Introduction
Smoking is the leading risk factor in the burden of disease and mortality in Europe (Lim et al.
2012), and one of the most important public health issues in Europe. Assessing smoking cessation determinants is important because smoking cessation significantly improves people’s health and reduces their mortality risk (Burns
2000; Doll et al.
2004; Taylor et al.
2002). Prior studies have shown that smoking cessation is consistently associated with demographic characteristics, such as sex and age; and that it is influenced by many factors that are amenable to change, including socio-economic factors, social support and the prevalence of smoking-related diseases (e.g. van den Putte et al.
2005; Broms et al.
2004; Margolis
2013; van Gool et al.
2007). Given the potential for changing many determinants of smoking cessation and the benefits of smoking cessation, gaining additional knowledge about the determinants of smoking cessation can aid policy-makers aiming to improve public health by promoting smoking cessation.
Previous studies on the determinants of smoking cessation mainly focused on the role of socio-economic status, social support or psychosocial determinants in smoking cessation (e.g. Osler and Prescott
1998; Janzon et al.
2005), and paid less attention to the associations between key life course transitions and smoking cessation. The studies that investigated the role of key life course transitions found that smoking cessation is associated with both marital transitions (Giordano and Lindstrom
2011; Kriegbaum et al.
2011; Lee et al.
2005; Nystedt
2006) and employment transitions (Blakely et al.
2014; De Vogli and Santinello
2005; Giordano and Lindstrom
2011; Kriegbaum et al.
2011; Lang et al.
2007). These analyses also showed that marital loss and employment loss influence smoking cessation via stress (De Vogli and Santinello
2005; Falba et al.
2005; Johnson and Wu
2002; Thomas et al.
2005), which usually hinders efforts to quit. However, because most of these studies examined all adult age groups combined, little is known about the associations between key life transitions and smoking cessation among older adults. In analysing the determinants of smoking cessation, and specifically the role of marital and employment transitions, it is essential that older adults are studied separately (Jarvis et al.
2013) for a number of reasons. First, although the benefits of smoking cessation are greatest for smokers who quit early in life, quitting later in life is also associated with lower mortality (Doll et al.
2004; Taylor et al.
2002). Second, older adults are of special concern because they tend to need more health care than younger people, and because the number of older people in Europe is growing. Third, older adults are more likely to quit smoking than younger adults (Osler et al.
1999), largely because older adults are more likely than younger adults to suffer from various smoking-related diseases, which is regarded as a motivation to quit smoking (Margolis
2013; van Gool et al.
2007). Fourth, marital and employment loss, particularly widowhood and retirement, are prevalent at older ages. Fifth, the effects of various stressful events on smoking have been shown to differ over the life course (Grotvedt and Stavem
2005; Jarvis et al.
2013; Umberson et al.
2008; Whitson et al.
2006).
Although the importance of studying smoking cessation separately for different age groups has been acknowledged in the literature, the existing evidence regarding the importance of marital and employment losses to smoking cessation at older ages is both scarce and mixed. We are aware of only two studies that specifically examined smoking cessation among middle-aged and older adults in relation to marital transitions: one study that focused on female nurses (Lee et al.
2005), and another that looked at male health professionals (Eng et al.
2005); both in the US. These studies found that becoming widowed and remaining unmarried were associated with lower smoking cessation rates among these older women (Lee et al.
2005), but these transitions were not found to result in significant differences in smoking behaviour among older men (Eng et al.
2005). Although these studies observed some interesting potential gender differences, the narrowness of the selected group and settings affected the generalisability of the results.
In the literature on the association between employment transitions and smoking cessation, there has been mixed evidence regarding two of the main work-related transitions experienced by older adults: retirement and unemployment. For example, a British study found a positive association between retirement and smoking cessation (Lang et al.
2007). However, no such association was found in research conducted in the US (Midanik et al.
1995) or in the Netherlands (Henkens et al.
2008). Being unemployed was found to be negatively associated with smoking cessation in a study among British adults (Giordano and Lindstrom
2011), but the transition from being employed to being unemployed was not shown to be associated with lower chances of smoking cessation in the samples of older British (Falba et al.
2005) or German adults (Schunck and Rogge
2012).
Most of the previous studies on this topic did not distinguish between men and women. Yet the masking of gender differences could affect the overall results (Grotvedt and Stavem
2005). Only a few studies have reported gender-specific results among adults (McKee et al.
2003; Nystedt
2006). On the one hand, gender differences were not found in the association between marital disruption and smoking cessation (Nystedt
2006), or in the association between interpersonal loss and smoking behaviour (McKee et al.
