Diabetes in Old Age - part 1 pps

This is particularly relevant to the prevalence of diabetes in older people who are more likely to be identi®ed as having diabetes by the 2-hour post-load glucose level than by the fasti

Section I

Epidemiology, Pathophysiology and Diagnosis

ISBNs: 0-471-49010-5 (Hardback); 0-470-84232-6 (Electronic)Copyright # 200I John Wiley & Sons Ltd

Diabetes Mellitus and Impaired Glucose Regulation in Old Age: The Scale of the

Problem Paul Finucane, Phil Popplewell

Flinders University and Flinders Medical Centre, Adelaide

INTRODUCTIONDiabetes mellitus is an important condition because it

is common in developed countries, is becoming

com-mon in developing countries, and places a very great

burden on individuals, healthcare systems and

socie-ties in all countries In 1997 it was estimated that 124

million (2.1%) of the world's 5.8 billion total

popula-tion had diabetes mellitus and it is projected that by

2010 this number will almost double to 221 million

(Amos, McCarthy and Zimmet 1997) Of the 124

million with diabetes in 1997, 120 million (97%) had

Type 2 diabetes

In this chapter, our main purpose is to describe the

incidence (i.e the number of new cases occurring

within a population over a speci®ed period of time) and

prevalence (i.e the proportion of people in a

popula-tion with that condipopula-tion at a given time) rates for

dia-betes We examine trends in incidence and prevalence

rates for diabetes over time and in different

popula-tions We also examine the epidemiology of impaired

glucose regulation in people without overt diabetes

DEFINITION AND CLASSIFICATION OF

DIABETES AND IMPAIRED GLUCOSE

REGULATIONThough de®nitions and classi®cation are dealt with in

detail in Chapter 3, it is necessary at this stage to

ex-plain brie¯y the terms used here This is particularly

important in view of some recent changes in diagnostic

criteria, which impact on the interpretation of

epide-miological studies Such changes are perhaps bestunderstood from a historical perspective

Up until the late 1970s, epidemiological research indiabetes was bedevilled by a lack of standardization inde®nitions, in classi®cation and in research method-ologies At that time, standardized diagnostic criteriawere proposed (National Diabetes Data Group, 1979),were essentially adopted by the World Health Orga-nization (WHO 1980) and were subsequently modi®ed

by that body (WHO 1985) All of these criteria placed

a degree of reliance on both the fasting glucose and hour post-load glucose levels to differentiate betweenthree groups of people: those with normal glucosetolerance, those with impaired glucose tolerance (IGT)and those with diabetes

2-In 1997, a modi®cation to the 1985 WHO criteriawas proposed (American Diabetes Association 1997)with the intention of moving away from reliance on the2-hour post-load glucose level and instead basing de-

®nitions and diagnosis on fasting blood glucose levelsalone The ADA also suggested a lowering of thethreshold for the diagnosis of diabetes from a fastinglevel of 7.8 mM(140 mg=dL) to 7.0 mM(126 mg=dL).The term `impaired fasting glucose' (IFG) was coined

to describe people whose fasting blood glucose levelswere above the reference range for normality but belowthat required for a diagnosis of diabetes (6.1±6.9 mM).The adoption of these criteria would have led to the2-hour post-load glucose level and to the term IGTboth becoming largely obsolete

For reasons that will be explored in Chapter 3, somelimitations to the ADA criteria soon became apparent.When the WHO further revised its classi®cations and

Diabetes in Old Age Second Edition Edited by A J Sinclair and P Finucane # 2001 John Wiley & Sons Ltd.

ISBNs: 0-471-49010-5 (Hardback); 0-470-84232-6 (Electronic)Copyright # 200I John Wiley & Sons Ltd

diagnostic criteria in 1999, it adopted the ADA

pro-posals in relation to the fasting blood glucose but

re-tained a de®nition of diabetes based on the 2-hour

post-load glucose level (WHO 1999) Thus the 2-hour

post-load glucose level and the term IGT still have

relevance While there is signi®cant overlap between

IFG and IGT, the terms are neither synonymous nor

mutually exclusive In accordance with the new WHO

guidelines, people with IFG and=or IGT can

collec-tively be considered to have impaired glucose

regula-tion

EPIDEMIOLOGYEffect of Altered De®nitions and

Classi®cation of Epidemiological Data

The effect of the above changes on the interpretation of

existing epidemiological studies of diabetes and

im-paired glucose regulation is somewhat confusing

Where data have been re-analysed using only fasting

blood glucose levels and disregarding 2-hour post-load

glucose levels, most studies show a fall in prevalence

rates for diabetes (Davies 1999) When both the fasting

blood glucose and the 2-hour post-load glucose levels

are considered, prevalence rates can only increase The

impact of the new diagnostic criteria on future

epide-miological studies will depend on the methodology

used This is particularly relevant to the prevalence of

diabetes in older people who are more likely to be

identi®ed as having diabetes by the 2-hour post-load

glucose level than by the fasting glucose level (Wahl et

al 1998)

