Article Text

Download PDFPDF

Do smoking cessation programmes influence geographical inequalities in health? An evaluation of the impact of the PEGS programme in Christchurch, New Zealand
  1. R Hiscock1,
  2. J Pearce2,
  3. R Barnett3,
  4. G Moon4,
  5. V Daley5
  1. 1
    UK Centre for Tobacco Control Studies, University of Bath, Bath BA2 7AY, UK
  2. 2
    Institute of Geography, School of GeoSciences, University of Edinburgh, Edinburgh EH8 9XP, UK
  3. 3
    Department of Geography, University of Canterbury, Christchurch, New Zealand
  4. 4
    School of Geography, University of Southampton, Southampton SO17 1BJ, UK
  5. 5
    Canterbury District Health Board, Christchurch, New Zealand
  1. Correspondence to Dr Jamie Pearce, Institute of Geography, School of GeoSciences, University of Edinburgh, Edinburgh EH8 9XP, UK; jamie.pearce{at}


Objective: To identify the impact of a smoking cessation programme on area-based social and ethnic inequalities in smoking rates through social and ethnic differences in enrolment and quitting.

Methods: Analysis of records of 11 325 patients who enrolled in an innovative smoking cessation programme in Christchurch, New Zealand between 2001 and 2006. We compare enrolment, follow-up, quitting and impact on population smoking rates in the most and least deprived neighbourhoods and the neighbourhoods with the lowest and highest proportions of Māori.

Results: Enrolment as a proportion of the population was higher from the most deprived areas but as a proportion of neighbourhood smokers, it was lower. Enrolees from the least deprived quintile were 40% more likely to quit than those from the most deprived quintile. Smoking rates were 2.84 (2.75 to 2.93) times higher in the most deprived neighbourhoods. If the programme had not been available we estimate that this differential would have reduced to 2.81 (2.72 to 2.90). In neighbourhoods with the highest proportion of Māori, smoking rates were 2.33 (2.26 to 2.41) times higher and we estimate that without the programme smoking rates would be 2.30 (2.23 to 2.37) times higher.

Conclusions: Although enrolees were drawn from a wide variety of backgrounds, those most likely to quit tended to reside in affluent areas or areas with a low proportion of Māori. There was no evidence that this smoking cessation programme increased or decreased inequalities within the Christchurch population. For smoking cessation programmes to have an impact on health inequalities more effort is required in targeting hard-to-reach groups and in encouraging them to quit.

View Full Text

Statistics from

Globally, smoking is well established as a major cause of ill health and mortality,1 2 3 4 contributing to approximately 5.4 million deaths each year.5 The consumption of tobacco is also an important determinant of social, ethnic and spatial inequalities in health.6 In New Zealand just over one-tenth of the residents living in the least socially deprived fifth of neighbourhoods are smokers compared to one-third in the most deprived quintile.7 Absolute socioeconomic deprivation and relative inequalities between New Zealanders are important in explaining the high rates of smoking among Māori, the indigenous people of New Zealand.8 9 10 As in other countries, people belonging to lower socioeconomic status and ethnic minority groups are less likely to quit,11 12 13 14 15 16 17 resulting in increased ethnic and social inequalities in rates of smoking.18

Many approaches have been taken to reduce smoking rates, including tobacco cessation treatments incorporated into primary care settings.5 While such approaches have increased quit rates,19 20 21 concerns have also been expressed that such approaches may increase rather than reduce health inequalities. If more affluent groups are the main beneficiaries of quit programmes then the relative difference in smoking rates between disadvantaged and affluent groups will grow. Moreover, such trends will be accentuated if disadvantaged groups are less likely to see their general practitioner (GP) for preventative treatment and if the costs of nicotine replacement therapy are a disincentive to enrolling in quit programmes.22 For smoking cessation to reduce inequalities there needs to be at least equity or even bias towards disadvantaged groups in two processes23—access to treatments, and in the desired outcome, quitting smoking. The evidence is inconclusive on whether smoking cessation programmes increase quitting among low-income groups, whether quitting among such groups is large enough to reduce inequalities or whether inequalities are actually increased because the quit rate is lower in these groups.9 24 25 26 27 Evaluations of smoking cessation programmes in New Zealand suggest that these programmes increase the quit rate.28 29 However there has been less focus on the impact of cessation programmes on health inequalities, and such evidence as exists has been equivocal.29 30 This research gap is perhaps surprising given the priority in addressing health inequalities by the New Zealand government. During the 1980s and 1990s there was a significant increase in social, ethnic and geographical inequalities in health in New Zealand.31 32 However, following the election of a Labour-led government in 1999, reducing health inequalities became enshrined in the 2000 New Zealand Health Strategy.33 34

