Socioeconomic inequalities in health and expectation of life have been found in many contemporary and past societies. In England although information based on an occupational definition of social class has only been available since 1921, other data identifying differences in longevity by position in society have been available for at least two hundred years. These differences have persisted despite the dramatic fall in mortality rates over the last century6.

Inequalities in health exist, whether measured in terms of mortality, life expectancy or health status; whether categorised by socioeconomic measures or by ethnic group or gender. Recent efforts to compare the level and nature of health inequalities in international terms indicate that Britain is generally around the middle of comparable western countries, depending on the socioeconomic and inequality indicators used22. Although in general disadvantage is associated with worse health, the patterns of inequalities vary by place, gender, age, year of birth and other factors, and differ according to which measure of health is used23.

General trends in health
Death rates in England have been falling over the last century, from a crude death rate of 18 per thousand people in 1896 to 11 per thousand in 199624,25. Over the last 25 years, there have been falls in death rates from a number of important causes of death, for example lung cancer (for men only), coronary heart disease and stroke25.

Life expectancy has risen over the last century26, but not all life is lived in good health. Healthy life expectancy - the measure of average length of life free from ill health and disability - has not been rising; the added years of life have been years with a chronic illness or disability27.

The proportion of people reporting a limiting long standing illness has risen from 15 per cent to 22 per cent since 1975. The proportion reporting illness in the two weeks previous to interview has nearly doubled from 9 per cent to 16 per cent. There is a slight increase in the proportion of people consulting the NHS28.

Measuring socioeconomic position
A number of different measures can be used to indicate socioeconomic position. These include occupation, amount and type of education, access to or ownership of various assets, and indices based on residential area characteristics. There has been much debate as to what each indicator actually measures, and how choice of indicator influences the pattern of inequalities observed. For example, measures based on occupation may reflect different facets of life for men compared to women, and for people of working age compared to older people or children.

Choice of measure is often dictated by what is available. In Britain occupational social class is frequently used, especially for data collected nationally. Table 1 shows examples of the occupations in each social class group.

Mortality
Over the last twenty years, death rates have fallen among both men and women and across all social groups25,29. However, the difference in rates between those at the top and bottom of the social scale has widened.

For example, in the early 1970s, the mortality rate among men of working age was almost twice as high for those in class V (unskilled) as for those in class I (professional). By the early 1990s, it was almost three times higher (table 2). This increasing differential is because, although rates fell overall, they fell more among the high social classes than the low social classes. Between the early 1970s and the early 1990s, rates fell by about 40 per cent for classes I and II, about 30 per cent for classes IIIN, IIIM and IV, but by only 10 per cent for class V. So not only did the differential between the top and the bottom increase, the increase happened across the whole spectrum of social classes29.

Both class I and class V cover only a small proportion of the population at the extremes of the social scale. Combining class I with class II and class IV with class V allows comparisons of larger sections of the population. Among both men and women aged 35 to 64, overall death rates fell for each group between 1976-81 and 1986-92 (table 3). At the same time, the gap between classes I and II and classes IV and V increased. In the late 1970s, death rates were 53 per cent higher among men in classes IV and V compared with those in classes I and II. In the late 1980s, they were 68 per cent higher. Among women, the differential increased from 50 per cent to 55 per cent30.

These growing differences across the social spectrum were apparent for many of the major causes of death, including coronary heart disease, stroke, lung cancer and suicides among men, and respiratory disease and lung cancer among women29,30.

Death rates can be summarised into average life expectancy at birth. For men in classes I and II combined, life expectancy increased by 2 years between the late 1970s and the late 1980s. For those in classes IV and V combined, the increase was smaller, 1.4 years. The difference between those at the top and bottom of the social class scale in the late 1980s was 5 years, 75 years compared with 70 years. For women, the differential was smaller, 80 years compared with 77 years. Improvements in life expectancy have been greater over the period from the late 1970s to the late 1980s for women in classes I and II than for those in classes IV and V, two years compared to one year31.

A good measure of inequality among older people is life expectancy at age 65. Again, in the late 1980s, this was considerably higher among those in higher social classes, and the differential increased over the period from the late 1970s to the late 1980s, particularly for women31.

