Impact of Alcohol Consumption on Use of Health Care Resources

The aim of the study is to describe the association between alcohol consumption and the use of health care resources and the health care costs in Catalonia, which is a region of Spain.

A cross-sectional study was done. Patients recruited in the study were those attended in the primary health care whose alcohol consumption was registered in the electronic medical record between 1st of January 2011 to 31st of December 2012. At Baseline (31st of December 2012), sociodemographic data and clinical data was recorded, and health care utilization and costs were obtained from year 2013.

Two different registers were used for sociodemographic, clinical and health service utilization indicators and costs. Information on life style factors (alcohol and tobacco consumption, body mass index), demographic information like the ecologic MEDEA index, and sick-leave costs were obtained from the Information System for the Development of Research in Primary Care (SIDIAP) database. This clinical database has anonymized records of almost the 80% of the Catalan population. General practitioners can record alcohol consumption in two different ways (quantitative and categorical information). A quantitative variable defined as grams of alcohol per week and a categorical variable measuring the risk of alcohol consumption (none, low risk, risky drinker). The type of risk was defined as follows (No drinker; Low drinker (men who drinks <280g per week of alcohol or women who drinks <179g per week); Riky drinker (men who drinks >=280g per week or women who drinks >=170g per week of alcohol, or men who drinks more than 5 drinks per occasion, women who drinks more than 4 drinks per occasion; men who drinks <280g per week of alcohol or women who drinks <179g per week and at the same time work with heavy machinery or are taking medication that could interfere with alcohol)).

Sociodemographic data (age and sex and socioeconomic status) and clinical data (hypertension, diabetes, hyperlipemia, obesity, psychiatric diseases, clinical risk groups and diagnosis wholly attributable and partly attributable to alcohol) were obtained from the central register (Morbidity Register) provided by the Catalan Health service. From the same register data on health service utilization was collected. These indicators include: the number and costs of outpatients visits from doctor, nurse and social worker in primary health centres; hospital admissions, emergency room visits, mental health outpatients visits, and specialists referrals. Data on acute and chronic hospitalizations were analysed using three indicators: costs, number of admissions and hospital days, defined as the number of days stayed in any hospital. Three types of hospitals were included in the administrative system: general (acute) hospital, rehabilitation hospital and psychiatric hospital. The patient's diagnosis at admission was also registered. Using the same categories as Jones and colleagues, we distinguished between entry diagnosis wholly attributable to alcohol (e.g., alcohol dependence or alcoholic liver cirrhosis), entry diagnosis partly attributable (acute and chronic), or not related to alcohol. Also information about medication expenses and seak-leave costs were included.

Quality checks, in SIDIAP database, to identify duplicate patient identification are performed centrally at the time of each SIDIAP database update. Furthermore, checks for logic values are performed and unexpected values are dismissed. A part from that, a number of papers have been published on the validity of SIDIAP coding for various conditions. The quality and consistency of the central Catalan register is very good with a built-in validation to detect problems and inconsistencies between variables. In addition, the system is periodically externally validated to ensure quality of data, as these data are used to control payment to service providers.

Descriptive analyses of sociodemographic and clinical data will be carried out. Analysis of variance (ANOVA) and t-students tests will be used to compare continuous data, and chi-square analysis for categorical variables. Bonferroni correction will be used for multiple comparisons. A probability level of 5% or less will be considered as statistically significant. Poisson regression and sensitivity analyses with negative binomial regression will be done. Linear regression will be done to analyze the relationship between grams per week of alcohol and health costs. Data were analysed with Statistical Analysis Software.

Inclusion criteria

• Patients over 17 years old
• Patients with alcohol consumption registered in the medical record

Exclusion criteria

• patients <= 17 years old.