eAppendix 1. Codes Used for Analysis
eAppendix 2. Sex and race-adjusted incidence of diabetes complications in older adults with short duration of type 2 diabetes and no prior history of diabetes complications
eAppendix 3. Sex and race-adjusted incidence of diabetes complications in older adults with long duration of type 2 diabetes and no prior history of diabetes complications
Huang ES, Laiteerapong N, Liu JY, John PM, Moffet HH, Karter AJ. Rates of Complications and Mortality in Older Patients With Diabetes MellitusThe Diabetes and Aging Study. JAMA Intern Med. 2014;174(2):251-258. doi:10.1001/jamainternmed.2013.12956
In the coming decades, the population of older adults with type 2 diabetes mellitus is expected to grow substantially. Understanding the clinical course of diabetes in this population is critical for establishing evidence-based clinical practice recommendations, identifying research priorities, allocating resources, and setting health care policies.
To contrast the rates of diabetes complications and mortality across age and diabetes duration categories.
Design, Setting, and Participants
This cohort study (2004-2010) included 72 310 older (≥60 years) patients with type 2 diabetes enrolled in a large, integrated health care delivery system. Incidence densities (events per 1000 person-years) were calculated for each age category (60-69, 70-79, and ≥80 years) and duration of diabetes (shorter [0-9 years] vs longer [≥10 years]).
Main Outcomes and Measures
Incident acute hyperglycemic events, acute hypoglycemic events (hypoglycemia), microvascular complications (end-stage renal disease, peripheral vascular disease, lower limb amputation, and diabetic eye disease), cardiovascular complications (coronary artery disease, cerebrovascular disease, and congestive heart failure), and all-cause mortality.
Among older adults with diabetes of short duration, cardiovascular complications followed by hypoglycemia were the most common nonfatal complications. For example, among individuals aged 70 to79 years with a short duration of diabetes, coronary artery disease and hypoglycemia rates were higher (11.47 per 1000 person-years and 5.03 per 1000 person-years, respectively) compared with end-stage renal disease (2.60 per 1000 person-years), lower limb amputation (1.28 per 1000 person-years), and acute hyperglycemic events (0.82 per 1000 person-years). We observed a similar pattern among patients in the same age group with a long duration of diabetes, with some of the highest incidence rates in coronary artery disease and hypoglycemia (18.98 per 1000 person-years and 15.88 per 1000 person-years, respectively) compared with end-stage renal disease (7.64 per 1000 person-years), lower limb amputation (4.26 per 1000 person-years), and acute hyperglycemic events (1.76 per 1000 person-years). For a given age group, the rates of each outcome, particularly hypoglycemia and microvascular complications, increased dramatically with longer duration of the disease. However, for a given duration of diabetes, rates of hypoglycemia, cardiovascular complications, and mortality increased steeply with advancing age, and rates of microvascular complications remained stable or declined.
Conclusions and Relevance
Duration of diabetes and advancing age independently predict diabetes morbidity and mortality rates. As long-term survivorship with diabetes increases and as the population ages, more research and public health efforts to reduce hypoglycemia will be needed to complement ongoing efforts to reduce cardiovascular and microvascular complications.
Nearly half of the 24 million patients with diabetes mellitus living in the United States are older than 60 years. In the next 2 decades, their numbers are expected to double and their direct medical costs are expected to triple because of the combined effects of an aging population and high rates of overweight and obesity.1 The clinical heterogeneity of these patients, in terms of characteristics such as duration of diabetes and comorbid illnesses, greatly increases the challenge of caring for older patients.2 Many older patients with diabetes are living longer than in previous years. Longer duration of diabetes is associated with more complications and more difficulty with maintaining glycemic control.3,4 Understanding the contemporary clinical course of diabetes in older patients is the critical first step needed to individualize and prioritize care as well as target support for future research efforts.
