaire for PsA.13–16 Slightly adapted ASAS classification criteria were used for axSpA.17 18

If a rheumatic disease was diagnosed, further assessments were performed to collect disease-specific and patient-related outcome parameters (table 1). Disease-specific remission was specified by the Disease Activity Score 28 for RA and PsA (<2.6) and Ankylosing Spondylitis Disease Activity Score score (<2.1) for axSpA.19 20

Table 1

Disease activity parameters, functional assessment and quality of life questionnaires

The Rheuma-VOR data were compared with a weighted cohort for effect evaluation (n=2139) and a matched reference cohort for economic evaluation (n=806) from the DRFZ, reflecting the standard of care. Since 1993, the DRFZ prospectively collects annual epidemiological cross-sectional and longitudinal data from rheumatic centres, such as university hospitals, acute care hospitals and rheumatologists across Germany.21 The data provided from the DRFZ cover the time period from 1 January 2015 to 31 June 2017 to ensure that patients are not included in both cohorts, as some of the rheumatologists participating in Rheuma-VOR also support the DRFZ’s annual documentation.

Time to first rheumatologist contact was documented retrospectively. Inclusion criteria were as follows:

Age at inclusion ≥18 years.

Signed informed consent form.

(Suspicion of) inflammatory rheumatic disease

Sufficient language skills to complete the questionnaire.

For comparisons with Rheuma-VOR, only persons affected by one of the three inflammatory rheumatic diseases for the first time and recorded in the National Database were selected.

Outcome measures

In addition to a qualitative process evaluation, Rheuma-VOR was evaluated on two aspects within the first year following diagnosis.

Disease-specific effects.

Health economic effects.

The reduction of the time to diagnosis is the focus in this study. Time from first medical contact to diagnosis between the early (1 July 2017–31 December 2018) and late (1 January 2019–31 December 2020) phases was the first coprimary endpoint. This was further evaluated by comparing time from symptom onset to diagnosis between Rheuma-VOR and standard care. The screening performance of the primary care physicians was the second coprimary endpoint (proportion of confirmed diagnoses).

The reduction of the disease activity, an increase in quality of life and overall well-being, improvement in fatigue, depression, functional ability, work ability, and reduction in drug and medical costs were analysed as secondary endpoints. Costs were determined by weighting the resource use with information from public cost databases. Drug costs were calculated using the current prices from the German Lauer-Taxe 4.0. Cost data were discounted in accordance with the recommendations for health economic evaluations.22

Only patients with both baseline and 1-year follow-up data were considered for the follow-up analysis.

Statistical analysis

All analyses were performed using R software version 4.0.3 and SPSS 26 . The first coprimary endpoint time to diagnosis was analysed using a Cox proportional hazards model with a random intercept to account for correlations within different regional coordinating centres. The second coprimary endpoint screening performance (percentage of patients with a suspected diagnosis confirmed by the rheumatologist) was evaluated using a logistic regression model considering site-specific effects by a random intercept model. Both models compare patients from the early with those from the late phase. We postulated an initial screening performance of 50% of the primary care physicians and an increase by 5% during the course of the study.

For the effect evaluation, the two cohorts were weighted. In this way, a probability of belonging to the introduction or consolidation phase was estimated for each patient and calculated with stabilised weights according to Robins et al.23

For the secondary endpoints, univariate and multivariable linear regression models with a random intercept to account for correlations within different regional coordinating centres and patients were performed comparing consultation 1 with consultation 2.

Two-sided 95% CIs and p values were calculated. The confirmatory significance level was p<0.025 for the two primary and descriptively set to p<0.05 for the secondary endpoints (no correction for multiple testing). Independent variables were imputed with multiple imputation algorithms using the mice-package in R.24

No alpha error correction was used to counteract multiple comparison problems of secondary endpoints.

To compare the health economic effects between the Rheuma-VOR and standard of care cohort, a case–control matching based on age (±5 years), gender, primary diagnosis and the number of consultations were conducted. Data for cost structure analysis are based on Huscher et al.7 χ2 tests and Mann-Whitney-U tests were used for the analysis. The significance level was set to p<0.05.

