Seeing the evidence: Tighter control of blood sugars in type 2 diabetics

Mainstream media paid little attention to this study, even though it provided a comprehensive look at the clinical trial evidence to date on whether keeping tighter control over blood sugars benefits people with type 2 diabetes. The facts could have extraordinary impact on support for population HbA1c surveillance programs with obligatory diabetes management that are being enacted by growing numbers of government health departments, health plans and employee wellness programs. It could also be important information for people with type 2 diabetes. But facts that few people hear about can’t help many people.

Researchers, led by Dr. Kausik K. Ray, M.D., with the Department of Public Health and Primary Care at Strangeways Research Laboratory in Worts Causeway, Cambridge, UK, noted that type 2 diabetes is widely seen as a risk factor for cardiovascular disease, and clinical management of the disease is focused on glycemic control in hopes of reducing cardiovascular and microvascular outcomes. Some have suggested that clinicians should aim for increasingly lower blood sugar levels, which has led to randomized controlled clinical trials trying to see if more intensive control of blood sugars can reduce long-term clinical events and lengthen lifetimes, compared to standard treatments. To date, however, “individually these trials have failed to show consistent beneficial effects on cardiovascular events,” they wrote.

Suggesting that, perhaps, these trials may have been too underpowered to show a clinical benefit, these researchers conducted a meta-analysis of randomized controlled clinical trials done between 1970 and 2009 that had studied the effect of tight blood sugar control on cardiovascular outcomes. The study was just published in The Lancet, but it deserves a closer look to distinguish what the data itself revealed from interpretations.

Methodology

The authors stated that their methodology was robust. They included only randomized, placebo-controlled trials that compared standard treatment to more intensive lowering of blood sugars, and which had cardiovascular events as their primary endpoint and reported measured clinical endpoints, including all-cause mortality. Their search to identify trials and obtain the needed clinical data was intense and included published and unpublished data, and they assessed the probability of publication bias with funnel plots and the Egger test.

They excluded six trials that failed to meet their criteria for fair tests, such as failing to use a placebo, failing to examine diabetics, or failing to report clinical endpoints. One study was excluded for not having cardiovascular events as its primary endpoint: the RECORD Study. This randomized controlled clinical trial on 4,447 type 2 diabetes patients had compared standard treatment to those receiving additional medications (rosiglitazone) for glycemic control. The interim analysis, cited by the Cambridge authors, reported that after nearly four years, there was no difference between the groups in deaths from any cause or from cardiovascular events. The rosiglitazone group also had more than a double risk for heart failure, compared to standard treatment.

The Cambridge authors also excluded trials conducted on hospitalized patients. The reasons for this may not be clear to the general public, but it is widely known among medical professionals that tighter blood sugar control among critically ill patients has never been shown to lower their mortality, the need for dialysis or mechanical ventilation, or the number of hospitalized days; but has even been shown to increase risks of dying and raise risks for serious hypoglycemia by up to 13-fold.

Last year, for example, the VA Outcomes Group had noted that recommendations for tight glucose control in critically ill patients are “based largely on one trial that shows decreased mortality in a surgical intensive care unit” but similar studies have not and are reporting that tight glucose control can cause dangerous hypoglycemia. They felt “the data underlying this recommendation should be critically evaluated.” They examined 29 randomized controlled trials involving 8,432 critically ill adult patients. Their results, published in the August 2008 issue of the Journal of the American Medical Association, found that tight glucose control was not associated with significant reduced mortality or need for dialysis, but with a five-fold higher risk of hypoglycemia.

A large international analysis of 26 randomized clinical trials on critically ill patients, just publishedCanadian Medical Association Journal, found no benefit in overall mortality with tighter blood sugar control, but it was associated with a six-fold increase in risks of severe hypoglycemia. last month in the

Doctors with Stamford Hospital, an adult ICU at a large university-affiliated hospital, examined about six years of patient records and found that a single episode of severe hypoglycemia was associated with more than doubled risk of dying among their critically ill patients.