The clinical course of diabetes in older patients today is presumably different from that reported in previous studies because of the rapid evolution of diabetes care. Quiz Ref IDMost of our current understanding about the clinical course of diabetes in older patients is based on studies of populations from the 1990s.5,6 These studies found that Medicare patients with diabetes had elevated risks of complications (microvascular and cardiovascular) and mortality compared with patients without diabetes5 and that cardiovascular events (ischemic heart disease [181.5 events per 1000 person-years] and stroke [126.2 per 1000 person-years]) were by far the most common complications, with hypoglycemia being much less common (28.3 per 1000 person-years).6 After publication of the UK Prospective Diabetes Study findings in the late 1990s,7,8 clinicians have pursued more aggressive risk factor control and have more treatment options at their disposal.9 For example, clinicians dramatically increased their prescribing of angiotensin-converting enzyme inhibitors and statins after the release of evidence demonstrating the benefits of these drugs10,11 for preventing diabetes complications.12
Apart from not reflecting the influence of recent changes in clinical practice, prior studies5,6 of older patients did not systematically examine how the course of diabetes differs by age and duration of the disease. These variables have been proposed as 2 key potential guides to the individualization of care goals and treatments.13 In a study by Bethel and colleagues,5 investigators monitored a diabetic population from a 5% sample of the Medicare population with newly diagnosed diabetes in 1994 and did not provide data on patients with longer diabetes duration. Bertoni and colleagues6 used the same Medicare sample to look at rates of complications by different age categories but did not evaluate differences in the course of the disease by duration of diabetes. A more nuanced description of the clinical course of diabetes is needed to inform medical decision making in older patients with diabetes. In this article, we describe contemporary rates of diabetes complications and mortality and contrast them across categories of age and duration of diabetes in a population of older adults with uniform access to integrated care.
The Kaiser Permanente Northern California (Kaiser) Diabetes Registry (Registry) is a well-characterized population maintained continuously since 1993.14- 16 Registry inclusion is based on a validated algorithm incorporating multiple sources of data, including pharmacy records, laboratory data, and outpatient, emergency department, and hospitalization diagnoses of diabetes.17 Data from clinical information systems (electronic medical records) are downloaded annually. The Diabetes and Aging Study has been approved by the review boards of the Kaiser Foundation Research Institute and the University of Chicago. A waiver of informed patient consent was granted by both review boards.
Our sampling frame included individuals with type 2 diabetes from the Registry who were aged 60 years or older on January 1, 2004 (baseline), with continuous Kaiser membership and available clinical data. Starting with the full Registry (N = 228 740), we excluded 75 673 individuals younger than 60 years, 61 284 lacking continuous Kaiser membership or pharmacy benefits in the 12 months prior to baseline, 11 107 with type 1 diabetes mellitus or unknown type of diabetes,18 and 8366 individuals with no hemoglobin A1c (HbA1c) test result during the year prior to baseline or no date of diabetes diagnosis (used to estimate duration). The remaining 72 310 patients were the basis for these analyses.
We monitored patients for a 7-year observation window, assigning person-time at risk starting on January 1, 2004, and censoring follow-up at the first occurrence of the following: any of the nonfatal outcomes of interest, death, discontinuation of Kaiser membership or pharmacy benefits (gap of at least 3 months in coverage), or the end of follow-up (December 31, 2010). The mean follow-up time for mortality was 5.44 years.
Patients were divided into 6 major subgroups defined jointly by age (60-69, 70-79, and ≥80 years) and duration of diabetes (shorter [0-9] vs longer [≥10] years) as of the baseline date (January 1, 2004). Date of initial diagnosis of diabetes was based preferentially on self-report (38.5%) and, if that date was missing, on administrative records (61.5%).
The occurrence and timing of each event were based on a combination of outpatient, emergency department, or inpatient primary diagnostic codes (International Classification of Diseases, Ninth Revision) or procedure codes (Current Procedural Technology, Fourth Edition) (Supplement [eAppendix 1] ).
Acute hyperglycemic event was defined as hospitalization for diabetes with hyperosmolarity, diabetes with ketoacidosis, or uncontrolled diabetes. Acute hypoglycemicevent (hypoglycemia) was defined based on hospitalization or emergency department diagnostic codes for hypoglycemia.
