Today’s guest post is by Dr. Rani Marx, and it is about pre-diabetes. It is easy to give Americans more ‘diagnoses’, but the correct question is: are they better off as a result of those labels? If we aggressively diagnose pre-diabetes, and if we aggressively counsel or treat Americans who meet his mark, are outcomes (living longer/ living better) superior to merely diagnosing diabetes & giving general lifestyle advice? The right answer is we don’t know, yet the medical industrial complex rages full steam ahead and that is the topic fo Rani’s post
-Vinay Prasad MD MPH
Everyone Wants Me to Have Prediabetes
I had a surprising interaction during my recent visit to the podiatrist As usual, the medical assistant took my vitals, confirmed my medications, but then inquired: “Do you have diabetes?” “No,” I replied. “Are you sure?” she persisted. “Yes,” I answered. “You don’t take any medications to lower your blood sugar?” “No, I don’t,” I responded. I mentioned my recent lab tests that indicated I did not have diabetes. She related that many patients in the practice have diabetes and numerous associated health problems, commenting “So sad...”
The odd exchange made an especially strong impression on me because of two recent diabetes-related discussions with my new primary care physician (PCP).
During a visit to my new PCP, we discussed my marginally high Hemoglobin A1c (HbA1c) from six months prior. She recommended repeating the test and expressed concern about the possibility of my having diabetes even in the absence of symptoms and risk factors (other than my maternal grandfather’s Type 2 diabetes). When I evinced surprise, she told me about one of her patients who, absent any risk factors or ill health, turned out to have full blown diabetes detected only by lab tests.
I repeated the HbA1c. This time, my reading was 5.9% instead of the previous 6.1%; the normal range is currently 4.8-5.6%. While awaiting test results my concerns were amplified by a website my PCP suggested on how to lower HbA1c. The recommendations were largely irrelevant: I’m a 62 year old White female, not overweight, get plenty of exercise, eat in alignment with best practices, and don’t smoke. My anxiety was stoked by the appearance in my electronic health record (EHR) of red exclamation points, “H” for high, and a “prediabetes” label next to my lab values. I lost sleep many nights worrying whether I might have diabetes and how that would play out in future. While my PCP was pleased with the second set of results and did not advise any further screening or intervention, the psychological effect of these marginally high test results and associated anxiety on my part turn out to result from a common and scientifically contentious medical “problem” or diagnosis: prediabetes.
My training as an epidemiologist and health services researcher caused me to question the necessity of the diabetes screening as well as the interpretation of results. I wondered when my HbA1c was last normal. It turns out my slightly elevated readings were nearly the same in 2007 (6.0%) and 2008 (5.9%) as in late 2021 and 2022. However, the American Diabetes Association (ADA) criterion for normal HbA1c changed from 4.5-6.1 in the early 2000’s to the current 4.8-5.6 (more on this below). My physicians in the early 2000’s had noted nothing amiss.
When reviewing my EHR lab test history I also discovered I had had fasting blood glucose screening over many years with far greater frequency than HbA1c, often linked to the annual physicals that my doctor of many years encouraged. Similar to my HbA1c results, four of 12 fasting glucose readings were flagged as (slightly) high from 2007 to the present, but never considered a problem by my physicians. The small variations were likely due to when or what I last ate.
So, I wondered, do these stable marginal findings in the context of shifting standards mean I have prediabetes? And what should I make of my new classification as a person with prediabetes?
Prediabetes prevalence and implications
Type 2 diabetes is a global health problem of considerable and rising proportions, with extremely serious associated morbidity and mortality (see here and here). An estimated 1 million persons in the world died from Type 2 diabetes in 2017, rising from the 18th leading cause of mortality in 1990 to the ninth. It is the most expensive chronic condition in the U.S., costing $327 billion in 2017 ($237 billion of which was in direct medical costs). Despite public health interventions, countries with higher socioeconomic status have increasing prevalence. The Centers for Disease Control and Prevention (CDC) estimates 37.3 million people or 11.3% of the U.S. population has diabetes, among whom 23% are undiagnosed. Diabetes is clearly linked to our national epidemic of obesity and disproportionately affects the poor and persons of color.