Power analysis

Sample size calculations were based on the coprimary endpoint describing the proportion of suspected diagnoses confirmed by rheumatologists. Due to lack of prior data, we adopted an initial worst-case scenario of 50% for the early phase and an improvement to 55% in the study’s late phase. Using a χ2 test with a two-sided significance level of 2.5%, a power of 90%, a ratio for patient inclusion of 1:1.25 (early to late phase), and a drop-out rate of 28% resulted in a minimum sample size of 6875 needed patients.

ResultsStudy population

During the study period, 7049 screening questionnaires were referred to the coordination centres. Following assessment by the multiprofessional team, the suspected diagnosis was upheld for 5124, of which 1363 patients were seen at the 15 min consultation in Rhineland-Palatine. Based on the information provided by the primary care physicians by fax, by app or the specialist assessment during the screening consultation, 4143 patients with a suspected diagnosis were referred to the rheumatology specialist level. Of the 3857 patients which were seen by a rheumatologist, the diagnosis of RA, PsA or axSpA was confirmed in 1778 patients. Finally, 1537 patients were included in the study (consultation 1), of which 592 had a follow-up after 1 year (consultation 2) (figure 1). Summarised, 37% of the primary suspected cases (n=2625) could be excluded from a consultation with the rheumatologist. Diagnostic performance was also significantly increased by the implementation of the screening consultation (OR 0.553, 95% CI: 0.453; 0.677, p<0.001). The drop-out rate was 10.2% (figure 1).

Nationwide geographical distribution of Rheuma-VOR across the four federal states is presented in figure 3.

Figure 3

Nationwide geographical distribution of Rheuma-VOR primary care physicians and rheumatologists.

Due to the study design, and study inclusion until the the end of the study (31 December 2020), a total of 404 patients which were included in Rheuma-VOR after the 1 January 2020 could not be considered for the follow-up analysis.

Based on the 1537 included patients, the main referring physicians were general practitioners (47%), followed by internists (16%), orthopaedists (16%) and dermatologists (9%). The remaining 12% were neurologists, surgeons, ophthalmologists, geriatricians or even rheumatologists who made an initial diagnosis and included patients in the study.

Additionally, the 700 excluded patients from the screening consultation level were used as the basis for a review of false-negative diagnoses. A total of 532 patients were eligible for follow-up. The difference of 168 patients resulted from definite rejections of the suspected diagnoses due to the presence of osteoarthritis, an orthopaedic or other rheumatic condition (eg, gout or systemic lupus erythematosus), an already existing rheumatic diagnosis, or from revocation of further participation in Rheuma-VOR. The period from the rejection of the diagnosis to the contact was at least 3 months and three attempts were made to contact the patient. Finally, 445 people were contacted, 30 of whom stated that they had been diagnosed with RA, PsA or axSpA. This corresponds to a share of just under 7%.

Patient characteristics

The most common diagnosis was RA (58%), followed by PsA (27%) and axSpA (15%).

The mean age of participants corresponded with the age of manifestation of the diseases (54±16 years); 57% of participants were female and 31% reported being smokers. The mean body mass index was 28 kg/m2, and 29% of the patients were already retired at consultation 1. On average, two comorbidities were present at diagnosis (table 2).

Table 2

Baseline characteristics and development in the first year after diagnosis

The average waiting time from registration at the Rheuma-VOR coordination centres until the rheumatological appointment was 30±37 days. In addition, 16±15 days of waiting time have to be added in Rhineland-Palatinate due to the 15 min screening consultation.

A main aim of the project was to support patients from rural regions, with 62% of included patients residing in towns with <5000 to 20 000 inhabitants (figure 3, table 2). Average distance to the rheumatologist was about 42 km.

Coprimary endpoints

The first coprimary endpoint time from first medical contact to diagnosis was significantly reduced between the two phases of the study (HR 1.27; (95% CI 1.17 to 1.37); p<0.001) and, similarly, from onset of symptoms to diagnosis in comparison to standard care (figure 4): RA: 0.55 years (mean) vs 2.31 years (p<0.001), PsA: 2.43 years vs 4.41 years (p<0.001) and axSpA: 3.92 years vs 8.41 years (p<0.001) (table 3).