Microvascular complications included end-stage renal disease (ESRD), severe diabetic eye disease, peripheral vascular disease, and lower limb amputation. IncidentESRD was defined as initiation of chronic dialysis therapy or kidney transplant identified through hospitalization records or Kaiser’s reporting systems for the US Renal Data System. Severe diabetic eye disease was identified based on outpatient diagnostic codes for blindness, proliferative retinopathy, macular edema, or outpatient photocoagulation procedures. The presence of peripheral vascular disease was based on inpatient codes for the disease as well as lower limb angioplasty. Amputation was identified through inpatient diagnostic and procedure codes.
Nonfatal cardiovascular complication was identified via hospitalization and/or emergency department records. These complications included coronary artery disease (myocardial infarction, coronary artery bypass surgery, and angioplasty), cerebrovascular disease (ischemic or hemorrhagic stroke and carotid endarterectomy), and congestive heart failure. Mortality and date of death were captured from the California State Mortality File, Social Security Death Records, or Kaiser administrative records when the death occurred within the health system.
Covariates ascertained at baseline included demographics (sex and race/ethnicity), most recent laboratory test results within 1 year before baseline (HbA1c, low-density lipoprotein cholesterol, and glomerular filtration rate), prevalent complications (acute hyperglycemic or hypoglycemic event within the previous year, or any history of ESRD, peripheral vascular disease, lower limb amputation, eye disease, coronary artery disease, cerebrovascular disease, and congestive heart failure), and diabetes-related medications the patient was receiving.
All analyses were performed with SAS, version 9.1 (SAS Institute Inc) and associations were considered statistically significant at the P < .05 level. However, given the large sample size, we also considered clinical significance in addition to statistically significant contrasts. We calculated crude and sex- as well as race-adjusted incidence densities for each outcome (number of events per 1000 person-years, based on the subpopulation at risk, ie, after excluding individuals with a prevalent history of each outcome of interest). We also conducted tests of trend to determine whether the incidence significantly increased or decreased by age group or duration of diabetes. In sensitivity analyses, rather than excluding patients with a prevalent history of only the outcome of interest, we repeated the above analyses in the subpopulation with no history of any of our outcomes of interest at baseline.
The mean age of the patients was 71 years, with 14.5% of the study population aged 80 years or older. For each age group, most patients had shorter duration (0-9 years) of diabetes (Table 1). There were 2.4 times more patients with shorter than longer duration of diabetes for patients aged 60 to 69 years and 1.5 times more patients with shorter duration for those 80 years or older.
There were several noteworthy baseline differences between patients with shorter vs longer duration of diabetes. Within each age category, patients with a longer duration of diabetes were more likely to have poorer glycemic control compared with those with a shorter diabetes duration. For example, 14.3% of patients aged 80 years or older with a longer duration of diabetes had a baseline HbA1c greater than 8.0% compared with only 6.4% of those in the same age group but with a shorter duration of diabetes (to convert HbA1c to a proportion of total Hb, multiply by 0.01). Compared with patients with a shorter duration of diabetes, those with a longer duration also had higher baseline prevalence rates of microvascular and cardiovascular complications and were more likely to use each class of glucose-lowering medication, particularly insulin. For example, the proportion of patients aged 80 years or older with a longer duration of the disease using insulin was 34.8% compared with 5.0% for those in the same age group but with a shorter duration of diabetes. Most patients were receiving statins and angiotensin-converting enzyme inhibitors, but the prevalence varied across age and duration subgroups (range, 56.9%-70.8% for statins; 50.6%-63.5% for angiotensin-converting enzyme inhibitors).