However, prediabetes is different, less clearly defined, and not universally accepted. Estimates of prediabetes prevalence worldwide vary greatly because of widely discrepant definitions and screening criteria, but a 2019 assessment found increasing trends. In the U.S. the CDC classifies 96 million adults, or fully 38% of the adult population, as having prediabetes based on HbA1c or fasting glucose.
Uncertainty surrounding prediabetes as a diagnostic category persists (Yudkin 2014). An evaluation of prediabetes prevalence in the 2015-16 U.S. adult population found that five different definitions using three different measures resulted in greatly divergent prevalence and implications for intervention. “If a combination of HbA1C 5.7–6.4%, FPG 100–125 mg/dl, and 2hBG 140–199 mg/dl was used—meaning all three criteria were satisfied—the prevalence was 2.5%. If any of the three definitions were used to define prediabetes, the prevalence was 51.3%.”
Interestingly, CDC reports that the age-adjusted prevalence of prediabetes did not change significantly between 2005-2008 and 2017-2020, while the proportion of people who were aware of the condition jumped from 6.5% to 17.4%. This might be attributed to the addition of a billable diagnostic code in 2017.
A little digging into the origins of the prediabetes classification is quite illuminating. It turns out there are conflicting opinions regarding the utility of testing, the best test to identify diabetes before it develops, the advisability of identifying and intervening with “prediabetes” patients (particularly for persons at low risk), and even the use of “prediabetes” as a diagnosis. I certainly knew nothing about these issues, nor was I informed of them, to help me evaluate my new diagnosis.
The invention of prediabetes
In 1974, the National Commission on Diabetes created a long range plan that set in motion U.S. research, programs, and treatment for diabetes and diabetes prevention. Before 1979 there were six different criteria for diagnosing diabetes that yielded discordant results. To address this and identify persons at risk of developing diabetes (also known as impaired glucose tolerance), national and international experts convened to review the data and set guidelines beginning in 1979 (see here and here).
Commercial assays for HbA1c, a measure of glycated hemoglobin (an indicator of blood glucose and the body’s ability to metabolize it), became available in the late 1970’s and popular in the 1980’s. The measure’s popularity stems from being stable, not requiring fasting, and reflecting the average blood glucose over the previous 3-4 months. While intended mainly to monitor blood glucose control among people with diabetes, it has become commonly used to detect prediabetes. However, screening for prediabetes using HbA1c is neither sensitive nor specific, while screening via fasting glucose is specific but not sensitive. Another complicating factor is the instability of HbA1c in individuals over time (more on this below).
Despite technical uncertainty, “prediabetes” entered the medical lexicon in 2001 (see here, here, and here). That year a public relations campaign was launched by the ADA to encourage physicians and patients to attend to slight increases in blood glucose that could increase the likelihood of developing diabetes. Prior monikers such as “impaired fasting glucose” and “impaired glucose tolerance” were replaced with “prediabetes” in all ADA literature, and the CDC took up the clarion call in its war on diabetes.
The condition or diagnosis of prediabetes is not uniformly endorsed. In 2009, the ADA reviewed the research on blood glucose testing using HbA1c. The three groups considering the evidence for prediabetes as measured by HbA1c diverged significantly in their opinions: the ADA enshrined the definition of prediabetes and set the intervention threshold at 5.7% (see here and here). Notably, the European Association for the Study of Diabetes (EASD) and the International Diabetes Federation (IDF) rejected the term prediabetes and considered a level of > 6% the intervention threshold. The World Health Organization (WHO) also rejected the new diagnostic category and standards. Rejection of the new diagnostic category was based on lack of evidence of routine progression from prediabetes to diabetes, and lack of evidence on treatment benefits (see here, here, here, and here).