In this chapter, much of the cited epidemiological

work predated the new diagnostic criteria for diabetes

and this should be kept in mind when absolute

in-cidence and prevalence rates are discussed For

ex-ample, most of the research on the epidemiology of

impaired glucose regulation deals with IGT rather than

with IFG However, such `older' studies are still

rel-evant, particularly in their ability to examine trends in

the epidemiology of diabetes and impaired glucose

regulation over time They also provide meaningful

comparisons of people from different age, gender,

ethnic, socioeconomic and other groups

Limitations of Epidemiological Research

Much of the epidemiological research is of little direct

relevance to elderly populations Even studies of Type

2 diabetes have tended to focus on relatively youngadult populations and some have excluded older peo-ple However, such studies are still of value as theyoung adult population with diabetes of today willform a major part of tomorrow's elderly populationwith diabetes Of the relatively few epidemiologicalstudies of diabetes in elderly people, most are cross-sectional prevalence studies Studies of disease in-cidence which require long-term follow-up of cohorts

of patients or repeated cross-sectional analyses arerelatively time-consuming and expensive and aretherefore less common

The Importance of Understanding the Scale

of the ProblemClinicians, educators, researchers and health plannersalike need to appreciate the current status and futuretrends in the epidemiology of diabetes This willpromote:

1 Rational health planning The magnitude of theclinical workload relevant to diabetes can be deter-mined together with the resources required to meetit

2 Placement of the disease in a proper perspective.Its importance relative to other disorders can bedetermined and this in turn can facilitate theequitable allocation of resources

3 The identi®cation of individuals, groups orcommunities who are at high risk for the develop-ment of diabetes This offers possibilities forresearch into the aetiology of the disease, and forhealth promotion and disease prevention programs

4 Awareness of any change in the nature of diabetesover time Furthermore, it will facilitate the evalua-tion of intervention programs

The epidemiology of IFG and IGT also needs to beconsidered as people with either of these disorders are

at high risk of developing diabetes in the future Forexample, a study involving Pima Indians from Arizona

in the US followed up those with IGT for a median of3.3 years and found that 31% developed overt diabetes,26% continued to have IGT, while 43% reverted

to normal glucose tolerance (Saad et al 1988) Thecumulative incidence of overt diabetes was 25% at

5 years and 61% at 10 years

PREVALENCE TRENDS OVER TIME

In developed and developing countries alike,

pre-valence rates for diabetes in the general population

have been on the increase since the early 1900s For

example, Harris (1982) had drawn attention to an

upward trend in US prevalence rates for diagnosed

diabetes between the 1930s and 1980 (Figure 1.1)

Prevalence rates rose in all age groups and in both

sexes, with the number of known people with diabetes

doubling between 1960 and 1980 (Bennett 1984)

In one Australian community (Glatthaar et al 1985),

the prevalence rate for diabetes increased by 50% over

the 15-year period 1966±1981, and in a UK population

prevalence rates rose by 60% between 1983 and 1996

(Gatling et al 1998)

However, in the US at least, prevalence rates for

diabetes now seem to be reaching a plateau The

an-nual review of 40 000 households comprising 120 000

US residents conducted by the National Health

Inter-view Survey (NHIS) indicates that the prevalence of

diabetes rose by 67% from 1959 to 1966; 41% from

1966 to 1973; 21% from 1973 to 1980; and 4% from

1980 to 1989 (Centers for Disease Control 1990a)

Incidence rates also increased in the US until the early

1980s and, at least in some populations, continue to

rise (Burke et al 1999)

Changes to the diagnostic criteria for diabetes and

IFG=IGT hamper direct comparisons of prevalence

rates between recent and earlier epidemiological dies However, using currently accepted criteria, theprevalence rate for diabetes in older adults (i.e aged40±74 years) rose from 8.9% during the years 1976±

stu-1980 to 12.3% during the years 1988±1994 (Harris et

al 1998)

The prevalence of any condition depends on both itsincidence and its duration At different times, thesefactors have made variable contributions to the in-creasing prevalence of diabetes, at least in Westerncountries For example, in the US between 1960 and

1970, incidence rates for diabetes increased due to agreater awareness of the condition, greater surveillanceand better diagnostic methods which all contributed tothe earlier diagnosis of milder cases (Harris 1982).Incidence rates for Type 2 diabetes continue to rise insome populations but not in others For example, theSan Antonio Heart Study found a three-fold increase inincidence between 1987 and 1996 (Burke et al 1999)while data from the Swedish Skaraborg DiabetesRegistry indicate no increase in the incidence of dia-betes between 1991 and 1995 (Berger, Stenstrom andSundkvist 1999)