The Christchurch PEGS programme

We evaluate the implications for inequality that arise from the implementation of the “Preparation Education Giving up and Staying smoke free” (PEGS) programme in Christchurch, New Zealand, the first multi-practice-based smoking cessation programme in the country. Established in 1995,35 PEGS represented an innovative educational approach to reduce smoking rates. However, only in later years was there more of a focus upon targeting deprived populations and, by implication, a reduction in health inequalities, an important goal of the primary health care strategy.36 In view of such trends, we compare the social and ethnic distribution by residence of enrolees in the programme with the social and ethnic distribution by residence of those who quit, and ask whether the programme has reduced differences in the population smoking rates. We also explore whether the social distribution of enrolment and quitting has changed over time.

PEGS is delivered by GPs and administered by Pegasus Health, an independent practitioner association that is the main part of the Partnership Health Primary Health Organisation.35 Over 90% of the Partnership Health population is enrolled with Pegasus Health GPs. The programme uses the “readiness to change” model.37 Enrolees are provided with different types of counselling and literature depending on how ready they are to quit.38 Those deemed most ready are also offered nicotine replacement therapy (NRT) and nominate a quit date.39 The delivery of the programme is not consistent across practices but face-to-face support tends to be given when the patient collects the NRT from the practice every one or two weeks. The NRT is heavily subsidised by the Ministry of Health for up to three months. A small part-charge of $NZ2.50 (£0.97; €1.1) per week is unlikely to be a significant disincentive, even to those on low incomes. Enrolees are followed up by their GP 6 months after their enrolment.

Data and methods

The residential address of each patient who enrolled in PEGS between January 2001 and December 2006 was geocoded to identify the 2001 census meshblock. We achieved 94.5% coverage. This provided a dataset of 11 413 enrolees who had 15 807 contacts with the PEGS programme. Excluding enrolees from outside the Christchurch urban area resulted in a final sample of 11 325 participants.

Patients were defined as quitters by their doctor or practice nurse if they were assessed as being non-smokers at the 6-month follow-up after completing the programme. Practices attempted to follow up all patients at 6 months and 7778 (69%) patients were re-contacted. All data were routinely collected rather than being collected as part of a research project. Some patients moved and others could not be contacted despite several telephone calls. Owing to the sizeable loss to follow-up we calculated two quit rates with different denominators—those followed up and the total enrolled. Our quit-rate calculations using the total enrolled as the denominator population therefore assume that those lost to follow-up had continued to smoke.

Our focus is on the impact of PEGS on geographical inequalities in smoking prevalence. Neighbourhood (census meshblock) measures of socioeconomic status (SES) and ethnicity were used for the main analysis. Each census meshblock includes about 60–90 households. Census meshblocks were classified by area-level deprivation in 2001 (measured using the New Zealand Deprivation Index (NZDep)40 and dividing all neighbourhoods into quintiles) and ethnic composition (quintiles according to the percentage of the population that were Māori; the quintile boundaries were 1.6%, 4.3%, 7.2% and 11.8% Māori). NZDep is a nationwide measure, whereas the Māori quintiles refer to Christchurch because there were too few meshblocks in the highest national Māori quintile in the study area. As there was no census question on smoking in 2001, we also linked smoking data from the 1996 and 2006 censuses to the 2001 meshblocks.

We adapted a methodology devised for English data27 to assess the impact of PEGS on health inequalities in the Christchurch population. We present results comparing enrolment and quitting among those who lived in meshblocks classified in the most deprived and least deprived quintiles and the likely effect on smoking rates in the Christchurch population. Second, we compare the quintile with the highest proportion of Māori with the quintile with the lowest proportion of Māori. Enrolees were grouped by the year of enrolment to explore changes in the social distribution of enrolment and quitting over time.