Figure 4

Symptom onset until time to diagnose of Rheuma-VOR and standard care. DRFZ, German Rheumatism Research Centre; PsA, psoriatic arthritis; RA, rheumatoid arthritis; SpA, spondyloarthritis.

Table 3

Symptom onset until time (years) to diagnosis of Rheuma-VOR and standard care

The second coprimary endpoint screening performance did not show a significant difference between early and late phase. However, in the early phase the screening performance was already much higher than originally assumed (75% vs 50%) (online supplemental file 1).

Secondary endpointsDisease activity

Patients in the Rheuma-VOR cohort had moderate to high disease activity at inclusion, which improved significantly at the follow-up consultation; 47% (n=228) of all patients achieved remission at consultation 2 (p<0.01) (table 2).

Detailed analysis revealed a comparable, uniform picture during the project: disease activity, measured with the respective RA-specific, PsA-specific and axSpA specific objective and subjective parameters decreased significantly (p<0.01). Functional impairment in the axSpA patients was minimal at consultation 1, reflecting early diagnosis. During the study, there was a numerical improvement to 0.9 at consultation 2 (p<0.06) (table 2).

Health-related quality of life, well-being and activities of daily living

Parallel with the decrease in disease activity, all patients reported significantly improved outcome parameters concerning functional ability, patient well-being, quality of life, fatigue and depression (p<0.01) (table 2).

Resource use analysis

Health economic evaluation revealed that optimising patient management with a reduction in the time to diagnosis resulted in considerable savings in resource use. Patients in the Rheuma-VOR cohort had fewer hospitalisations than standard care at consultation 1 (7.2% vs 15.7%) and consultation 2 (9.3% vs 12.7%). Regarding the ability to work, 6.1% of the Rheuma-VOR cohort patients were on sick leave at consultation 2, compared with 9.3% in the control group. The major part of the costs arose from drug therapy. In the Rheuma-VOR cohort, 2.4% (n=34) received biologics therapy at consultation 1, in contrast to 11.8% (n=165) in standard care (online supplemental file 1).

Incremental cost-effectiveness ratio

Cost-effectiveness of Rheuma-VOR was analysed by comparing the cost–utility ratio for both cohorts. At lower care costs (including the costs for the coordination centres), a higher benefit is achieved at consultation 2.

Further calculations were conducted for patients with data available for both consultations. Costs necessary to increase the quality of life by 0.1 points, that is, by 10%, (consultation 1 to consultation 2) were calculated. In the Rheuma-VOR cohort, on average, €2445, 95% CI (€1955.79 to €3041.50) per patient was spent compared with €3107, 95% CI (€1958.28 to €3997.62) for the standard cohort.

Thus, the cost–benefit ratio in the Rheuma-VOR cohort showed higher effectiveness than the standard cohort since improvement in quality of life is associated with lower costs.

Discussion

The Rheuma-VOR study has highlighted that coordination between rheumatic healthcare providers significantly improved time to diagnosis, especially from symptom onset until diagnosis. The provided DRFZ data show comparable results to general and former published DRFZ-data.21 Patients suffering from the first symptoms of a chronic rheumatic disease immediately benefited from earlier diagnosis with improved health-related quality of life and increased daily activities and work ability. The benefit and relevance of early diagnosis and early medication were already shown and are included in current guidelines.2 3 5 8 25

Contrary to our assumptions, 73% of the initially referred suspected patients were confirmed by the coordination centres. The screening questionnaires may have contributed to this unexpected high rate. It has to be mentioned that only <5% of all participating GPs received training during the Rheuma-VOR study due to pandemic reasons. All primary care physicians received a feedback for their screening questionnaires to improve the accuracy of the screening. Yet, the accuracy of the screening questionnaire was much better than expected. Only 15% of screening questionnaires needed to be revised by the primary care physician.