Since the early 1970s, however, it is clear that much

of the rise in prevalence rates for diabetes is table to enhanced survival in those affected Whilesome two-thirds of people with diabetes die fromcardiovascular disease, mortality rates in diabetes arefalling, though not as rapidly as mortality rates from

attribu-Figure 1.1 Trends in the prevalence an rate of diagnosed diabetes in the USA Open and closed circles represent data from National Interview Surveys of the US Public Health Service Rates from seven community-based surveys are included for comparison M, Maryland (State); O, Oxford, Massachusetts; H, Hargerstown, Maryland; K, Kansas City, Kansas; T, Tecumseh, Michigan; S, Sudbury, Massachusetts;

R, Rochester, Minnesota

cardiovascular disease in non-diabetics (Gu, Cowie

and Harris 1999) Skaraborg Diabetes Registry data

indicate that between 1991 and 1995, the median age

of death for people with diabetes increased from 77.2

to 80.2 years (Berger et al 1999)

In developed countries in particular, the increased

prevalence of diabetes and IFG=IGT over time can be

further attributed to:

 aging of the population

 greater number of people from ethnic minority

backgrounds who are adopting a `transitional'

life-style

 greater levels of overweight and obesity

 more sedentary lifestyles

The importance of these risk factors in the

pathophy-siology of diabetes is discussed in more detail in

Chapter 2 However, from an epidemiological

per-spective, aging is a crucially important factor and

nowadays, individuals have a longer life span during

which to develop diabetes, to live with the condition

and to develop its complications (Wilson, Anderson

and Kannel 1986) The population with diabetes is

increasingly elderly Even twenty-®ve years ago, 40%

of newly diagnosed people with diabetes in a US

sample were aged over 65 years (Palumbo et al 1976)

The NHIS survey already cited documented a greater

than 100% increase in the number of people with

diabetes aged over 75 years in the US between 1980

and 1987 (Centers for Disease Control 1990a)

Diabetes is fast becoming a signi®cant problem in

many developing countries where previously it was

little recognized This can be explained in part by all of

the factors mentioned above: increased detection rates,

improved survival rates and an aging society More

importantly, however, and as will be discussed in

Chapter 2, people in many developing countries are

switching from a traditional to a Western lifestyle and

in the process adopting diets and exercise patterns that

lead to the development of diabetes

FACTORS INFLUENCING THE

PREVALENCE OF DIABETES AND

IMPAIRED GLUCOSE REGULATION

The prevalence of diabetes and IFG=IGT varies

con-siderably in different populations and in different

sub-groups within populations The most important

vari-ables (see Table 1.1) are now discussed more fully

Some factors, for example lifestyle and obesity, are

both closely interrelated and dif®cult to measure cisely This makes it dif®cult to disentangle one fromthe other when analysing their relative contributions tothe development of diabetes (King and Zimmet 1988)

pre-AgeAge is the single most important variable in¯uencingthe prevalence of diabetes and IFG=IGT Almost everyepidemiological study, whether cross-sectional orlongitudinal, shows that the prevalence of both dia-betes and IFG=IGT initially increases with advancingaging, reaches a plateau and subsequently declines.However, the time of onset of the increase, the rate ofincrease, the time of peak prevalence and rate of sub-sequent decline differ in the various groups studied.There is general agreement that the rise in pre-valence begins in early adulthood For example, PimaIndians aged 25±34 years are 10 times more likely tohave diabetes than those aged 15±24 years (Knowler et

al 1978) In Americans aged 45±55 years, diabetes isover four times more common than in those aged 20±

44 years (Harris et al 1987) The subsequent rate ofincrease with aging is variable, being greatest in so-cieties with the highest prevalence of glucose intoler-ance (King and Rewers 1993)

In Pima Indians, the prevalence of abnormal glucosetolerance peaks at 40 years for men and 50 years forwomen and declines in men after the age of 65 yearsand in women after the age of 55 years (Knowler et al1978) In other populations, prevalence rates peak inthe sixth decade and subsequently decline (King andRewers 1993) However, in a study of elderly Finnishmen, prevalence peaked in those aged 75±79, fallingoff in 80±84 years olds (Tuomilehto et al 1986) Insome populations however, the highest prevalencerates are found in the oldest age groups (Glatthaar et al1985; King and Rewers 1993)

Table 1.1 Factors in¯uencing the prevalence of diabetes and IFG=IGT Age

Sex Country of residence Place of residence Race and ethnicity Socioeconomic status and lifestyle Obesity

