A third of insured adults with diabetic-range lab results had no recorded diagnosis or no treatment a year on

A blood test that comes back in the diabetic range is supposed to start something: a diagnosis entered in the record, and usually a treatment plan. A study of 27,650 insured American adults suggests that for about a third of them, at least one of those steps left no trace in the following year.

The study, published on October 5 in Diabetes Care, found that roughly one in ten had neither a diabetes diagnosis nor a diabetes medicine on record 12 months after their results. It relied on insurance claims, which capture what was billed and can miss what was said in the exam room, so the true size of the gap is uncertain.

Why lab results in the diabetes range do not always become a diagnosis

Diabetes is defined by numbers, and the two most used come from fasting blood glucose, measured after a night without food, and HbA1c, a blood test that reflects average blood sugar over roughly three months. In US research the thresholds are a fasting glucose of 126 milligrams per deciliter or more and an HbA1c of 6.5% or more.

One raised number is not the end of the process. Clinical definitions require repeated blood work to confirm the first result, as a 2018 paper in Annals of Internal Medicine noted, and the second test is there to reduce the possibility of a false-positive diagnosis, a result that wrongly signals disease.

That requirement creates a gap in which people can be lost. Someone has to notice the first result, order the second, tell the patient, write the diagnosis into the record and decide on treatment, and none of it happens automatically.

Most diabetes in the United States is found. An analysis of national health surveys covering 30,492 adults estimated that the share of diabetes cases going undiagnosed fell from about a third in 1988 to 1994 to under a fifth by 2017 to 2020, and to about one in ten when a stricter two-test definition was used. The authors read this as suggesting major improvements in diabetes screening and detection, and concluded that up to 90% of all cases are diagnosed. The gap was uneven, though, with the highest proportion of undiagnosed cases among Asian Americans and people without access to health care.

Those survey figures count people who have never been tested as well as people who have. The new study asks something narrower: among people who did have the blood test, and whose results did meet the definition, what happened next?

How the diabetes lab results study worked

Jim Stimpson led the research team, with Joshua Liao as senior author, working from a national database of health insurance claims from 2017 to 2023 that had been linked to laboratory results. A claim is the bill a clinic, hospital or pharmacy sends to an insurer, and it carries codes for the diagnoses and services involved. The database covered people with commercial insurance and people in Medicare, the US public insurance program for those aged 65 and over.

From it the researchers picked out adults with what the paper calls laboratory-confirmed diabetes who had no earlier record of diabetes or of prediabetes, the stage of raised blood sugar below the diabetes threshold. That left 27,650 people, each followed in the claims for one year.

In that year the team looked for two things, which the paper calls diagnostic coding and pharmacologic treatment initiation: a type 2 diabetes code appearing on a claim, and a first prescription for a diabetes medicine. Because it looked back through records that already existed, the study is what researchers call retrospective.

What the study found a year after the lab results

The 27,650 people fell into four groups:

• Both. 66.6% received both a diagnosis code and a prescription for a diabetes medicine within the year.
• Diagnosis only. 19.7% had a diabetes code on a claim and no prescription for a diabetes medicine.
• Treatment only. 4.1% were prescribed a diabetes medicine with no diabetes code recorded.
• Neither. 9.6% had no code and no prescription.

Added together, 33.4% were missing at least one of the two, the basis for the paper’s conclusion that one third lacked claims-documented diagnostic coding, treatment or both. Put another way, 13.7% had no diagnosis code within the year, and 29.3% had no diabetes medicine. In head counts, the group with neither works out to roughly 2,650 people, and the group missing one or both to roughly 9,200.

The four groups are not equally worrying. A person with a recorded diagnosis and no prescription may be managing the condition through changes to diet and activity, which leaves no claim. A person with a prescription and no code has at least been given a diabetes medicine. The group with neither is the one where, as far as the insurance record shows, nothing followed the result.

Who was less likely to get a diabetes diagnosis or treatment

Two patterns stood out, and the first was the blood sugar level itself. Compared with people whose HbA1c sat just over the line, at 6.5% to 6.9%, those with higher HbA1c had progressively higher probabilities of diagnostic coding and treatment. The clearer the result, the more likely it was to be acted on, and the borderline results were the ones most often left without a code or a prescription.