Thus, I discovered, U.S.standards and guidelines are not shared across the OECD and have major implications for our population.
Does prediabetes progress to diabetes?
The CDC’s website asserts but does not quantify the increased risk of developing diabetes among those with prediabetes. What do the data show?
Already in 2006, a WHO report on diabetes and intermediate hyperglycemia (now known as prediabetes) stated that HbA1c is an unsuitable diagnostic test and recommended establishing a prediabetes risk score that combines measures of glucose tolerance with known risk factors.
A Cochrane systematic review of 103 prospective cohort studies with 250,000 participants and follow-up ranging from 1 to 24 years evaluated the likelihood of Type 2 diabetes (using three different lab tests including HbA1c) among persons with prediabetes compared to persons without. While cumulative diabetes incidence increased over time among persons with prediabetes (moderate certainty of evidence) and perhaps was more likely among those with prediabetes than persons with normal blood glucose (low certainty of evidence), the results varied widely between studies and by lab test used. Furthermore, many with prediabetes reverted to normal levels, even after up to 11 years of follow-up. None of the measures demonstrated a linear increase over time.
A large prospective community-based cohort of older adults with high prediabetes prevalence (using several different criteria) found only 12% developed diabetes no matter which of several definitions were applied; reversion to normal glycemia or death was more common than progression to disease after 6.5 years. The authors concluded that prediabetes among older individuals is not a “robust diagnostic entity” for predicting progression to disease and thus characterize aggressive diabetes screening in this population as “futile.”
A meta-analysis of patient-level longitudinal data evaluated the predictive ability of five different definitions of prediabetes on progression over five years. The definitions were similarly good at identifying people at high risk of progression. However, the authors concluded that setting lower thresholds for classifying prediabetes will increase the number of patients who qualify for intervention, but risks treatment for many who are healthy and unlikely to progress.
A 2021 evidence review for the United States Preventive Services Task Force (USPSTF) found insufficient follow-up on long term health outcomes, few documented events, and no mortality benefit of prediabetes interventions at 10 years. In addition, after 21 years of follow-up, a randomized controlled trial (RCT) comparing intensive lifestyle intervention, metformin, and placebo among 3,234 persons at high risk for diabetes in 27 U.S. centers found a reduction in weight, diabetes incidence, and cardiovascular risk factors but no reduction in cause-specific and all-cause mortality.
So, maybe “prediabetes” isn’t a thing, or isn’t a useful thing, particularly for persons at low risk?
Should we treat prediabetes?
The ADA’s 2020 guidelines on diabetes prevention or delay recommend annual risk factor screening and intensive behavioral lifestyle intervention that includes weight loss (achieving and maintaining 7% loss of body weight), moderate intensity exercise, a healthy diet, and smoking cessation among persons with prediabetes. The guidelines also recommend considering pharmacologic treatment with metformin (for those with BMI ≥35 kg/m2, age <60 years, and women with prior gestational diabetes), as well as considering other diabetes drugs that are not FDA-approved for prediabetes. The guidelines align with the large-scale RCT in 27 U.S. centers among persons at very high risk for developing diabetes (i.e., more than just prediabetes) that found a 27% decrease in diabetes incidence for lifestyle versus an 18% decrease for metformin compared with placebo after 15-years average follow-up. A USPSTF 2021 systematic review reported moderate certainty and moderate net benefit of prediabetes screening and referral to effective prevention interventions among overweight and obese persons. But should we screen everyone regardless of risk, and what is the evidence for implementing and sustaining interventions outside of clinical trials?
Many researchers question the utility and the advisability of prediabetes labelling and intervention altogether. The term itself is stigmatizing and erroneously implies progression. A systematic review and meta-analysis of interventions with people identified via screening demonstrate some efficacy at preventing or delaying onset, but due to inaccurate screening diagnoses and people who remain undiagnosed, the authors conclude that one should not screen and treat. Given the evidence on progression and the efficacy of intervention, prediabetes has been characterized as a “risk factor for developing a risk factor” that simply represents a lower threshold in diagnostic criteria, and perhaps “just a condition defined by laboratory measurements” rather than “a real risk factor for diabetes.”
