Bio-fluid Glucose Correlation

Bio-fluid Glucose Correlation

Introduction

A continuous glucose monitor (CGM) doesn't read blood. Its sensor sits in interstitial fluid (ISF), the thin layer of fluid that bathes the cells under the skin, and glucose has to diffuse out of the capillaries into that fluid before the sensor can see it [1]. That trip takes a few minutes. When a Mayo Clinic team traced labeled glucose from the bloodstream into the fluid under the skin of the abdomen, it showed up after about 5 to 6 minutes in healthy adults who had fasted overnight [2]. In adults with type 1 diabetes, the median was 6.8 minutes, and every participant's sample had picked it up by 9.8 minutes [3]. Older studies, summarized in a 2009 review, report an average lag of 8 to 10 minutes, with a much wider spread between individual reports [1].

Physiological Lag and Its Implications

The gap matters most when glucose moves fast, such as after eating or during exercise. During rapid change, sensor readings can trail blood glucose by 5 to 10 minutes [1]. Diffusion explains much of the steady-state lag, but the same review notes that during fast swings, differences in glucose concentration between tissues also play a part [1]. Sensor makers, including Dexcom and Abbott, build software to offset part of this delay [1]. Here's what surprised us: in Abbott's accuracy study of the FreeStyle Libre 2 system with its updated algorithm, sensor readings lagged lab reference glucose by an average of only 2.4 minutes in adults and 2.1 minutes in children [4]. That study and the Mayo work used different methods, so the numbers aren't directly comparable. The Libre 2 sensor itself has since been replaced by the Libre 2 Plus sensor; Abbott kept the original available only through September 30, 2025 [5]. Whether to double-check a reading with a fingerstick while glucose is changing quickly is a decision for your care team.

Sweat and tears give a sensor far less glucose to work with. A 2017 review tabulated glucose in healthy people at 4.9 to 6.9 mmol/L in blood, 0.06 to 0.11 mmol/L in sweat, and 0.05 to 0.5 mmol/L in tear fluid [6]. That puts sweat at roughly 1/50 to 1/100 of blood, and tears anywhere from about 1/10 to under 1/100. The single "1/100th" figure often quoted for both fits sweat better than tears.

How well these fluids track blood is still debated. A small 2012 study of seven people with diabetes found that sweat glucose followed blood glucose closely, but only when sweat was collected so that glucose already sitting on the skin couldn't contaminate the sample [7]. For tears, a 2007 review of 80 years of research found that measured tear glucose depended heavily on how the tears were collected, and that researchers still disagreed on whether it follows blood glucose at all [8]. A 2019 review in Nature Biotechnology argued that the fluid itself is often the hardest problem, with tiny sample volumes, uneven secretion rates and shifting pH [9]. None of this has reached a product you can buy with FDA backing. The FDA's February 2024 safety communication, still posted when we checked in October 2026, says the agency has not authorized any smartwatch or smart ring that measures glucose without piercing the skin, and it contrasts them with CGMs, which do pierce the skin [10].

Innovation and Future Directions

Since the fluid isn't changing, most accuracy work goes into the sensor and its software. Factory calibration is one piece of that: the sensor is calibrated during manufacturing under controlled lab conditions, so users no longer enter fingerstick values to calibrate it [11]. Abbott's engineers point to a second benefit beyond convenience. Taking the user out of calibration removes calibration mistakes, which can make a sensor inaccurate [11]. That ties back to lag, because the 2009 review found lag's biggest effect on long-term sensor performance was the error it introduces during calibration [1].

Accuracy is usually reported as MARD, the mean absolute relative difference: the average percentage gap between sensor readings and reference blood glucose values [12]. In Abbott's Libre 2 study, the system with the updated algorithm had a MARD of 9.2% in adults and 9.7% in children, and the authors reported better accuracy than the previous-generation device [4]. MARD has limits. A 2020 review cautioned that published MARD values depend heavily on how each study was designed, and that MARD shouldn't be the only number used to judge a CGM [12].

Conclusion

The fluid a sensor reads decides what it can tell you. ISF follows blood closely enough for CGMs to work, with a delay of minutes that shows up mostly when glucose changes quickly [1][2]. Sweat and tears hold far less glucose, and their link to blood glucose is still unsettled [6][8]. The open question is whether better sampling can make those readings consistent; the 2019 Nature Biotechnology review argued that understanding how these fluids are secreted is the key to faster progress [9].