Cigarette smoking is the leading cause of fires that result in death worldwide1 and it is thought that cigarettes are responsible for around 25% of all residential fire deaths in the US.2^–5 Recent research suggests a strong ecological relationship between smoking and residential fire death in the US.6 Smoking-related fires are almost always caused by cigarettes, rather than other modes of tobacco smoking such as cigars and pipes.7 8 The materials most frequently set alight by smouldering cigarettes are bedding, mattresses and upholstered furniture9^–11 and ignition often occurs in the context of smoking in bed.9 11^–14 Fires caused by cigarettes are frequently lethal. In a case–control study of fatal and non-fatal fires in North Carolina, USA, the odds ratio for fatal fires caused by smoking vs all other causes was 7.7.15 In another study from Washington, USA, smoking materials were the ignition source in 11 of 17 residential fires where fatalities occurred and around 20% (29 of 122) of residential fires where non-fatal injury occurred.16

In 2004, there were approximately 10 000 emergency department visits in the US due to injuries sustained from lit tobacco products.17 About half of these injuries were thermal burns, including burns from direct contact with lit tobacco and burns sustained in fires caused by smoking materials. Around 15% of hospital burns admissions in the US have historically been due to fires caused by smoking.18 19 Individuals who receive serious burn injuries in fires caused by cigarettes are likely to have a worse prognosis than those who experience burns from other causes, with substantially longer hospital stays and much greater mortality.4

Beyond official records of fires and fire deaths where cigarettes were the responsible heat source, little attention has been given to examining the occurrence of cigarette fire and burn events among smokers or the characteristics of the smokers who were involved in these events. In the present work, we report on the lifetime prevalence of cigarette-caused fires and burns in a group of regular smokers in New Zealand. By lifetime prevalence, we refer to the proportion of individuals within a population who have experienced an event or condition at any time in their lives up to the point of sampling.20^–22 We also report on the characteristics of smokers that were associated with an increased risk of experiencing fires and burn injuries.

METHODS

We used baseline data from 1097 of 1100 participants in the Pre-Quitting Nicotine to Increase Quitting (PQNIQ) trial, a randomised controlled trial of pre-cessation nicotine replacement therapy compared with usual care that recruited callers to the New Zealand Quitline who were aged ⩾18 years with high nicotine dependency (⩾10 cigarettes per day, first cigarette within 30 min of waking) between March 2006 and May 2007. The baseline questionnaire included questions about participants’ sociodemographic characteristics and past experiences of cigarette-caused fires and burns, including questions about the frequency of these events and characteristics of the most serious events. Ethics approval for the trial was granted by the Northern Region Ethics Committee. All statistical analyses were conducted using SPSS V. 15.0 (SPSS Inc, Chicago, Illinois, USA) and Microsoft Excel (Microsoft, Redmond, Washington, USA). Associations between smoker characteristics and cigarette fires and burn injuries were examined using logistic regression, with variables included in the final models only if significant on stepwise regression. Because participants were able to self-identify with one or more ethnic category (known as total response ethnicity, the approach used in the New Zealand population census) the sum of proportions for ethnicity exceeds 100%. In our analyses, we simply compared Māori (indigenous New Zealanders, who comprise 15% of the national population and among whom 42% of adults are smokers)23 with non-Māori. Tests of statistical significance used a 5% cutoff. Where appropriate, estimates are presented with 95% CIs.

RESULTS

Participant characteristics

Key characteristics of the 1097 participants are summarised in table 1. The mean age of participants was 40 years and 40% were males. Almost 28% of study participants identified as Māori. The mean duration of smoking was 22 years. Just under half of the sample reported personal annual incomes of $20 000 or less, below the national average of NZ$30 000 (New Zealand dollars).

Prevalence and characteristics of fires and burns

A total of 75 (6.8%, 95% CI 5.3% to 8.3%) participants had previously experienced ⩾1 fires caused by cigarettes. At least 96 separate cigarette-caused fires were reported, 8.8 fire events per 100 smokers. A total of 13 participants had experienced more than 1 cigarette-caused fire. Lifetime prevalence of cigarette fires varied by age group (table 2) and contrary to expectations was highest in the youngest age group. Participants who reported previous cigarette fires described the most serious of these fires: 50 occurred indoors in homes, 20 occurred outdoors and 5 were in vehicles. Emergency services were called to 22 (29.3%) fires. Table 3 shows the situations that led to fire ignition in each of the 75 most serious cases.