In general, the nonfatal complications with the highest incidence were cardiovascular (coronary artery disease, congestive heart failure, and cerebrovascular disease), followed by diabetic eye disease and acute hypoglycemic events (Table 2). Congestive heart failure and cerebrovascular disease increased most dramatically with advancing age (255% and 229% increase between ages 60-69 and ≥80, respectively); the other outcomes increased less markedly. End-stage renal disease, lower limb amputation, and acute hyperglycemic events were consistently rare complications. Quiz Ref IDAmong individual microvascular complications, diabetic eye disease had the highest incidence in all age groups. For example, in patients aged 70 to 79 years, the incidence of eye disease was 6.16 per 1000 person-years and the incidences of ESRD and amputation were 2.60 per 1000 person-years and 1.28 per 1000 person-years, respectively.
Not unexpectedly, the mortality rates increased steeply with advancing age among those with a shorter duration of diabetes (60-69, 19.61 per 1000 person-years; 70-79, 42.69 per 1000 person-years; and ≥80, 105.20 per 1000 person-years). In patients aged 80 years or older, the mortality rate was 4.3 times the rate of congestive heart failure, the most common nonfatal event.
The incidence rates in older patients with a longer duration of diabetes (≥10 years) shared some similarities with rates of those with a shorter durations, with some exceptions (Table 3). For patients with a longer duration of diabetes, hypoglycemia had incidence rates that were similar to those of coronary artery disease and cerebrovascular disease. As a result, hypoglycemia had the fourth highest incidence of all individual nonfatal complications among patients aged 60 to 69 years and the third highest incidence among patients aged 70 years or older. Rates of incident coronary artery disease, cerebrovascular disease, congestive heart failure, and hypoglycemic events increased more steeply with advancing age than did the other nonfatal outcomes. Unlike those with a shorter duration of diabetes, patients with a longer duration had modest declines in the rates of microvascular complications (ESRD, eye disease, and lower limb amputation) with advancing age (−27%, −27%, and −1% between ages 60-69 and ≥80 years, respectively). The rarest complications among all age groups were lower limb amputation and acute hyperglycemic events.
The largest difference across age groups among those with the longer duration of diabetes was in the mortality rates. The mortality rates were significantly higher among patients aged 80 years or older in comparison with those in the younger patients (60-69, 33.21 per 1000 person-years; 70-79, 65.87 per 1000 person-years; and ≥80, 132.90 per 1000 person-years). In patients aged 80 years or older, the mortality rate was almost 4.0 times the rate of congestive heart failure, the most common nonfatal event.
For a given age group, incidence rates were consistently higher for all outcomes (nonfatal complications as well as mortality) in patients with a longer compared with those with a shorter duration of diabetes. This pattern was most evident in those aged 60 to 69 years. The incidence of microvascular outcomes was much greater among patients with a longer duration of diabetes compared with those with a shorter duration. For example, ESRD, eye disease, and lower limb amputation incidences were 296%, 248%, and 290% greater, respectively, comparing longer with shorter duration among patients aged 60 to 69 years. Much smaller differences (142%, 86%, and 128%, respectively) were observed in the 80 years or older group. The other complications differed less dramatically by duration. Among patients in the 80 years or older group, the nonfatal complication with the greatest elevated risk with longer duration was hypoglycemia (215%).
For patients with no prevalent history of any complications, we found that, in general, complication rates were markedly lower compared with those in the overall cohort (Supplement [eAppendix 2 and eAppendix 3]). However, we found very similar rankings of individual complications as well as similar patterns between patients with shorter and longer durations of diabetes and across age groups.
Recent recommendations regarding diabetes care in older patients19,20 have emphasized the importance of individualization of treatment, although this remains poorly defined. Based on clinical trial results,13,21 some experts suggest the importance of considering age and duration of diabetes as criteria for risk stratification when individualizing care.13 However, the contemporary epidemiologic patterns have not been adequately documented in usual care settings.
Quiz Ref IDIn this evaluation of patients with type 2 diabetes, we found that age and duration of diabetes were independent predictors of the clinical course of the disease.Quiz Ref IDMoreover, their interaction (ie, age × duration) was significant for ESRD, eye disease, lower limb amputation, stroke, heart failure, and mortality, albeit not for hyperglycemia. The significant interactions justified evaluating effect sizes separately for each stratum. For consistency, we used the stratified approach for all outcomes.
