Rural community views were elicited from 2,977 surveys completed and returned in 2005 by randomly selected members of the 20 AARC communities (39% response rate) [²⁹,³⁰]. The sample comprised 18-62 year olds, selected using the age and gender distribution of these communities to optimise its representativeness [³¹]: 18 coincides with the minimum age for voting and legal drinking in Australia and those over 62 contribute relatively little to alcohol-related harm [³²].

As part of the survey, respondents were asked to allocate a $1,000 budget (a reasonable household contribution over a lifespan and an easily divisible amount) across eight possible intervention types. This budget allocation exercise aimed to identify the types of interventions most commonly selected and the extent of support for them. The specific question was: "Think about all problems related to alcohol in your community. These may include relationship difficulties, health issues, car accidents and crime. The next 3 questions ask you to consider what you would be prepared to do to reduce these problems. Your community is given $1,000 to spend on programs to reduce alcohol problems. It is your job to allocate this money. You can spend it all on one program (100%) or a combination of programs. Please enter answers in percentages and make sure it adds up to 100%." Intervention options were: promotion of safer drinking through media and licensed venues (promote safer drinking); policies to reduce work-related drinking (workplace); information on alcohol harms provided by pharmacists (pharmacists); community-wide strategies to help local communities work together more effectively (community); advice from General Practitioners (GPs); school-based information (school); legal strategies, such as random breath testing and enforcing liquor licensing laws (police); and advice from hospital staff (hospital/emergency departments). These broad intervention areas were chosen because it was considered unlikely that the majority of the public would have knowledge about specific strategies. The order in which the first and last four interventions were presented was reversed in two different versions of the survey, to measure order response bias.

Similarly, professionals were asked to allocate a $100,000 budget (a reasonable government contribution and an easily divisible amount) to 23 interventions over 3 years (the maximum amount of time likely to be available to implement interventions in the AARC project) for a hypothetical rural community, the characteristics of which are summarised in Table 1. These characteristics were modelled on actual data from two rural communities in NSW, which were specified to standardise definitions given intervention preferences may change depending on community characteristics.

Interventions were identified and categorised in a three step process. First, a list of potentially effective interventions was compiled from the existing literature relevant to community-based alcohol trials [^6-20,^33-40], excluding those not practically feasible for communities to implement (eg. alcohol taxation policy [⁴¹,⁴²]. Second, this list of interventions was reviewed for its comprehensiveness and modified by five alcohol professionals. Third, similar interventions were grouped into categories to reduce the total number of interventions to a practical number (eg. codes of practice for hotels/bars and training programs for hotel staff were grouped together as 'promote safer drinking'). The interventions and intervention categories are listed in Table 2.

Both the community and professionals' surveys also asked about respondents' personal and professional characteristics (specified in the data analyses section), to identify potential influences on their views.

For both the communities and professionals, budget allocations erroneously reported as greater than 95% but less than 105% were proportionately re-scaled to ensure the total equalled 100%. Responses with errors outside this range were excluded.

The percentage of community and professional respondents who selected an intervention, and the average amount allocated to the intervention, are reported separately. Interventions selected by less than 25% of professionals were excluded from further analysis, as it was deemed unlikely that they would be included in an aggregated, optimal set of professionals' choices.

Tobit regression models were estimated to identify characteristics which explain differences in intervention preferences. Tobit models are appropriate because the budget allocation is constrained by a minimum of $0 (not selecting the intervention) and a maximum of either $1,000 (community survey) or $100,000 (professionals' survey). The outcome variable in all models is the average level of preference for an intervention (both the frequency with which it is selected and the budget amount allocated). The community preferences model was estimated with three levels of explanatory variables. First, the change in order of questions (order). Second, the extent of heterogeneity between communities (19 dummy variables). Third, individual-level factors: age; sex; education level; number in the household aged at least 14 years; frequency and quantity of alcohol consumption; having a family member/friend they perceive drinks too much; gross annual household income (from the mid-point of the selected income band); and dummy variables for gross annual household income greater than $78,000, don't know and prefer not to say. The professionals' preferences model was estimated with explanatory variables for whether the professional had lived or worked in a rural community, the number of years worked in the alcohol and other drugs field and whether the professional works for a government or other organisation.