Years of life lost
Premature mortality, that is death before age 65, is higher among people who are unskilled. Table 4 illustrates this with an analysis of deaths in men aged 20 to 64 years. If all men in this age group had the same death rates as those in classes I and II, it is estimated that there would have been over 17,000 fewer deaths each year from 1991 to 1993. Deaths from accidents and suicide occur at relatively young ages and each contribute nearly as much to overall years of working life lost as coronary heart disease. Death rates from all three causes are higher among those in the lower social classes, and markedly so among those in class V32,33.

These major differences in death rates and life expectancy between social classes do not just apply to those people already well into adulthood. Infant mortality rates are also lower among babies born to those of higher social classes. In 1994-96, nearly 5 out of every thousand babies born to parents in class I and II died in their first year. For those babies born in to families in classes IV and V, the infant mortality rate was over 7 per thousand babies. As with mortality at other ages, infant mortality rates in each class have been decreasing over the last twenty years. However, there is no evidence that the class differential in infant mortality has decreased over this period34.

Morbidity
Although death rates have fallen and life expectancy increased, there is little evidence that the population is experiencing less morbidity or disability than 10 or 20 years ago. There has been a slight increase in self-reported long standing illness and limiting long standing illness, and socioeconomic differences are substantial. For example, in 1996 among the 45 to 64 age group, 17 per cent of professional men reported a limiting long standing illness compared to 48 per cent of unskilled men. Among women, 25 per cent of professional women and 45 per cent of unskilled women reported such a condition. These patterns were similar among younger adults, older men and among children28.

In adulthood, being overweight is a measure of possible ill health, with obesity a risk factor for many chronic diseases. There is a marked social class gradient in obesity which is greater among women than among men35-37. In 1996, 25 per cent of women in class V were classified as obese compared to 14 per cent of women in class I. For men, there was no clear difference in the proportions reported as obese except that men in class I had lower rates of obesity, 11 per cent, compared to about 18 per cent in other groups. Overall, rates of obesity are rising. For men, 13 per cent were classified as obese in 1993 compared to 16 per cent in 1996. For women, the rise was from 16 per cent to 18 per cent37.

Another indicator of poor health is raised blood pressure. There is a clear social class differential among women, with those in higher classes being less likely than those in the manual classes to have hypertension. In 1996, 17 per cent of women in class I and 24 per cent in class V had hypertension. There was no such difference for men where the comparable proportions were 20 per cent and 21 per cent respectively37.

Among men, major accidents are more common in the manual classes for those aged under 55. Between 55 and 64, the non-manual classes have higher major accident rates (figure 3). For women, there are no differences in accident rates until after the age of 75 when those women in the non-manual group have higher rates of major accidents37.

Mental health also varies markedly by social class. In 1993/4, all neurotic disorders, such as anxiety, depression and phobias, were more common among women in class IV and V than those in classes I and II - 24 per cent and 15 per cent respectively38. This difference was not seen among men. However, there were striking gradients for alcohol and drug dependence among men, but not women. For example, 10 per cent of men in classes IV and V were dependent on alcohol compared to 5 per cent in classes I and II, (figure 4)38.

Trends in socioeconomic determinants of health

Income distribution
Over the last twenty years, household disposable income per head of population has grown both in actual and in real terms. Between 1961 and 1994, average household disposable income (in real terms) rose by 72 per cent39. However, this was not experienced to the same extent across the whole of the income distribution.
The median real household disposable income, before housing costs, rose over the period 1961 to 1994 from £136 per week, to £234 per week (figure 5). The top decile point more than doubled, from £233 per week to £473 per week. The bottom decile point rose by 62 per cent from £74 per week to £119 per week.

Households below average income
The proportion of people whose income is below average has been at about 60 per cent for the last 35 years (figure 6). However, the proportion of people below half of the average income (the European Union definition of poverty) has grown over this period from 10 per cent in 1961 to 20 per cent in 1991. It has decreased since then and was at 17 per cent in 199540.

Education
Since the early 1970s, the proportion of children aged 3 or 4 who attend school has trebled from 20 per cent to nearly 60 per cent40. The proportion who attend school (as opposed to playgroups) varies from 84 per cent in the North East to 43 per cent in the South West41.