The most recently reported randomized clinical intervention study was the large international NICE-SUGAR Study. In this study, 6,104 patients with type 2 diabetes admitted to intensive care units were randomized to receive either standard blood sugar control (with a blood sugar target of 180 mg/dl or less) or tighter control (81-108 mg/dl). Regardless of the different conditions among the critically-ill patients (medical or surgical), the 90-day mortality was higher among those whose blood sugars were more intensively controlled: 27.5% died compared to 24.9% receiving standard control. There was no difference between the groups in the number of days in the ICU or in the hospital, or those needing dialysis or mechanical ventilation. But life-threatening severe hypoglycemic episodes occurred in 6.8% of the intensive control group compared to only 0.5% in the standard care group.

In the end, five trials were included in the Cambridge authors’ meta-analysis, all varying greatly in the demographics of the participants, progression of their diabetes, duration of follow-up, and drugs used for intensive glucose control.

Findings

While mainstream media didn’t widely report Dr. Ray and colleagues’ study, industry trade publications for medical professionals did. They consistently reported the study as supporting intensive glucose control, all emphasizing that lowering mean A1c levels by nearly 1% had been shown to significantly reduce cardiovascular events.

The study authors, themselves, concluded: “Our quantitative analysis of randomised controlled trials provides reliable large-scale evidence of a consistent beneficial effect of intensive treatment on non-fatal myocardial infarction and coronary heart disease, without increased risk of all-cause mortality.”

The review articles reported the authors’ findings, for example, that the intensive medical and pharmacological blood glucose management was associated with a 15% lower risk for all coronary heart disease events, compared to standard treatment. The authors had calculated odds ratios, which makes risks seem more significant, while actual compiled incidences of coronary heart disease events were 6.8 percent in the intensive treatment group and 7.2 percent among the standard treatment groups, with a difference of a mere 0.4%. Absolute values aren’t headline material.

Heartwire said this study reinforces the recommendations of the American Diabetes Association, American Heart Association, and American College of Cardiology, to get diabetics’ HbA1c levels below 7.0%. No one drug is capable of doing that in the vast majority of patients, and the use of three or more drugs in combination is common. Dr. Ray told Heartwire, “what this study does is help those individuals out there with diabetes and their caregivers and lets them know that what they’ve been doing is likely safe and is justified.”

The full picture

But is that really what this study demonstrated? Did you catch the more important piece of the story — the one that patients care about?

If handfuls of drugs might lower your risk of dying from one thing, but raise your risks of dying from something else, that’s an important part of the story. Medical professionals care for the whole patient and the most important clinical endpoint is death. The more intensive drug regimens were not shown to improve lifespans. Overall deaths were slightly higher in the intensive treatment group compared to the standard treatment. It was an untenable 2% higher risk (0.8% higher actual incidences), but clearly this meta-analysis didn’t provide evidence to justify exposing patients to more drugs to lower a single health index, which comes with additional financial costs and concomitant side effects.

This paper didn’t examine what the causes of the higher deaths were among the patients receiving more intensive management or side effects that may have impacted the patients’ quality of life, although the authors did note that severe hypoglycemic episodes were doubled in the intensive treatment group.

In contrast to the industry coverage of this meta-analysis, an evidence-based perspective might more accurately have said: "The clinical trial evidence to date fails to show that tight glucose control offers significant benefits to patients."

Instead, medical professionals saw headlines focused on reduced nonfatal heart attacks and coronary heart disease events, leading most readers to come away thinking overall benefits had been shown.

In fact, the meta-analysis found greater risks of dying among those with the most intensive pharmacological management of blood sugars. The two trials (VADT and ACCORD) showing the highest risks for overall deaths were on patients with the more advanced diseases (10-12 years from first diagnosed compared to 8 years in all but one* of the other trials). The patients also had the highest HbA1c levels at the start of the trials (9.4 - 8.3% compared to 7.9 - 7.1% in the other trials), meaning they also required more medications to bring their HbA1c levels below 7%. More intensive pharmacological management and advanced disease was associated with greater risks of dying. The authors said they didn’t have sufficient data when calculating their odds ratios to control for confounding factors, such as age, gender, duration of diabetes, baseline HbA1c, and comorbidities. In other words, they couldn’t credibly conclude that tighter glucose control was behind the purported lower risks seen primarily in the type 2 diabetes whose disease was less advanced.

A second look

Did you see the controversial study included in their analysis? The trial that the authors showed with the most significant reduction in risks of dying (-21% odds ratio) associated with intensive glucose management to get HbA1c levels below 7% was the UKPDS Study (United Kingdom Prospective Diabetes Study). Without this trial, the reported risks of dying associated with tighter glycemic control would have been higher.

