Performance of a fast-track pathway for giant cell arteritis in Waikato, Aotearoa New Zealand

Giant cell arteritis (GCA) is a large vessel vasculitis, which is the most common primary vasculitis in adults over 50 years of age. Ischaemic complications can arise, including vision loss that can be permanent. Early accurate diagnosis and prompt treatment is therefore critical.

Historically, temporal artery biopsy (TAB) has been the primary means of diagnosis; however, in recent years, there has been focus on colour Doppler ultrasound (CDUS) of the temporal and axillary arteries with particular interest in its low cost and availability.^1,2 Numerous meta-analyses support the performance of CDUS in the diagnosis of giant cell arteritis.^1,3–9

Fast-track pathways in GCA aim to have a risk assessment by a specialist (usually a rheumatologist or ophthalmologist) performed at the time of referral with a CDUS organised for the same day or next working day. Decisions regarding further tests and the need to continue corticosteroids are made rapidly, reducing harms of treatment or investigations to patients. The benefits of fast-track pathways include less vision loss, reduced time to diagnosis and fewer TAB requests.^2,10–13 Fast-track pathways also support primary care physicians, who have highlighted access issues in the rapid investigation of patients with suspected GCA.¹⁴

Most of the data on fast-track pathways do not take into consideration “real-world” practice. This is where our study is beneficial in reflecting the practical aspects of implementing a fast-track pathway for GCA in the hope of guiding other healthcare centres in the design of their own pathways.

Methods

The Rheumatology Department at the Waikato Hospital set up a GCA fast-track pathway at the end of 2013 (Figure 1) in collaboration with ophthalmologists. Patients suspected of having GCA were referred via telephone from primary care physicians, specialists or inpatient teams to either Ophthalmology (if visual symptoms were present) or Rheumatology (if no visual symptoms).

View Figure 1–2, Table 1–3.

If the risk of GCA was assessed as sufficient, a CDUS of the temporal arteries was requested, which would usually be done on the day of or next day after the referral. If the referral was made out-of-hours, corticosteroids would be commenced at the discretion of the on-call specialist. Following the CDUS, the specialist would make a decision regarding further investigation (i.e., TAB) and on appropriate follow-up.

Ultrasound examinations were performed by experienced vascular ultrasonographers with postgraduate- or master’s-level qualifications. The protocol involved scanning both temporal and axillary arteries. Ultrasounds early in the study period were performed using the GE Logiq 9 or Philips IU22 ultrasound system with transducers of operating frequency at 3–9MHz or 5–17MHz. Over the course of the study period, newer ultrasound systems and transducers were used. By 2021, there was an upgrade to GE Logiq 10 and Philips EPIQ (with higher frequency transducers of 4–18MHz or 3–12MHz). When scanning temporal or axillary arteries, the highest frequency transducer for adequate penetration was chosen.

Cases in the fast-track pathway have been collected prospectively from January 2014 using the referral request for CDUS. The requests from January 2014 to December 2022 were reviewed retrospectively to gather data on clinical symptoms, laboratory results and to generate a list of all cases in the fast-track pathway. The electronic records for each case were searched to collect information from referrals, results of investigations, treatment received by patients and final clinical diagnoses noted by treating physicians. In order to capture all patients going through the fast-track pathway, we also searched ultrasonography lists for patients who had a CDUS requested over this period. If the CDUS occurred alone or prior to a TAB, it was considered part of the fast-track pathway. TAB lists were also collected for the 3 years prior to the pathway to appreciate the rate of biopsy use prior to implementation of the pathway.

Ethnicities have been reported in conjunction with Statistics New Zealand (Stats NZ) reporting,¹⁵ and where patients identified with two different ethnicities (commonly Māori and New Zealand European), both were counted, thus giving a total percent as greater than 100%. This is consistent with how Stats NZ report their census data.

Given the pre-test probability of GCA was not often clear from the records, a risk score was applied, dividing patients into risk categories using an externally validated probability score established by Ing et al.¹⁶ This includes age, gender, clinical symptoms (i.e., new headache, temporal artery tenderness or reduced pulse, jaw or tongue claudication, diplopia or typical vision loss) with erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) and elevated platelets above upper limit of normal. Based on these factors, a “probability of having GCA” score was generated, organising patients into very low risk (<2.7%), low risk (<7%), moderate risk (<23%), high risk (<43%) and very high risk (>43%). This score appears to assist in the triage of patients with suspected GCA, though not without fault.^16,17

CDUS was classified as positive if the halo sign was present (defined as a hypoechoic ring around the lumen of the temporal or axillary artery) and indeterminate if only wall thickening was noted according to the radiology report. TAB was classified as positive if there was evidence of active inflammation consistent with GCA and the histopathologist’s report was supportive of active GCA. If there was a suggestion of possible past arteritis, this was not included as a positive TAB.

