Epidemiology of Diabetic Retinopathy

3.1 Introduction and Definitions

Epidemiology is the study of factors that determine the occurrence and distribution of disease in a population. In studying the epidemiology of diabetic retinopathy (DR), the literature is large and can be difficult for the clinician to mentally organize. To help in this regard, we note that reports generally display information in grids that are multidimensional. These dimensions can be categorized as follows:

Outcome variables – indices of visual loss or improvement

Prevalence and incidence rates

Risk factors for outcome variables

Types of diabetes

Populations sampled

Eras of sampling

In any single report, it is typical that only a slice of information through this multidimensional epidemiologic study space is reported. In this chapter, we will attempt to provide a broad view of the many dimensions of the topic. For a more detailed study of parts of the topic, the interested reader can pursue the cited references.

The outcome variables reported typically are presence of any DR, presence of proliferative diabetic retinopathy (PDR), presence of diabetic macular edema (DME), and measures of vision loss. There are many other outcome variables, and

A.R. Bhavsar (*)

Clinical Research, Retina Center, University of Minnesota, Posterior Segment Research, Phillips Eye Institute, Minneapolis, MN 55404, USA

e-mail: bhavs001@umn.edu

their meanings often differ slightly among studies, but the most common ones found in the literature are defined as follows:

Presence of Diabetic Retinopathy – the presence of a threshold level of fundus change indicating that the eye has diabetic retinopathy (DR). This threshold level varies among studies. Examples include the presence of a certain number of microaneurysms detected in a specified region of the fundus or the presence of any characteristic lesion of diabetic retinopathy.1 The Eye Diseases Prevalence Research Group classified presence of DR as mild, moderate, or severe DR, diabetic macular edema, or any combination of these.2

Progression of Diabetic Retinopathy – there are a number of diabetic retinopathy severity scales, the best known being the Early Treatment Diabetic Retinopathy Study (ETDRS) classification system and its modifications. Progression means that the level of retinopathy severity has increased on such a scale by a threshold amount, often two steps (e.g., the Wisconsin Epidemiologic Study of Diabetic Retinopathy [WESDR]) or three steps (e.g.,

the Diabetes Control and Complications Trial [DCCT]).3,4

Progression to Proliferative Diabetic Retinopathy

– this variable applies to an incidence study and refers to those participants without PDR at the baseline examination who had PDR at the follow-up examination.5

Presence of Diabetic Macular Edema or Clinically Significant Macular Edema – DME is the presence of macular thickening within a certain distance from the center of the macula, often

two disk diameters, as determined by stereoscopic fundus photographs grading. Clinically significant macular edema (CSME) is thickening at or within 500 mm of the center of the macula, presence of lipid with adjacent thickening within 500 mm of the center of the macula, or presence of thickening of at least one disk area in extent, any part of which comes to within one disk diameter of the center of the macula.6

Requirement for Laser Photocoagulation – sightthreatening DR due to DME and PDR is treated with laser photocoagulation. This outcome variable therefore is a practical index of progression to a clinically important level of retinopathy. In the United Kingdom Prospective Diabetic Retinopathy Study (UKPDS), 78% of laser treatment was for DME and 22% was for PDR.7

Moderate Visual Loss – on the ETDRS visual acuity chart, a loss of visual acuity of three lines or 15 letters; a synonymous term is doubling of the visual angle.

Visual Impairment – best corrected visual acuity in the better seeing eye worse than 20/40.8

Moderate Visual Impairment – best corrected visual acuity in the better seeing eye worse than 20/40 but better than 20/2009; some studies use different cutpoints for the definition.5

Severe Visual Impairment – best corrected visual acuity in the better seeing eye of 20/200 or worse.9

Severe Visual Loss – on the ETDRS visual acuity chart, severe visual loss refers to visual acuity of 5/200 on two consecutive visits separated by 4 months or more.

Blindness – legal blindness, synonymously termed blindness, means that the visual acuity has dropped to the level of 20/200 or worse.10 Some studies use a 20/400 threshold for the definition.11

Vision Threatening Retinopathy – presence of severe nonproliferative diabetic retinopathy

(NPDR), PDR, or clinically significant macular edema (CSME).1,2

Epidemiologic studies report prevalence rates, incidence rates, or both. These are measures of frequency of disease at a single time or over

time. The definitions of these terms are often confused.

