Resetting the relationship between local and national government. Read our Local Government White Paper

Future health challenges: public health projections - smoking

Smoking prevalence analysis: methodology summary by the LGA Research and Information Team

Purpose

The purpose of this analysis is to produce a set of forecasts at local authority level for prevalence of smoking among adults. It is hoped that these forecasts will allow local authorities to anticipate likely future levels of smoking in their area, assuming that trends in smoking prevalence continue in their current trajectories.

Data source

The data used to produce these forecasts comes from the Annual Population Survey (APS), published by the Office for National Statistics (ONS). These statistics provide local authority level figures for smoking prevalence on an annual (calendar year) basis. The prevalence is calculated by dividing the weighted number of self-reported smokers aged 18 and over by total number of respondents (with a valid smoking status) aged 18 and over, expressed as a percentage. At the time of writing, the latest reliable statistics relate to the calendar year 2019, due to disruption to the 2020 figures resulting from the COVID-19 pandemic.

Data collected

The rates of smoking prevalence were extracted for all English single tier and county councils (that is, all English principal authorities excluding district councils), and for England as a whole. The local authorities of the City of London and the Isles of Scilly were not present in the data, due to having unusually small populations and often forming outliers compared to all other local authorities.

In some cases, local authorities or their boundaries changed during or after the seven-year period for which data was extracted. The measures taken to adjust for those changes was as follows:

  • Bournemouth, Christchurch and Poole Council: replaced Bournemouth and Poole unitaries and part of Dorset County Council on 1 April 2019, with the earliest available figures for the new council starting with 2020. As such, since the 2020 figures are not comparable with previous years due to the pandemic’s disruption, all figures used relate to the original Bournemouth unitary authority.
  • Dorset Council: replaced most of the area previously governed by Dorset County Council on 1 April 2019. As above, the earliest available figures for the new unitary authority start with 2020, so all figures used have been drawn from the county council.
  • Buckinghamshire Council: replaced Buckinghamshire County Council on 1 April 2020. All of the NCMP figures extracted relate to the former county council rather than the new unitary council, but as their boundaries are identical these estimates can be used to estimate future smoking prevalence for the unitary authority.
  • North Northamptonshire and West Northamptonshire Councils: replaced Northamptonshire County Council on 1 April 2021. All NCMP figures extracted relate to the former county council rather than the new unitaries. As such, it is not possible based on this analysis to arrive at individual projections for either of the two new councils, but only for Northamptonshire as a whole.

As a result, whilst there are 152 single-tier and county councils in existence in England, this forecasting approach produced estimates for 149 local authority areas (excluding the Isles of Scilly, City of London, North Northamptonshire and West Northamptonshire, and including one set of projections for the former Northamptonshire County Council).

The time range for which statistics were extracted covered the nine calendar years starting from 2011 and ending with 2019.

Projection methods

These historic smoking figures have been projected forward using a regression-based approach where, for each authority, a line of best fit is calculated for its existing trend using “least squares” methodology. This best fit line was then projected forwards, to the years 2025, 2030, 2035 and 2040.

The regression modelling was also used to calculate 95% confidence intervals for each set of projections, giving an indication of their level of uncertainty and the range in which the actual levels of prevalence might be expected to fall.

By definition, this approach assumes that past trends are representative of future trends; as such, these forecasts estimate future smoking prevalence under a scenario in which current factors influencing smoking continue unmitigated.

This approach does not make use of individual-level data or demographic projections; it simply takes a local authority area’s trajectory and extrapolates it into the future. As such, this approach is not as nuanced as other methodologies which may be based on more detailed information.

A further limitation of this approach is its assumption of a linear model; in certain cases where local authority areas have experienced unusually large increases or decreases in smoking in previous years, this may result in their future forecasts containing implausibly high or low values. The limitations of these projections should be borne in mind when interpreting their findings, and due consideration should be given to the level and trajectory of historical figures and any insights offered by other available data sources.

Overall results

Table 1 shows the results of these projections for English local authorities. Of particular note is the insight provided as to whether each area will reach the “Smokefree by 2030” target; that is, whether each English local authority area will have a smoking prevalence of below five per cent by the calendar year 2030.

The overall projections across England show that overall, the country is on track to reach below five per cent prevalence by 2030, provided that trends continue in their current trajectory. Smoking prevalence is set to reach approximately 4.9 per cent by 2030 and reach negligible levels by 2040. It should be noted, however, that current trajectories are unlikely to continue indefinitely, projections further in the future are less reliable, and it is likely that diminishing returns will lead to a reduction in the rate of decrease before a nationwide prevalence of zero is attained. The chief purpose of this analysis is to provide projections for local authority areas, not England as a whole, and it is likely that more reliable nationwide projections are available by other means.

The overall prevalence of smoking across England is in essence an average, masking considerable variation between local areas. By 2030, 77 of the 149 available local authority areas, around 52 per cent, are projected to reach smoking prevalence of below five per cent; by 2040 this figure is set to reach 139 authority areas, over 95 per cent. A noticeable difference in this distribution exists among the 37 authorities in the most deprived quartile of authority areas according to the Index of Multiple Deprivation (IMD): 15 of these authorities (41 per cent) are set to reach smokefree levels by 2030, and 33 (around 89 per cent) by 2040. This shows that the most deprived local authority areas are likely to be slower to reach smokefree prevalence levels by 2030, although they are likely to catch up somewhat by 2040. Very similar results exist with regard to the top quartile for the IMD Health Deprivation sub-domain.

Table 1; Projected smoking prevalence across England, 2020-2040

Table 1: Projected smoking prevalence across England, 2020-2040

Reception

2019

2025

2030

2035

2040

Overall England prevalence

13.9%

8.8%

4.9%

0.9%

0.0%

Number of councils smokefree (below 5%)

0

(0.0%)

13

(8.7%)

77

(51.7%)

116

(77.9%)

139

(93.3%)

Number of councils not smokefree (5% or above)

149

(100.0%)

136

(91.3%)

72

(48.3%)

33

(22.1%)

10

(6.7%)

Number of councils in the most deprived IMD quartile smokefree

0

(0.0%)

2

(5.4%)

15

(40.5%)

23

(62.2%)

33

(89.2%)

Number of councils in the most deprived IMD quartile not smokefree

37

(100.0%)

35

(94.6%)

22

(59.5%)

14

(37.8%)

4

(10.8%)