Factory-made (manufactured) cigarettes were responsible for the majority of the 75 most serious fires, causing more than 10 times as many of these fires as hand-rolled (“roll-your-own”) cigarettes. Of these fires, 61 (81.3%) were caused by factory-made cigarettes, while hand-rolled cigarettes were responsible for only 6 fires (8.0%). The type of cigarette responsible was unknown in eight fires (10.7%).

A total of 658 participants (60.0%, 95% CI 57.1% to 62.9%) had past experience of ⩾1 burns caused by cigarettes. Analysis of lifetime prevalence of cigarette burn injuries by age group shows that burns, like fires, were more common in younger smokers (table 4). Many of these individuals had experienced more than one burn (table 5).

Participants were asked to provide details of the most serious of the burns they had experienced. Of 658 burns, 57 (8.7%, 95% CI 6.5% to 10.7%) required medical care. Many burns were sustained while cigarettes were held in the hand, or after a cigarette was dropped, thrown or left unattended (table 3). As with fires, factory-made cigarettes were responsible for the majority of cigarette-caused burns, causing 2.7 times the number of burns than hand-rolled cigarettes in the 658 most serious injuries for which details were available. Of these injuries, factory-made cigarettes were responsible for 470 (71.4%), whereas hand-rolled cigarettes were responsible for only 172 (26.1%). Cigarette type was unknown for 16 burns (2.4%). In 16 of the 658 burn incidents, the burn was related to a cigarette-caused fire, of which 14 (87.5%) were caused by factory-made cigarettes, while only one was caused by a hand-rolled cigarette; in the other case the type was unknown.

Characteristics of smokers associated with fires and burns

Table 6 shows the statistical association between male sex (p = 0.02), Māori ethnicity (p = 0.01) and risk of cigarette fires. None of the other 11 variables assessed (age, daily cigarette exposure, education level, employment status, age of smoking initiation, duration of continuous smoking, marital status, smoking status of partner, medication use, cigarette type usually smoked at the time of interview and annual personal income) were associated with cigarette fires. In the final model for burns risk, seven variables were associated with cigarette-caused burns (table 7).

The risk of burn injuries reduced with increasing age (p<0.001) and an association was found between male sex and risk of burn injuries (p<0.001). Other characteristics associated with risk of burn injury included: Higher level of education (p = 0.02), being unmarried (p = 0.03), having a partner who was a current or former smoker (p = 0.01) and having an annual personal income of $20 000 or more (p = 0.01). In a separate model for burns sufficiently serious to require medical attention, the strongest associations were with male sex (odds ratio (OR) 1.98, 95% CI 1.10 to 3.55), Māori ethnicity (OR 1.85, 95% CI 1.03 to 3.31) and higher education level (OR 2.01, 95% CI 1.12 to 3.61).

DISCUSSION

This is the first study to describe the lifetime prevalence of cigarette burn injuries and only the second to document the lifetime prevalence of cigarette fires in a population of regular smokers. Cigarette-caused fires and burns appear to be commonplace among New Zealand smokers. Burns were a remarkably common experience: 60% reported at least one previous cigarette burn and almost 50% had experienced two or more burns. Almost 9% of those with past burns reported sustaining burn injuries serious enough to warrant medical attention and 16 reported burns directly related to cigarette-caused fires. If our finding that almost 7% of participants had experienced at least one fire caused by cigarettes is assumed to be a valid reflection of lifetime prevalence in smokers, and given a crude prevalence of current smokers (aged ⩾15 years) in New Zealand in 2006/07 of 19.9%23 from a total population (⩾15 years) of 3.2 million,24 we estimate that 33 000 to 52 000 adult New Zealand smokers have experienced cigarette-caused fires. Thus, deaths from cigarette fires almost certainly under-represent a much larger burden of cigarette-related injury. Our finding differs substantially from that of O’Connor et al in which only 1.7% of 596 Canadian smokers described past experience of cigarette-caused fires.25 However, we asked participants about any past experience of cigarette fires whereas in the Canadian study the question addressed only fires in the home. We did not collect data on the effects on health, but for 22 participants (2%) the fires were of sufficient severity that emergency services were called.