The level of statistical significance was set at p≤0.01 for the rural community sample, due to its relatively large sample size, and p ≤ 0.05 for the professionals' sample.

Of the 200 questionnaires mailed to APSAD members, five were returned with an incorrect address and 41 completed surveys were returned (21.0%). The average number of years respondents had worked in the alcohol and other drugs field was 14.6. The majority worked for a government organization (44%), 20% worked for a treatment organization, 17% worked for a university or research organization, 10% worked for a non-government organization, and 10% worked across multiple institutions. Sixty-one percent indicated that they had lived or worked in a rural town.

As summarised in Table 4, all 23 interventions were allocated at least some funds by at least one professional.

Training GPs was selected most often (61.0%) and allocated the largest proportion of funds when selected ($17,900), followed by school programs (36.6% selected, $10,900 allocated), community programs (33.8% selected, $15,300 allocated), promoting safer drinking (31.7% selected, $8,700 allocated), hospital/emergency department interventions (24.4% selected, $11,200 allocated), police activity (28.4% selected, $15,000 allocated) and pharmacists (14.6% selected, $6,000 allocated).

The Tobit regression showed no statistically significant relationships between professionals' characteristics and their intervention preferences, although there was a trend for professionals with more years of experience to allocate fewer resources to school-based interventions (p ≤ 0.06).

This study shows broad agreement between the views of rural communities and alcohol and other drug professionals about preferred strategies for reducing alcohol misuse and related harms: two of the four strategies selected by at least 50% of respondents to both surveys were the same, namely, promotion of safer drinking and community-wide activity.

Promotion of safer drinking generally refers to attempts to control drinking environments, through mechanisms such as codes of practice or training hotel staff and, more generally, promoting more responsible drinking in the community through increased awareness of alcohol harms through media advocacy. Although these views do broadly align with the limited research evidence for media advocacy [^9-14], enforced point-of-sale legislation [¹¹,¹²,¹⁵,¹⁶] and increased police visibility [¹⁴,¹⁷], they do not appear to adequately reflect the need for concomitant enforcement of codes of practice: increased police activity was rated fourth by communities and fifth by professionals. Given the concept of codes of practice in alcohol outlets are well ingrained internationally [⁴³], it may be possible to improve the cost-effectiveness of this approach by increasing the sophistication of tailoring of police visibility and enforcement of point-of-sale legislation, beyond the typical targeting of high-risk times, such as weekend nights, community events and seasonal effects. Alternatively, this idea could be expanded to other defined settings where alcohol may be regularly consumed, such as workplaces. Although many workplaces do have alcohol policies and counselling programs in place, these could be strengthened to reflect the high level of workplace-related alcohol harm [⁴⁴]. There also appears to be potential to further utilise media advocacy: data on alcohol consumption and harms that is specific to a community could be provided to communities via the media and key stakeholders, as a behaviour change strategy that has been shown to be effective in promoting behaviour change among individuals [⁴⁵] and health care professionals [⁴⁶].

The apparent popularity of community activity reflects support for broad-based services, such as an increased number of counselling type services (family interventions, social work services, drug and alcohol counsellors) and community development programs (sports and arts facilities). These may reflect increased awareness of the socio-economic determinants of health [⁴⁷] and the negative impact of alcohol misuse on people other than drinkers themselves [⁴], although improvements in community development programs need to be independent of alcohol industry advertising and support, to avoid deleterious impacts from increased alcohol advertising and supply. Community activity also encompassed the idea of improved integration between various initiatives to potentially improve their combined effect across the community. This is analogous to the introduction of random-breath testing in Australia, the effectiveness of which is attributable to the combined effect of the testing itself, along with a sustained awareness-raising media campaign, high visibility of testing and enforced consequences for being caught [⁴⁸]. Despite the popularity of this idea with rural communities and professionals, and its centrality to the concept of community-wide interventions [⁷,²⁶], there is little research evidence to date for its likely cost-benefit.

These rural communities ranked school-based interventions as their most preferred strategy, although the research evidence for its effectiveness in changing young people's drinking behaviour is equivocal [³⁸,^49-52]. Given its popularity with these rural communities, however, it is highly likely that the acceptability to a community of any type of community-wide approach would depend on the inclusion of a school-specific intervention. Given the lack of evidence, it seems reasonable that such an intervention could have a harm-reduction focus (that is, equipping young people with strategies for staying safe when they are drinking), in addition to attempting to reduce the frequency and quantity of their drinking.