For the smoking cessation programme to have the potential to reduce health inequalities, the smoking rates of low SES groups and Māori should decline towards a level similar to the general population. Although, without individual-level socioeconomic and ethnicity data we cannot directly measure the extent to which this is the case, we can examine the neighbourhoods in which enrolees live (we must of course make the caveat however that some residents of poor areas are of high SES and vice versa). Given the high prevalence of smoking in neighbourhoods that are more deprived and have a high proportion Māori, a reduction in inequalities will require enrolees to be drawn disproportionately from such areas.

We explore first whether the PEGS programme draws enrolees from areas with higher proportions of deprived residents. Superficially it appears that coverage from the most deprived quintile is reasonably good: 2.0% of Christchurch residents from the least deprived quintile enrolled and 5.2% did so from the most deprived quintile: a higher proportion from the most deprived quintile (table 1). A higher proportion of PEGS enrolees came from the most deprived quintile compared to the least deprived quintile (19.5% compared to 12.6%).

Table 1

Enrolment and quitting by deprivation and ethnicity

We then considered enrolees in terms of the smoking population. Smoking data were collected in the 1996 and 2006 censuses. We estimated the number of smokers in 2001 (the mid-points between 1996 and 2006) to be 6467 in the least deprived quintile and 10 661 in the most deprived quintile. Although there were more enrolees from the most deprived quintile there were also more smokers in the most deprived quintile. A slightly higher proportion (22.0%) of smokers were estimated to have enrolled from the least deprived quintile in comparison with the most deprived quintile (20.7%).

To reduce inequalities in smoking-related morbidity and mortality, quit rates from a smoking cessation programme should favour enrolees from deprived areas. There was little difference between the highest and lowest quartiles in the proportion for whom quit rates were available (the follow-up population) but there was a marked difference in the quit rates. The quit rates of those followed up were 36.1% for enrolees from the least deprived quintile and 25.6% for enrolees from the most deprived quintile. If we assume that all the enrolees who were not followed up had also not quit, then the quit rates fall to 25.2% and 17.5%, respectively.

To assess whether the larger number of quitters from the most deprived quintile compensates for the lower quit rate we have to consider the impact of PEGS on population smoking rates. Bauld and colleagues assessed the role of English cessation programmes in reducing inequalities by comparing smoking rates in the population at the start and end of the assessment period in deprived and non-deprived areas.27 We do not have data on Christchurch smoking rates in 2001 but we do have census data for 2006. We can compare the actual number of smokers in the population in 2006 with an estimate of the number of smokers that there would have been had the PEGS programme not been available. In order to do this, we have to make some adjustments to the number of quitters before making population level calculations. Six months is the shorter of the two recommended long-term follow-up periods,41 the other being one year. In an English study, 75% of those who had quit when followed up one month after participating in a smoking cessation programme, had relapsed by one year,42 at all levels of deprivation.27 Our follow-up is at 6 months so we assume that half of those who are going to relapse within a year would have already relapsed, but that a further 37.5% of the 6-month quitters will relapse. This leaves 224.4 quitters from the least deprived areas and 240.6 quitters from the most deprived quintile. In New Zealand the latent or background quit rate is likely to be about 2.5%43; thus it is likely that about six quitters from both quintiles would have quit without assistance from PEGS. If we exclude latent quitters we estimate that there were 219 quitters from the least deprived quintile and 235 quitters from the most deprived quintile who would have been recorded as ex-smokers in the 2006 census because of PEGS; without PEGS they would have been recorded as smokers.

The smoking rate of the Christchurch population from the 2006 census was 8.3% for the least deprived quintile and 23.7% for the most deprived quintile. If PEGS had not been available we estimate that the percentages would have been 8.6% for the least deprived quintile and 24.2% for the most deprived quintile. The actual absolute gap between the smoking rates for the most deprived and most affluent areas is 15.4 (table 2), whereas without PEGS we estimate the gap would be 15.6. The smoking rate was 2.84 times higher in the most deprived compared to the least deprived areas. We estimate the smoking rate would have been 2.81 times higher in the most deprived areas without PEGS. Confidence intervals overlap so we can conclude that PEGS neither increased nor decreased deprivation-related inequalities in the smoking rate.

