The prevalence of type 2 diabetes mellitus (DM) increases with age and reaches 25% in those older than age 65 years. Pre-diabetes status is also very common in the elderly, and is present in about half of those age 75 years and older. Many physicians care for elderly patients with diabetes and pre-diabetes, dealing with the challenge of controlling glucose levels and improving health with minimal adverse events. Over the last decade, research on diabetes among the elderly population has proliferated, adding new information on this topic. This review summarizes the updated medical literature on diabetes and pre-diabetes in the elderly, including the significance of pre-diabetic conditions, new-onset DM in the elderly and long-standing DM. The role of therapeutic intervention and the level of glycemic control for this population are discussed in particular.

The concept of individualized treatment for type 2 diabetes mellitus (DM) is becoming established and replaces previous recommendations for tight glucose control for all diabetic patients. One of the main criteria in constructing personalized care for the patient is the chronologic and biologic age.

The incidence and prevalence of DM increase with age (http://www.cdc.gov/diabetes/statistics). Pre-diabetes states, including impaired fasting glucose (IFG), impaired glucose tolerance (IGT) and elevated HbA1c are even more prevalent among the elderly[1-3].

However, the clinical impact of glucose levels on microvascular and macrovascular complications, and mortality is not well established. The commonly used clinical guidelines do not provide separate recommendations for elderly individuals with pre-diabetes, and do not differentiate between elderly with long-standing or new-onset DM[4].

This review summarizes the data in the literature regarding the effect of glycemia in different stages on morbidity and mortality in the elderly population. It will address the aspects of the clinical impact of glucose levels in pre-diabetes, diabetes that was first diagnosed in old age and long-standing diabetes separately.

According to American Diabetes Association (ADA) guidelines, pre-diabetes may be diagnosed as IFG, IGT and/or by elevated HbA1c values of 5.7%-6.4%[4]. The rates of pre-diabetes states, including IFG and IGT, are very common in the general population, and increase with age[1-3].

It is well-established that pre-diabetes states are a significant risk factor for developing type 2 DM, as well as for diabetic complications and mortality in younger adults[5,6]. It is also well known that lifestyle changes, including loss of at least 7% of body weight and ≥ 150 min/wk of activity, delay or even prevent development of DM, and may potentially reduce its complications among persons with pre-diabetes[7]. Thus, pre-diabetic patients are an important target group for primary prevention interventions.

Understanding the clinical impact of pre-diabetes in older adults is very important, as the prevalence of pre-diabetes increases with age and reaches about 50% in those age 75 years and older[2]. Moreover, lifestyle interventions are more successful in decreasing hyperglycemia in the elderly than in younger adults. This was demonstrated in the Diabetes Prevention Program (DPP) trial, which included subjects with combinations of IGT and IFG, considered to be at high risk for developing DM. The oldest age group, 60-85 years at enrollment, had the greatest benefit from the program, both in terms of weight loss and decreased incidence of DM over time[7,8].

Yet, there are some important, unanswered questions for the clinician. First, what is the clinical impact of pre-diabetes state in the elderly? Second, do glucose lowering interventions improve morbidity and mortality in this population?

The answer to the first question, regarding the clinical significance of pre-diabetes in elderly subjects, is based on a small number of studies (Table 1). A prospective, observational study followed 1466 elderly subjects with IGT and compared their mortality rate to subjects with normal glucose levels and overt diabetes. The age of enrolled participants was 55-74 years and median follow-up was 8.8 years. Mortality rates were almost equal in the pre-diabetes and normal glucose groups. Nevertheless, within the non-diabetic range (i.e., normal and pre-diabetic glucose levels), a J-shaped association was demonstrated between glycemia and all-cause mortality, even after adjustment for multiple risk factors. The lowest mortality rates were documented in subjects with fasting plasma glucose 88-93 mg/dL and HbA1c 5.4%-5.5%. Participants with glucose levels at the upper pre-diabetes range had a higher mortality rate[9].

On the other hand, a recent prospective cohort study of 8365 older subjects, 50-74 years old, revealed that the increased cardiovascular risk in pre-diabetes (defined as IFG or HbA1c 5.7%-6.4%) can mainly be explained by other concurrent cardiovascular risk factors and not by the hyperglycemia itself[10]. Similar results arose from the Cardiovascular Health Study of 4602 community-dwelling elderly participants, 65 years-of-age and older. This study found no evidence that pre-diabetes is an independent risk factor for a variety of cardiovascular outcomes, including heart failure, myocardial infarction, stroke and all-cause mortality[11]. Pre-diabetes increased the risk of developing DM, but the absolute rate was low and not related to increased cardiovascular risk.