The majority of alcohol and other drug professionals also selected greater support and training for GPs and hospital/ED staff. Although the popularity of GP interventions most likely reflects the wealth of evidence for the cost-effectiveness of screening and brief intervention delivered in primary care [⁵³], it is unclear if increased GP effort would be effective across a whole population, that is, if GP-based interventions are a cost-beneficial community-level strategy. Increasing hospital/ED-based effort is problematic because of the time and resources restrictions, especially in ED settings [³⁴]. Nevertheless, given evidence that the proportion of risky drinkers among ED patients is significantly higher than in the general community [⁵⁴] and that those who report higher rates of consumption also report worse health [³⁰], interventions delivered in health settings seem appropriate and there is some evidence for their effectiveness [⁵⁴]. Screening and brief intervention (SBI) for alcohol could also be implemented in a wider variety of settings, particularly as SBI has evidence for its cost-effectiveness [⁵⁵] and does not require a high-level of technical skill to implement. Community pharmacies have been recognised as one relatively novel and promising health setting in which SBI could be delivered [⁵⁶], although there have been only two intervention outcome studies, the results of which were equivocal [⁵⁷,⁵⁸]. Both communities (Figure 1) and professionals (Table 4) ranked pharmacy-based interventions last. Some of the practical limitations of time and resources constraints could potentially be addressed by simultaneously offering internet-based SBI, evidence for which is emerging [⁵⁹].

The methodological limitations of this study are unlikely to have substantially impacted on the results. Although the response rate of 39% may have restricted the extent to which the responses are representative of all community members, it is higher than for the 2004 (34.1%) and 2007 (33.2%) Australian national household drug and alcohol surveys [⁶⁰,⁶¹]. While females and older people were over-represented, the Tobit regression results show their preferences effectively counteracted each other for the intervention type most susceptible to bias: a hospital/ED intervention was supported more by females and less by older people. School and pharmacy interventions were supported more and less frequently by older people, but given these were easily the most and least supported strategies overall, the responses of older people are unlikely to alter the ordinal preferences of these extreme interventions. The other Tobit regression results were as expected, providing reasonable evidence for the face-validity of both consumer and professionals' preferences. Frequent drinkers prefer health, rather than law enforcement, interventions, higher income earners are more likely to prefer school-and pharmacy-based interventions, those who have a relative or friend who they think drinks too much are more likely to select community programs, more educated respondents preferred media advocacy strategies and preferences for hospital/ED interventions varied by community.

The lack of any statistically significant predictors of professionals' views may reflect the abstract nature of the budget allocation task or the relatively large number of potential interventions. However, the community characteristics were both specified and based on actual rural communities and the majority of professionals agreed or strongly agreed that the budget allocation was a realistic task facing those trying to reduce alcohol related harm (66.0%), and expressed confidence that their allocation would reduce alcohol related harm (58.5%). It is most likely this finding reflects the reasonable expectation that professionals' views are less susceptible to bias than those of consumers. Although the 21% response rate to the professionals' survey creates some uncertainty about the generalisability of their views, and we were unable to obtain any further information about non-responders due to APSAD's privacy concerns, the extent of any bias is most likely to be limited for a number of reasons: unlike community views, the Tobit regressions found no statistically significant predictors of professionals' views which would indicate substantial bias; their views are generally aligned with existing evidence (the majority supported GPs, about a third supported school, community or safer drinking interventions, and pharmacy interventions received the least support); they selected a range of strategies; and they worked for a variety of organisations. The less than ideal response rate to both surveys highlights the need for more effective methods of identifying professionals' and community views in the short-term and the need to increase the range of community-based alcohol interventions with demonstrable evidence for their effectiveness over the longer term.

*Statistically significant where p ≤ 0.01

^aNeither high quantity drinking nor number of persons in the household aged ≥ 14 years were statistically significant predictors of intervention preference

^bCommunity is the maximum difference in average allocation preference between all twenty communities - significance is based on a joint test for significance of all community dummy variables