Table 2

Absolute and relative smoking rate differences (95% confidence intervals) between the most affluent and deprived neighbourhoods and neighbourhoods with the highest and lowest proportions of Māori

The pattern shown for deprivation was mirrored by that for ethnicity (tables 1 and 2). More enrolees were drawn from the quintile with the highest proportion of Māori (2972) than the quintile with the lowest proportion of Māori (1165). The proportion of smokers enrolled was similar (21.5% and 22.7%, respectively). Enrolees from the lowest Māori quintile were two-fifths more likely to quit (36.4%) than enrolees from the highest quintile (26.3%). We estimate that there were an extra 179 people from the lowest Māori quintile and 324 people from the highest Māori quintile recorded as ex-smokers in the 2006 census who would been smokers if they had not taken part in the PEGS programme. Again there was no significant difference in the actual gaps between the smoking rates in the highest and lowest Māori quintiles and the difference if PEGS had not been available. PEGS made no difference to area-based ethnic inequalities.

How has the social distribution of enrolment and quitting changed over time? Given the similar conclusions drawn above for area inequalities by deprivation and ethnicity, we focus on deprivation. Enrolments in general declined from 2781 in 2001 to 1373 in 2006 (table 3). About a quarter of participants enrolled in 2001 whereas only just over 12% enrolled in 2005 and 2006. The social distribution of enrolees was stable throughout. Between 11.6% and 13.5% of the participants who enrolled each year were from the least deprived quintile and between 17.5% and 20.6% were from the most deprived quintile.

Table 3

Enrolment and quitting over time and by deprivation

The number quitting declined from 406 in 2001 to 175 in the final three years. The numbers quitting in the most deprived and the least deprived quintiles were similar: about 100 in 2001 and about 40 in 2006. Quit rates of those followed up remained stable, ranging from 32.3% in 2001 to 29.1% in 2003. Although numbers quitting from the most deprived and least deprived quintiles were similar, quit rates differed because of the higher enrolment from the most deprived quintile with ratios of 0.6 or 0.7 for each year. In the least deprived quintile quit rates varied from 40.6% in 2001 to 30.2% in 2005 and from 27.7% in 2004 to 22.2% in 2003 in the most deprived quintile. There was a little more fluctuation in the ratios for quitting than for enrolment. For every 1% quitting from the most deprived quintile there were between 1.1% (2005) and 1.6% (2001) quitting from the least deprived quintile. The same pattern was followed when quit rates for all enrolees were calculated.


The PEGS programme represents an innovative approach to reducing smoking rates by engaging with GPs as the key drivers of behaviour change. Compared to the initial evaluation of the programme in the 1990s,39 our results indicate that, since 2001, PEGS has consistently attracted more enrolees from deprived neighbourhoods and from areas with a greater proportion of Māori. However, despite this success, the impact on inequality has been limited. When considered against the population of smokers, the proportion of smokers enrolled, if anything, favoured affluent neighbourhoods with fewer Māori. It appears that targeting needs to be further improved.

Quit rates were higher among enrolees from the most affluent neighbourhoods than enrolees from the most deprived neighbourhoods. Over a quarter of enrolees from the most affluent neighbourhoods quit compared to less than one-sixth from the most deprived neighbourhoods. The figures were similar for neighbourhoods with the highest and lowest proportion of Māori. When we reanalysed the results using individual ethnicity data rather than neighbourhood data, our results confirmed that more Māori enrol yet fewer quit suggesting that our results are not the result of an ecological fallacy (results not shown). This pattern did not change over the study period.

The finding that enrolees from more deprived areas were relatively less likely to quit is perhaps a reflection of the various contextual drivers, such as more opportunities to purchase cigarettes, higher levels of crime or feelings of relative deprivation that may operate in highly deprived neighbourhoods.44 45 46 The result is more stressful lives.47 Under such circumstances, the adverse health consequences of smoking pale in significance in comparison to other adverse neighbourhood conditions.44 45 46 If people do quit in the face of these circumstances, it has been found that they experience less financial stress and improved material wellbeing48 thus underlining the importance of cessation programmes.

The quit rate from the PEGS programme is similar to a cessation programme for Māori women29 and slightly less favourable than the rate given in an earlier PEGS evaluation.38 The pattern of quitters being more likely to reside in affluent neighbourhoods or areas with few Māori was not found in the earlier evaluation of PEGS,38 but in this study a much larger sample was extracted.

Overall, despite the innovative nature of the PEGS programme and the relatively high quit rate, the differences in smoking rates between the most and least deprived neighbourhoods and neighbourhoods with higher and lower proportions of Māori were effectively unchanged by PEGS. A similar study in England27 found that higher recruitment in more deprived areas could compensate for the lower quit rate among residents of such areas and, furthermore, at the population level, inequalities between more deprived and less deprived areas were reduced. However the most deprived areas were compared with all other areas rather than just the most affluent areas.