An interesting pooled analysis examined the age specific effect of different metabolic risk factors on cardiovascular diseases. The analysis for plasma glucose included 372000 participants in 116 cohorts. The authors calculated the impact of mildly elevated glucose on the relative risk of ischemic heart disease and stroke. They concluded that the proportional effect of elevated fasting glucose declined with age[12].

For the second question - whether glucose lowering intervention would improve morbidity and mortality outcomes in elderly population - there is still no satisfactory answer. The DPP study, mentioned above, found a better response of elderly subjects to lifestyle interventions, in particular weight loss and DM prevention. However, clinical outcome data were not reported[7,8]. The effect of lifestyle interventions on mortality and cardiovascular disease in pre-diabetes patients is questionable, even among younger adults[13].

In summary, existing data regarding the clinical impact of pre-diabetes on morbidity and mortality among elderly individuals are limited and study results are conflicting. There is an association between pre-diabetes and mortality, which increases with higher glucose levels within the pre-diabetic range. However, it is unclear whether pre-diabetes is a marker of poor metabolic condition or an independent risk factor. There is almost no information regarding the influence of lifestyle or medical interventions on morbidity and mortality in this population.

The ADA does not have separate clinical guidelines for pre-diabetes in the elderly population[4]. The 2012 consensus report of the ADA and the American Geriatrics Society provided vague advice on this issue. They recommended screening for pre-diabetes in elderly patients who are likely to benefit from diagnosis of pre-diabetes and from subsequent intervention. Similarly, they recommended lifestyle interventions for elderly individuals with pre-diabetes who are able to participate and are likely to benefit from DM prevention[14].

The pathophysiology of the appearance of DM in the elderly is a combination of age-related changes in carbohydrate metabolism, pancreatic endocrine dysfunction and adverse lifestyle factors[15,16]. The epidemiology of incident DM in relation to population age is interesting. The incidence of new-onset DM increases with age until age 65 years, after which both incidence and prevalence of DM seem to level off (http://www.cdc.gov/diabetes/statistics).

The natural history of new-onset DM in the elderly seems to have a benign course in comparison to that of long-standing DM. Interesting information comes up from a study of centenarian subjects, ages 100-109, compared to elderly subjects aged 65-84 years. The centenarians had relatively low prevalence of DM (7.64%), and almost exclusively had senile DM, that is DM diagnosed after 65 years of age. The authors suggest that long-standing DM is not compatible with extreme longevity, while senile DM does not change the clinical outcomes significantly[17]. New onset DM in older age is associated with better glycemic control[18] and with less frequent microvascular complications compared to long standing DM[19]. Data from the National Health and Nutrition Examination Survey database found that although elderly with new-onset DM were 5 years older in average, they had much lower prevalence of retinopathy and a similar burden of macrovascular disease compared with long-standing DM. The difference in retinopathy rate may reflect the difference in DM duration between the two groups[20].

The few studies that compared elderly subjects with new-onset DM to non-diabetic patients demonstrated short term elevation in all-cause and cardiovascular mortality and long term elevation of microvascular and macrovascular complications[21-25]. None of these studies checked the association of glucose levels with mortality or diabetic complications and did not consider the influence of other cardiovascular risk factors on morbidity and mortality.

A large, observational study focused on the association between glycemic control and mortality in elderly patients with new-onset DM. This study followed almost 3000 elderly patients with new-onset DM for 7 years. A J-shaped relationship was found between HbA1c level and mortality rate. A HbA1c level above 7.5% was associated with significantly higher all-cause mortality, while the lowest mortality rate was found in subjects with HbA1c levels from 6.5% to 6.99%. This association remained statistically significant after adjustment for other conventional cardiovascular risk factors[26].

In summary, the existing data suggest that new-onset DM in the elderly is associated with better glycemic control and better prognosis compared to long-standing DM in this population. However, when compared to elderly people with normal glucose levels, the new-onset DM patients have higher rates of morbidity and mortality (Table 2). There is some evidence that higher glucose levels within the diabetic range are associated with increased mortality.

The ADA guidelines do not deal separately with new-onset DM in elderly individuals, but mention the duration of the disease as a parameter that should be considered when choosing HbA1c target levels[4]. Vacante et al[27] suggested combining the current age of the patient with the duration of DM.

As mentioned above, long-standing DM in the elderly has higher morbidity rates compared to new-onset DM. Therefore, the question is whether good glycemic control in elderly people with long-standing DM will influence the course of the disease.

In young and middle-aged diabetic patients, the role of tight glycemic control is crucial, as was proven in the Diabetes Control and Complications Study (DCCT) and United Kingdom Prospective Diabetes Study (UKPDS) in type 1 and type 2 DM, respectively[28-30]. These two studies confirmed the benefit of intensive glycemic control in reducing microvascular and macrovascular complications[31]. Nonetheless, these studies included only new-onset diabetic patients and excluded patients ages 65 years and above at the time of enrollment.