Educational attainment - as measured by the proportion of children gaining 5 or more GCSEs at grades A star to C - has risen from less than 25 per cent in 1975/76 to about 45 per cent in 1995/9640,42. This measure of attainment varies not only by gender, but also by geographical area and by measures of deprivation.

As well as looking at the future workforce and their qualifications, it is useful to look at the educational attainment of those presently of working age40. In 1997, 16 per cent of men and 21 per cent of women of working age had no qualifications. There were also large differences between ethnic groups (figure 7).

Employment
The seasonally adjusted unemployment rate for those aged 16 and over stood at 6.2 per cent in summer 1998, almost three times the level of 30 years ago43. Although rates have been falling since 1993, there have been changes in the patterns of unemployment over the last thirty years, well beyond what might have been expected from seasonal and cyclical variations (figure 8). Youth unemployment is still at higher rates now than it was in 1991 and unemployment rates are four times higher among unskilled workers than among professional groups44.

Across different ethnic groups, there are very different rates of unemployment (table 5). Those from minority ethnic groups have higher rates than the white population. Black men have particularly high unemployment rates as do Pakistani and Bangladeshi women45.

Housing
Over the last sixty years, the number of dwellings has doubled from 10.6 million in 1938 to 20.7 million in 199746,47. Housing tenure has also changed dramatically over this period with a doubling of the proportion of owner-occupied dwellings46,47 and a dramatic fall in the proportion of privately rented dwellings (table 6).

There has also been a growth in the number of one-person households over the last ten years from 4.4 million in 1984 to 5.5 million in 1995/9648. The proportion of all households which had only one person rose from 25 per cent to 28 per cent over this period. In 1984, 46 per cent of one-person households were owner occupied. By 1995/96, this had grown to 54 per cent (table 7).

Between 1991 and 2016, the number of households is expected to rise from 19.2 million to 23.6 million - a rise of 4.4 million households49.

Conditions of the housing stock vary considerably. In 1996 about 14 per cent of all households were living in poor conditions. About 8 per cent of dwellings in England were unfit, and about 7 per cent of households were living in unfit dwellings. The proportions of households in unfit dwellings varied with the type of tenure, from 4 per cent in the Registered Social Landlord sector to 18 per cent of households who rented from private landlords. In urban areas, 8 per cent of dwellings were deemed unfit whereas in rural areas, 5 per cent were deemed unfit50.

Homelessness
Between 1982 and 1992, there was a steep increase in the number of households accepted by Local Authorities as homeless. Since then, there has been a decrease of about a quarter. Of the 166,000 households classified as homeless in 1997, over 103,000 were accepted by local authorities to be unintentionally homeless and in priority need. Over half of households accepted by local authorities as homeless had dependent children and a further tenth had a pregnant household member51.

Public safety
The crime rate has nearly trebled since 1971. In 1996, the crime rate in England was nearly one crime for every ten people40. Crime rates were highest in areas with large conurbations - the North East, Yorkshire/Humberside and London41. There were also different crime rates in different types of areas - lowest in affluent suburban and rural areas and highest in council estates and low income areas (table 8).

Different areas of the country have very different rates of particular types of crime. London has the highest rate of fraud and forgery, robbery and sexual offences. The North East has the highest rate of criminal damage and the lowest rate of sexual offences. Yorkshire and the Humber has the highest burglary rate. The East has the lowest overall crime rate41.

Transport
Access to private means of transport has increased in recent years. In 1996, 70 per cent of households had access to a car or a van. This compared with just over half of households in 1972. About a quarter of households had access to two or more cars and vans compared to only 9 per cent in 1972 (figure 9)28,52,53.

Those with access to two or more cars or vans were not only more likely to be economically active, but also tended to be in the higher socioeconomic groups. Only seven per cent of households had access to two or more vehicles when the head of household was economically inactive compared to 36 per cent of households with an economically active head28. In 1991, those who lived in the social rented sector had the highest proportion with no access to a car, 68 per cent, while those in the owner occupied sector had the smallest proportion with no access, 19 per cent52.

How people travel to work differs depending on whether the areas in which they live are urban or rural54. In England in 1991, 60 per cent of people travelled to work by car in urban areas and 69 per cent in rural areas. Rail and bus accounted for 17 per cent of journeys to work for those in urban areas but only for five per cent for rural areas. A higher proportion of people work at home in rural areas, 12 per cent compared to four per cent in urban areas (table 9).