2003). On the other hand, women were found to be less likely than men to quit smoking after financial or health events (McKee et al.
2003).
Various mechanisms are likely to affect the smoking cessation chances of men and women differently. Stress levels linked to marital and employment loss may not be the same for men and women, as men tend to report experiencing more work-related stressful events while women tend to report experiencing more interpersonal stressful events (Kendler et al.
2001). These gender differences may be due to differences between men and women in terms of their role configurations (Ensminger and Celentano
1990), and in their strategies for coping with stressful life course circumstances (Kessler and McLeod
1984). Furthermore, it is also well known that marriage has a more protective effect for men than for women in terms of the adoption of unhealthy behaviours (Schone and Weinick
1998), adverse health outcomes and even mortality (Lillard and Panis
1996). Similarly, the health and mortality outcomes of men have been found to deteriorate more than those of women after widowhood (Moon et al.
2011).
In this study, we seek to add to the existing knowledge on the determinants of smoking cessation by studying the gender-specific associations between marital and employment losses and smoking cessation among older adults (i.e. individuals aged 50 and over) in Europe. We use a longitudinal perspective to examine these transitions over a two-year period (2011–2013).
We hypothesise that potential stressful transitions in life, such as marital and employment losses, could have different effects on the smoking cessation chances of men and women. We also expect that our results for older adults will differ from the findings of previous research that did not focus on this specific age category.
Results
In our study sample of 6460 smokers (3345 men and 3115 women) aged 50 and over in 2011, 613 men (18.3 %) and 556 women (17.8 %) had quit smoking by 2013. Transitions from marriage were more common among women (70 out of 3115; 2.2 %) than among men (49 out of 3345; 1.5 %). By contrast, men experienced slightly more employment transitions (5.4 % retired and 1.5 % became unemployed) than women (4.5 and 1.6 %, respectively) (Table
1).
Table 1
Background characteristics of smokers in 2011 and percentages of smokers who had stopped smoking by 2013.
Source Own estimations based on waves 4 and 5 of SHARE
Marital transitions |
Stayed in a union | 2468 | 19.9 | 1762 | 18.7 | 4230 | 19.4 |
Became widowed or divorced | 49 | 10.2 | 70 | 11.4 | 119 | 10.9 |
Not in a union | 828 | 14.0 | 1283 | 17.1 | 2111 | 15.9 |
Employment transitions |
Stayed employed | 909 | 16.6 | 853 | 15.8 | 1762 | 16.2 |
Became retired | 179 | 15.6 | 139 | 16.5 | 318 | 16.0 |
Stayed retired | 1532 | 21.7 | 1223 | 19.1 | 2755 | 20.6 |
From employed to unemployed | 51 | 11.8 | 49 | 8.2 | 100 | 10.0 |
Sick/disabled or other unemployed | 298 | 10.1 | 199 | 16.6 | 497 | 12.7 |
Stayed homemaker | 0 | – | 220 | 18.2 | 220 | 18.2 |
Others | 376 | 17.3 | 432 | 20.1 | 808 | 18.8 |
Age (baseline) |
50–59 | 1458 | 13.7 | 1597 | 14.9 | 3055 | 14.3 |
60–69 | 1307 | 18.5 | 1061 | 18.8 | 2368 | 18.6 |
70+ | 580 | 29.5 | 457 | 26.0 | 1037 | 28.0 |
Education (ISCED-97) |
0–2: Lower secondary school | 1272 | 19.8 | 1188 | 17.7 | 2460 | 18.8 |
3: Upper secondary school | 1277 | 14.6 | 1215 | 18.4 | 2492 | 16.4 |
4–6: Post-secondary school | 796 | 22.0 | 712 | 17.3 | 1508 | 19.8 |
Newly diagnosed diseases |
Hypertension | 334 | 22.8 | 280 | 21.4 | 614 | 22.1 |
Cholesterol | 289 | 20.8 | 274 | 18.2 | 563 | 19.5 |
Diabetes | 114 | 26.3 | 99 | 21.2 | 213 | 23.9 |
Lung disease | 133 | 23.3 | 132 | 22.0 | 265 | 22.6 |
Heart attack | 170 | 32.4 | 91 | 22.0 | 261 | 28.7 |
Cancer | 86 | 29.1 | 73 | 31.5 | 159 | 30.2 |
Country |
Austria | 358 | 17.9 | 370 | 14.6 | 728 | 16.2 |
Belgium | 324 | 14.5 | 327 | 14.4 | 651 | 14.4 |
Czech Republic | 394 | 16.8 | 472 | 15.3 | 866 | 15.9 |
Denmark | 204 | 17.2 | 200 | 22.5 | 404 | 19.8 |
Estonia | 638 | 15.7 | 394 | 17.3 | 1032 | 16.3 |
France | 274 | 11.7 | 246 | 16.7 | 520 | 14.0 |
Germany | 81 | 22.2 | 73 | 24.7 | 154 | 23.4 |
Italy | 215 | 27.4 | 191 | 29.3 | 406 | 28.3 |
Netherlands | 139 | 22.3 | 192 | 15.1 | 331 | 18.1 |