Prevalence – the number of cases of the disease divided by the population at a specified time, expressed as a percentage.12 A prevalence study is sometimes called a cross-sectional study.13

Incidence – the number of new cases that arise during a time divided by the population at risk but disease-free at the beginning of that time.12

Prevalence studies are easier to perform than incidence studies because data are collected once rather than twice. Although simpler to perform, they have the drawback that the temporal sequence of the associations cannot be determined.14 Incidence studies can determine temporal relationships, but have the problem of nonparticipation in the follow-up examination which can potentially bias the data.5 If the incidence study collects data at widely separated times, the problem of spontaneous onset and resolution of the condition can arise, leading to underestimation of annual incidence. For example, WESDR collected data at 4- and 10-year intervals and arrived at annual incidences of DME smaller than studies with annual data collection.13

Epidemiologic results for DR are often presented with reference to the type of diabetes mellitus. There are always two and sometimes three types. All studies distinguish types 1 and 2 diabetes mellitus. In WESDR, the type 1 group is called younger onset diabetes, referring to patients with disease onset at less than 30 years of age.5 WESDR uses the term older onset, not taking insulin to refer to some of the patients with type 2 diabetes mellitus.5 In the WESDR, a third group is defined as older onset, taking insulin; this is a mixed group containing patients with types 1 and 2 diabetes mellitus.5 The unique WESDR nomenclature is important to keep in mind because of the large number of published WESDR reports.

Many variables have been explored in the search for risk and protective associations with aspects of diabetic retinopathy. The risk and protective factors may be different for prevalence rates and incidence rates.13 For example, male sex was a risk factor for

prevalence of CSME but not for incidence of CSME in WESDR.15,16 The following demographic variables have been found to be informative – type of diabetes, requirement for insulin, duration of diabetes, age at baseline examination, ethnicity, and inconsistently gender. Systemic variables found to be important include the quality of glycemic control, quality of control of any concomitant arterial hypertension and dyslipidemia, effects of certain types of medications, certain genotypes, and era of therapy. There are a few ocular factors of importance, such as level of retinopathy at baseline and presence of modifying concomitant ocular conditions such as glaucoma or optic atrophy. In the present chapter, we will review the effects of demographic factors on the clinically important indices of diabetic retinopathy. The effect of systemic and

ocular variables will be discussed in Chapter 4. Genetic variables important for risk and protection are reviewed in Chapter 2.

The last two dimensions considered in many epidemiologic studies characterize the different populations studied and the different eras for which studies have been done. These are important because populations across the world differ and management of diabetes and diabetic retinopathy changes over time. As an example of the former, an estimated 50% of Pima Indians have diabetes, compared to 6.2% of adult Chinese in Taiwan. As an example of the latter, the prevalence rates for diabetic retinopathy among US Whites in 1980–1982 reported in WESDR are higher than in more recent reports, probably reflecting tighter glycemic control over the past 25 years.

Comparison of epidemiologic studies is difficult. Methodologies differ among studies, such as the criteria for diagnosing diabetes, the tests for ascertaining diabetic retinopathy, and the formats for reporting data.8 Populations may differ in levels of interacting variables, such as blood pressure, glycosylated hemoglobin, and level of serum lipids. In general terms, many epidemiologic studies must be

reviewed to look for a consistent pattern of concordance or discordance of risk and protective factors.1 Moreover, epidemiologic studies are unable to determine causation. Rather, they reveal associations that may be explored in studies aimed at defining causation and effective treatment. Many associations in epidemiologic studies are univariate and vanish when multivariate analyses are carried out.14

That is, the predictive power of a variable in a univariate analysis examining prevalence of DR may be accounted for by some other variable. An example of this is the reported higher prevalence of diabetic retinopathy in blacks than in whites, a difference that in some studies vanishes when the effects of glycemic control and blood pressure between these ethnic groups are taken into account.14

3.2 Epidemiology of Diabetes Mellitus

Diabetes mellitus is classified into two types – 1 and 2. In type 1, pancreatic beta cells are destroyed through an autoimmune process leading to an absolute insulin deficiency. In type 2, the peripheral tissues exhibit insulin resistance, and a relative rather than absolute secretory defect of the beta cell is present.19 Types 1 and 2 diabetes comprise variable fractions of diabetes mellitus depending on the country and region, perhaps reflecting a survivorship effect. In population-based studies in rural

China and Barbados, type 1 diabetes comprised <1%.20,21 In contrast, in the United States, type 1

diabetes comprised 11.9% of the participants in WESDR in 1980–1982.22 Estimates for the

prevalence of types 1 and 2 diabetes in persons over the age of 18 in the United States for 2004– 2007 are 0.4 and 7.6% of the population, respectively.23,24 Thus, currently in the United States, approximately 5 and 95% of patients have types 1 and 2 disease, respectively.

In the United States in 2008, it was estimated that 8% of the population, or 24 million persons, had diabetes mellitus. From 1980 to 2006 the number of persons with diabetes in the United States tripled.24

Similar statistics apply worldwide.25 The prevalence of diabetes increases with age (Fig. 3.1).25,26 Con-

tinued growth in prevalence of diabetes mellitus in both developed and less developed countries is projected (Fig. 3.2 and Table 3.1).27 In the United States, it is estimated that 30% of the cases of diabetes mellitus are undiagnosed, and analogous

estimates as high as 76% are reported in other countries.9,19,20,25

An increasing prevalence of diabetes has been documented widely (example for India in Fig. 3.3) and is projected for all countries, but the increase is expected to be higher in developing countries.19 The increase in prevalence is associated with changes in diet, exercise, lifestyle, and increased life expectancy.19,30 Incidence data on diabetes mellitus worldwide are scarce relative to prevalence data.