Life expectancy: intermediate geographies and locally defined areas

Small areas and exclusion rules

Life expectancy (LE) in small geographical areas has been analysed using intermediate geography and locally defined area. Locally defined areas are aggregates of data zones determined by the relevant NHS board, but note that not all boards have chosen to define such areas and therefore these areas do not cover the whole of Scotland. (For further details of intermediate geographies and locally defined areas please see the technical paper  (155KB), section 6.) For most of these small areas, data are available for LE at birth for 5-year periods. However, to avoid the least robust situations, LE for a sex and area is not presented where:

a) the 5-year total population for that sex was less than 5,000 people; and/or

b) there were fewer than 40 deaths for that sex over the 5-year period.

These cut-offs are arbitrary, although other researchers have previously advocated rule a) (see technical paper (155KB), section 2.1).

In December 2008, as part of the Scottish health and wellbeing profiles ('new CHP profiles'), LE data were published for the four time periods 1994-98, 1996-2000, 1999-2003 and 2001-05, and the cut-offs excluded:

• for intermediate geographies, males: 162 of the potential 4,940 records (3.3%)
• for intermediate geographies, females: 177 of the potential 4,940 records (3.6%)
• for locally defined areas, males: 61 of the potential 452 records (13.5%)
• for locally defined areas, females: 60 of the potential 452 records (13.3%).

Using the same methodology, in November 2009 the General Register Office for Scotland published updated data on LE in intermediate zones (intermediate geographies) for the period 2003-07.

Interpreting the LE findings

Excluding records means that some extreme LE estimates are excluded from the full range for intermediate geography within Scotland and from the dataset for locally defined areas. It is therefore not possible to state which intermediate geography or locally defined area in Scotland has the highest or lowest LE.

Over the four time periods 1994-98, 1996-2000, 1999-2003 and 2001-05, amongst the remaining data for intermediate geographies, LE at birth ranged from 58.2 to 87.5 years for males, and from 68.4 to 90.1 years for females. Amongst the remaining data for locally defined areas, LE at birth ranged from 60.2 to 77.8 years for males, and from 68.8 to 83.6 years for females. Please note that some of these data have large 95% confidence intervals (up to 15.9 years). The implications of this are best explained using an example. The lowest remaining male LE at birth (58.2 years) has a confidence interval of nearly 5 years, from 55.7 to 60.6 years. This means that there is a 1 in 20 chance that the true LE at birth lies either above 60.6 years or below 55.7 years (although the most likely true value remains at 58.2 years). The wide range of possible values indicates that caution is required in drawing conclusions about the apparent ranking of small areas.

As small numbers are involved, these estimates for LE at birth in small areas are subject to random fluctuations in the number of deaths and the age at death. As a result, conclusions about time trends for any specific small area may not be reliable. Also, in any small area where the population is changing significantly (rising or falling), the calculated mortality rates and LE estimates may be less reliable as the population estimates may not yet fully reflect this change.

Therefore the results should be interpreted as providing a general indication of LE estimates over time, rather than precise and robust figures. The 95% confidence intervals give some indication of the stability of the estimates, in terms of the likely range of random year-to-year statistical variability in the data from which the estimates are produced. However, it should be emphasised that it is not possible to quantify the effects of any unusual circumstances which affect the figures for a particular area or year.

If a small area has a low LE at birth, there may be particular reasons for this. For example, the area may include nursing homes, hostels or other long-term care establishments which may serve people with poorer than average health and lower LE, and many of these people may have come from other parts of Scotland.

Note that healthy life expectancy (HLE) is not calculated for intermediate geographies and locally defined areas, due to the severe instability of the very small numbers that would be used to estimate the proportion of Census respondents with ‘good’ health in each age group.