The management of type 1 diabetes (T1D) is a dynamic and evolving field. Glycated haemoglobin (HbA1c) has long been the standard measure of long-term glycaemic control and remains the gold-standard biomarker for assessing the risk of diabetes-related complications (1–3). However, in some individuals, HbA1c does not accurately reflect mean glucose levels (4), which may lead to suboptimal treatment decisions and potential under- or overtreatment (5). Moreover, HbA1c represents only an average value over a prolonged period, limiting its usefulness when detailed glucose patterns are required for individualised management (6).

In recent years, continuous glucose monitoring (CGM) has become increasingly common in T1D care, enabling more detailed assessments of glucose patterns (7, 8). CGM data are typically summarised in the Ambulatory Glucose Profile (AGP) report (9), which presents the means of eight CGM metrics (although two of these, mean glucose and glucose management indicator [GMI] are linearly related), including time in range (TIR). While the AGP report is a valuable tool, the presentation of multiple metrics can be challenging to interpret, particularly when some metrics improve while others deteriorate.

TIR provides a straightforward summary of overall glycaemic exposure, accounting for both low and high glucose levels. However, TIR alone does not capture the frequency or severity of glycaemic excursions (10, 11), nor does it incorporate variability, and should therefore be interpreted alongside complementary metrics such as time below range (TBR) and time above range (TAR) (8). To address these limitations, composite metrics have been proposed to provide a more holistic evaluation of glycaemic control (11–13). Nevertheless, such metrics may be challenging to implement in clinical practice, particularly in primary care settings where clinicians must manage multiple responsibilities and monitor a wide range of health indicators beyond glucose control.

The Glycaemic Risk Index (GRI) (14) provides a single numerical summary of glycaemic quality, derived from expert evaluations of the risks associated with hypo- and hyperglycaemia. It is based on time spent in very low (<3.0 mmol/L), low (3.0–3.9 mmol/L), high (10.0–13.9 mmol/L), and very high (≥13.9 mmol/L) glucose ranges, with greater weight assigned to more pronounced deviations and to hypoglycaemia than to hyperglycaemia (14). By emphasising more pronounced hypo- and hyperglycaemia, the GRI may offer a more clinically relevant assessment of glycaemic risk, particularly in individuals experiencing frequent or severe excursions.

In the GOLD trial, we previously demonstrated that transitioning from self-monitoring of blood glucose (SMBG) to CGM improved HbA1c, TIR, treatment satisfaction, hypoglycaemia confidence, and well-being in adults with T1D using multiple daily insulin injection (MDI) therapy (15, 16). Further analyses showed that individuals with higher baseline HbA1c, greater TAR stage 2, lower TBR stage 1 and 2, and lower coefficient of variation (CV) experienced greater HbA1c reductions after switching to CGM (17). However, conventional metrics such as HbA1c and standard CGM parameters demonstrated limited associations with improvements in participant-reported outcomes (17) and only marginally explained the improvements in well-being (18). Given its broader dynamic range and weighting of more pronounced glucose values, the GRI may be more sensitive in capturing such associations.

The aim of this study was to evaluate the effects of switching from SMBG to CGM on GRI in adults with T1D using MDI therapy, and the associations between participant-reported outcomes and GRI.

Analyses from the GOLD randomised trial demonstrate that transitioning from SMBG to CGM significantly improved GRI in adults with T1D. Both the total score and its hypo- and hyper- glycaemia components decreased, with the most pronounced reduction in hypoglycaemia risk. The standardised effect size (mean difference divided by the SD) for GRI was more than twice that of TIR, likely reflecting the additional impact of the intervention on hypoglycaemia. Greater improvements in GRI were observed among females and in individuals with longer diabetes duration, lower HbA1c, and more baseline hypoglycaemia. Baseline behavioural and psychological characteristics further appeared to modulate the impact of CGM on GRI.

The results align with previous GOLD trial results showing CGM benefits for HbA1c and other CGM metrics (15), and that CGM more effectively reduces hypoglycaemia than hyperglycaemia (16). An earlier study has also shown that GRI more effectively reflects hypoglycaemia than TIR among individuals with diabetes (27). GRI’s relatively strong negative correlation with TIR (r = –0.47) confirms its validity, but this corresponds to an explained variance of only about 22%. In other words, more than three-quarters of the variation in GRI reflects aspects of glycaemic control not captured by TIR. This includes the fact that TIR does not distinguish hypo- from hyperglycaemia. The same TIR can exist with a large amount of hypoglycaemia or no hypoglycaemia. Further, the GRI hypoglycaemia component assigns greater weight to more severe or prolonged episodes (<3.0 mmol/L).