A particular strength of our study is its large size, conferring a high degree of precision around the estimates of lifetime prevalence and good statistical power to detect smoker characteristics associated with increased risk of fires and burns. Participants were smokers willing to quit smoking and to enrol in a clinical trial, meaning the study would not have been entirely representative of New Zealand smokers. The study also contained a greater proportion of females than was found among Quitline callers and New Zealand smokers.26 27 However, smoking characteristics and distributions of age and ethnicity in this sample were otherwise similar to those of Quitline callers and New Zealand smokers in general.26^–29

Male smokers and those identifying as Māori had greater risk of cigarette fires in the regression model. In a US case–control study, males and people of “non-white race” were found to have greater risk of cigarette fires.8 30 Males31 and Māori32 33 experience a greater burden of fire injury in general compared with females and non-Māori respectively. Ethnic composition and the “meaning” and impact of ethnicity on risk behaviours and life chances differ between the US and New Zealand, and ethnicity in itself is seldom an adequate explanatory concept for differences in outcomes.34 Further qualitative research is required to gain greater understanding of the drivers behind the relationships found in our analysis. Our finding of an inverse relationship between age and risk of cigarette burn injury is counter to the expectation that people who are older and who have smoked for longer would have had more time to experience burns. Similar proportions of younger and older smokers in the sample used factory-made cigarettes regularly, while use of hand-rolled cigarettes was more common among young people. Younger smokers in the sample also had slightly lower daily exposure to cigarettes than older smokers. It may be that younger smokers had better recall of burn events, or perhaps they had different risk profiles to older smokers that put them at greater risk of burns and that were not accounted for in the analysis. For example, age-related differences in alcohol use, an important factor in cigarette-caused fires10 17 35^–37 could be relevant. However, as no data on exposure to alcohol or illicit drugs were collected in the baseline questionnaire, we were unable to pursue this line of enquiry.

Measurement of lifetime prevalence provides valuable insight into the experience of cigarette fire and burn events among smokers and shows that fires and burns are common events in this group. However, lifetime prevalence under-estimated the total number of cigarette-caused fires and burns, as many smokers had experienced more than one episode of each event. Level of exposure to lit cigarettes is also dependent on the age structure and duration of smoking in the sample population.

Those with experience of burns or fires may have recalled answers to questions around cigarette consumption and past smoking behaviour differently to those without such experiences. However, demographic characteristics and current smoking patterns are less likely to be affected by recall bias. Furthermore, only data on the most serious fire and burn events were analysed in order to focus on events with the most accurate recall.

Until such time as cigarettes are eliminated, greater effort is needed to minimise the risk of fires and burn injuries among those who smoke, their families and neighbours. While there are multiple ways to limit cigarette ignition propensity,2 4 13 38^–40 the use of cigarette wrapping paper without citrate impregnation is an effective, simple and practical method to achieve this goal.41 Cigarette papers used for hand-rolling are typically citrate-free and have significantly superior self-extinguishing characteristics to factory-made cigarettes (which have substantial amounts of citrate added to the wrapping paper), after adjusting for potential confounders such as tobacco type and density.41 In our study, factory-made cigarettes were responsible for more than 10 times as many fires as hand-rolled cigarettes and more than 2.5 times as many burns, despite the fact that similar numbers of participants in the study sample smoked each type of cigarette. Without data on past smoking habits and other variables such as tobacco density in hand-rolled cigarettes it is difficult to draw conclusions from these stark differences. However, legislation mandating the sale of reduced ignition propensity cigarettes is in force in several states in the US42 and Canada,43 44 and New Zealanders support the introduction of self-extinguishing cigarettes.41 45 The results of our study support legislation banning the sale of cigarettes designed to keep burning when unattended. Self-extinguishing cigarettes, such as those manufactured without citrate added to the wrapping paper, provide a safer alternative.

In conclusion, fires and burns caused by cigarettes are sufficiently common and serious to warrant greater attention and action to reduce mortality and potentially under-estimated morbidity.