CGM in India: Who Actually Needs One

A sensor on the back of your arm, two weeks of graphs, and someone explaining which foods spiked you. Continuous glucose monitoring is being sold hard in India as of 2026, usually bundled into a weight-loss programme. If you do not have diabetes, no guideline recommends it for you.

The evidence supports CGM in type 1 diabetes, insulin-treated type 2 diabetes and pregnancy complicated by diabetes. Everyone else is buying high-frequency data whose accuracy in their own glucose range has barely been measured, and the companies selling sensors and coaching bundles are structurally unable to say so.

Key Takeaways:

• A CGM measures interstitial glucose, not blood glucose. Total device lag is typically 10 to 20 minutes, and is largest exactly when people watch most closely, during a rapid post-meal rise.
• MARD figures used in CGM marketing are measured mainly in people with diabetes. In the normal glucose range that most buyers sit in, one located study found MARD of 27.5%, against roughly 9% in diabetes cohorts.
• Brief post-meal peaks above 140 mg/dL occur in people with entirely normal glucose metabolism. They are not, alone, a finding.
• No guideline, Indian or international, recommends CGM for adults without diabetes seeking weight loss or metabolic insight.
• Two weeks of CGM data describes your glucose pattern. It does not diagnose anything and cannot replace clinical interpretation.

A CGM does not measure blood glucose. It measures glucose in the interstitial fluid under the skin with an electrochemical sensor, then applies an algorithm to infer a blood-equivalent value. Interstitial glucose tracks blood glucose with a physiological delay of roughly 5 to 6 minutes, but total lag at the device output is typically 10 to 20 minutes once sensor and algorithm delays are added.

Lag is smallest when glucose is stable and largest exactly when people watch most closely: during a rapid rise or fall after a meal, when differences of tens of mg/dL between a fingerstick and the sensor are common. Calibration does not remove it. Factory-calibrated systems, which include the sensors marketed in India, reduce variability between sensors without eliminating lag or artefacts.

Newer sensors need 30 to 60 minutes of warm-up before showing anything, and wear runs from 10 to 15 days depending on the system.

MARD, the mean absolute relative difference, is the average absolute percentage difference between sensor readings and a reference glucose value, usually venous plasma measured on a YSI analyser, across every paired measurement in a validation study. It is the number quoted in almost all CGM marketing.

The limits of MARD are well described. It cannot separate a consistent offset from random scatter, does not reflect transient large errors, and is heavily influenced by study design and by how many readings fall in each glucose range.

Overall MARD is dominated by performance across the wide 40 to 300+ mg/dL span found in people with diabetes. A person without diabetes lives in a narrow band of roughly 70 to 140 mg/dL. The one located study of a sensor worn by healthy non-diabetic women reported an overall MARD of 27.5%, against roughly 9% for the same sensor family in diabetes cohorts.

What validation studies show for the main sensors sold in India:

• FreeStyle Libre 2: 9.2% MARD (95% CI 8.7 to 9.9). Tested in 144 adults with diabetes, 6,319 paired readings against YSI. Source: peer-reviewed validation.
• FreeStyle Libre 2 in healthy non-diabetic females: 27.5% MARD. Sample size not stated in the available record. Source: peer-reviewed performance study.
• FreeStyle Libre 3: 7.8% MARD. Tested in 100 participants aged 6 and above, 6,845 matched pairs against YSI. Source: peer-reviewed validation.
• Dexcom G7: Around 8 to 9% MARD. Tested largely in people with diabetes. Source: regulatory and manufacturer data.
• GlucoRx Vixxa 2: 8.66% MARD. Testing population not published in detail. Source: manufacturer clinical data, no independent validation located.

These figures come from different studies, cohorts and comparators, so they cannot be ranked against each other. Two of the five originate with the company selling the device.

Accuracy is not constant across a sensor's life. In one Libre 2 validation, day-one performance was worse than later days, with 87.9% of adult readings within 20% or 20 mg/dL of reference against 93.2% overall. A Libre 3 study showed the same pattern, with MARD falling from 8.6% across days one to three to 6.4% by days nine to twelve as the tissue settled.

Two artefacts recur in troubleshooting documents. Compression lows, caused by lying on the sensor or by tight clothing, produce falsely low readings unrelated to blood glucose. Sensor-to-sensor disagreement appears where people wore several devices at once, with peaks arriving minutes apart and mean biases of several mg/dL, widest after meals.

Set that against the signal. Validation studies typically report median absolute differences of 8 to 14 mg/dL in the euglycaemic range, overlapping the 10 to 30 mg/dL post-meal bumps non-diabetic users are told to read as feedback on a food.

Most of what a non-diabetic wearer sees is normal physiology. In a prospective study of 153 healthy children and adults without diabetes or islet autoantibodies, mean 24-hour glucose was around 99 mg/dL, with a median 96% of time between 70 and 140 mg/dL, about 2% of the day above 180 mg/dL and about 1% below 70 mg/dL.