Figure 1.2 shows the prevalence rates of diabetes in

males and females and in different age groups, taken

from the third National Health and Nutrition

Ex-amination Survey (NHANES III) carried out in the US

(Harris et al 1998) This survey is the most extensive

and up-to-date currently available, being conducted

from 1988 to 1994 and involving some 19 000 adult

Americans (i.e aged over 20 years) Prevalence rates

rise with advancing age until a plateau is reached at age

60±74 years

GenderThere is evidence to suggest that diabetes was once

more common in females than males In recent years

however, a disproportionate increase in the number of

males known to have diabetes has resulted in equal

prevalence rates being found in some societies while

males predominate in others Possible explanations for

this change include a disproportionate increase in the

incidence of diabetes in males, increased detection in

males and reduced mortality in diabetic males

Between 1980 and 1987, NHIS data showed a 33%

increase in the prevalence of self-reported diabetes

among white males but no increase among white

fe-males (Centers for Disease Control, 1990) Althoughthere was a 16% increase among black males and a24% increase among black females, this had a smallerimpact as non-whites constitute only 15% of the USpopulation When interpreting this data, one shouldremember the limitations of self-reporting which theNHIS used as a measure of the prevalence of diabetes.Recent studies involving predominantly non-elderlypeople have found the prevalence of Type 2 diabetes inmales to exceed that in females in Australia (Glatthaar

et al 1985; Welborn et al 1989) and Finland lehto et al 1991) However, similar prevalence rateshave been reported from New Zealand (Scragg et al1991) and Japan (Sekikawa et al 1993) The NHANESIII survey from the US already cited (Harris et al 1998)found no signi®cant overall difference in the pre-valence of Type 2 diabetes between the sexes (Figure1.2) However, in that survey, prevalence rates wereslightly higher in males than in females in both the 60±

(Tuomi-74 years sub-group (20.2% vs 17.8%) and in the 75years and over sub-group (21.1% vs 17.5%) This re-presented a change from previous surveys in whichelderly females had predominated The prevalence ofIFG was also greater in males than in females in the60±74 years sub-group (16.2% vs 12.3%) and in the 75years and over sub-group (17.9% vs 11.9%)

Figure 1.2 Prevalence of diabetes in men and women in the U.S population age 20 years, based on NHANES III Diabetes includes previously diagnosed and undiagnosed diabetics de®ned by fasting plasma glucose 126 mg=dl Age-std, age-standardized Reproduced by permission from Harris et al (1998)

A review of the prevalence of diabetes from 75

communities in 32 countries, found the sex ratio for

diabetes to vary widely (King and Rewers, 1993)

Some studies found an excess of males while females

predominated in others A regional trend was apparent,

whereby in Africa=Asia and the Americas there was a

trend to male excess, whereas in the Paci®c regions

females predominated IGT was generally found to be

more common in women

The few studies that have focussed on prevalence

rates in elderly populations have either not reported a

sex difference, or found either a male (Lintott et al1992) or female (Mykkanen et al 1990) excess

Country of residenceKing and Rewers (1993) have collated data on theprevalence of abnormal glucose tolerance in over

150 000 people from 75 communities in 32 countries(Figure 1.3) As diabetes is an age-related disorder, itsprevalence in individual countries varied according to

Figure 1.3 Prevalence (%) of abnormal glucose tolerance (diabetes and impaired glucose tolerance) in selected populations in the age range

of 30±64 years, age standardized to the world population of Segi, sexes combined *, Upper income; #, middle income; d, low income; j diabetes mellitus; u impaired glucose tolerance Reproduced by permission from King and Rewers (1993)

the age structure of that society Thus developed

countries with a large elderly population have high

prevalence rates; conversely, low rates are found in

developing countries with few elderly people

Age-standardized rather than true prevalence rates are

therefore used to allow valid comparisons between

countries As King & Rewers used a truncated age

range of 30±64 years, their ®ndings cannot be

auto-matically extrapolated to elderly populations

Diabetes was found to be absent or rare (less than

3% of people affected) in some traditional

commu-nities in developing countries Prevalence rates in

Europe were 3±10%, while some Arab, Asian Indian,

Chinese and Hispanic American populations had rates

of 14±20% The highest rates were found in natives of

the South Paci®c island of Nauru and in Pima=Papago

Indians in the USA who had prevalence rates as high

as 50%

Migrant populations are at particular risk of

devel-oping diabetes A study of Japanese-American men

who had retained their racial and cultural identity,

found that 56% had abnormal glucose tolerance and

that a third had diabetes (Fujimoto et al 1987) This

rate is far higher than among white Americans with a

similar socioeconomic pro®le in terms of education,

occupation and income It is also higher than the rate

among the native population of Japan Chinese and

Indian migrants have a particularly high prevalence of

abnormal glucose tolerance when compared with

in-digenous communities (King and Rewers 1993) These

studies emphasize the importance of environmental

factors, largely absent in the indigenous population but

acquired in the migrant setting, in the development of

diabetes

Place of ResidenceThe prevalence of diabetes differs between regions in

the same country In the US, for example, there is more

self-reported diabetes in Hawaii and in states east of

the Mississippi river (Centers for Disease Control

1990b) Even when differences in age, sex and

racial=ethnic differences between states were taken

into account, a greater than three-fold difference

existed between the state with the highest rate and that

with the lowest

A study of people aged 18±50 years and living in

nine towns in England and Wales, chosen to represent

different latitude and socioeconomic status, found a

greater than two-fold difference in the numbers

re-ceiving hospital treatment for newly-diagnosed Type 2diabetes (Barker, Gardner and Power 1982) Type 2diabetes was found more frequently in towns with thepoorest socioeconomic environment, irrespective oflatitude Caution must be exercised when using such