Statistical analysis

Descriptive data are presented as frequencies for categorical variables. Continuous variables are presented as a mean with standard deviation (SD) and a median with interquartile range (IQR). Where symptoms were not reported, this was reported as missing data; thus, percentage calculations are a valid percent rather than a total percent. Student t-Tests or Welch’s t-Test were performed where appropriate, as were non-parametric tests (Mann–Whitney U test). All analyses were conducted in IBM SPSS 29 (New York, United States). All significance tests were two-tailed and p-values of less than 0.05 were considered significant.

Ethics statement

National ethics approval was granted by HDEC (Reference: 2023 EXP 15448) and there was a local assessment through the Waikato Hospital who also approved the project (RD023025), which included review by Te Puna Oranga Māori Consultation Research Review Committee.

Results

Between January 2014 and December 2022, there were 664 patients who were referred through the fast-track pathway, with 648 individual patients. There were 16 duplicate episodes that have been excluded from analysis but are detailed in the Appendix.

Baseline characteristics

Patients managed through the fast-track pathway had a mean (SD) age of 70.5 (11.0) years and 69.3% were female. The age range was 17 to 96 years, with 25 patients (3.8%) being under 50 years of age. Ethnicity, clinical and laboratory features are detailed further in Table 1.

Of those referred through the fast-track pathway, 511/648 (78.8%) patients were managed by Rheumatology, 39/648 (6.0%) by Ophthalmology, 45/648 (6.9%) by a combination of both and 53/648 (8.2%) by other teams, which included General Medicine or Neurology.

Colour Doppler ultrasound

All patients in the fast-track pathway had CDUS performed. Out of 648 CDUS scans, 18 (2.8%) were reported as positive with the halo sign, 102/648 (15.7%) had abnormal vessel wall thickening noted and were thus labelled as indeterminate and 528/648 (81.5%) were negative. Axillary involvement was noted with vessel wall thickening (no axillary halos noted) in 52/648 (8.0%) of patients. A final diagnosis of GCA was made in 166/648 (25.6%) of patients.

Figure 2 illustrates the flow of patients through the pathway, outlining investigations performed. It shows how patients exited the pathway if they did not have GCA.

For patients with a halo sign (all of which were in the temporal artery), 17/18 (94.4%) had a final diagnosis of GCA. There was one patient with a halo sign on CDUS but clinical review assessed them as not having GCA. Out of the 102 patients with an indeterminate CDUS, 60/102 (58.8%) had a final diagnosis of GCA. Out of those with a negative CDUS, 89/528 (16.8%) had a final diagnosis of GCA. The sensitivity and specificity of CDUS compared to different reference standards are summarised in Table 2.

Non-specific vessel wall thickening was noted in 102 patients (i.e., indeterminate scans) and in 21/102 (20.5%) of these patients, this was supportive enough for a final diagnosis of GCA without a TAB needing to be performed. A further 61/102 (59.8%) patients required a TAB, of which 20/61 (32.8%) were positive and thus labelled as GCA, 41/61 (67.2%) were negative and of these, 19/41 (46.3%) were diagnosed with GCA. Thus, in total, 60/102 (58.8%) of indeterminate CDUS were associated with a diagnosis of GCA.

The sensitivity of a non-negative CDUS (i.e., either halo sign present or vessel wall thickening) was 46.7% (95% confidence interval [CI] 39.1–54.3%) and specificity was 91.1% (95% CI 88.3–93.4%). The positive predictive value was 64.2% (95% CI 55.4–72.4%) and negative predictive value was 83.3% (95% CI 80.0–86.3%).

Corticosteroids were started in 73.5% of patients. The mean (SD) was 13.8 (55.6) days of corticosteroids prior to CDUS and the median (IQR) was 1 (1–4) day. For patients with a final diagnosis of GCA, those with a positive CDUS had a mean (SD) duration pre-ultrasound of 0.7 (SD 1.1) days and a median (IQR) of 0 (0–1.3) days. This compared to a mean (SD) of 14.2 (56.2) days and median (IQR) of 1 (0–4) days in patients with a negative CDUS (p=0.007). When patients who were already on corticosteroids for polymyalgia rheumatica were excluded from analysis, the mean and median did not change for the positive CDUS group but was reduced to a mean (SD) 3.1 (13.1) and median (IQR) 1 (0–3) days for the negative CDUS group, which remained statistically different between the two groups (p=0.022).