* This was also the one trial with newly diagnosed patients (<1 year from diagnosis) and with starting HbA1c levels of only 7.1%. These HbA1c levels were much lower than those seen in the VADT study participants (average 9.4%) and in the ACCORD patients (average 8.3%).

But the UKPDS Study had reported no statistically significant (p=0.44) reduction of all-cause mortality among the intensive blood sugar managed group compared to standard treatment. As covered here in “Are you sure about that?”, the evidence found in the UKPDS Study had contradicted the claims of benefits that were being reported by reviews in industry publications and written by experts in the medical community. The UKPDS data itself showed that early and intensive management was not associated with significant reduction in any adverse outcome. The trial showed no reduction in any macrovascular (large blood vessel) complication, such as cardiovascular events (heart attacks, heart failure, strokes or amputations), and no reduction in deaths.

As Dr. Kenneth G. Marshall, M.D., with the Department of Family Medicine at the University of Western Ontario, wrote in the Journal of the Canadian Medical Association, the UKPDS documented that intensive pharmacological treatment of type 2 diabetes did not lessen illness or reduce deaths from macrovascular causes. A sub analysis on ‘obese’ diabetics was widely reported has showing that the intensive therapy reduced cardiovascular disease in these patients, but few heard that it had no effect on microvascular or cardiovascular outcomes or that these patients had higher rates of death compared to obese controls receiving standard care (RR=1.60).

“The UKPDS reported a 25% reduction of microvascular [small blood vessel] disease with intensive treatment,” noted Dr. Marshall, but this was defined by using a surrogate outcome: progression of retinopathy as identified by ophthalmologic examination. The study was not blinded, lending further caution with such subjective assessments. The claim was not based on clinical endpoints related to microvascular disease, such as visual acuity or blindness. “No difference was seen in the more important clinical outcome of vision loss between patients treated intensively and those who received conventional treatment,” Dr. Marshall wrote.

Never the less, the UKPDS Study’s abstract concluded: “Intensive blood-glucose control by either sulphonylureas or insulin substantially decreases the risk of microvascular complications.” And this is the source of beliefs that tighter blood sugar control reduces the complications of the disease.

As a busy family GP of 17 years , I have many NIDDM patients and get to know them all quite well. I have looked at the actual evidence ie UKPDS to see if I can decide for myself how to advise the diabetic who is torturing themselves physically and psychologically to seek the supposed targets promoted in Diabetic Education Brochures. I have read UKPDS 33 over and over and over and am just astounded at the rampant interpretative bias…On a final cynical note I wonder how the BMJ/Lancet could have ever allowed the abstract for UKPDS33 to include the comment, "SUBSTANTIALLY reduces the risk of diabetic complications." It is the abstract that is quoted and requoted, and there is the origin of the mythology. — Dr. Paul C. Neeskens, “UKPDS—Emperors New Clothes,” British Medical Journal, September 11, 2003.

To this day, no sound clinical study has ever shown that treating type 2 diabetics to achieve even lower blood glucose levels provides added benefits that outweigh the harms. Treating a number that is a symptom of a disease doesn’t mean the disease process has been changed. Lowering health indices in elderly patients to match those of healthy 20 year olds doesn’t mean their risks will be lowered to those of 20-year olds again. And minimizing the risks associated with extremely high lab values doesn’t mean that “how low can you go” is better for patients.

Busy medical practitioners rely heavily on experts’ assessments of research findings, but those assessments are fraught with biases. As Dr. John P. Ioannidis, M.D., at the University of Ioannina School of Medicine in Ioannina, Greece, and with the Institute for Clinical Research and Health Policy Studies at Tufts-New England Medical Center, Tufts University School of Medicine in Boston, cautioned: “Empirical evidence on expert opinion shows that it is extremely unreliable.” As we also see time and again, the analyses and conclusions made by study authors and industry experts often differ from what the data actually shows. Bias doesn’t always come from financial conflicts, but can come simply from a belief in a popular scientific theory. It can lead even medical professionals to see only what supports a theory: confirmation bias.

Myths can take on lives of their own even in medicine unless we look objectively and carefully at the evidence. Only with unbiased discussions can we ever hope to turn evidence-based medicine into evidence-based medicine.

© 2009 Sandy Szwarc