What A1c Misses — and Why It Matters

We covered A1c in depth in Post 4 — Complications and Targets. The short version: A1c reflects average blood glucose over approximately three months, but two people can have identical A1c values while living in completely different glucose realities. One person might have stable, consistent glucose throughout the day. Another might be oscillating between dangerous lows and dangerous highs, with the highs and lows averaging out to the same number. The first person is doing well. The second is at elevated risk for hypoglycemic emergencies, the oxidative damage caused by glucose spikes, and the anxiety and impaired quality of life that come with an unpredictable glucose profile.

This is the problem that continuous glucose monitoring (CGM) was designed to solve. Rather than a single number every three months, CGM provides a glucose reading every few minutes — showing trends, peaks, valleys, and patterns over days and weeks. The key metric that has emerged from CGM data is time-in-range: the percentage of time glucose stays within a target window, typically 70–180 mg/dL. Time-in-range correlates with complication risk in ways that A1c alone cannot capture.

CGM: What It Is and How It Works

A CGM consists of a small sensor inserted just under the skin — typically on the upper arm or abdomen — that measures glucose in the fluid surrounding cells (interstitial fluid) every 1–5 minutes. This glucose measurement correlates closely with blood glucose but lags behind by approximately 5–15 minutes, which matters when glucose is changing rapidly. The sensor transmits readings wirelessly to a smartphone or receiver, and most systems display the current glucose level, the rate and direction of change (a falling arrow means glucose is dropping, a rising arrow means it is rising), and a graph of the last few hours.

There are two main types. Real-time CGM (rtCGM) continuously displays the reading and alerts the user when glucose goes out of range — this is the current standard and what most new devices use. Intermittently scanned CGM (isCGM, such as the original FreeStyle Libre) is an older format that requires the user to actively scan the sensor with a reader or phone to get a reading, and does not alert automatically unless glucose goes critically low. isCGM is largely being replaced by real-time systems as the technology matures, but it remains available and is a reasonable choice for patients who prefer a dedicated reader over a smartphone, or who simply want trend data without active alerts. For patients who experience hypoglycemia unawareness or need active low-glucose alerts, rtCGM is strongly preferred.

CGM in Type 2 diabetes

CGM is now standard of care for Type 1 diabetes and is increasingly used and supported in Type 2. The evidence in insulin-treated Type 2 shows A1c reductions of approximately −0.35–0.45% and consistent improvements in time-in-range. A large Veterans Affairs study found that initiating CGM in insulin-treated Type 2 patients was associated with a 13% reduction in hyperglycemia-related events and a 25% reduction in all-cause hospitalization. Even in non-insulin-treated Type 2, RCT data now show meaningful A1c and time-in-range improvements — partly from the behavioral feedback CGM provides, helping patients understand in real time how specific foods, meals, and activities affect their glucose.

Time-in-Range: The Metric That Changes Management

Time-in-range (TIR) — the percentage of time glucose stays between 70 and 180 mg/dL — has become the primary CGM-derived outcome in clinical trials and increasingly in clinical practice. Higher TIR is associated with lower rates of retinopathy, albuminuria, cardiovascular mortality, and all-cause mortality. Peripheral neuropathy correlates most strongly with glucose variability metrics (how much glucose swings around), reinforcing that controlling the swings matters as much as controlling the average.

The international consensus targets, endorsed by the ADA, are shown below. These are not arbitrary thresholds — they were chosen to approximate an A1c of approximately 7% when achieved, while providing additional information about hypoglycemia and variability that A1c cannot.