Our study also showed that individuals with identical TIR values may have markedly different GRI scores depending on TBR and TAR. This supports earlier observations that TIR alone may not fully capture risk and needs to be interpreted together with TBR and TAR (4, 8). The combination of TIR and TBR offers a clear and intuitive overview of glycaemic control, which remains central in clinical care. While exposure to hypoglycaemia and hyperglycaemia can be individually calculated to evaluate the two domains separately, GRI integrates them both into a single composite score, weighting extreme deviations more heavily. This integration may explain its greater sensitivity to CGM-related changes.

Previous research has linked higher GRI to greater glycaemic variability (28). Since both TIR and coefficient of variation (CV) are associated with vascular complications (29–31), GRI may similarly reflect long-term risk, although this requires further research. In the present study, changes in GRI also correlated more strongly with glycaemic variability than did changes in TIR, suggesting that GRI captures dynamic aspects of glucose fluctuations beyond overall time in range. In this study, the hyperglycaemia component emerged as the main barrier to improved overall glycaemic quality. The GRI Grid’s two-dimensional display may help clinicians quickly determine whether poor glycaemic control is driven mainly by low or high glucose excursions, supporting more targeted treatment adjustments. Visual summaries of GRI and other CGM metrics could accelerate both risk assessment and the evaluation of therapeutic responses (32).

Baseline characteristics such as female sex, longer diabetes duration, lower HbA1c, and more hypoglycaemia were associated with improved GRI, while behavioural and psychological factors also played a role. For example, greater reductions in hypoglycaemia risk when switching to CGM were observed in individuals describing themselves as thorough, while distractibility and self-described laziness were linked to smaller improvements. A slight increase in hypoglycaemia risk among those trained in carbohydrate counting may reflect greater clinical complexity or overconfidence in insulin dosing. Higher fear of hypoglycaemia at baseline was associated with improvements in hyperglycaemia risk, suggesting that CGM may provide reassurance and support more balanced glucose management in this group. These exploratory findings indicate that individual traits may influence CGM effectiveness. However, the lack of consistent associations between changes in GRI and general participant-reported outcomes such as well-being or quality of life indicates that objective improvements in glycaemic control may not directly translate into perceived improvements in daily life. This echoes previous findings in the GOLD trial (18), and is an area that warrants further investigation.

GRI may serve as a valuable single metric for summarising glycaemic quality in clinical care, complementing HbA1c and TIR. Its distinct hypo- and hyperglycaemia components, together with visual representation, provide actionable insights into underlying glucose patterns. The greater sensitivity to certain treatment effects compared with TIR suggests that GRI could be particularly useful for identifying patients most likely to benefit from CGM, especially those at risk of both glycaemic extremes. It may also be suitable as an endpoint in certain clinical trials to capture effects on various glucose patterns that TIR alone might miss. Given the relatively low baseline exposure to hypoglycaemia in this cohort, the potential utility of GRI may be even greater in populations with problematic hypoglycaemia, where reducing extreme low-glucose events is a key clinical goal. However, in the clinical management of T1D, conventional CGM metrics such as TIR, TBR, and TAR remain highly valuable because of their simplicity and ease of interpretation, as they directly express the percentage of time spent in different glucose ranges. Further, for health professionals working with diabetes, a relatively good overview can often be achieved using only a few key metrics such as TIR in combination with time in hypoglycaemia. Thus, while GRI may have several potential applications, conventional CGM metrics remain essential for everyday clinical decision-making. Further research on the relationship between GRI and diabetes complications is essential, particularly if specific GRI levels are used as key metrics. This knowledge is critical for understanding how complication incidence is associated with GRI levels when monitoring patients in clinical practice.

Key strengths of this study include the randomised crossover design, within-subject comparisons, and standardised assessments where all participants used the same CGM device, central laboratory for HbA1c measurement, and procedures for collecting research data. The GRI may be particularly valuable in primary care, where familiarity with multiple CGM metrics may be limited, as it provides a single summary measure of glycaemic quality.

Limitations of this study included the exploratory nature of the psychosocial associations, the lack of validated clinical cut-off points for GRI, the restriction of the study population to adults with moderately elevated HbA1c receiving MDI therapy, and the absence of data on long-term diabetes complications.

Switching from SMBG to CGM significantly reduced the GRI in adults with T1D on MDI therapy, with improvements in both the hypo- and hyperglycaemia components. GRI correlated moderately to strongly with established CGM metrics, and the observed effect sizes were larger than for TIR in this dataset. As a composite metric that integrates both hypo- and hyperglycaemia, GRI may serve as a valuable endpoint for intervention studies and as a complementary tool in clinical practice. Further research is needed to validate its clinical utility and generalisability across broader populations and relation to diabetes complications.