Almost 10 years after the publication of the UKPDS study, and 15 years after the DCCT study, 3 large randomized controlled trials examined the influence of intensive glycemic control on microvascular and macrovascular complications in older subjects with long standing DM. The Action to Control Cardiovascular Risk in Diabetes trial[32] enrolled diabetic patients (mean age 62.2 ± 6.8 years) 10 years after diagnosis, 35% with previous cardiovascular disease. This trial was terminated after 3 years because of excessive deaths in the intensive glycemic control arm. The Action in Diabetes and Vascular Disease trial[33], which had also enrolled people with advanced disease (mean age 66 ± 6 years, average duration of diabetes 8 years, 32% with previous major macrovascular disease), showed no significant effect of tight glycemic control on major macrovascular events or death from any cause, but there was significant reduction in nephropathy incidence and as a result, reduction in the incidence of combined microvascular and macrovascular events. This reduction was proven only for patients younger than 65 years, according to a sub-group analysis published as part of the trial. The Veterans Affairs Diabetes Trial trial[34] enrolled similar diabetic patients (mean age 60.5 ± 9 years, 11.5 years after diagnosis, 41% with major macrovascular disease) and also showed no significant effect of tight glycemic control on major macrovascular events or death from any cause. As one would expect, adverse events related to intensive glycemic control, such as hypoglycemia, were more common in the elderly[35].

Few studies are directed to glycemic control in the old age. For example, the retrospective diabetes and aging study[36] enrolled participants above 60 years of age (38% between 70-79 years and 15% age 80 or older); 57% had diabetes for more than 4 years at enrollment. There was a U-shaped relationship between mortality and HbA1c, with the lowest mortality rate at HbA1c 6%-8%, in all age-groups (Table 3).

The benefit of tight glycemic control for prevention of microvascular complications is not immediate. As proven in the UKPDS studies, the difference in outcome between the tight glycemic control group and the control group only appeared 9 years after randomization[14,31]. In other words, for patients with life expectancy of 7 years or less, the benefit in this area is doubtful.

The complexities of diabetes care in old age, with benefits alongside potentially serious adverse events, led researchers to quality-adjusted life year (QALY) trials. Vijan et al[37] compared the QALY gained with intensive glycemic control versus moderate glycemic control, in different age-groups. They concluded that older patients, age 75 years and older, experience smaller benefit from glycemic control compared to younger patients, and their expected gain in QALYs for a 1-point change in HbA1c was minimal, even with the favorable assumption that the benefits of glycemic control extend to the elderly.

The results of the aforementioned studies resulted in changes in the clinical guidelines regarding treatment goals for elderly with long-standing DM. However, the guidelines offer general instructions and leave a large margin for clinical judgment.

The 2014 ADA guidelines[4] recommend a standard glycemic goal of HbA1c below 7% for adults, and a less stringent goal, such as < 8% for patients with a more complex status, which is defined according to disease duration, life expectancy, important comorbidities, risk for adverse events and existing vascular complications. The International Association of Gerontology and Geriatrics and the European Diabetes Working Party for Older People published similar recommendations[38].

The consensus report of the ADA and the American Geriatrics Society from 2012[14], offered 3 levels of glycemic control for the old patient: HbA1c < 7.5% for healthy patients, < 8% for patients with intermediate health status (multiple chronic illnesses, 2+ instrumental impairments or mild cognitive impairment), and < 8.5% for patients with poor health status (end-stage chronic illnesses or moderate-to-severe cognitive impairment or 2+ activities of daily living dependencies). The International Diabetes Federation 2013 guidelines[39], suggest a similar categorization. They also emphasize that for end-of-life situations, the goal should be merely to avoid symptomatic hyperglycemia (Table 4).

This review summarizes the current evidence about glycemic control in the elderly. Similar to young and middle-aged adults with DM, it seems that the elderly patient with diabetes has higher risks for morbidity and mortality compared to a non-diabetic person of the same age. Even so, new-onset DM is less severe in elderly patients compared to young adults and easier to control. The impact of pre-diabetes state on morbidity and mortality risk in the elderly is doubtful, and the role of screening and treatment in these patients is questionable. Finally, the importance of tight glycemic control on long-standing DM in the elderly in not well-established and the preferred level of glycemic control should be considered in the overall context of the patient’s health status. The optimal level of control among elderly patient subgroups requires further evaluation.

Beyond all the above, the heterogeneity of the elderly population presents a significant challenge in clinical decision making. Old diabetic patients can be healthy or with much comorbidity and the risks of adverse events from medications increases with age. The decision regarding an individual patient’s glycemic goal should be made, ideally with the patient himself, after considering all the comorbidities, together with current cognitive state, risk of adverse events, quality of life aspects and life expectancy.

The authors are grateful to Prof. MS Shapiro for his assistance.