Roll-your-own (RYO) tobacco or crolliesd are hand-rolled cigarettes made of loose tobacco and cigarette papers, and may be smoked with or without a filter. Smoking prevalence among people aged 15 years and over in New Zealand is around 21%. Among them, 61% smoke RYO with or without tailor-made (TM) cigarettes.1 The prevalence of RYO use differs by ethnicity, being more popular among M ori and European/Other than Pacific and Asian smokers.1 Overall, use of RYO is more common in New Zealand than other developed countries such as United States, Canada, Australia, and United Kingdom (e.g. 7-25% of current smokers).2Previous reports suggested that RYO smokers are more sociodemographically disadvantaged and nicotine-dependent.1,3,4 They are therefore in a less likely position to quitting smoking successfully.5-8 The purposes of this study are to confirm the different characteristics of RYO users compared to TM and mixed tobacco users, as well as to investigate whether tobacco type has an additional explanatory power for quit success when sociodemographic and smoking behaviours are controlled for. This study fills a gap of limited research on RYO smokers and is especially relevant to this country with a high prevalence of RYO smoking.Method Data for this analysis comes from a 12-month cohort study which formed part of a wider evaluation programme of the national Quitline, focusing on customer satisfaction, programme effectiveness and cost-effectiveness. The Quitline is the largest smoking cessation provider in New Zealand, and the most common source of quit advice received by regular smokers who have quit or tried to quit.9 As this study is considered as a client satisfaction survey, ethic approval was not required. Between March and September 2002, all Quitline callers were asked to register their interest to participate in an evaluation of the service. As the provision of subsidized nicotine replacement therapy (NRT) is a core component of the service, callers were automatically excluded from the sampling frame if they were ineligible for NRT (due mainly to medical reasons) although they were still provided telephone support by the Quitline. Among those who were eligible for the study, 65% registered their interest in participating although not all of them were recruited (Figure 1). Figure 1. Flowchart for subject recruitment Ethnicity was self-reported during callers initial call to the Quitline, and total responses were recorded. Callers identifying themselves as M ori ethnicity were classified as M ori and otherwise non-M ori. Two different sampling frames were used to randomly select M ori and non-M ori callers from all consenting callers, in order to recruit an adequate size of M ori callers for some focused analyses. In 2002, 22% of overall Quitline callers were M ori, while M ori were over-sampled in this evaluation to comprise almost 50% of the sample. Randomly selected participants were contacted by an external marketing and social research company via the telephone 3 weeks after starting the Quitline programme. Among them, 57% of M ori and 63% of non-M ori callers contacted agreed to participate and completed the initial survey, resulting in a total of 2002 participants. The participants were followed-up via the telephone 6-months and 12-months after they started the Quitline programme. Participants lost to the 6-month survey were not contacted again in the subsequent survey. Among those who completed the initial survey at 3-weeks, 64% completed the 6-month follow-up and 42% completed the entire study. The interviewers were not blind to the type of tobacco the participants smoked, but they were unaware that the quit rate would be analysed by tobacco type. The three surveys covered a range of topics which included sociodemographics, smoking behaviours, and smoking status. Two participants were excluded from this analysis as they smoked neither TM nor RYO (i.e. pipe or cigar). Among the 2000 participants included, 52% smoked TM, 39% smoked RYO and 9% used both TM and RYO regularly. TM cigarette smokers were asked for the number of cigarettes they smoked on a typical day, and RYO smokers were asked for the amount of tobacco (in grams) they smoked in a typical week. To compute a single index for daily cigarette consumption of TM, RYO and mixed tobacco users, it is assumed that each RYO cigarette equates to 0.51 gram,10 and that an equal number of RYO cigarettes were smoked each day during a week (i.e. number of cigarettes smoked per day=grams smoked per week/7/0.51). Smoking status was self-reported at the 6-month and 12-month follow-ups, and quit status was assessed using seven-day abstinence to define quitters, i.e. not having a single puff in the last seven days. We used a conservative measure whereby participants lost to follow-up are assumed to be smoking. This assumption is commonly used in smoking cessation research. The proportions lost to follow-up were similar across the three tobacco types (40% for RYO, 35% for TM and 38% for mixed tobacco users at 6-months, and 61%, 57% and 63% at 12-months). Therefore this