CGM Targets — International Consensus (ADA 2026)
Metric
Most Adults
Older / High-Risk
What It Means
Time in range (70–180 mg/dL)
>70%
>50%
The primary goal. >70% approximates an A1c of ~7%. Each 10% increase in TIR corresponds to roughly 0.5% A1c reduction.
Time below 70 mg/dL (low)
<4%
<1%
Level 1 hypoglycemia — symptomatic low. Goal is to minimize this entirely in older and frail patients.
Time below 54 mg/dL (very low)
<1%
<1%
Level 2 hypoglycemia — clinically significant, dangerous. Less than 15 minutes per day at this level is the target.
Time above 180 mg/dL (high)
<25%
<50%
Level 1 hyperglycemia. Less than 6 hours per day above 180 in most adults.
Time above 250 mg/dL (very high)
<5%
<10%
Level 2 hyperglycemia — significant complication risk. Less than ~70 minutes per day above this threshold.
Coefficient of variation (CV)
≤36%
≤36%
Measures glucose variability — how much glucose swings around the average. CV above 36% predicts higher hypoglycemia risk and correlates with neuropathy progression.
My Synthesis

The coefficient of variation is the metric I find most useful for identifying patients whose glucose is technically "controlled" by A1c but who are actually living with dangerous instability. A patient with A1c 7.2% and CV of 45% is not well-controlled — they are oscillating widely, spending time in significant lows and highs that are averaging to an acceptable number. My approach is to address variability first: stabilize the swings before pushing aggressively toward a lower average. A stable A1c of 7.5% is clinically safer than a volatile A1c of 6.9%.

Fingerstick Glucose Monitoring: When It's Actually Needed

Traditional fingerstick blood glucose monitoring — pricking a finger and placing a drop of blood on a test strip — remains useful in specific situations. CGM measures glucose in the fluid between cells, not directly in blood, and lags behind blood glucose by 5–15 minutes when glucose is changing rapidly. During steep post-meal rises, rapid hypoglycemia correction, or exercise, CGM readings can be meaningfully inaccurate. Most manufacturers recommend confirming with a fingerstick before treating hypoglycemia if symptoms don't match the CGM reading.

In my practice, I don't routinely recommend fingerstick monitoring for patients who are not on insulin or hypoglycemia-causing medications (sulfonylureas). For those patients, routine fingerstick testing rarely changes management and adds burden without proportional benefit. The exception is patients who want to use real-time glucose data to understand how specific foods or meals affect them — in that context, even occasional targeted checking can be genuinely informative and help build intuition about dietary choices.

For patients on insulin who cannot access CGM due to cost or preference, structured fingerstick monitoring — before meals, two hours after meals, at bedtime, and whenever symptoms occur — still provides the actionable pattern data needed to adjust doses safely. CGM is strongly preferred for all insulin-treated patients, but fingerstick monitoring remains a viable and practical option when CGM is not accessible.

Insulin Types and What They Do

For patients requiring insulin — all Type 1, and many Type 2 — understanding the different insulin types is essential for managing glucose effectively. Modern insulin therapy aims to mimic the normal pattern of insulin secretion: a steady background level (basal) supplemented by meal-related doses (bolus). Different insulin preparations achieve different parts of this profile.

The chart below shows the approximate action profiles of the main insulin types — when each starts working, when it peaks, and how long it lasts. These are averages; individual response varies.

Insulin Action Profiles — Onset, Peak, and Duration
Insulin action profiles 0h 2h 4h 6h 8h 12h 24h Hours after injection Insulin activity Ultra-rapid (Fiasp, Lyumjev) — onset 10–15 min, peak ~1h, duration ~3–4h Rapid-acting (Humalog, NovoLog) — onset 15–30 min, peak ~1–2h, duration ~4–5h Regular human insulin — onset 30–60 min, peak ~2–4h, duration ~6–8h NPH insulin — onset ~1–2h, pronounced peak ~4–8h, duration ~14–16h (twice daily dosing) Glargine U-100 / Detemir — onset ~1h, low flat peak, duration ~20–24h Degludec / Glargine U-300 — onset ~1h, near-peakless, duration >24h

Profiles are approximate averages — individual response varies with injection site, dose, temperature, and metabolic factors.