`surrogate' markers of prevalence and incidence, asillustrated by a Finnish study which found that theprevalence of known diabetes in a cohort of elderlymen was 11% in the east of the country and 5% in thewest When a glucose challenge and then currentWHO criteria were used to measure the true prevalencerate, it was identical at 24% in both regions (Tuomi-lehto et al 1986) Regional differences are not alwaysfound For example, a study of over 6000 Tanzanianmen showed that prevalence rates for diabetes, whichwere generally low, were similar in six villages despitehaving geographical, socioeconomic and dietary dif-ferences (McLarty et al 1989)

Diabetes is considered to be a disease of nization and urbanization (Welborn 1994) and severalstudies have found signi®cantly higher prevalencerates in urban than in rural environments (King andRewers 1993) Comparisons of migrant populationsliving in rural and urban settings in the same countryalso consistently show an excess of diabetes and IGT

moder-in urban migrants

Finally, it should be remembered that particularsub-groups of the population, such as those living ininstitutional care, will have a particularly highprevalence of diabetes (Grobin 1970) This is notsurprising, given the advanced ages of such people andthe fact that diabetic complications place them in need

of residential care

Race and EthnicityStudies from multicultural societies provide compel-ling evidence that racial background impacts greatly

on the incidence and prevalence of diabetes andIFG=IGT Here again, the NHANES III study from the

US provides the best American epidemiological data.Three racial groups are identi®ed in NHANES III:non-Hispanic whites, non-Hispanic blacks and Mex-ican-Americans Compared with non-Hispanic whites,non-Hispanic blacks had a 1.6 times and Mexican-Americans a 1.9 times higher prevalence of diabetes(Harris 1998) In absolute terms, age- and sex-stan-dardized prevalence rates for diabetes and IFG com-bined were 14.1% in adult non-Hispanic whites, 18.8%

in non-Hispanic blacks and 22.7% in

Mexican-Americans

Racial differences in prevalence rates of diabetes are

also apparent in older people NHANEs III reported

that for people aged 60±74 years, prevalence rates for

diabetes were 17.3% in non-Hispanic whites, 28.6%

in non-Hispanic blacks and 29.3% in

Mexican-Americans For those aged over 75 years,

corre-sponding ®gures were 17.5%, 22.4% and 29.7%

Diabetes was found to be over twice as common in

Aboriginal Australians than in non-Aboriginals living

in the same community (Guest et al 1992) In the same

study, both groups had similar prevalence rates for

IGT In a large multiracial New Zealand workforce the

relative risk of having diabetes was 4±6 times greater

in Maori, Paci®c Islanders and Asians, than in people

of European backgrounds (Scragg et al 1991) This

increased risk remained signi®cant after controlling for

age, income and body mass index

Attention has already been drawn to the high

pre-valence of abnormal glucose tolerance among the

Pima Indians of Arizona in the US (Knowler 1978)

For indigenous North Americans, susceptibility to

Type 2 diabetes is related to the degree of racial

ad-mixing; thus Americans of mixed ethnicity have rates

of diabetes intermediate between those of full native

Americans and of Caucasians (Gardner 1984)

In a survey of the Southall district of London, which

has a large Asian population, the overall age-adjusted

prevalence of self-reported diabetes was almost four

times higher in Asians than in Europeans (Mather and

Keen 1985) It is also of interest that the excess

pre-valence of diabetes among Asians was greatest in the

older age groups However, this survey also relied on

self-reporting to measure the prevalence of diabetes

Another UK study also showed that diabetes was four

times higher in Asian men than white men and twice as

high in Asian women as white women (Simmons,

Williams and Powell 1989)

Socioeconomic Status and Lifestyle

It is dif®cult to disentangle the effect of socioeconomic

status and lifestyle on the prevalence of diabetes from

confounding factors such as country of residence,

place of residence and racial origin The evidence,

suggests however, that these are independent risk

factors Certain ethnic groups are particularly

suscep-tible to developing abnormal glucose tolerance when

they forsake a traditional for an urbanzized lifestyle

(Dowse et al 1990) This has been documented in

North American Indians, Mexican-Americans, tralian Aborigines, Micronesian and Polynesian Paci®cIslanders and Asian Indians For example, urbandwellers on the Paci®c island of Kiribati have rates ofType 2 diabetes three times greater than those living