Slovenia | 145 | 15.2 | 128 | 9.4 | 273 | 12.5 |
Spain | 236 | 24.6 | 146 | 21.9 | 382 | 23.6 |
Sweden | 60 | 41.7 | 109 | 30.3 | 169 | 34.3 |
Switzerland | 277 | 20.2 | 267 | 18.4 | 544 | 19.3 |
Significant and positive ORs of quitting smoking by age were observed among both genders (p value < 0.001) (see Online Supplementary Tables 1, 2). The expected educational gradient in quitting smoking was not found to be significant although the ORs for the highest educational group were larger than the ORs of the reference category: the lowest educated group (men: OR 1.22, 95 % CI 0.94–1.56; women: OR 1.08, 95 % CI 0.82–1.41). The results for disease incidence indicated that all of the newly diagnosed diseases were positively associated with smoking cessation as compared to not having the disease, but only the incidence of the following diseases was statistically significantly associated with higher smoking cessation ORs: hypertension and heart attack for men and cancer for both men and women. For example, the chances of quitting smoking were twice as high for the individuals with cancer incidence relative to the individuals who did not suffer cancer (OR 1.81, 95 % CI 1.10–2.98, and OR 2.06, 95 % CI 1.22–3.46, for men and women, respectively). Finally, relative to the country with lower share of smokers who stopped by 2013 (Poland for men and Slovenia for women) we found that Italy, the Netherlands, Spain and Sweden had statistically higher ORs of smoking cessation for men, and that Denmark, Germany, Spain and Sweden had higher ORs for women.
The effect of marital losses
Becoming widowed or divorced was negatively associated with smoking cessation compared to remaining in a union for both men (OR 0.36, 95 % CI 0.14–0.94) and women (OR 0.46, 95 % CI 0.21–0.99) (Table
2). The ORs for remaining not in a union were greater but also negatively associated with smoking cessation among men (OR 0.72, 95 % CI 0.57–0.90). This association was not significant among women although the ORs pointed in the same direction than for men (OR 0.82, 95 % CI 0.67–1.00). To further examine whether our non-significant results were observed due to a relatively small sample size we have performed identical analysis but grouping together data from previous SHARE waves (1–2, 2–4, 4–5), which indeed confirmed that gender differences remained insignificant among older adults (see Online Supplementary Table 3).
Table 2
Odds ratios (OR) of smoking cessation between 2011 and 2013 the SHARE sample men and women aged 50 and over.
Source Own estimation based on SHARE, waves 4 and 5
Marital transitions |
Stayed in a union (ref.) | | | | |
Became widowed or divorced |
0.359
| (0.137–0.937) |
0.458
| (0.213–0.987) |
Not in a union |
0.718
| (0.570–0.904) | 0.819 | (0.668–1.004) |
Employment transitions |
Stayed (self-)employed (ref.) | | | | |
Became retired | 0.677 | (0.426–1.074) | 0.845 | (0.511–1.396) |
Stayed retired |
0.723
| (0.534–0.979) |
0.728
| (0.530–0.999) |
From (self-)employed to unemployed | 0.624 | (0.257–1.516) | 0.507 | (0.178-1.446) |
Sick/dis or other unemployed |
0.582
| (0.376–0.900) | 1.053 | (0.682–1.626) |
Stayed Homemaker | | | 0.786 | (0.507–1.219) |
Others | 0.885 | (0.628–1.248) | 1.014 | (0.732–1.405) |
The effect of employment losses
The transitions from employed to unemployed or retirement were not statistically significantly associated with smoking cessation for any of the two genders, although the ORs were above 1. Remaining unemployed (category “sick disabled or other unemployed”) was significantly associated with lower ORs among men (OR 0.58, 95 % CI 0.68–0.90) but not among women (OR 1.05, 95 % CI 0.68–1.63). Stayed retired had a significant negative association with smoking cessation among both men (OR 0.72, 95 % CI 0.53–0.98) and women (OR 0.73, 95 % CI 0.53–1.00) as compared to those individuals who remained employed throughout the analysed period. All other categories of the variable employment transitions were not significant, and only those for the categories “Sick/disabled or other unemployed” and “others” were (slightly) above 1 among women. Again these insignificant effects seem not to be the result of the sample size (see Online Supplementary Table 3).