The policy implication that flows from our research is that to reduce health inequalities even more work needs to be done within cessation programmes to attract smokers in the most deprived neighbourhoods to compensate for the lower quit rate. In England, smokers from more deprived areas are attracted by the wide publicity of services and the locating of services in non-healthcare settings in addition to primary care.27 More fundamentally, alterations need to be made to PEGS to increase the quit rate among groups where smoking is more entrenched, perhaps using ideas from other programmes that have been successful in targeting disadvantaged groups.27 29 Finally we note that enrolment in general dropped in the study period. The impetus of the programme needs to be reinforced.

There are of course limitations in our analysis. First, a third of the enrolees were not able to be re-contacted to establish their quit status. This reduces the efficacy of our analysis as we can only establish that the overall quit rate was between 20% and 30%. Further work could usefully locate those lost to follow-up. The 6-month follow-up period is itself short for a long-term addiction. It may be possible to return to participants and collect longer-term follow-up data. Second, we based the long-term relapse rate on English data. It may be that the relapse rate is different in New Zealand. We were also unable to consider the possibility that latent quit rates vary with area deprivation or ethnicity. Third, the use of neighbourhood SES means that we cannot say that enrolees themselves were deprived; individual data were available on ethnicity but not deprivation. The individual ethnicity data substantiate our conclusions. Fourth, at the neighbourhood level, smoking questions were not asked in the 2001 census. Our estimate of 2001 smokers (the mid-point between the 1996 and 2006 census numbers) may be inaccurate if PEGS did increase the number of quitters. Fifth, although self-reports of quitting were routinely collected, they were not validated by using objective biochemical markers. However, self-reported smoking status is generally considered to be an accurate measure of actual smoking status.49 Sixth, patients were recruited into PEGS by various GPs and nurses who may have made different judgments as to when an individual was “ready” for the cessation programme rather than actively attempt to enrol all eligible patients. Finally, we only present bivariate analyses. In future work, we intend to undertake multivariate modelling so that the relative strengths of predictors of quitting can be compared. This is a rich dataset that includes details of the GP surgeries that provide the programme as well as the enrolee’s residential neighbourhood. We therefore hope to compare and contrast surgery and neighbourhood variances in quitting using multilevel modelling.

In conclusion, we have established that after receiving the PEGS smoking cessation treatment there was a significant reduction in smoking rates. While there was some evidence of variation in quitting by levels of neighbourhood deprivation and ethnic composition of enrolees’ residential neighbourhoods, the effect on relative differences between neighbourhoods differentiated by social deprivation and ethnicity was small and non-significant. Programme coverage in socially deprived and neighbourhoods with a high proportion of Māori, although sizeable, could be further improved. Despite helping a significant number of people to quit smoking, the PEGS programme is not yet making a noteworthy contribution to reducing health inequalities.

What this paper adds

  • Although many smoking cessation programmes have previously been evaluated there has been less research on whether such programmes can influence inequalities in health.

  • Data involving more than 11 000 patients over 6 years were extracted from a major smoking cessation programme in New Zealand—a world leader in smoking cessation programmes.

  • This paper explored whether the smoking cessation programme had reduced inequalities in health through the coverage of the programme and the quit rates.

  • More enrolees resided in the most deprived areas than the least deprived areas.

  • Just over 20% of smokers enrolled from both the most deprived and the least deprived areas.

  • Quit rates were worse for enrolees from deprived areas compared to affluent areas.

  • Smoking rates in the most deprived neighbourhoods were nearly three times higher than in the most affluent areas and the PEGS cessation programme did not change this.

  • Similarly, although there was substantial enrolment from neighbourhoods with higher proportions of Māori, quit rates among such neighbourhoods were lower. The smoking cessation programme did not change the 13% gap in smoking rates between the neighbourhoods with the highest and lowest proportion of Māori.


We thank Pegasus Health for preparing the PEGS dataset.


View Abstract


  • Funding This research was funded through the GeoHealth Laboratory, a joint venture between the University of Canterbury and the New Zealand Ministry of Health.

  • Competing interests VD was formerly employed by Pegasus Health, which administers PEGS.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.