Single versus repeat diabetes testing in older adults: Observations from the STAREE clinical trial

The growing burden of diabetes in older adults has increased the need for accurate and pragmatic screening tools. Current methods of screening for diabetes include the use of fasting plasma glucose (FPG), 2-hour plasma glucose following a 75 g oral glucose tolerance test (OGTT), and glycated haemoglobin levels (HbA1c). There is no current consensus on the recommended method to diagnose diabetes in asymptomatic individuals. While the World Health Organisation (WHO) recommends the use of OGTT as the primary screening tool, the American Diabetes Association (ADA) advocates for the use of FPG.1, 2 In older adults, evidence around the accuracy of these screening tests is even more limited. FPG and HbA1c may be preferentially used as the OGTT may be less convenient or well tolerated, both with their own advantages and disadvantages.3 FPG assays are inexpensive and readily available, but limited by the need for fasting and influenced by diurnal variation.1, 2 Intercurrent illness, acute stress and physical activity may also transiently elevate glucose concentrations. Conversely, HbA1c does not require fasting, is less prone to day-to-day fluctuations, but is influenced by factors affecting erythrocyte lifespan and haemoglobin concentration. For instance, iron deficiency and uraemia can cause false HbA1c elevation in individuals without diabetes. These situations are often more prevalent in older adults, thereby further impacting diagnostic accuracy. It has been recently proposed that having both elevated FPG and HbA1c assessed in a single sample at one timepoint may be sufficient to confirm diabetes diagnosis.4 This approach has been endorsed by the ADA. This pragmatic approach, while more convenient and reducing the need for repeat testing, is not without criticism due to the potential for transient elevations of glucose concentration in individuals without diabetes.1, 2 Given these concerns, repeat testing of an initially elevated or abnormal diabetes screening test at a subsequent timepoint is conventionally recommended to ensure appropriate confirmation of diabetes diagnosis and avoid inappropriate management, although the recommended time interval between initial and repeat testing is not clearly stated.1, 2 To evaluate the approach of utilising abnormal diabetes screening tests at a single timepoint versus repeat testing in community dwelling older adults, we utilised data collected from the screening visits for the large-scale STAtin therapy for Reducing Events in the Elderly (STAREE) trial (ClinicalTrials.gov Identifier: NCT02099123). STAREE is a double-blind, randomised-controlled trial examining the effects of statin therapy (atorvastatin 40 mg oral daily) versus matching placebo on the primary endpoints of disability-free survival and major adverse cardiovascular events over an average six-year follow-up period.5 Adults aged ≥70 years living independently without a history of cardiovascular events, diabetes, dementia or other life-limiting conditions were recruited from 1583 general practices across Australia between July 2015 and March 2023. Full details regarding the trial protocol and design including recruitment, inclusion and exclusion criteria have been previously published.6 Potential participants underwent an initial screening visit for trial inclusion and exclusion criteria. If eligible, individuals were invited to take part in a baseline screening visit which included routine pathology testing by their usual general practitioner (GP) for FPG and HbA1c. Diabetes diagnosis was suspected based on the World Health Organisation (WHO) diagnostic criteria for asymptomatic individuals, with the cut-off limits for FPG being ≥7.0 mmol/L (126 mg/dL) and HbA1c ≥6.5% (48 mmol/mol). These thresholds were applied at both initial and repeat testing. If both initial tests (FPG and HbA1c) were abnormal, diabetes diagnosis was suspected. If only one of FPG or HbA1c was abnormal, the abnormal test was repeated with recommendation to promptly attend follow-up testing. Individuals who had an abnormal result at repeat testing were also suspected to have diabetes. Those with a suspected diabetes diagnosis were excluded and returned to their GP's usual care for further workup. Individuals who did not perform repeat testing after an initial abnormal test were also excluded as the diagnosis of diabetes (an exclusion criterion of the trial) could not be confirmed. A total of 11 557 individuals without a history of diabetes underwent initial laboratory measurements for FPG and HbA1c. A total of 11 348 individuals (98.2%) had normal FPG and HbA1c, while 209 individuals (1.8%) had either: (i) abnormal FPG but normal HbA1c (124 individuals), (ii) normal FPG but abnormal HbA1c (41 individuals), or (iii) both abnormal FPG and HbA1c (44 individuals) (Figure 1). A total of 96 individuals (58.1%) from those with a single initial abnormal test (either FPG or HbA1c) had repeat testing (66.1% of those with initial abnormal FPG, 34.1% of those with initial abnormal HbA1c). On repeat testing, 24 out of 82 individuals (29.3%) had persistently abnormal FPG, while 4 out of 14 (28.6%) had persistently abnormal HbA1c. The average time intervals between the initial (1st) and second (2nd) tests for FPG and HbA1c were 54.2 and 115.4 days, respectively. Among those with a normal or abnormal repeat test, no significant differences in characteristics were observed (Table 1), but of those with a normal repeat test, a higher proportion had mild FPG elevation (7.0–7.4 mmol/L). 