Pooled analyses of euglycaemic cohorts land in the same place: roughly 95 to 96% of time within 70 to 180 mg/dL, about 75% within the tighter 70 to 100 mg/dL band, and median glycaemic variability around 15%. Brief peaks above 140 mg/dL and short dips below 70 mg/dL occur in people with entirely normal glucose metabolism and are not, alone, a finding.

Large India-specific CGM profiles for non-diabetic adults have not been published in mainstream sources. Indian consensus documents apply international normative targets, so an Indian wearer is measured against a curve nobody has drawn for this population.

The prospective evidence that CGM-derived variability predicts disease in people without diabetes is thin.

For CGM itself, a systematic review of CGM in non-diabetic populations covering 23 studies including seven randomised controlled trials found:

• CGM reduced mean blood glucose and improved adherence to lifestyle advice, most clearly in prediabetes and comparable at-risk groups.
• There was no significant effect on body mass index.
• There was limited evidence on long-term outcomes.
• There was no clear glycaemic benefit in normoglycaemic individuals.

The prediabetes signal is the one worth holding onto in a country where the at-risk population runs into nine figures, with the caveat that the trials behind it were not Indian.

Anxiety is the most consistently reported harm. In one study of CGM users, around two-thirds reported anxiety when confronted with high readings, regardless of diabetes status, with distress most common among younger people, people with obesity, and those with higher baseline eating-disorder symptoms.

Qualitative research on glucose monitoring describes the mechanism. Frequent checks and strict food rules can medicalise eating, feed disordered-eating thoughts, and produce guilt and shame around food, even where monitoring is clinically indicated and supervised. A wellness programme built on avoiding "spikes" reproduces those conditions without the supervision.

The readings themselves fail in two directions. Artefacts can generate alarming values that trigger repeated laboratory tests for someone never at risk. A reassuring two-week trace can delay proper assessment in someone with a family history, a high BMI or a rising HbA1c.

Indian and international guidance restrict CGM to defined clinical situations. The RSSDI expert consensus of 2024 and ICMR's type 2 diabetes guidelines both frame it as a diabetes-management tool, not a wellness product.

What guidance says for each population group:

• Type 1 diabetes: Recommended for paediatric and adult patients on insulin who can afford it. Continuous use preferred. Minimum two to four sensors a year.
• Insulin-treated type 2 diabetes: At least twice yearly when well controlled. More often with lower time in range, on intensive insulin regimens, or with high hypoglycaemia risk.
• Pregnancy with type 1, insulin-treated type 2, or gestational diabetes: Real-time CGM recommended to improve neonatal outcomes. Target 70% or more of time within 70 to 140 mg/dL.
• Complex or high-risk diabetes: Professional CGM for recurrent severe hypoglycaemia, hypoglycaemia unawareness, frequent ketoacidosis, or a large gap between clinic glucose and HbA1c.
• Adults without diabetes, for weight or metabolic insight: Not covered by any Indian or international recommendation located.

International position statements sit in the same place, adding that readings which do not match symptoms should be confirmed with a fingerstick. None of it decides whether an individual reader qualifies, which is a call for a registered medical practitioner.

CGMs are regulated as medical devices under India's Medical Devices Rules 2017, administered by CDSCO. Class assignments and registration details for individual models are not consistently visible in public databases, and clear public guidance on whether a prescription is required for non-diabetic use was not traceable. Several sensors can be bought without a prescription through Indian online platforms as of 2026, which sits awkwardly against guidance treating CGM as a diabetes-management tool.

Sensors on sale in India as of 2026 include the FreeStyle Libre family, Dexcom G7 and the GlucoRx Vixxa 2, with Medtronic Guardian sensors appearing mainly in specialist endocrine settings alongside pumps. The honest cost answer is a band, not a price list: roughly ₹3,000 to ₹12,000 per sensor for 10 to 15 days of wear, or between roughly ₹6,500 and ₹36,000 a month depending on the system. Those figures are assembled from publicly visible retail listings as of 2026 and almost all of it is paid out of pocket.

Nothing a non-diabetic wearer sees over 14 days changes a diagnosis. Diabetes and prediabetes are still identified from fasting glucose, an oral glucose tolerance test and HbA1c, read by a registered medical practitioner against Indian reference points rather than a graph on a phone.

A CGM reading needs clinical context to be actionable. A Voy clinician can read your glucose data alongside your HbA1c, fasting insulin and full metabolic picture. Book a consultation.

This article is for general information and education only and is not medical advice, diagnosis, or treatment. GLP-1 and other medications referenced are prescription-only and are appropriate only for certain people under the supervision of a qualified clinician. Do not start, stop, or change any medication based on this article. Please consult a registered medical practitioner about your individual circumstances. Information reflects what was available at the time of review and may change.