Aus-in a rural settAus-ing; Aus-in the over 65 population, there is

a four-fold urban-rural difference (King et al 1984).Large variations in the prevalence of diabetes in dif-ferent Australian Aboriginal communities have beenreported (Guest and O'Dea 1992) with urbanizedAboriginals having the highest rates (Cameron, Mof®tand Williams 1986) Migration is a potent stimulus tolifestyle change; the higher prevalence of diabetes inmigrant communities when compared with those leftbehind has been explained by socioeconomic ad-vantage which migration tends to confer (King andZimmet 1988)

Socioeconomic deprivation, which is associatedwith poor diet and other adverse lifestyle factors is alsolinked to high rates of diabetes In the US, the 1973National Household Interview Survey documented aninverse relationship between income and the pre-valence of known diabetes (US Dept of Health, Edu-cation and Welfare 1978) A study of nine towns inEngland and Wales, chosen to represent differentlatitude and socioeconomic status, found that thedetection rate for newly diagnosed Type 2 diabetes wasgreatest in towns with a `poor' socioeconomic pro®leand least in towns with `good' pro®les (Barker et al1982) In a survey of a large multiracial New Zealandworkforce, the relative risk for glucose intolerance wasinversely related to income but not to other markers ofsocioeconomic status (Scragg et al 1991)

A study of over 1100 Hindu Indians living in es-Salaam, Tanzania, looked at the prevalence of dia-betes and IGT in seven sub-communities of differentcaste The age- and sex-adjusted prevalence of diabetesdiffered more than ®ve-fold (Ramaiya et al 1991).Similar differences were noted in the prevalence ofIGT These sub-communities differed in socio-economic characteristics and lifestyle and may alsohave differed genetically and in their diet Studies such

Dar-as this highlight the danger of regarding people from asingle geographical area or with similar racial origins

as homogenous and of grouping them under a singlelabel (e.g `Asians')

The effect of physical exercise on the pathogenesis

of diabetes is discussed in Chapter 2; there is demiological evidence that exercise in¯uences pre-valence rates For example, migrant Indians in Fiji whowere physically active had half the risk of diabetes than

epi-those who were inactive (Taylor et al 1984) Physical

activity was also implicated as an environmental risk

factor for diabetes mellitus in a multi-racial

commu-nity in Mauritius (Dowse et al 1990)

ObesityThis section outlines the epidemiological evidence for

obesity as a risk factor for diabetes and IFG=IGT; its

importance in the pathogenesis of Type 2 diabetes is

discussed in more detail in Chapter 2 There is clear

evidence that obesity is an independent risk factor for

diabetes In the NHANES II study cited earlier, obesity

doubled the probability of having diabetes and was

also an independent risk factor for IGT (Harris et al

1987) The Framingham study has had broadly similar

®ndings, with people overweight by >40% having

twice the prevalence of diabetes than others (Wilson et

al 1986) A study of 1300 Finns aged 65±74 years

found an association between diabetes and obesity and

particularly between diabetes and central obesity

(Mykkanen et al 1990) Central obesity, recognized by

a high waist=hip girth ratio, correlates with

intra-ab-dominal visceral fat mass The importance of central

obesity in the pathogenesis of diabetes is explained in

Chapter 2 A study of elderly Hong Kong Chinese also

found diabetes to be more common in overweight and

obese subjects (Woo et al 1987)

In other studies, the association between obesity and

diabetes has been less impressive Among elderly New

Zealanders, a positive association was found in newly

diagnosed people with diabetes but not in those with

known diabetes (Lintott et al 1992) Racial factors may

play a part, though the evidence is somewhat

confus-ing For example, in a survey of a large multiracial

New Zealand workforce, the increased prevalence of

glucose intolerance in Maori and Paci®c Islanders over

people of European origin was partly attributable to

obesity (Scragg et al 1991) Obesity has been

im-plicated in the high prevalence of diabetes in Pima

Indians; furthermore, in those with IGT, obesity

pre-dicted subsequent development of diabetes, though it

was not an independent risk factor (Saad et al 1988)

On the other hand, a study of over 6000 young

Tan-zanians found only a modest increase in the prevalence

of diabetes with increasing body mass (McLarty et al

1989) Furthermore, obesity was not prevalent among

elderly Finnish men, many of whom had diabetes

(Tuomilehto et al 1986) In this study, the BMI

de-creased with age in those with diabetes, IGT and mal glucose tolerance alike

nor-PREVALENCE OF ABNORMAL GLUCOSETOLERANCE IN DIFFERENT COUNTRIESFrom all that has been stated above, it follows thatprevalence rates for diabetes and IFG=IGT are speci®c

to the population from which the study sample isdrawn and cannot easily be extrapolated to other po-pulations However, it is still possible to pro®le acommunity in which the prevalence of diabetes islikely to particularly high It will have both a largeelderly and migrant population and be located in anurban setting in a `developed' country A high per-centage of people will be at either extreme of the so-cioeconomic scale, many will have sedentary lifestylesand will be overweight or obese In communities thatlack these characteristics, the prevalence of glucoseintolerance will be relatively low

