Alcohol consumption and health-related quality of life in regional, rural and metropolitan Australia: analysis of cross-sectional data from the Community Health and Rural/Regional Medicine (CHARM) study

Data were collected as part of the CHARM (Community Health and Rural/Regional Medicine) project between August 2020 and March 2022, which overlapped with the COVID-19 pandemic. For context, a brief timeline of the COVID-19 pandemic in Australia is that the first case in Australia was recorded in January 2020, with the first COVID-19 wave occurring between March and May 2020, the second between June and November 2020, a Delta wave occurring between July and December 2021 and an Omicron wave during 2022. All states experienced a range of measures to contain the virus including social distancing and lockdowns [26]. This is a cross-sectional, online, health survey aiming to gain a better understanding of the health needs, community issues and perspectives of Australians residing in rural, regional and metropolitan areas. The survey took approximately 15–20 min to complete and covered multiple aspects of health, wellbeing and satisfaction with aspects of life and community. The survey was promoted through social media groups located Australia wide, including those in regional, rural and metropolitan areas. A prize draw of three $100 store vouchers was offered to encourage participation.

Demographic information, such as age, gender and postcode, was collected. Postcodes were coded into remoteness areas (RA) based on the Australian Bureau of Statistics ‘Australian Statistical Geography Standard (ASGS): Volume 5—Remoteness Structure, July 2016’ [27]. There are five classes of RA that allow consistent analysis over time. These include RA1: major cities of Australia, RA2: inner regional Australia, RA3: outer regional Australia, RA4: remote Australia and RA5: very remote Australia [27].

Data were collected using an online self-administered survey that comprised of several validated questionnaires. For the purpose of the current study, the World Health Organization Quality of Life-brief version (WHOQOL-BREF) and the Alcohol Use Disorders Identification Test-Concise (AUDIT-C) [28, 29] were analysed.

The WHOQOL-BREF is a brief version of the WHOQOL-100 [30]. The WHOQOL-BREF contains 26 questions and is used to assess respondents’ health-related QoL across physical, psychological, environmental, and social wellbeing domains [28]. Overall QoL is measured by a single item (Question 1; How would you rate your quality of life?). Overall satisfaction with health is measured by a single item (Question 2; How satisfied are you with your health?). Each question is scored on a Likert scale from 1 to 5. Domain scores are scaled in a positive direction (i.e. higher scores indicate higher QoL) [28]. The Physical Health domain includes questions about activities of daily living, dependence on medicinal substances and medical aids, energy and fatigue, mobility, pain, sleep and work capacity. The Psychological domain includes bodily image and appearance, positive and negative feelings, self-esteem, spirituality and cognition. The Social Relationships domain covers satisfaction with personal relationships, social support and sex life. The Environmental domain includes financial resources, freedom, physical safety and security, accessibility and quality of health and social care, home environment, opportunities for acquiring new information and skills, opportunities for recreation, physical environment and transport. Participants with more than 20% of responses missing were excluded from analysis [30].

AUDIT-C is a shortened version of the original AUDIT. It was developed by the WHO as a screening tool to help identify individuals that are drinking excessively and are at an increased risk of having an alcohol use disorder [29].The AUDIT-C consists of three questions which measure alcohol consumption over the previous year, with scores ranging from 0 to 12 [29]. Score of zero reflect no alcohol use and higher scores reflect higher alcohol use. Risk categories were then calculated by gender. For males a score of 0–3 was categorised as low risk, 4–5 moderate risk, 6–7 high risk and 8–12 severe risk. For females, a score of 0–2 was categorised low risk, 3–5 moderate risk, 6–7 high risk and 8–12 severe risk [31]. Participants who did not complete all required questions were excluded from analyses.

Data were collected on Qualtrics™ and statistical analyses were conducted using the Statistical Package for Social Sciences (IBM: SPSS Version 25) and R studio [32‐34]. Prior to analysis, data were cleaned by removing participants with fewer than 80% of responses completed. The total number of survey responses was 2416, which was reduced to 1717 after removal of incomplete responses. We employed recommended steps to reduce and identify potential data integrity issues such as computer-generated (bot) responses [35]. Instead of a guaranteed payment, survey completers could enter a prize draw and were required to follow a link to another Qualtrics survey to enter their personal details; this approach has been found to be the most effective way of deterring bot activity [35]. Summary statistics and descriptive analyses were performed to check whether expected values, distributions and relationships were present, while checking for anomalies such as inconsistent responding and exact duplication of responses to free text answers [35]. Additionally, survey respondents were required to enter an Australian postcode, and these were verified by mapping responses to the Australian Bureau of Statistics database. Qualtrics also collects meta-data which provide the general location of the respondent, accurate to the nearest city. Next, we conducted Mahalanobis distance analyses on study variables including time to complete the survey, which is one of the most effective ways of detecting outlier respondents, non-human response patterns and atypical response sets that may indicate careless responding [36].Mahalanobis distance analyses indicated the presence of 14 multivariate outliers in the dataset. The analyses were run with and without the outliers and because the results were equivalent it was decided to retain them in the final analyses.