7.4 10 (14.7) 13 (19.1) 8 (11.8) 12 (17.7) 2 (2.9) 4 (5.9) 19 (27.9) 4 (14.3) 1 (3.6) 5 (17.9) 1 (3.6) 2 (7.1) 3 (10.7) 12 (42.9) 6.9 58 (85.3) 7 (10.3) 1 (1.5) 0 (0) 1 (1.5) 0 (0) 1 (1.5) 24 (85.7) 1 (3.6) 2 (7.1) 0 (0) 0 (0) 1 (3.6) 0 (0) In the community-dwelling older population recruited for the STAREE trial, initial diabetes screening identified 209 individuals with possible diabetes, with the most common glycaemic abnormality being elevated FPG. Only 21.1% had both abnormal FPG and HbA1c at initial testing. While repeat testing was recommended for individuals with one initial abnormal measure (FPG or HbA1c), only 58.1% were found to have had it done. We observed that repeat testing was more likely to occur after an abnormal FPG than after an abnormal HbA1c, although the reason for this is unclear and not evaluated in our study. The similarly low rates of diabetes confirmation upon repeat testing, approximately 30% across both diagnostic modalities, emphasise the importance of reinforcing repeat testing. This is essential to enhance diagnostic accuracy and mitigate the risk of inappropriate treatment, particularly in older adults who may be more vulnerable to treatment-related harms. There are limitations to our study. As this was an observation generated from the STAREE screening visits, the variables presented were limited to those collected for the parent trial; therefore, they may not fully account for all factors that specifically influence FPG and HbA1c. Information around comorbidities was also based on participant self-report and GP medical reports, which were subject to recall bias. Additionally, STAREE involved community-dwelling older adults in Australia without a history of diabetes, cardiovascular disease, and functional impairment, which was reflected in the low prevalence of newly suspected diabetes (1.8%). Therefore, confirmation rates and testing behaviour may differ in older adults with a higher comorbidity burden or in other healthcare systems. Next, the diabetes status of individuals who did not have repeat testing remains unknown. Identifying potential barriers to repeat testing in these individuals would be crucial to improving future screening practice; however, this information was not available. Lastly, the diabetes status of individuals who initially had both an abnormal FPG and HbA1c, who were excluded from study participation and not followed up, was also unknown. Given our observation that a significant proportion of individuals had normal tests after an initially abnormal test, there is a possibility that these individuals may also return with normal tests if repeated at a different timepoint. Further longitudinal studies focused on evaluating this approach in older adults are required. In summary, while FPG and HbA1c are suitable and convenient screening tools, repeat testing likely remains important to ensure accurate diagnosis. Further longitudinal studies are required to assess for potential differences in the accuracy of these screening tests in older adults. This study was conceptualised and designed by Jia Yong Tan, Clara Nalbandian, Chris Moran, Zhen Zhou and Sophia Zoungas, while the STAREE clinical trial was conceptualised and designed by Mark R. Nelson and Sophia Zoungas. Data analysis was performed by Jia Yong Tan, Clara Nalbandian, Zhen Zhou, Chloe Dawson and Sophia Zoungas. The initial drafts of the manuscript were written by Jia Yong Tan, Clara Nalbandian and Sophia Zoungas. All authors reviewed further drafts of the manuscript and approved the final manuscript. Jia Yong Tan is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. The authors thank the STAREE participants and study partners including the registered general practitioners. STAREE is sponsored by Monash University and has received funding from the National Health and Medical Research Council (NHMRC APP1068146 and APP1161503) and the National Heart Foundation of Australia (HF Stroke Prevention grant). The funders have had no role in the study design, collection, management, analysis, and interpretation of data or in writing or submission of reports or publications. Mark R. Nelson has received a speaker's fee from Medtronic. Sophia Zoungas is employed by Monash University and receives funding from NHMRC, MRFF, and the Commonwealth Department of Health and Aged Care. She reports previous payments to the institution (Monash University) from AstraZeneca, Boehringer Ingelheim, CSL Seqirus, Eli Lilly Australia, GSK, Moderna, and Novo Nordisk for participation in advisory and educational meetings. She is a Board Director of the Australian Clinical Trials Alliance. She declares no relevant declarations of interest with regard to this manuscript. The remaining authors have no disclosures to report. The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/dom.70539. The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.