Amos et al (1997) estimated that in 1997 diabetesaffected:

 66 million people in Asia

 22 million people in Europe

 13 million people in North America

 13 million people in Latin America

 8 million people in Africa

 1 million people in Oceania

Future increases in the prevalence of diabetes are likely

to affect Asia and Africa more than other regions By

2010, its prevalence in these areas will become two tothree times more common than in 1997, at which timemore than 60% of all people with diabetes will live inAsia (Amos et al 1997)

The US and CanadaThe key epidemiological studies of diabetes andIFG=IGT in the US in the past 25 years have been theperiodic National Health and Nutrition ExaminationSurveys conducted by the National Center for HealthStatistics of the Centers for Disease Control and Pre-vention The second national survey (NHANES II)covered the period 1976±1980 (Harris et al 1987) andthe third (NHANES III) covered the period 1988±1994(Harris et al 1998) These data have been supple-mented by the Hispanic Health and Nutrition Ex-amination Survey (Hispanic HANES) which surveyed

a representative sample of Mexican-Americans in the

south-western US during 1982±1984 (Flegal et al

1991) All three studies used the recommended

diag-nostic criteria of the ADA and WHO

In the NHANES III study, some 19 000 adult

Americans aged over 20 years were questioned about

having previously diagnosed diabetes Over 6000 of

this group also had measurement of fasting plasma

glucose and almost 3000 of those aged 40±74 years

had a formal oral glucose tolerance test A glucose

challenge was not offered to people 75 years and over

NHANES III estimated that for 1997, 15.6 million

adult Americans (8.1% of the total population) had

diabetes mellitus Of these, 10.2 million (5.3%) had

diagnosed diabetes and 5.4 million (2.8%) had

un-diagnosed diabetes

Using ADA and the recently updated WHO criteria,

a further 13.4 million (6.9%) adult Americans have

IFG It is not possible to accurately estimate the

number of additional adult Americans with impaired

glucose regulation (i.e those with a normal IFG but

with IGT on the basis of a 2-hour post-load glucose

level) A subset of the NHANES III study population

did receive a glucose challenge; these were all aged

40±74 years and 15.8% had IGT In summary,

there-fore, 8.1% of adult Americans have diabetes, an

ad-ditional 6.9% have IFG and some others have impaired

glucose regulation, identi®ed only by a 2-hour

post-load glucose level

With regard to elderly Americans, 18.8% of those

aged 60±74 years and 18.9% of those aged over 75

years have diabetes An additional 14% of those over

60 years have IFG and still some others have IGT but

not IFG (Harris et al 1998) Some years ago it was

estimated that almost 20% of white North Americans

can expect to develop Type 2 diabetes if they survive

into their seventh decade (King and Zimmet, 1988)

The 1991 Canadian Study of Health and Aging

in-volved over 10 000 elderly people (aged 65±106 years)

and estimated the prevalence of diabetes at 12.4%

(Rockwood et al 1998) However, this estimate is at

best approximate as it largely relied on self-reporting

and health records and only some 700 subjects

un-derwent random glucose estimations

The UKTwo studies from the UK contain data that provide an

estimate of the prevalence of diabetes and IGT in

elderly people Both studies predated the new ADA=

WHO criteria for diabetes In two London general

practice populations, 10.3% of men and 9.5% ofwomen aged 65±69 years had diabetes; the corre-sponding ®gures were 11.4% and 9.4% for peopleaged 70±75 years (Yudkin et al 1993) A further 6.5%

of men and 5.6% of women aged 65±69 years had IGT

as had 8.4% of men and 3.6% of women aged 70±75years The second study, from Melton Mowbray, esti-mated that 9% of people aged over 65 years had dia-betes (Croxson et al 1991)

Other European CountriesThe prevalence of diabetes and IGT in elderly Finns isremarkably high, with 30% of men aged 65±84 yearshaving Type 2 diabetes and another 32% having IGT(Tuomilehto et al 1986) Less impressive rates werefound in a study of 1300 younger subjects (aged 65±74years), where 16% of men and 19% of women haddiabetes and another 18% and 19% respectively hadIGT (Mykkanen et al 1990) A Swedish communitystudy estimated that 15% of people aged over 65 yearshad diabetes (Andersson, Svardsudd and Tibblin 1991)and similar prevalence rates have been reported fromDenmark (Agner, Thorsteinsson and Eriksen 1982) Inmore southern parts of Europe, the prevalence rate forType 2 diabetes has been estimated at 22.8% in Italiansaged 65±84 years (Rosso et al 1998) and at 8.5% inFrench people aged over 65 years (Bourdel-March-asson et al 1997) However, as the French study largelyrelied on self-reporting, the true prevalence rate islikely to be substantially higher