Depending on the subgroup, cardiovascular complications and hypoglycemia traded positions (depending on age and duration of diabetes) among the most frequent nonfatal complications of diabetes, and microvascular complications and acute hyperglycemic events occurred at much lower rates. Quiz Ref IDMost notably, the risk of hypoglycemia rose markedly and independently with advancing age and duration of diabetes, and hypoglycemia was the third most frequent complication among patients aged 70 years or older with a longer duration of diabetes. Within specific age groups, microvascular complications increased most dramatically with a longer duration of diabetes. Selected groups, particularly those aged 80 years or older with a longer duration of diabetes, were at the highest risk for developing almost all individual complications and, consequently, multiple complications.
These findings should be informative for the management of diabetes in older people. Diabetes management has been classically centered on glycemic control. Moderate glycemic control (ie, HbA1c 7.0%) most clearly prevents microvascular complications and potentially reduces long-term cardiovascular complications in middle-aged patients.7,22 However, very intensive glycemic control (ie, with targets for HbA1c <6.0%) has been found to increase mortality risk in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial in older patients with long-standing type 2 diabetes.23 In addition, the collective experience from the ACCORD, Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation, and Veterans Affairs Diabetes Trials suggests that very intensive control in type 2 diabetes does not appear to provide significant benefits for cardiovascular outcomes24,25 in older patients, especially those with or at high risk for heart disease. There are also doubts regarding the effect of intensive glucose control on preventing ESRD.26 Our observations that cardiovascular complications and hypoglycemia are common provides additional support for the reorientation of care of older patients with diabetes away from intensive glycemic control as the core focus of management. The distinctive clinical course of different patient strata supports recommendations to individualize glycemic targets among older people.19
These findings are based on observations in a contemporary, ethnically diverse population of older diabetic patients with uniform access to care in a large, integrated health care delivery system. The incidence rates of acute hyperglycemic events, microvascular complications, and cardiovascular complications in our study are all dramatically lower than those reported previously. In the study by Bertoni et al6 of Medicare recipients enrolled in 1995, the incidence density of diabetic ketoacidosis was 24.3 per 1000 person-years, the incidence of ischemic heart disease was 181.5 per 1000 person-years, and the incidence of amputation was 9.8 per 1000 person-years. We observed much lower rates of those complications in the present study. For example, among the oldest patients (≥80 years) with a longer duration of diabetes, the rates for acute hyperglycemic events, coronary artery disease, and lower limb amputation were only 2.35 per 1000 person-years, 24.09 per 1000 person-years, and 3.92 per 1000 person-years, respectively. Our lower rates of cardiovascular events are consistent with a general decline in these events in the past 30 years for both the diabetic and nondiabetic populations.27 Our hypoglycemia rates varied from a low of 3.03 per 1000 person-years for patients aged 60 to 69 years with a diabetes duration of less than 10 years to 19.60 per 1000 person-years for those aged 80 years or older with 10 or more years of diabetes. Because these rates were very sensitive to age and diabetes duration, it is difficult to contrast the hypoglycemia rates we observed with those reported by Bertoni et al6 (28.3 per 1000 person-years).