Long-Acting Basal Insulin

Glargine (Lantus, Basaglar, Toujeo U-300) · Degludec (Tresiba) · Detemir (Levemir)
Basal

Basal insulin provides a steady, low-level background of insulin over 12–24+ hours, suppressing the liver's glucose output between meals and overnight. It does not cover meals — it is the foundation that keeps glucose stable when you're not eating. In Type 1 diabetes, the basal component typically accounts for 40–50% of total daily insulin. In Type 2, basal insulin alone is often the starting point before adding mealtime coverage if needed.

The clinically meaningful difference within this class is hypoglycemia risk, not glucose lowering. All long-acting insulins produce equivalent A1c reductions — the trials consistently show no meaningful A1c difference between agents. What differs is the stability and safety of the action profile. Second-generation agents — degludec (Tresiba) and glargine U-300 (Toujeo) — have longer, flatter, near-peakless profiles. The DEVOTE trial showed degludec reduced severe hypoglycemia by 40% and nocturnal severe hypoglycemia by 53% compared to glargine U-100, with identical A1c outcomes. For patients experiencing recurrent or nocturnal hypoglycemia on first-generation basal insulin, upgrading to a second-generation agent is evidence-based and clinically meaningful. For stable patients without hypoglycemia concerns, the less expensive first-generation agents are entirely appropriate.

NPH insulin — an older intermediate-acting human insulin — is substantially less expensive and remains a viable option in Type 2 diabetes for cost-constrained patients who are not experiencing significant nocturnal hypoglycemia. It requires twice-daily dosing and has a more pronounced peak than the long-acting analogs, making it somewhat less convenient and slightly higher-risk for overnight lows, but large real-world studies have not found dramatic differences in hypoglycemia-related hospitalizations when used in standard Type 2 management.

Rapid-Acting Bolus Insulin

Lispro (Humalog) · Aspart (NovoLog) · Glulisine (Apidra) · Faster aspart (Fiasp) · Lispro-aabc (Lyumjev)
Bolus

Rapid-acting insulin is taken at meals to cover the glucose rise from carbohydrate absorption. It begins working within 15–30 minutes, peaks at 60–90 minutes, and lasts approximately 3–5 hours. The dose is calculated based on how many carbohydrates are being eaten (insulin-to-carbohydrate ratio, or ICR) and whether a correction is needed for a high pre-meal glucose (using the insulin sensitivity factor, or ISF).

Within this class, lispro, aspart, and glulisine are clinically interchangeable. Meta-analysis of head-to-head trials finds no meaningful difference in A1c, hypoglycemia rates, or weight between them — choose based on cost and availability. The ultra-rapid formulations (Fiasp, Lyumjev) offer a modestly faster onset and a ~20 mg/dL improvement in 1-hour post-meal glucose compared to standard rapid-acting insulins, but produce no meaningful A1c difference (−0.02% in meta-analysis). They are a reasonable choice for patients with significant post-meal glucose spikes or those using closed-loop AID systems, but do not justify cost-driven switching if standard agents are working well. Timing matters: standard rapid-acting insulin is typically taken 15–20 minutes before a meal; ultra-rapid formulations can be taken at the start of or even during a meal.

Regular human insulin (available over the counter at Walmart as ReliOn, ~$25/vial without prescription) has a slower onset (30–60 min) and longer duration (6–8 hours) than rapid-acting analogs, making meal timing more challenging and late post-meal hypoglycemia more common. It is a legitimate option when cost is the primary barrier, but it requires taking the dose 30–45 minutes before eating — a practical challenge for many patients.

Biosimilar insulins are therapeutically equivalent to their brand-name reference products. Meta-analysis of 25 RCTs (10,617 patients) shows a mean A1c difference of 0.01% — effectively zero. A 1:1 switch from brand to biosimilar is appropriate and strongly supported by evidence. Select based on cost.

Correction Doses

Using rapid-acting insulin to bring elevated glucose back to target
Correction

A correction dose uses rapid-acting insulin to lower glucose that is above target, independent of a meal. The calculation uses the insulin sensitivity factor (ISF) — how many mg/dL one unit of insulin typically lowers blood glucose for that individual. For example, if an ISF is 50 mg/dL per unit, and glucose is 250 mg/dL with a target of 100 mg/dL, the correction dose would be (250 − 100) ÷ 50 = 3 units.