approach in dealing with missing cases should not bias the findings. Self-reported smoking status was not bio-chemically verified with cotinine levels. Nonetheless, participants were encouraged to provide truthful responses and reassured that their responses would only be used to assess the effectiveness of the Quitline programme and would not affect the current or future services they received from the Quitline. Sociodemographic and smoking characteristics of RYO, TM, and mixed tobacco smokers were compared using SPSS 15.0. Statistical significance was indicated by the Chi-squared test of proportion. A logistic regression model was fitted for each outcome variable: 7-day point prevalence quit at 6-months and 12-months, using STATA. The model was adjusted for all the significant variables identified in the univariate analysis. Results Callers characteristics by tobacco typeThe proportion who smoked RYO, TM and mixed tobacco varied by sociodemographic factors. Thus RYO use was more common among M ori, male, younger (<40 years), low income, and less educated smokers. RYO use was also more common among those who were married or in a partnership, and those who were not in full time employment. In general, the profile of the mixed group resembled RYO smokers more closely than TM smokers. It is also noteworthy to point out the high proportion of mixed tobacco use in participants aged 15-19 at 17%, compared to only at 4-9% in all older age groups. Table 1. Distribution of RYO use by sociodemographic and smoking characteristics, weighted by age and ethnicity Sociodemographics characteristics RYO (%) TM (%) Mixed (%) Ethnicity (n=2000)** Non-M ori (n=974) 37.9 55.1 6.9 M ori (n=1026) 40.6 49.2 10.2 Gender (n=2000)** Male (n=778) 41.4 50.4 8.2 Female (n=1222) 36.3 56.6 7.1 Age (n=1968)** 15-19 (n=92) 42.9 40.0 17.1 20-29 (n=503) 40.6 50.0 9.4 30-39 (n=600) 45.1 48.6 6.3 40-49 (n=432) 32.1 60.7 7.3 50-59 (n=252) 29.5 65.6 4.9 60 or over (n=89) 37.7 58.0 4.2 Income (n=1857)** <$20,000 (n=534) 46.5 44.2 9.3 $20,000-$40,000 (n=672) 40.8 51.6 7.6 $40,001 or over (n=651) 31.1 62.6 6.3 Employment (n=1996)** Full time (n=1029) 34.8 58.3 6.8 Part time (n=300) 40.6 51.4 8.0 Other (n=667) 44.4 46.8 8.8 Qualification (n=1980)** No secondary school (n=735) 46.0 46.7 7.3 Secondary school (n=506) 35.5 56.8 7.6 National certificate/trade certificate (n=310) 41.8 46.8 11.4 Tertiary (n=429) 29.4 65.5 5.2 Marital Status (n=1994)** Married/living with partner (n=1118) 40.6 52.5 6.9 Never married (n=497) 36.0 57.5 6.5 Other (n=379) 34.9 55.0 10.1 * denotes p < 0.05 ** denotes p < 0.01 Note: Data are weighted to represent the profile of overall Quitline callers who were eligible for the study In comparison to TM and mixed tobacco smokers, RYO smokers had a shorter smoking history but the relationship between tobacco type and nicotine dependency was not clear. Specifically, while the first cigarette measure indicated that RYO smokers were more nicotine dependent, the number of cigarettes smoked per day showed the contrary. This inconsistency may be due to the method used in converting the weight of loose tobacco into cigarettes, which is discussed in further detail in the limitations section. Table 2. Smoking characteristics by tobacco use, weighted by age and ethnicity Smoking characteristics RYO (%) TM (%) Mixed (%) Years Smoked (n=1970)* <5 years (n=209) 7.8 11.2 15.4 5-15 years (n=606) 32.7 29.7 36.0 16 years or over (n=1155) 59.5 59.1 48.6 First Cigarette (n=1971)* <30 minutes after waking (n=1481) 78.3 70.8 75.9 30 minutes or longer after waking (n=490) 21.7 29.2 24.1 Number of cigarettes per day (n=1981)** <10 (n=791) 85.6 10.2 39.2 10-19 (n=492) 13.0 33.0 25.5 20 or over (n=662) 1.5 56.8 35.3 * denotes p < 0.05 ** denotes p < 0.01 Distribution of tobacco type is also compared by the degree of Quitline support received by callers. Full intervention is defined as those who had spoken to a Quitline Advisor at least twice, had read some of the Quitline quitting resources, and had redeemed at least one voucher for subsidised NRT. Remaining clients belonged to the partial intervention category. The difference in distribution of tobacco type between these two groups was not statistically significant. Table 3. Distribution of tobacco type by degree of Quitline support Quitline support (n=2000) RYO (%) TM (%) Mixed (%) Full intervention (n=856) 39.9 53.2 6.9 Partial intervention (n=1144) 37.4 54.6 8.0 Quit outcomes by tobacco typeTwo logistic regression models were fitted on two outcome variables, being 7-day abstinence at 6- and 12-months. Apart from the demographic and smoking characteristics mentioned in Tables 1 and 2, the degree of Quitline service provided to clients is likely to have an impact on quit success. Thus this additional variable was included in the model. Considering the results from the logistic regression models, the type of tobacco Quitline callers smoked prior to their quit attempt was not a contributing factor to their likelihood of quitting 6- and 12-months after registering with the cessation programme (Table 4 & 5). Instead, a numb

