The Be Our Ally Beat Smoking (BOABS) study, a randomised controlled trial of an intensive smoking cessation intervention in a remote aboriginal Australian health care setting

In 2008, the age-standardised prevalence of current smoking among Indigenous people was more than double that among other Australians (49.8% compared with 20.5% of those aged 18 years and over) [1]. Much of the health disparity between Indigenous and non-Indigenous Australians is attributable to conditions that are either due to or exacerbated by tobacco smoking. Ischaemic heart disease and type 2 diabetes contributes 14% and 12%, respectively, to the Indigenous health gap [2]. Indigenous men and women 35 – 54 years of age die from ischaemic heart disease at 7.2 and 16.6 times, respectively, the rate in the non-Indigenous Australian population [3]. In addition similarly aged Indigenous men and women die from chronic lung disease at a rate 9.7 and 13.9 times greater respectively, and from lung cancer at a rate twice that of non-Indigenous Australians [3]. Smoking substantially increases the risk of both macrovascular and microvascular complications of diabetes and may have a role in the development of type 2 diabetes [4]. Stopping smoking has been described as the single most important individual change Indigenous Australian smokers could make to improve their health.

While many smokers quit unassisted, for others helping them to become and remain non-smokers is difficult. Randomised controlled trials (RCTs) aimed at the general smoking population demonstrate interventions with success rate of at most 25% at 12 months and more typically substantially lower. In the general population smoking cessation interventions have an odds ratio (OR) of between 1.42 and 2.17, depending on the type of intervention used [5]. There have been several smoking cessation interventions conducted amongst disadvantaged and Indigenous groups [6, 7]. However, there is a paucity of published RCTs [7].

Smoking cessation programs are a major priority in Indigenous Australian health. A range of strategies have been used to encourage Indigenous Australians to quit smoking however there have been few good quality studies that show which approaches work best. The only published Indigenous smoking cessation RCT investigating the benefit of a personal support intervention with validated smoking cessation showed a doubling of the smoking cessation rate after providing extra support (7% v 3%), although this improvement was not statistically significant [8].

While smoking cessation rates in primary health care settings are likely to be lower than in settings where patients have specifically sought help, particularly when health care providers approach patients to quit [8, 9], more evidence of effective strategies are needed if effective policy is to be developed and implemented. Aboriginal Community Controlled Health Services (ACCHS) provide an appropriate setting for such interventions as they provide a culturally safe environment that has been shown to be successful in delivering interventions in other health care areas [10–12].

The Be Our Ally Beat Smoking (BOABS) Study aimed to test the efficacy of a culturally appropriate multidimensional intensive smoking cessation intervention provided by Aboriginal researchers in helping Aboriginal and Torres Strait Islander people to become and remain non-smokers at 12 months.

Participant recruitment occurred from January 2009 to May 2011 and follow-up occurred from February 2010 to July 2012. Common reasons that potential participants were unwilling to consent were a lack of willingness to be involved in a research project or not being interested in quitting smoking. Of the 168 participants who were randomised to the study, 163 continued in the study (Figure 1) and their baseline characteristics, and smoking and related risk behaviours were generally similar between the intervention and usual care groups (Tables 1 and 2). The small difference in mean age (38.3 v 41.9 years) between the intervention and usual care group was statistically but not clinically significant. Participants in both groups reported high levels of alcohol, cannabis and other illicit drug use (Table 2).

An average of 7 of the 12 planned formal smoking cessation sessions were provided by the Aboriginal researchers to participants in the intervention group. Additional informal contact occurred throughout the study period (e.g. Aboriginal researchers meeting participants while out in the community and stopping to discuss how they are going with their quit smoking attempt). Twenty two (40%) participants in the intervention group had documented action plans. In addition to the content delivered by Aboriginal researchers during smoking cessation counselling sessions (see Methods), they also provided service mapping and linkages to other services. Individual case conferences with clinic staff did not happen as frequently as planned and only occurred as required. Few participants attended group sessions.

