We read with great interest the study by Yao et al. 1 in Diabetes, Obesity and Metabolism and commend the authors for leveraging a very large continuous glucose monitoring dataset to formalize the relationship between time in range (TIR) and time in tight range (TITR). Their effort is timely, yet several methodological and translational issues warrant discussion. First, the central claim that a unified equation can ‘bridge’ TIR and TITR should be interpreted cautiously because this relationship was already shown to be nonlinear and strongly conditioned by glycemic variability rather than fixed across populations 2. Beck et al. 2 demonstrated that TIR and TITR are highly correlated, but that TITR at a given TIR differs between type 1 and type 2 diabetes largely because coefficient of variation (CV) and time below range alter the glucose distribution. Dunn et al. 3 subsequently showed in more than 20 000 real-world CGM users that identical TIR values can correspond to materially different TITR values when CV changes, indicating that the conversion is distribution-dependent rather than constant. Xu et al. 4 extended this observation by showing that a gamma-distribution model captured the interaction among average glucose, TIR and TITR better than simpler fixed mappings, again underscoring that variability reshapes the apparent equivalence between metrics. Consistently, Burckhardt et al. 5 reported in a multinational youth registry that TTR estimates predicted from TIR were significantly higher in high-CV profiles, and adjusted models actually fitted HbA1c slightly better from TIR than from TTR. In this context, the very high correlation reported by Yao et al. 1 may reflect, at least in part, the deterministic nesting of TITR within TIR rather than a new clinically transportable rule. Second, the proposed unified equation is difficult to interpret as a deployable clinical model because it is not mathematically constrained to the 0%–100% range of TITR 1. Using the published formula, a patient with TIR 100% and CV 24% would have a predicted TITR of 100.4%, and with CV 33% the prediction rises to 108.8%, which is physiologically impossible 1. By contrast, Burckhardt et al. 5 modelled the same family of bounded glucose percentages with beta regression, a choice that preserves the natural limits of the outcome. Moreover, Yao et al. 1 observed substantial systematic bias during external validation, with intercept shifts of −10.24 and −8.18 and markedly improved fit only after recalibration, which weakens the notion of a single universal equation. The 2023 international consensus on CGM metrics for clinical trials emphasized rigorous endpoint qualification and standardization across study settings, rather than assuming direct transportability between devices and cohorts 6. That caution is particularly relevant here because head-to-head studies have shown significant differences in TIR, time below range and CV when glycaemic metrics are derived from different CGM systems worn simultaneously 7. Third, translating TIR 70% into TITR around 50% may be clinically premature because TITR is not yet anchored to one universal endpoint across diabetes types and monitoring platforms 1. In adults with type 1 diabetes, De Meulemeester et al. 8 linked lower TITR to a higher burden of chronic complications, but their data positioned TITR as a related risk marker rather than a direct replacement for TIR. In adults with type 2 diabetes, Wang et al. 9 showed that TITR remained associated with incident retinopathy even within subgroups already achieving TIR above 70%, suggesting that TITR may add information precisely where TIR is already high. Cai et al. 10 likewise reported that lower TITR predicted all-cause and cardiovascular mortality in type 2 diabetes, yet that evidence arose from a separate Chinese prospective cohort and does not establish a device-independent conversion rule. Importantly, Kim et al. 11 found that TITR was a more reliable predictor of clinically relevant HbA1c targets such as 6.5% and 7.0% in insulin-treated adults. Burckhardt et al. 5, however, found slightly better HbA1c fit from TIR than from TTR in a multinational youth registry, indicating that the preferred metric remains endpoint- and cohort-specific rather than universally exchangeable. Accordingly, a bounded modelling strategy, prespecified recalibration across CGM platforms, and prospective validation against hard clinical outcomes would substantially strengthen this otherwise valuable dataset. Until then, the proposed equation may be better viewed as a population-specific descriptive approximation than as a unified target-setting tool for T1D and T2D. The author has nothing to report. The author declares no conflicts of interest. The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/dom.70782.
Lei Hu (Tue,) studied this question.