The second pattern ran the other way from what might be expected. People who were sicker in other respects were less likely to have a diabetes code or a prescription on record: having more other illnesses (a greater comorbidity burden, in the paper’s terms) went with lower probabilities of both outcomes. Care, the authors conclude, followed patterns defined by glycemic severity and comorbidity burden, meaning by how high blood sugar ran and how many other illnesses a person had.

What earlier research shows about acting on diabetes test results

Other recent studies have followed what happens after an abnormal result, in very different settings, and each found a leak at some point.

One US health system screened patients at its four emergency departments with an HbA1c test and found diabetes in 615 (6.2%) of the 9,897 people screened. When researchers later reached 307 of them by phone, most had seen a clinician since and about half had started a diabetes medicine, but fewer than half remembered being given the diagnosis in the emergency department. Those who never followed up reported barriers including awareness of the screening result, time and scheduling, and money.

In Denmark, researchers read the charts of 416 patients of psychiatric services whose blood tests, ordered by those services, were in the diabetic range for the first time, and checked whether anything was done within three months, such as a referral, a repeat test or a prescription. Only 37% received a clinical intervention of any kind.

The Danish study also carries a warning for anyone reading the American figures. When the team first selected patients by laboratory result and the absence of a recorded diagnosis or treatment, about one in six turned out to be something else: a manual review of clinical notes showed they were already known to have diabetes, and the diagnosis simply had not been registered. The authors count this mismatch between what the lab shows and what the record says, the discrepancy between biochemical evidence of disease and its formal registration, as a persistent problem in psychiatric and general care alike.

Delay after diagnosis is a known pattern too. A UK study of 81,573 people with type 2 diabetes found that when blood sugar stayed above target on one tablet, the median wait before a second was added was 1.6 to 2.9 years, and its authors concluded that there are delays in treatment intensification, or stepping treatment up, even when blood sugar control is poor. The pattern has a name that has stuck, clinical inertia, which the paper describes as a failure to intensify treatment in a timely manner.

Why would delay matter? According to the US Preventive Services Task Force, an independent panel that reviews evidence on preventive care, diabetes is the leading cause of kidney failure and of new cases of blindness among US adults. Its 2021 statement covers screening for prediabetes and type 2 diabetes in people aged 35 to 70 with overweight or obesity.

No trial has tested delaying treatment, for ethical reasons, but a 2015 computer simulation built on a European screening trial estimated the cost. It found that a three-year delay in diagnosis and routine treatment raised the five-year risk of a heart attack, stroke or similar event by 3.3 percentage points, and a six-year delay by 4.9 points. The authors concluded that major benefits are likely to accrue from diagnosing and treating raised blood sugar and other cardiovascular risk factors early, and that the intensity of treatment after diagnosis matters less than the time of its initiation. A simulation is a model, and its output depends on its assumptions.

What the diabetes lab results study cannot show

Claims are not conversations. The study counted codes and prescriptions billed to insurers. A clinician may have told a patient about the result, repeated the test or given advice without a diabetes code ever reaching a claim, and the Danish chart review shows how often records understate what clinicians know.

No medicine is not the same as no care. Diet and activity changes are a recognized first step, and the UK paper notes that British guidelines of the period placed lifestyle measures as the first step in treatment. Such care leaves no pharmacy claim, which matters most for the 19.7% with a diagnosis and no prescription, and for people just over the threshold.

Only prescriptions billed to insurance were visible. Medicines paid for out of pocket would not appear.

Insured adults only. Everyone had commercial or Medicare coverage and had a blood test done, so all were already in contact with health care. The national survey data suggest gaps are wider among people without access to care, who are not in this study.

One year. A diagnosis or prescription arriving in month 13 was not counted.

Why, and what happened next. The study describes who was less likely to be coded or treated. It does not establish the reasons, and it did not follow people to see whether those without a recorded diagnosis went on to worse health.

Borderline results. Some people just over the line on one occasion would test below it on another, which is why confirmation exists. The people least likely to have a code or a prescription were those just over the line.

Do not stop or change a prescribed medicine without talking to your doctor. Whether a set of blood test results in a particular person amounts to diabetes, and what treatment follows, is assessed by their doctor.

For two thirds of these adults a diabetic-range result led to a diagnosis and a prescription within a year, and for about one in ten the insurance record shows neither.