Health related behaviour
Over the last twenty years, the proportion of people who report that they smoke cigarettes has fallen. In 1974, roughly a half of men and two fifths of women smoked cigarettes, compared with less than 30 per cent of men and women in 1996. The trends in drinking alcohol are broadly unchanged over this period. However, the proportion of women who drank more than 14 units of alcohol a week rose from 9 per cent in 1984 to 14 per cent in 199628.

There is a clear social class gradient for both men and women in the proportion who smoke. In 1996, this ranged from 12 per cent of professional men to 41 per cent of men in unskilled manual occupations and from 11 per cent to 36 per cent for women28. In spite of the major class differences in dependence on alcohol in men38, there are very small differences in the reported quantities consumed. This is not the case among women where higher consumption is related to higher social class28.

Among women, there are no differences in levels of physical activity across the social classes. Among men, higher proportions in the manual classes have a high level of physical activity than in the non-manual classes. However, some of this difference is due to work related physical activity. Men in non-manual occupations have higher rates of leisure time physical activity35.

People in lower socioeconomic groups tend to eat less fruit and vegetables, and less food which is rich in dietary fibre. As a consequence, they have lower intakes of anti-oxidant and other vitamins, and some minerals, than those in higher socioeconomic groups35,55-58.

One aspect of dietary behaviour that affects the health of infants is the incidence of breastfeeding. Six weeks after birth, almost three quarters of babies in class I households are still breastfed. This declines with class to less than one quarter of babies in class V. The differences between classes in rates of breastfeeding at six weeks has narrowed slightly between 1985 and 199559.

Trends in health differences between minority ethnic groups
There are many indications of poorer health among the minority ethnic groups in England. For example, people in Black (Caribbean, African and other) groups and Indians have higher rates of limiting long standing illness than white people. Those of Pakistani or Bangladeshi origin have the highest rates. In contrast, the Chinese and "other Asians" have rates lower than the white population60.

Although in analysing mortality rates we have to use country of birth as a proxy for ethnicity, a similar pattern emerges61. There is excess mortality among men and women born in Africa and men born on the Indian sub-continent and men and women born in Scotland or Ireland (table 10).

Many women from minority ethnic groups giving birth in the 1990s were born in the United Kingdom. Because country of birth of the mother, and not ethnicity, is recorded at birth registration, it is not possible to estimate infant mortality rates by minority ethnic group. However, among mothers who were born in countries outside the UK, those from the Caribbean and Pakistan have infant mortality rates about double the national average. Perinatal mortality rates have also been consistently higher for babies of mothers born outside the UK. The differences between groups have not decreased over the last twenty years34.

Trends in health differences between the sexes
Death rates have been falling for both males and for females (figure 10). Since 1971, these have decreased by 29 per cent for males and by 25 per cent for females, narrowing the differential in death rates very slightly. Cancers and coronary heart disease account for 55 per cent of the deaths of men and 42 per cent of the deaths of women25.

At each age in childhood, and on into adulthood, the age-specific mortality rates for boys is higher than for girls (figure 11)62. For the under 5s, nearly half of the difference is due to external causes, in particular accidental drowning and submersion. For children aged 5 to 14, external causes, chiefly motor vehicle traffic accidents, account for nearly 70 per cent of the difference25.

Although the life expectancy gap between males and females is decreasing26, this is not the case for healthy life expectancy. Healthy life expectancy of females is only two to three years more than that of males27. Overall, there is little difference in the proportions of males and females reporting a limiting long standing illness53. Women report more illness of many different types than men during the reproductive years53.

For both children and adults of working ages, males have higher major accident rates than females (figure 12). At older ages, women have higher major accident rates than men37.

The proportion of smokers is higher among girls than boys63. By adulthood, the proportions of men and women smoking are about the same (29 and 28 per cent), compared with 51 per cent of men and 41 per cent of women in 197453. For both children and adults, males are more likely to drink alcohol heavily than females53.

Conclusion
Inequalities by socioeconomic group, ethnic group and gender can be demonstrated across a wide range of measures of health and the determinants of health. Analysis of these patterns and trends in inequalities has informed the development of areas for future policy development, which are considered in Part 2.