Clinic staff were reported to have provided the same level of routine care relating to smoking cessation to both groups: 23% of participants reported that they spoke to clinic staff about smoking during the study. Overall participants reported that clinic staff provided advice on risks of smoking (75%), recommended quitting smoking (71%), recommended medication to assist with quitting (54%), discussed passive smoking (61%), provided practical advice on quitting (50%), provided written advice on quitting (46%), arranged for a follow-up appointment (46%), and set a quit date (32%). There was no difference between groups. Participants were predominantly offered NRT Patches (39%) or varenicline (14%).

One hundred and forty-four (88%) participants were followed-up after 12 months, with 92 (56%) providing urine samples for validation of smoking cessation. Loss to follow-up was due to participants moving home and Aboriginal researchers not knowing where they moved to so they were unable to contact them.

For four participants there was a discrepancy between self-report smoking status and cotinine levels. Thirteen participants reported quitting smoking and 13 had a cotinine level consistent with not smoking. Two smokers (usual care group) claimed they had quit smoking at the final follow-up, however both reported quitting within the last 24 hours and had high cotinine levels. Two participants (intervention group) who self-reported continuing smoking had cotinine levels consistent with not smoking (below the level of detection of 20 ng/mL). One of these participants had the urine sample collected three months after the final questionnaire was administered and reported not smoking at the time of urine collection, no reason was identified for the other discrepancy as the participant was clear she was still smoking. As participants were only classified as not smoking if they satisfied both criteria, these four participants were classified as still smoking at final follow-up.

Participants in the intervention group quit smoking sooner than those in the usual care group: all intervention group participants reported quitting at least two months before the final follow-up, as against only three of five in the usual care group. There was no difference in the median time between enrolling and final follow-up for both groups (13 (IQR 12–15) months).

The proportions quitting smoking at final follow-up are shown in Table 3. Based on cotinine results alone at final follow-up the intervention group had a significantly higher proportion who quit smoking than the usual care group. However, based on the combination of self-report and cotinine level at final follow-up, the final results (2.4; 95% CI, 0.8-7.4, p = 0.131) were not statistically significant. A meta-analysis of the BOABS Study and the only other published RCT of a personal support intervention with validated smoking cessation [8] demonstrated a statistically significant higher smoking cessation rate for participants in intervention compared to usual care groups (Figure 2).

Outcome	Usual care group	Intervention group	P*	RR (95% CI)†
Urinary cotinine <50 ng/mL	5/64 (8%)	8/28 (29%)	0.009	3.7 (1.3–10.2)
Self-report quitting and urinary cotinine <50 ng/mL for participants with complete outcome data	5/95 (5%)	6/49 (12%)	0.135	2.3 (0.7–7.2)
Self-report quitting and urinary cotinine <50 ng/mL assuming those lost to follow-up were still smoking	5/108 (5%)	6/55 (11%)	0.131	2.4 (0.8–7.4)

Data are number (%) based on the study arm to which participants were originally assigned;

*Fishers exact;

^†Compared with usual care group;

No-one who had been recently incarcerated, chewed tobacco, or drank alcohol daily quit smoking. Age and gender were not independently associated with quitting. There was no relationship between the easily quantifiable measures of the intervention (documented action plan, or plan to stop smoking, setting a quit date and number of smoking cessation counselling sessions) and success in stopping smoking at final follow-up. We could not measure the quality of the relationships between the researchers and participants, which was likely to be important.

The current study did not demonstrate a statistically significant benefit from the BOABS intervention and due to difficulties recruiting participants did not have the power to do so. Pooling our data in a meta-analysis of the only other reported Indigenous smoking cessation RCT using a personal support intervention and validated smoking cessation, demonstrated a statistically significant increase in quitting (OR 2.4, 95% CI 1.01-5.5). This is similar to smoking cessation interventions in non-Indigenous settings (OR 1.4-2.2) [5]. Doubling the quit rate is a clinically important improvement and based on these pooled data one-on-one intensive interventions delivered by, and provided to, Indigenous Australians through a primary healthcare setting are more effective than usual care in encouraging people to stop smoking.

Both studies while more intensive than usual care were less intensive than originally planned. The participants in the BOABS intervention group received an average of seven of 12 smoking cessation sessions over 12 months and 40% had documented action plans for helping them to quit smoking. In contrast only 23% of participants in the usual care group reported discussing smoking cessation with clinic staff. In the non-Indigenous setting a systematic review of physician advice found increasing the intensity of the intervention showed a small advantage over minimal advice (RR 1.37; 95% CI 1.2-1.6) [27]. The level of intensity in these studies varied from brief intervention to repeated advice to quit as an inpatient with follow up in a special clinic.