Pharmacologic intervention for prediabetes is contentious. Numerous reviews caution that the poor quality of study methods and reporting limit conclusions about pharmacologic benefit to treating prediabetes, particularly when reversion to normal glycemia over time is common. Authors of a Cochrane review of prospective cohort studies on prediabetes progression state that the evidence, albeit based on wide variation in study methods and precision, suggests treatment may result in more harm than benefit, specifically because of reversion to normal glycemia in many individuals over time. An umbrella review of randomized trials reported some benefit of metformin in preventing diabetes but note study quality was low or critically low. A Cochrane review of RCT’s of one or more years duration (1-5 year range) evaluated whether metformin prevented or delayed diabetes onset in those at increased risk; authors concluded it was unclear if prediabetes resulted in better “patient important” outcomes. They found that metformin was no better than intensive diet and exercise alone and there were no benefits of metformin versus other drugs (although there was little reporting of side effects). A comprehensive review of the evidence recommends against metformin use (or any other drug used to lower the slightly elevated glucose levels of prediabetes) for treatment of prediabetes because two thirds of such persons don’t develop diabetes even after many years, and a third return to normal glucose levels. Additionally, the slightly elevated glucose levels of prediabetes are not associated with the small vessel complications of diabetes which affect the eyes, kidneys, and nerves so that lowering them further with drugs will not affect these complications. If drugs are used to treat prediabetes but then stopped, the chances of developing diabetes become the same as if the drugs had never been used. This suggests that the drugs will have to be used throughout a person’s lifetime even though two-thirds of people would not develop diabetes if drugs had not been used in the first place.
The author advocates weight loss (especially in overweight and obese persons) and other lifestyle interventions instead of medication, with close follow-up and treatment as soon as those at highest risk are diagnosed with diabetes. Weight loss is extremely important in overweight/obese people with prediabetes to prevent the development of diabetes. If diabetes should develop, only at this point should metformin or other anti-hyperglycemic drugs be used (Mayer Davidson, personal communication 2022). However, losing and then maintaining significant weight is very difficult. Recently, several drugs (which are also used to treat diabetes) have been shown to cause robust weight loss and one has already been approved to treat obesity (regardless of the presence of diabetes). With the weight loss, many people no longer have prediabetes. Unfortunately, when the drugs are discontinued, weight regain occurs necessitating continued or ongoing intermittent treatment.
Unrelenting prediabetes messaging:
Even after my unusual exchange at the podiatrist’s office and despite putting concerns about my lab test results to rest, the messages reaching me on prediabetes continued.
A few weeks after my podiatry consultation, I was sitting in slow-moving traffic on a major thoroughfare in town when a billboard in all caps caught my eye:
“RISK OF SHARK ATTACK:
1 IN 11.5 MILLION
RISK OF PREDIABETES:
1 IN 3 ADULTS
DoIHavePreDiabetes.org”
On the left side of the billboard was a graphic depicting a woman in a bikini on an inflatable swim donut; below her was what looked to be a Great White Shark aiming vertically toward her. On the bottom right were the sponsors of the message: the Ad Council, the AMA (American Medical Association), and the CDC.
Not long thereafter I visited my dentist. As usual, we chatted about a wide range of topics. She mentioned that recently it seemed all her patients were telling her they had prediabetes. The volume was so high she was beginning to tune out these announcements entirely. “What is going on?” she inquired.
Where do we go from here?
Can you imagine if over a third of the U.S. population was classified as “precancerous”? The anxiety levels in the community would skyrocket, as would the demand for consultations (both primary and specialty care), testing, intervention (whether medication, lifestyle changes, hardware for monitoring, etc.). This is akin to what is afoot with prediabetes.