A critical principle: give correction doses time to work before giving another. Rapid-acting insulin takes 15–30 minutes to begin acting and peaks at 60–90 minutes. In practice, I recommend waiting at least 1–1.5 hours before reassessing whether a further correction is needed. Taking repeated corrections before the first has fully acted — called insulin stacking — is one of the most common causes of severe hypoglycemia in insulin-managed patients. CGM trend arrows are useful here: a flat or falling arrow means the correction is working; don't add more.

In my practice, I don't generally recommend manual correction doses for patients on multiple daily injections (MDI) unless they are detail-oriented and comfortable with the calculations involved — the margin for error is meaningful, and CGM and pump-based AID systems automate this far more safely. For appropriate patients who want this level of control, it is a legitimate tool, but it requires clear understanding of active insulin on board before each correction.

Insulin Pumps (Continuous Subcutaneous Insulin Infusion)

Delivers rapid-acting insulin continuously via a small under-skin catheter
Pump

An insulin pump replaces the need for long-acting insulin by delivering small, continuous doses of rapid-acting insulin throughout the day and night (the basal rate), with larger user-triggered doses at meals (boluses). Pumps allow basal rates to be programmed differently at different times of day — accounting for the dawn phenomenon (early-morning glucose rise driven by cortisol and growth hormone) and exercise periods, which multiple daily injections cannot achieve as precisely.

The major advance over the past decade is the automated insulin delivery (AID) system — sometimes called a closed-loop or artificial pancreas system — which connects the pump to a CGM and adjusts insulin delivery automatically based on glucose trends. These systems have produced the largest improvements in time-in-range of any diabetes technology to date.

Insulin Affordability: What Patients Should Know

Insulin cost is a genuine barrier to care for many patients, and the landscape of affordability programs has changed significantly in recent years. Every major insulin manufacturer now offers programs that cap monthly costs at $35 for eligible patients, and several pathways exist depending on insurance status.

⚠ Cost Should Not Be a Reason to Ration Insulin

Insulin rationing — taking less insulin than prescribed to make a supply last longer — is dangerous and contributes to DKA hospitalizations and deaths. If cost is a barrier, talk to your physician or pharmacist before reducing doses. Multiple programs exist to help.

📊 $35/Month Insulin Programs — Current Options

Long-acting (basal) insulin: Sanofi's Valyou Savings Program caps Lantus (glargine U-100) and Toujeo (glargine U-300) at $35/month for up to 10 packs per fill for uninsured or commercially insured patients paying cash. Generic glargine (Basaglar, biosimilar) is often available for $35 or less at major pharmacies with GoodRx or similar discount cards.

Rapid-acting insulin — Eli Lilly (Humalog, Insulin Lispro): Lilly Insulin Value Program — $35/month for up to a 30-day supply. Available at participating pharmacies for commercially insured or uninsured patients. Visit InsulinAffordability.com or call 833-808-1234.

Rapid-acting — Novo Nordisk (NovoLog, Fiasp): My$99Insulin or Copay Savings Card — as low as $25–35/month for up to 3 vials or 2 packs of pens. Commercially insured or uninsured. Visit NovoCare.com or text SAVE to 97430.

Rapid-acting — Sanofi (Apidra, Admelog): Valyou Savings Program — $35/month for uninsured patients or commercially insured patients paying cash. Visit TeamingUpForDiabetes.com or call 800-633-1610.

Medicare patients: The Inflation Reduction Act caps insulin costs at $35/month for Medicare Part D beneficiaries for all covered insulins — no manufacturer program needed.

Over-the-counter option: ReliOn regular human insulin at Walmart costs approximately $25/vial without a prescription. This is regular (not rapid-acting analog) insulin with a different action profile requiring adjusted meal timing, but it is a viable option when other programs are inaccessible. Discuss the timing differences with your physician before switching.