In 2017, the Interdisciplinary Commission on Healthcare Quality of the German Society for Rheumatology (DGRh) updated the 2008 memorandum on rheumatological healthcare in Germany.2 The update examined the need for rheumatologists and determined the gap between needs and supply. According to the analysis, at least two rheumatologists are required for the outpatient care of 100 000 adult inhabitants. This is equivalent to 1350 rheumatologists in Germany, which currently has only 812 rheumatologists.2 26 While some larger cities (eg, Hamburg or Hannover) meet these criteria, many rheumatologically underserved areas, especially in rural regions such as Rhineland-Palatinate and Saarland, do not. Therefore, an additional bottleneck is caused by a ruralurban gap.2 27 The Rheuma-VOR study allowed referral to rheumatologists only by urgency and not by place of residence, and therefore, balanced the existing rural–urban gap in care since 62% of the included patients lived in cities with up to 20 000 inhabitants. In addition, the involvement of primary care physicians in therapy decisions, immediate support via consultation with the coordinating centres, and continuous training will strengthen the competence of the primary care physicians and consequently further relieve specialist resources.

The Rheuma-VOR study was quickly communicated among GPs via the Association of Statutory Health Insurance Physicians. The fact that there was a way to quickly send a patient with rheumatic complaints to a coordination centre that takes care of this patient is another explanation for the success of Rheuma-VOR, given regular waiting times of 1 year and longer for an appointment with a rheumatologist. It can be noted that all patients were seen by a rheumatologists within an average waiting time of about 4 weeks (29.87±37.24), respective approximately 2 weeks (16.28±15.11 days) considering the 15 min screening consultation in Rhineland-Palantine.

Optimising coordination by reducing waiting time until diagnosis leads to significant savings in resource use and favourable cost-effectiveness compared with standard care. Studies revealed a positive correlation between the rising severity of the disease and increasing costs.28 Regarding the number and duration of hospital stays, the periods of incapacity to work, and the medication used, further positive effects could occur in the long term due to the early and close-meshed initiation and control of therapy. It has to be mentioned that the Rheuma-VOR does not consider the time from symptom onset until the visit to the primary care physicians and initial referral to the rheumatologists which can extend the time to rheumatologist significantly. Some of the National Health Service data published in 2019 showed an average of four visits at the general practitioner/primary care physicians with a median waiting time of 6.9 weeks (IQR 2.3–20.3) for referral to a rheumatologist in patients suffering from RA. Patients who purchased over-the-counter medications took longer to seek help.29 Thus, an additional focus should be on rising awareness in different sectors of the healthcare systems.

Defining the DRFZ cohort as the standard of care is not entirely accurate, since the National Database of the DRFZ is not a registry, and the aim is not to achieve full coverage. Participating facilities are encouraged to enrol persons in the National Database on an unselected basis. However, within Germany, the National Database currently represents the best possible data collection in the field of rheumatology to reflect the standard of rheumatological care.

Additionally, due to data protection reasons, it was possible to control potential false-negative diagnosis only at the level of the screening consultation following the triage level ‘coordination centre’. As a result, the actual proportion could be higher due to this bias.

In Lower Saxony and Rhineland-Palatine screening questionnaire for collagenosis and vasculitides were developed and implemented. Follow-up time was extended in Rhineland-Palatinate for additional 2 years to obtain further data and optimisation approaches.

Although Rheuma-VOR has significantly improved care and rheumatologists were relieved of 37% of patients, the problem of the insufficient number of rheumatologists still exists. Therefore, we envision Rheuma-VOR as the core of an optimisation strategy around which further initiatives must be established, for example, such as the effect of nurse-led care, an independent study in Rheuma-VOR.30 Several campaigns are ongoing to increase interest in rheumatology.31

This novel concept has shown an improvement in rheumatological care, patient-reported outcome parameters, and cost savings by coordinating the cooperation of general practitioners, rheumatologists and patients in a nationwide approach. Additionally, the study shows effectiveness to decrease inappropriate rheumatological referrals by 37% as well as an approach to reduce the time to diagnosis. The Rheuma-VOR concept of ‘coordinated cooperation’ is not restricted to RMDs but can also be transferred to other clinical critical courses and bottlenecks in care (eg, neurology, psychiatry). Rheuma-VOR represents a general model for action and structure for the comprehensive care of diseases if the time factor for the response to therapy and the overall outcome is critical.