Population-based lifestyle advice would be beneficial for many non-communicable diseases, rather than turning many healthy people categorized as having prediabetes into patients. While lifestyle interventions that reduce overweight, increase regular exercise, and eliminate smoking would be salubrious for multitudes of Americans, such interventions are extraordinarily challenging to implement and maintain. The ADA’s lower cut-point identifies a very large group of people who are not at risk of and are unlikely to develop diabetes, the majority of whom are not motivated to undertake or sustain lifestyle changes.
A preponderance of prediabetes labelling, messaging, and interventions are misleading at best, and misguided and harmful at worst. The science does not support classifying prediabetes as a disease requiring intervention without considering risk factors, and there are serious questions regarding optimal intervention. A 2015 review on diagnosis and treatment of prediabetes concludes “…there is no concrete evidence to formulate clinical guidelines for treatment of prediabetes.” Interestingly, the ADA’s 2021 guidelines states that “Prediabetes should not be viewed as a clinical entity in its own right but rather as an increased risk for diabetes and cardiovascular disease (CVD).” However, the ADA’s actions would indicate otherwise. Furthermore, intervening with all persons classified as having prediabetes is infeasible (see here, here, and here), distracts from and usurps resources for persons who are truly at risk and should be the focus of intervention, and ignores the lack of evidence that clinical trials can be replicated at a population level (see here, here, and here). A large cohort study on prediabetes progression among older adults recommends a focus on cardiovascular disease and mortality instead.
Assessments of the cost-effectiveness of community-based diabetes prevention programs have been based on underestimates of lifestyle interventions, inadequate length of follow-up and thus inaccurate projections (10-15 years vs. 30+ years needed), weight loss magnitude and sustainability not documented outside clinical trials, and (with one exception) an inappropriate mathematical model (Markov). Extrapolating from RCT’s with highly selected risk groups and intensive interventions is misleading. There have been no real-life observations of population level diabetes prevention program cost-effectiveness.
The focus should shift to interventions among persons who have or are much closer to diabetes (see here and here), with resources directed toward slowing the epidemics of obesity and diabetes. A 2021 report by the National Clinical Care Commission determined that treatment and prevention must focus on the “socioecological” nature of diabetes, one that requires much more than a medical approach (still poorly integrated in routine clinical practice), and includes agriculture, housing, environment, and industry. The psychological effects of diagnosing patients with prediabetes, costs, and potential harms are significant, but inadequately quantified.
It is worth placing the U.S. in context. We are out of step: our expansive view of prediabetes is not shared. The ADA’s lower cut-point, intended to identify a large group at higher risk of progression to diabetes and cardiovascular disease, simply identifies multitudes who are not at risk and suggests we embrace the international standards. Factors that influence maintaining ADA criteria may include funding, profit motives, risk aversion, and liability concerns. Pharmaceutical companies make large donations to the ADA and to individual physicians, resulting in serious potential conflicts of interest. The prediabetes classification has been immensely profitable increasing screening and physician visits, marketing of specific foods and supplements, demand for fitness coaching and glucose monitors, CDC’s diabetes prevention budget, diabetes drug marketing for prediabetes, and fierce competition to test prediabetes drugs. As noted above, prediabetes became a billable diagnostic code in 2017.
This led me to ask if prediabetes is a business plan in search of a diagnosis.
The decision to flag the EHR patient portal in alarmist fashion (red lettering, exclamation points, labelling) could and, in my opinion should, be rectified. “Marginally elevated” might be a better term for glycemia that is, in fact, just that. Furthermore, depiction of screening results over time should immediately pop up alongside the most recent result, something I had to hunt for in my EHR. This would temper distress at seeing a slightly elevated finding and place it in the context of past results.