Australia and New ZealandNational studies on the epidemiology of diabetes arelacking in both countries What data exists are mainlyderived from relatively small studies of rural or semi-rural communities In the 1989±90 National HealthSurvey in Australia, the prevalence of self-reporteddiabetes in people aged 65±74 years was 7.8% in malesand 6.5% in females (Welborn et al 1995) Prevalencerates for people aged over 75 years were 8.7% formales and 7.5% for females Another study whichrelied on self-reporting and fasting blood glucose le-vels found that 16.4% of people aged 60±69 years,16.7% of people aged 70±79 years and 16% of thoseaged 80 years or over had diabetes (Mitchell et al1998) In a random sample of some 600 New Zeal-anders aged over 65, the age-adjusted prevalence ofdiabetes was 15% (Lintott et al 1992) However, as

people living in residential care were excluded, this

rate is also likely to be an underestimate

Japan

It has been estimated that over 10% of people aged

over 45 years in Japan have diabetes while another

15% have IGT (Sekikawa et al 1993) This prevalence

rate is far higher than earlier estimates that used less

satisfactory methodologies Another study found that

13% of people aged 60±79 years had diabetes while an

additional 25% had IGT (Ohmura et al 1993)

Other CountriesReaders with a particular interest in the world-wide

prevalence of diabetes and IFG=IGT are referred to the

reviews by King and Zimmett (1988) and King and

Rewers (1993) An even more detailed account is

provided by Amos et al (1997) who estimate 1997

prevalence rates in various counties in Asia, Europe,

the Americas, Africa, and Oceania As can be seen

from Figure 1.3, some communities and countries have

remarkably high prevalence rates for diabetes and IGT

The highest recorded rates are among the Pima Indians

of Arizona, US, where 40% of those aged 65±74 years

have diabetes (Knowler et al 1978) However, new

challengers for this dubious distinction are now

appearing from populations in Papua New Guinea

(Dowse et al 1994) and Micronesians in Nauru

(Zimmet, Humphrey and Dowse 1994)

Countries with large populations also deserve

spe-cial mention even if the prevalence of diabetes is not

particularly high In Hong Kong Chinese, 10% of

those aged over 60 and 17% aged over 75 have

dia-betes (Woo et al 1987) In Taiwan, one study showed

that 26% of people aged over 60 years had diabetes and

another 22% had IGT (Lu et al 1998) Demographic

and socioeconomic changes in mainland China

sug-gest that a similar prevalence rate can be expected there

in time

THE ECONOMIC COST OF ABNORMAL

GLUCOSE TOLERANCE

Diabetes incurs both direct and indirect costs Direct

patient costs are the sum of what is spent on

diag-nosing and treating diabetes itself and on managing its

acute and chronic complications; it is estimated that

people with diabetes use hospital and primary care services two to three times as much as the generalpopulation (Damsgaard et al, 1987a,b) Lost pro-ductivity due to short term illness, disability and pre-mature mortality accounts for the indirect costs.Estimates of the economic cost of diabetes are largelydrawn from a few Western countries and cannot beextrapolated to other countries or healthcare systems

health-In the US, the National Diabetes Data Group mated the total cost of diabetes at $13.8 billion or 4%

esti-of the US health budget in 1984 (Entmacher et al1985) The Center for Economic Studies in Medicineput the cost at $20.4 billion for 1987 (Fox and Jacobs1988) Both ®gures are probable underestimates asthey either failed to include the cost of diabetic com-plications or underestimated the prevalence of dia-betes For similar reasons, the ®nding that Type 2diabetes alone cost the US economy $19.8 billion in

1986 is also an underestimate (Huse et al 1989) Even

if the absolute ®gures are inaccurate, the breakdown ofthe expenditure on Type 2 diabetes in the US in 1986 isstill of interest (Table 1.2) In the UK, by the late 1980s

it was conservatively estimated that diabetes cost

£1.2billion per annum or 5% of the NHS budget(Laing and Williams, 1989) In Australia, the directcost of diabetes was estimated at between $0.44 billionand $1.4 billion in 1995 and this is projected to rise tobetween $0.9 billion and $2.3 billion by 2010(McCarthy et al 1996)

There are age and sex differences in the economiccost of diabetes, with men aged under 65 years ac-counting for 35% of cost, women aged over 65 yearsfor 30%, men aged over 65 years for 18% and womenaged under 65 years for 17% (Huse et al 1989) Ithas been further estimated that per capita health

Table 1.2 Breakdown of estimated $19.8 billion spent on Type 2 diabetes in the US in 1986 a

Source: Modi®ed from Huse et al (1989).