Roll-your-own (RYO) tobacco or crolliesd are hand-rolled cigarettes made of loose tobacco and cigarette papers, and may be smoked with or without a filter. Smoking prevalence among people aged 15 years and over in New Zealand is around 21%. Among them, 61% smoke RYO with or without tailor-made (TM) cigarettes.1 The prevalence of RYO use differs by ethnicity, being more popular among M ori and European/Other than Pacific and Asian smokers.1 Overall, use of RYO is more common in New Zealand than other developed countries such as United States, Canada, Australia, and United Kingdom (e.g. 7-25% of current smokers).2Previous reports suggested that RYO smokers are more sociodemographically disadvantaged and nicotine-dependent.1,3,4 They are therefore in a less likely position to quitting smoking successfully.5-8 The purposes of this study are to confirm the different characteristics of RYO users compared to TM and mixed tobacco users, as well as to investigate whether tobacco type has an additional explanatory power for quit success when sociodemographic and smoking behaviours are controlled for. This study fills a gap of limited research on RYO smokers and is especially relevant to this country with a high prevalence of RYO smoking.Method Data for this analysis comes from a 12-month cohort study which formed part of a wider evaluation programme of the national Quitline, focusing on customer satisfaction, programme effectiveness and cost-effectiveness. The Quitline is the largest smoking cessation provider in New Zealand, and the most common source of quit advice received by regular smokers who have quit or tried to quit.9 As this study is considered as a client satisfaction survey, ethic approval was not required. Between March and September 2002, all Quitline callers were asked to register their interest to participate in an evaluation of the service. As the provision of subsidized nicotine replacement therapy (NRT) is a core component of the service, callers were automatically excluded from the sampling frame if they were ineligible for NRT (due mainly to medical reasons) although they were still provided telephone support by the Quitline. Among those who were eligible for the study, 65% registered their interest in participating although not all of them were recruited (Figure 1). Figure 1. Flowchart for subject recruitment Ethnicity was self-reported during callers initial call to the Quitline, and total responses were recorded. Callers identifying themselves as M ori ethnicity were classified as M ori and otherwise non-M ori. Two different sampling frames were used to randomly select M ori and non-M ori callers from all consenting callers, in order to recruit an adequate size of M ori callers for some focused analyses. In 2002, 22% of overall Quitline callers were M ori, while M ori were over-sampled in this evaluation to comprise almost 50% of the sample. Randomly selected participants were contacted by an external marketing and social research company via the telephone 3 weeks after starting the Quitline programme. Among them, 57% of M ori and 63% of non-M ori callers contacted agreed to participate and completed the initial survey, resulting in a total of 2002 participants. The participants were followed-up via the telephone 6-months and 12-months after they started the Quitline programme. Participants lost to the 6-month survey were not contacted again in the subsequent survey. Among those who completed the initial survey at 3-weeks, 64% completed the 6-month follow-up and 42% completed the entire study. The interviewers were not blind to the type of tobacco the participants smoked, but they were unaware that the quit rate would be analysed by tobacco type. The three surveys covered a range of topics which included sociodemographics, smoking behaviours, and smoking status. Two participants were excluded from this analysis as they smoked neither TM nor RYO (i.e. pipe or cigar). Among the 2000 participants included, 52% smoked TM, 39% smoked RYO and 9% used both TM and RYO regularly. TM cigarette smokers were asked for the number of cigarettes they smoked on a typical day, and RYO smokers were asked for the amount of tobacco (in grams) they smoked in a typical week. To compute a single index for daily cigarette consumption of TM, RYO and mixed tobacco users, it is assumed that each RYO cigarette equates to 0.51 gram,10 and that an equal number of RYO cigarettes were smoked each day during a week (i.e. number of cigarettes smoked per day=grams smoked per week/7/0.51). Smoking status was self-reported at the 6-month and 12-month follow-ups, and quit status was assessed using seven-day abstinence to define quitters, i.e. not having a single puff in the last seven days. We used a conservative measure whereby participants lost to follow-up are assumed to be smoking. This assumption is commonly used in smoking cessation research. The proportions lost to follow-up were similar across the three tobacco types (40% for RYO, 35% for TM and 38% for mixed tobacco users at 6-months, and 61%, 57% and 63% at 12-months). Therefore this