Two-way ANOVAs were performed to compare AUDIT-C and WHOQOL-BREF scores by gender and alcohol risk category. A Chi-square test of independence was performed to examine the relation between gender and alcohol use risk category. Bonferroni-corrected post hoc analyses were then performed to identify any significant differences in alcohol use risk category between genders [37, 38]. Spearman’s correlations were performed to assess the relationships between alcohol consumption, gender and QoL in Australia. As previous studies have identified that the direction of relationships between alcohol consumption and health may differ for people in different risk categories, correlations between AUDIT-C scores and QoL domain scores were performed separately for participants in each AUDIT risk category (low, medium, high, severe). An ANOVA was used to compare alcohol consumption and QoL by rurality. Lastly, a multiple linear regression was used to evaluate the relationships between alcohol intake and domain-specific QoL, adjusting for sociodemographic characteristics.

Due to the small cell sizes in the more remote areas of Australia, the RA areas were combined to create three categories. ‘Major cities’ included RA1, ‘inner regional’ included RA2 and ‘outer regional and remote’ included RA3—outer regional, RA4—remote and RA5—very remote.

The age range of male participants (n = 289) was 18–80 years, with a mean of 51.8 years (SD = 15.2). Female participants (n = 1410) accounted for 83% of the study sample, with an age range of 18 to 89 years and a mean age of 50.6 years (SD = 14.1). Across the five RA in Australia, 453 (21.6%) study participants were living in major cities' areas (RA1), 1004 (47.8%) participants were living in inner regional areas (RA2), 260 (12.4%) were residing in outer regional areas (RA3) and 2 (0.1%) were residing in remote and very remote areas (RA4 and RA5).

Two-way ANOVAs indicated that males had higher AUDIT-C scores than females. There was no significant difference between males and females for overall QoL. Males scored higher than females in general health, physical and psychological QoL. Females scored higher than males for social relationships QoL (Table 1). There were significant main effects of AUD risk category on Audit-C score, overall QoL, general health and all QoL domains (Table 1 and Fig. 1). There were no significant interaction effects between gender and AUD risk category with regard to the study variables.

QoL domain scores by gender, stratified by alcohol risk category, are plotted in Fig. 1. Across each domain of QoL, generally higher levels of QoL were reported by those with moderate than very high or low drinking risk categories. Bonferroni-corrected post hoc comparisons indicated that Physical Health QoL was lower in the low AUD category than all other categories, and higher in the moderate than the severe AUD category. Psychological QoL was lower in the low than the moderate AUD category and higher in the moderate than the severe category. Psychological QoL was lower in the low than the moderate AUD category, and higher in the moderate than the severe AUD category. Environmental QoL was higher in the moderate and severe categories than the low AUD category. Overall QoL and general health were higher for respondents in the moderate and high than the low AUD categories. General health was lower for those in the severe than the moderate-risk AUD category.

The Chi-square test of independence showed a strong relationship between the risk of alcohol use disorders and gender, X² (3, n = 1616) = 860.24, p =  < 0.01 (Table 2). The Bonferroni-corrected post hoc analysis identified that males were more likely to have a high (X²(1, n = 1616) = 8.82, p =  < 0.01) or severe risk (X²(1, n = 1616) = 41.22, p =  < 0.01) for alcohol use disorders than females.

Due to gender differences displayed in Table 2, the correlations between alcohol use disorder risk categories and QoL were performed separately for males and females (Table 3). Overall, the correlations between alcohol consumption and QoL were positive or non-significant in those at low risk of an alcohol use disorder, non-significant or positive for those at medium or high risk of an alcohol use disorder, and negative for those at severe risk of alcohol use disorder, particularly for females.

In males with low risk of alcohol use disorders, alcohol consumption had a moderate positive correlation with psychological QoL. In females with low risk of alcohol use disorders, alcohol consumption was positively but weakly correlated with overall QoL, general health, physical, psychological and environmental QoL domains. There were no significant correlations between alcohol consumption and QoL in males with medium or high risk for alcohol use disorders. In females with a medium risk for alcohol use disorders, there was a positive but weak correlation between alcohol consumption and overall QoL. There were no significant correlations between amount of alcohol consumed and QoL in females with high risk of an alcohol use disorder. In males with severe risk of an alcohol use disorder, there was a negative weak correlation with general health and a negative moderate correlation with social QoL. In females with a severe risk of alcohol use disorders, there were negative but weak correlations between the amount of alcohol consumed and general health, physical, psychological and social QoL domains (Table 3).

AUDIT-C scores were significantly higher in males residing in outer regional, remote and very remote areas than major cities and inner regional Australia (mean difference = -2.4, p < 0.01) (Table 4). There was no difference observed between the AUDIT-C scores of females by location. Rurality was not associated with overall QoL, general health, or the domains of physical, psychological and social QoL. Respondents residing in major cities had higher levels of environmental QoL than people residing in outer regional, remote and very remote Australia (mean difference = 6.6, p < 0.01). To better understand this result, the responses to individual items from the WHOQOL-BREF Environmental domain were examined and the question with the largest difference between the two locations was question 24: How satisfied are you with your access to health services? (Mean difference = 1.09, p < 0.01).

Regression analyses identified that household income, education and age were positively related to QoL across all domains. After adjusting for these sociodemographic variables, alcohol consumption was positively related to overall (B = 0.04, p < 0.01), environmental (B = 0.73, p < 0.01) and physical QoL (B = 0.81, p < 0.01); and general health (B = 0.04, p < 0.01). Remoteness area was only significantly related to environmental QoL (B = − 0.25, p < 0.01). Females had higher overall QoL and higher social QoL after adjusting for sociodemographic variables (Table 5).