Increasing the intensity of the intervention may have had an impact on the quit rate in the intervention groups of both the Eades study and BOABS. Although the Australian Indigenous smoking prevalence is high, there was only a 3% quit rate in a local program delivered at one of the sites prior to the BOABS intervention (personal communication CN). The quality of the relationships between the BOABS Aboriginal researchers and participants may well have had an impact. A qualitative evaluation of the observations and notes from the BOABS Study is required to further elucidate this.

There was repeated clinic and research staff training about roles and responsibilities to minimise contamination between groups. The process of regular follow up, the core of the intervention, appeared to be relatively free from contamination. Nonetheless even if some contamination did occur it was likely to reduce the difference between the intervention and usual care groups.

A limitation of both this study and the study by Eades and others was the lower than expected numbers of participants and an attendant lack of statistical power. While the BOABS Study had a high follow-up rate for a relatively long-term community-based RCT (88% actual follow-up v 90% planned), we were unable to recruit the planned number of participants. In this study there were a number of reasons why people chose not to enrol even though they wanted assistance to help quit smoking and most of these related to the consent and/or the randomisation processes. Therefore, we believe that more people will be interested in joining a program than a research study. It is unclear whether the people who did not want to participate in research would be more or less likely to quit than those who did participate.

Despite more than doubling the time for recruitment only about half of the expected number of participants were recruited. While we have demonstrated it is possible to undertake high quality RCTs in Aboriginal primary health care settings [28] the difficulty of enrolling participants in future studies should not be underestimated and needs to be carefully considered in planning.

Solutions to the problems of insufficient numbers and contamination include using more sites in multicentre trials and randomising at the cluster level. Individually randomised multicentre drug trials have been demonstrated to be possible in Aboriginal and Torres Strait Islander health [28]. Cluster randomised multicentre trials may be another solution for more complex non-placebo controllable interventions such as BOABS, however the risks of: 1) heterogeneity between sites; 2) the intervention only working in some locations; 3) added complexity of both design and analysis; 4) a substantially larger sample size required to obtain the same statistical power; 5) post-randomisation selection bias if randomisation occurs prior to participant recruitment; and 6) a potential slowdown of recruitment in the clusters that are to receive the less interesting treatment such as “usual care” [29–31], remain.

In addition to the risks listed above, RCTs are not well placed to take into account the need for operational changes of health service providers to deliver an intervention such as BOABS [30]. In the BOABS Study, because it was an RCT and by definition could not be delivered as part of usual health care, the intervention was delivered by individual workers and by design was not intended to create substantial change to the actions of other health care staff or the day to day operation of the clinics. We believe this had a negative effect on both recruitment and the intervention.

Although RCTs are generally considered the gold standard in research, it has been argued that in complex public health interventions other methodologies should be pursued [30]. Similarly RCTs may not be the best design for complex clinical individualised interventions such as BOABS. Alternative methodologies include plausibility (observational design with a comparison group) and adequacy (process indicators and outcome data are used to suggest if the intervention is having an important effect) evaluations [30]. While these methodologies provide lower levels of evidence, recruitment would be expected to be higher as it is not restricted in the same way it is in RCTs, and they should have greater potential to be generalisable. Future and ongoing evaluation of smoking cessation programs in this setting should therefore consider other methodologies beyond that of RCTs.

Whilst the BOABS study did not demonstrate a significant benefit from a multi-dimensional smoking cessation intervention the meta-analysis of our and Eades’ earlier study does demonstrate that one-on-one intensive intervention delivered by and provided to Aboriginal and Torres Strait Islander peoples in a primary healthcare setting is more effective than usual care in encouraging smoking cessation. Taking into account the feasibility and cost-effectiveness of RCTs and cluster randomised trials, the priority should be to demonstrate the effectiveness of programs based on personal support to quit smoking in a real world setting. The questions should be whether real world interventions work in practice, whether they are cost effective and sustainable and to identify enablers and barriers to the integration of such programs into primary health care for Aboriginal and Torres Strait Islander peoples.