TAB

TAB was performed in 229/648 patients and was positive in 42/229 (18.3%) patients. The mean (SD) size of biopsy was 16.0 (9.8) mm. The British Society of Rheumatology guidelines recommend a biopsy length of at least 10mm.¹⁸ In this cohort, 48/229 patients (21.1%) had a TAB length of less than 10mm, with 5/48 (10.4%) of these biopsies being positive.

Time to TAB was mean (SD) 14.5 (20.6) days with a median (IQR) of 10 (5–17) days. Duration of corticosteroids prior to TAB was mean (SD) 25.1 (46.1) days with a median (IQR) of 12 (4–25.5) days. The sensitivity and specificity of TAB with clinical diagnosis as a reference standard were 34.4% and 100% respectively.

TAB was avoided in 43/648 (6.6%) of patients with GCA after having a CDUS performed.

Prior to the pathway being established (i.e., 2011 to 2013), the mean annual incidence of all TAB requests was 28.1 (95% CI 23.0–33.3) per 100,000 people over 50 years. This reduced to 21.5 (95% CI 19.1–23.8) per 100,000 people over 50 years. Given the slight overlap of CIs, this was not statistically significant but was clearly trending downwards in conjunction with the introduction of the fast-track pathway.

Patients discharged from the pathway

Focussing on patients exiting the pathway, there were 376/648 patients (58.0%) who were discharged after a CDUS who did not have a final diagnosis of GCA. These patients had a mean (SD) of 15.6 (66.9) days and a median (IQR) of 1 (0–4) days of corticosteroids. Patients without GCA but who had a TAB and a CDUS performed had a mean (SD) duration of corticosteroids of 23.7 (SD 36.6) and a median of 12 (3–27) days. This is significantly longer than those patients who were discharged after only a CDUS (p<0.001). When patients who were previously on corticosteroids for polymyalgia rheumatica were excluded, the mean (SD) and median (IQR) for the CDUS only group was 3.0 (16.0) days and 0 (0–2) days respectively. In the CDUS and TAB group, this was a mean (SD) of 12.4 (12) and median 9 (2–19) days, remaining a statistically significant difference (p<0.001).

COVID-19 pandemic

Aotearoa New Zealand had an initial lockdown period due to the COVID-19 pandemic in March–May 2020, and a subsequent lockdown in August–September 2021 when there was community transmission. During these periods, the service remained in place where patients and physicians had access to colour Doppler ultrasound and TAB. Reviews occurred via telephone consultation unless the patient was unwell, in which case a face-to-face review was organised.

Discussion

This real-world study of the Waikato GCA fast-track pathway is the largest cohort published to date, alongside the study by Pinnell et al.¹⁹ who had 620 patients. The benefit of a fast-track pathway for GCA is clear with significant numbers of patients avoiding TAB, an invasive and costly investigation for both patients and healthcare systems.¹ This was evident for low-risk patients without GCA who had a non-positive CDUS result. This could also be appreciated by the down-trending rate of TAB requests with implementation of the pathway. The reduction in exposure to corticosteroids by only having CDUS to investigate GCA rather than needing a TAB was significant. This would likely translate to reduced corticosteroid toxicity for patients.

We note that our data is reflective of TAB access in Aotearoa New Zealand but may not be as significant a finding in other centres with rapid access to TAB. Rheumatology patients with suspected GCA are referred to vascular surgery, an under-resourced service where waiting times to TAB are often outside of the optimal window. For a portion of patients with a higher probability of GCA, the positive or indeterminate CDUS result was supportive enough to commit to the diagnosis and avoid the need for biopsy.

This study helps assess the performance of colour Doppler ultrasound in a real-world setting where, due to practical and safety reasons, corticosteroids are commenced at the time of referral. The sensitivity in our study is significantly lower than that reported in numerous meta-analyses. Table 3 provides a comparison of our data to other meta-analyses. Corticosteroid use appears to significantly decrease the chance of a positive colour Doppler ultrasound and may be part of the reason for our lower sensitivity. Pinnell et al. also performed a real-world study that had a lower sensitivity for colour Doppler ultrasound. They demonstrated the impact that corticosteroids had on detecting a positive CDUS and an increase in sensitivity when ultrasound was performed without corticosteroids.¹⁹

Corticosteroids appear to contribute to the disappearance of the halo sign.^1,20–22 Hauenstein et al.²⁰ noted that if colour Doppler ultrasound was performed on the first day of corticosteroid treatment, the sensitivity of the ultrasound was 88%. It dropped to 50% after 2–4 days of corticosteroids and 50% if patients had more than 4 days of corticosteroids.