The effect of gender
The association between gender and smoking cessation was not statistically significant (OR 1.08, 95 % CI 0.94–1.24) (Table
3). The results of the interaction terms were not statistically significant at 95 % confidence level, but the interaction between gender and the category “Sick/disabled or other unemployed” was statistically significant at 90 % confidence level (OR 1.80, 95 % CI 0.99–3.27), suggesting women from this category have higher risks of smoking cessation, as compared to men.
Table 3
Odds ratios (OR) of smoking cessation between 2011 and 2013 in the SHARE sample men and women aged 50 and over: gender effects.
Source Own estimation based on SHARE, waves 4 and 5
Marital transitions |
Stayed in a union (ref.) |
Became widowed or divorced |
0.425
| (0.234–0.771) |
0.381
| (0.147–0.985) |
0.426
| (0.235–0.773) |
Not in a union |
0.765
| (0.659–0.888) |
0.712
| (0.567–0.893) |
0.777
| (0.669–0.903) |
Employment transitions |
Stayed (self-)employed (ref.) | | | | | | |
Became retired | 0.744 | (0.530–1.043) | 0.746 | (0.532–1.047) | 0.680 | (0.432–1.069) |
Stayed retired |
0.729
| (0.587–0.906) |
0.732
| (0.589–0.910) |
0.751
| (0.578–0.974) |
From (self-)employed to unemployed | 0.571 | (0.291–1.122) | 0.572 | (0.291–1.123) | 0.617 | (0.255–1.494) |
Sick/dis or other unemployed | 0.764 | (0.563–1.037) | 0.770 | (0.567–1.045) | 0.584 | (0.381–0.895) |
Stayed Homemaker | 0.735 | (0.493–1.096) | 0.747 | (0.500–1.117) | 0.749 | (0.494–1.135) |
Others | 0.935 | (0.741–1.181) | 0.732 | (0.589–0.910) | 0.751 | (0.578–0.974) |
Gender |
Men (ref.) | | | | | | |
Women | 1.079 | (0.939–1.240) | 1.036 | (0.876–1.224) | 1.051 | (0.810–1.362) |
Women * became widowed or divorced | | | 1.223 | (0.362–4.134) | | |
Women * Not in a union | | | 1.138 | (0.842–1.538) | | |
Women * Became retired | | | | | 1.230 | (0.631–2.397) |
Women * Stayed retired | | | | | 0.932 | (0.674–1.288) |
Women * From (self-)employed to unemployed | | | | | 0.832 | (0.211–3.278) |
Women * Sick/dis or other unemployed | | | | | 1.801 | (0.991–3.273) |
Women * Stayed Homemaker | | | | | | |
Women * Others | | | | | 1.108 | (0.706–1.737) |
Acknowledgments
This research was initiated through the European Doctoral School of Demography (2013/14) at the Warsaw School of Economics, which the first author attended with a fellowship from the Centre for Demographic Studies. The research was finalised at the Population Research Centre, University of Groningen, where the first author holds a Ph.D. position within the research project “Smoking, alcohol and obesity - ingredients for improved and robust mortality projections”, which was financed by Dutch Scientific Research (NWO), Grant no. 452-13-001. This paper uses data from SHARE wave 4 release 1.1.1, as of March 28th 2013(DOI:
10.6103/SHARE.w4.111) and SHARE wave 1 and 2 release 2.6.0, as of November 29 2013 (DOI:
10.6103/SHARE.w1.260 and
10.6103/SHARE.w2.260). The SHARE data collection has been primarily funded by the European Commission through the 5th Framework Programme (Project QLK6-CT-2001-00360 in the thematic programme Quality of Life), through the 6th Framework Programme (projects SHARE-I3, RII-CT-2006-062193, COMPARE, CIT5- CT-2005-028857, and SHARELIFE, CIT4-CT-2006-028812) and through the 7th Framework Programme (SHARE-PREP, N° 211909, SHARE-LEAP, N° 227822 and SHARE M4, N° 261982). Additional funding from the U.S. National Institute on Aging (U01 AG09740-13S2, P01 AG005842, P01 AG08291, P30 AG12815, R21 AG025169, Y1-AG-4553-01, IAG BSR06-11 and OGHA 04-064) and the German Ministry of Education and Research as well as from various national sources is gratefully acknowledged (see
www.share-project.org for a full list of funding institutions).