approach in dealing with missing cases should not bias the findings. Self-reported smoking status was not bio-chemically verified with cotinine levels. Nonetheless, participants were encouraged to provide truthful responses and reassured that their responses would only be used to assess the effectiveness of the Quitline programme and would not affect the current or future services they received from the Quitline. Sociodemographic and smoking characteristics of RYO, TM, and mixed tobacco smokers were compared using SPSS 15.0. Statistical significance was indicated by the Chi-squared test of proportion. A logistic regression model was fitted for each outcome variable: 7-day point prevalence quit at 6-months and 12-months, using STATA. The model was adjusted for all the significant variables identified in the univariate analysis. Results Callers characteristics by tobacco typeThe proportion who smoked RYO, TM and mixed tobacco varied by sociodemographic factors. Thus RYO use was more common among M ori, male, younger (<40 years), low income, and less educated smokers. RYO use was also more common among those who were married or in a partnership, and those who were not in full time employment. In general, the profile of the mixed group resembled RYO smokers more closely than TM smokers. It is also noteworthy to point out the high proportion of mixed tobacco use in participants aged 15-19 at 17%, compared to only at 4-9% in all older age groups. Table 1. Distribution of RYO use by sociodemographic and smoking characteristics, weighted by age and ethnicity Sociodemographics characteristics RYO (%) TM (%) Mixed (%) Ethnicity (n=2000)** Non-M ori (n=974) 37.9 55.1 6.9 M ori (n=1026) 40.6 49.2 10.2 Gender (n=2000)** Male (n=778) 41.4 50.4 8.2 Female (n=1222) 36.3 56.6 7.1 Age (n=1968)** 15-19 (n=92) 42.9 40.0 17.1 20-29 (n=503) 40.6 50.0 9.4 30-39 (n=600) 45.1 48.6 6.3 40-49 (n=432) 32.1 60.7 7.3 50-59 (n=252) 29.5 65.6 4.9 60 or over (n=89) 37.7 58.0 4.2 Income (n=1857)** <$20,000 (n=534) 46.5 44.2 9.3 $20,000-$40,000 (n=672) 40.8 51.6 7.6 $40,001 or over (n=651) 31.1 62.6 6.3 Employment (n=1996)** Full time (n=1029) 34.8 58.3 6.8 Part time (n=300) 40.6 51.4 8.0 Other (n=667) 44.4 46.8 8.8 Qualification (n=1980)** No secondary school (n=735) 46.0 46.7 7.3 Secondary school (n=506) 35.5 56.8 7.6 National certificate/trade certificate (n=310) 41.8 46.8 11.4 Tertiary (n=429) 29.4 65.5 5.2 Marital Status (n=1994)** Married/living with partner (n=1118) 40.6 52.5 6.9 Never married (n=497) 36.0 57.5 6.5 Other (n=379) 34.9 55.0 10.1 * denotes p < 0.05 ** denotes p < 0.01 Note: Data are weighted to represent the profile of overall Quitline callers who were eligible for the study In comparison to TM and mixed tobacco smokers, RYO smokers had a shorter smoking history but the relationship between tobacco type and nicotine dependency was not clear. Specifically, while the first cigarette measure indicated that RYO smokers were more nicotine dependent, the number of cigarettes smoked per day showed the contrary. This inconsistency may be due to the method used in converting the weight of loose tobacco into cigarettes, which is discussed in further detail in the limitations section. Table 2. Smoking characteristics by tobacco use, weighted by age and ethnicity Smoking characteristics RYO (%) TM (%) Mixed (%) Years Smoked (n=1970)* <5 years (n=209) 7.8 11.2 15.4 5-15 years (n=606) 32.7 29.7 36.0 16 years or over (n=1155) 59.5 59.1 48.6 First Cigarette (n=1971)* <30 minutes after waking (n=1481) 78.3 70.8 75.9 30 minutes or longer after waking (n=490) 21.7 29.2 24.1 Number of cigarettes per day (n=1981)** <10 (n=791) 85.6 10.2 39.2 10-19 (n=492) 13.0 33.0 25.5 20 or over (n=662) 1.5 56.8 35.3 * denotes p < 0.05 ** denotes p < 0.01 Distribution of tobacco type is also compared by the degree of Quitline support received by callers. Full intervention is defined as those who had spoken to a Quitline Advisor at least twice, had read some of the Quitline quitting resources, and had redeemed at least one voucher for subsidised NRT. Remaining clients belonged to the partial intervention category. The difference in distribution of tobacco type between these two groups was not statistically significant. Table 3. Distribution of tobacco type by degree of Quitline support Quitline support (n=2000) RYO (%) TM (%) Mixed (%) Full intervention (n=856) 39.9 53.2 6.9 Partial intervention (n=1144) 37.4 54.6 8.0 Quit outcomes by tobacco typeTwo logistic regression models were fitted on two outcome variables, being 7-day abstinence at 6- and 12-months. Apart from the demographic and smoking characteristics mentioned in Tables 1 and 2, the degree of Quitline service provided to clients is likely to have an impact on quit success. Thus this additional variable was included in the model. Considering the results from the logistic regression models, the type of tobacco Quitline callers smoked prior to their quit attempt was not a contributing factor to their likelihood of quitting 6- and 12-months after registering with the cessation programme (Table 4 & 5). Instead, a numb