While the number of positive colour Doppler ultrasounds is small, there is a larger number with increased thickening of the blood vessel wall. It remains unknown if any of these would have manifested a halo sign if corticosteroids had been withheld until after the ultrasound was performed. In the development of our fast-track pathway in Waikato, urgent corticosteroid treatment is mandatory to avoid consequences. Our protocol design and the restraints on our healthcare system cannot always guarantee a same-day CDUS.

Despite this emerging association, corticosteroid exposure may not entirely explain the discrepancy in the number of positive colour Doppler ultrasounds in our study compared to others, and the reasons are probably multi-factorial. There may be a larger number of low-risk patients entering the pathway, which reduces the number of true GCA cases and is reflected by 37.6% of patients being categorised as very low risk or low risk through the prediction score. Of note, Sebastian et al.⁴ and Melville et al.² had similar risk profiles in their study and yet had 37.6% and 30.2% positive scans respectively. The total number of cases of clinically diagnosed GCA in our study was 19.7%, which is smaller than to Sebastian et al.’s and Melville et al.’s studies of 25% and 34.1% respectively.^2,4

Technical factors including ultrasound machines and probes may play a role given that Aotearoa New Zealand has a resource-limited healthcare system with ultrasonographers using older, less advanced equipment at the start of the study period. Our ultrasonographers are experienced in vascular ultrasound; however, it remains unclear as to how this compares to experts internationally, specifically trained in the features of GCA on ultrasound. We are currently undertaking a retrospective audit on the ultrasounds to look for any features of GCA that had not been reported in the final report and will use this to further improve the fast-track pathway.

The majority of patients in this cohort are referred to Rheumatology with smaller proportions of referrals to Ophthalmology. Other studies have not noted this discrepancy between specialties.¹⁹ Our fast-track pathway protocol recommends that patients with any vision symptoms are referred to Ophthalmology. Rheumatology had 25% of their patients reporting any vision symptoms, suggesting some of their patients should be seen by Ophthalmology instead.

As we reflect on the implementation of this fast-track pathway in Waikato, we can visualise potential improvements. The fact that 3.8% of patients were under 50 years raises the question that many low-risk patients were entering the pathway, perhaps inappropriately. There is a significant proportion (189/648 [37.6%]) of patients who are low or very low risk entering the pathway, though 12/189 (6.3%) of these had a final diagnosis of GCA. It is difficult to know if this reflects the weaknesses of prediction scores for GCA or the entry of too many low-risk patients to the pathway. As a real-world study, physicians have varying levels of confidence in excluding GCA in low-risk patients and this is reflected in our data. We must note that when the pathway was developed in 2013, in order to validate the safety, efficacy and accuracy of the pathway for the investigation of GCA, all patients regardless of risk needed to go through the pathway. The CDUS was a new test to the department and caution was exercised. Given the pathway is now well established, review of entry criteria and exploring other prediction tools would be appropriate to reduce unnecessary patients going through. Clearly, ultrasound access has improved but access to TAB is delayed and more focus could be on improving this aspect of the pathway.

We acknowledge other limitations to this study. It is a retrospective study and there was missing data due to inadequate documentation. We have not included patients from the private health community in Aotearoa New Zealand. However, the private sector does not have the same rapid access to colour Doppler ultrasound as this pathway does; thus, most patients would have entered the public health system to access the pathway. Lastly, we acknowledge that there are no conventional pathway data to provide a true reflection of the benefit that this pathway has had on the Waikato community.

Conclusion

Fast-track pathways using temporal artery colour Doppler ultrasound in the investigation of GCA are beneficial to patients and our healthcare systems. There is a reduction in the number of temporal artery biopsies required and subsequent reduction in exposure to corticosteroids in patients without GCA. Corticosteroid exposure, while often mandatory in preventing serious complications, appears to reduce the sensitivity of colour Doppler ultrasound and remains an issue to consider when designing a fast-track pathway. Reflection around entry criteria to such pathway is also crucial.

View Appendix.