In addition, communication between physicians and patients requires serious improvement. Education and discussion should include information about discrepancies between methods of measuring glycemia and variability over time, as well as reversion to normal levels in the majority of patients. Patients need not be frightened into believing that progression is inevitable. Those at high risk (e.g., obese) certainly should be encouraged to engage in behavioral lifestyle interventions, but these interventions must be tailored, affordable or covered by insurance, and realistic (including ongoing follow-up and support). Patients who are not at high risk need not be worried or necessarily receive intervention. Thoughtful talking points for physicians have been suggested. Pharmacologic solutions for prediabetes appear to be misguided and carry substantial risks. Thankfully, despite ADA recommendations, prevalence of prescribing metformin for patients with prediabetes remains low. The time to prescribe diabetes drugs seems to be when a patient has diabetes.
Finally, public messaging needs to be driven by science, not fear-mongering or profit. It is time to retire the term prediabetes and implement the international criteria embraced by WHO, Canada, and Europe.
Rani Marx, PhD, MPH is an epidemiologist and health services researcher dedicated to exploring how the U.S. health system currently fails patients and why that need not be the case. She writes about her own patient experiences through her professional lens, publishing on such topics as overscreening for women’s cancer, continuity of care, the EHR, and slow medicine. Rani dreams of a day when universal health care coverage and slow medicine are a reality. ranimarx1960@gmail.com
Acknowledgements
I am grateful for the constructive feedback provided by Drs. Mayer Davidson, John Yudkin, Richard Kahn, and James. G. Kahn.
As always, an extremely important and timely topic. When my A1c was elevated by one or two points I was frantic and I thought that I was going to become diabetic at any minute. Thank goodness my doctor explained to me that the range from one point to the other is very very different and no need to worry particularly since I had no intervening issues. Too bad I hadn’t read this sub stack a year ago. I would’ve saved myself a lot of aggravation. Thank you for posting this so important and timely.
As an internist and having a public health background, I think the author has missed a point while using her own example, as she is actually likely to be an outlier in the population who actually have optimal metabolic health. What I see in my daily practice, is a completely different reality, where I would say 70-80 % of my patients have overall poor metabolic health, which is the underlying root cause etiology for all their conditions.
As doctors, and in current "in and out" model of medicine, metrics and numbers are unfortunately what are used to assess performance which has so many downstream second order effects. Rather than focusing on labs like A1c, as the only measure and lipid panel, we forget the big picture of what is causing them to have these numbers in the first place. On one side, there is over treatment of elderly patients in terms of A1c, which can lead to risk of having risks of hypoglycemia, but on the other spectrum, there are too many patients who have completely normal A1c < 5.6, on whom I cannot put them on medications that can help improve their metabolic health, (eg, GLP-1 analogues), that have had a profound effect of helping patients lose weight, and have subsequently helped with their overall mental and physical wellbeing.
A recent NHANES study, found only 6.3 % of adults in optimal cardiometabolic health, which is just appalling. Considering that we are reaching 20 % of GDP, as health expenditure, and most states, the biggest employers are either retail giants like Walmart, and Amazon, and hospital systems, the market incentives are complete opposite to what we perceive as a good health.
"Less is more" model of healthcare doesnt work, as hospitals rely on sick patients being seen consistently, without poor biopsychosocial care provided outside. No wonder, we spend the highest $$ in the last year of patient's life, trying to fix something that was broken a long time ago. What I strive for in my patients, is how to convince them to change their daily behaviors, which I thing arguably is the most difficult thing to do, and unfortunately as part of training, we spend the least amount of time on it. We just assume that patients arent going to change, and its all their bad decisions that have led them to be like this, which is a very naive way to look at the systemic issues present. I occcasionally joke, the primary care DR stands for "Downstream Revenue", as it is just a referral based system to incentivize specialist interventions, so that they have to justify their existence. Obviously, this is an overgeneralization, but we spend less on primary care, and its in a dire state because of how it is perceived as low status in the healthcare world. In one year of my practice, my assumptions of thinking that primary care is boring are completely proven wrong, and I have never felt so grateful and proud of my patients, that have actually gotten better, and improved their behaviors if provided with the correct tools, and actually "LISTENING" to them.
Vandan Panchal, MD MPH