Progression of diabetes complications is influenced by myriad factors, including delays in diagnosis, the natural history of type 2 diabetes, evolving diabetes care, patient self-management, and genotypic variation. For example, the incidence of hypoglycemia among the elderly may increase because of defective counterregulation and hypoglycemic unawareness that occurs with aging and longer duration of diabetes,28 or it may be the result of increased reliance on specific agents (eg, insulin, sulfonylureas) associated with a higher risk of hypoglycemia to achieve lower glycemic targets. Psychosocial factors and functional limitations are also likely to be important determinants of hypoglycemia risk.29 Hypoglycemia may also occur more frequently than traditional diabetes complications because the rates of microvascular and cardiovascular complications have declined from historic highs with more effective primary prevention efforts and better risk factor control. The use of more intensive diabetes treatments, which became more common from the 1990s to the 2000s,30 may put patients at greater risk for hypoglycemia. As we become more successful at increasing long-term survivorship with type 2 diabetes, hypoglycemia may emerge as a dominant nonfatal complication of older patients. To the extent that hypoglycemia is an adverse effect of treatment, its emergence as a dominant “complication” raises serious concerns about the acceptable limits of iatrogenesis; it has been suggested that, for some patients, preventing hypoglycemia is more important than tight glycemic control.31
The event rates of outcomes observed in this study are influenced not only by short-term clinical exposures and treatments but also by those received over a lifetime. For example, the older patients in our sample with a duration of diabetes of more than 10 years may have received their initial treatments during the early 1990s; thus, their early treatment may have been less intensive because they predated the influence of UK Prospective Diabetes Study and Diabetes Control and Complications Trial32 findings. In accordance with the concept that early control of glucose has long-lasting effects (legacy effect or metabolic memory),22,33 it is likely that the rates of complications observed in patients with a longer duration of diabetes will be quite different in the coming decade resulting from changes that have occurred in treatment patterns.
The clinical course of disease among patients in the Registry is also, in part, a product of the access and quality of diabetes care delivered within an integrated health care delivery system. In this population, the blood glucose levels were, on average, well controlled (which may have influenced hypoglycemia rates) and the use of statins and angiotensin-converting enzyme inhibitors was high. The patterns observed in this setting may differ from those in other clinical settings and populations, particularly the uninsured and underinsured. Although there are geriatric diabetic populations with limited access to quality diabetes care (eg, geographically isolated patients), our findings should be broadly applicable to many older people with diabetes because the vast majority receive Medicare-supported coverage. In addition, the event rates in this study were based on a particular coding strategy that we applied systematically to all patient subgroups (Supplement [eAppendix 1]).
This 4-year cohort study describes the clinical course of diabetes in older adults. These findings will be relevant and informative for clinicians, researchers, and policymakers. For clinicians, the study details how the expected course of diabetes may differ by age and duration of the disease. Although we have witnessed great advances in the prevention of microvascular and cardiovascular disease among patients with diabetes, our findings suggest the need for increased clinical and research focus on reducing and understanding the incidence of hypoglycemia in older adults with diabetes. For policymakers, the study provides important data that may be used for projecting health care expenditures for a large and growing segment of the Medicare population.1 More important, the data from this study may inform the design and scope of public policy interventions that meet the unique needs of elderly patients with the disease.
Accepted for Publication: July 31, 2013.
Corresponding Author: Elbert S. Huang, MD, MPH, Section of General Internal Medicine, University of Chicago, 5841 S Maryland Ave, MC 2007, Chicago, IL 60637 (firstname.lastname@example.org).
Published Online: December 9, 2013. doi:10.1001/jamainternmed.2013.12956.
Author Contributions: Dr Huang had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Huang, John, Moffet, Karter.
Acquisition of data: Liu, John, Moffet, Karter.
Analysis and interpretation of data: All authors.
Drafting of the manuscript: Huang, John.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Liu, Karter.
Obtained funding: Huang, Moffet, Karter.
Administrative, technical, and material support: John, Moffet.
Study supervision: Moffet, Karter.
Conflict of Interest Disclosures: None reported.
Funding/Support: This research was funded by the National Institute of Diabetes and Digestive and Kidney Diseases grants R01 DK081796 (Dr Huang, Ms Liu, Ms John, Mr Moffet, and Dr Karter), R01 DK65664 (Ms Liu, Mr Moffet, and Dr Karter), K23DK097283 (Dr Laiteerapong), P30 DK092949 (Drs Huang and Laiteerapong and Ms John), P30 DK092924 (Ms Liu, Mr Moffet, and Dr Karter), and a University of Chicago John A. Hartford Centers of Excellence Award (Dr Laiteerapong).
Role of the Sponsor: The National Institute of Diabetes and Digestive and Kidney Diseases had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.