Prevalence of simultaneous use of alcohol and prescription medication in older adults: findings from a cross-sectional survey (Health Survey for England 2013).

OBJECTIVES: To investigate the concurrent use of 'at-risk' (AR) drinking (>10 units of alcohol per week) and prescription medications, while controlling for sociodemographic, and health-related factors, among older adults (aged 65-89 years).
DESIGN: Cross-sectional survey.
SETTING: Data from Health Survey of England, 2013.
INTERVENTIONS: None. PARTICIPANTS: General population survey of 2169 adults aged 65-89 years. PRIMARY OUTCOME MEASURES: AR drinking (>10 units per week). Secondary outcome was AR drinking defined as >14 units of alcohol per week limit (the cut-off used by the Department of Health for AT drinking).
RESULTS: Twenty-seven per cent (n=568) of the sample were AR drinkers. Factors associated with alcohol consumption were gender, age, social class, marital status, rurality of dwelling, deprivation index, self-reported general health, cigarette smoking, body mass index, exercise level, health and well-being scores' and number of prescription drugs. Logistic regression analysis showed that males were more likely to be AR drinkers (OR 3.44, 95% CI 2.59 to 4.57, p<0.0001) than females. Each year increase in age, lowered the probability of AR drinking by a factor of 0.95 (95% CI 0.93 to 0.98, p<0.0001). Using prescription drugs reduced AR drinking by a factor of 0.92 (95% CI 0.85 to 0.93, p=0.033), after controlling for age, sex and rurality of dwelling. No other predictors were significant. Similar results were obtained for AR drinking of >14units per week.
CONCLUSION: AR drinking is more likely in older men than women. The odds of AR drinking lessens, as individuals age, and using prescription drugs also reduces AR drinking. © Author(s) (or their employer(s)) 2019. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

The research uses data from a gold-standard general population health survey.

To our knowledge, this is the first study that has shown there is a negative relationship between being prescribed medication and at-risk drinking in older adults (aged 65–89 years).

Health Survey of England (HSE) relies on self-report when assessing alcohol consumption.

There is a high probability of under-reporting of alcohol consumption in the HSE.

Introduction

In the UK, it is estimated that around 17% of the population are now over the age of 65 years.1 The general trend is for alcohol consumption to decline with age.2. However, a recent paper3 has stated that individuals within the age band 55–74 years (now often referred to as baby boomers), is the only one where the percentage of individuals drinking above the daily limits have increased during the last 5 years. The sensible drinking limits in the UK have recently been reduced to 14 units per week for both men and women with a further recommendation of two alcohol-free days.4 Prior to this, it was 14 units per week for women and 21 units for men.5 During the last decade, there has been a 20% increase in the number of people over 65 years drinking above the daily limits at least 1 day per week.6 The Royal College of Psychiatrists (RCP)7 recommended guidelines in the UK should be lowered to 1.5 units of alcohol per day (10.5 units per week), with at least two alcohol-free days per week, for individuals over 65 years. A second edition of the report8 suggested that the revised national guidelines4 may still be too high for older people with physical or mental health problems who are taking medications.

According to the 2015 Health Survey of England (HSE), around half of the people over the age of 65 years in the England regularly take prescription medication.9 As the body ages, it becomes more susceptible to alcohol–drug interactions, and metabolises both alcohol and drugs less efficiently. Even a small amount of alcohol (2–3 units), when taken in combination with some prescription drugs, can cause adverse effects.10 A systematic review has shown that prescription medication including psychotropic drugs are frequently used in combination with alcohol.11 A US study found that the prescription of cardiovascular, central nervous system (CNS) and metabolic agents was commonplace in current drinkers over 65 years of age,12 and similar work from Ireland found that heavy drinking was associated with the prescription of anticoagulants/antiplatelets, cardiovascular and CNS agents.13 Non-steroidal anti-inflammatory drugs, which are commonly used by older adults, can cause stomach bleeding, gastric inflammation and kidney damage if taken in combination with alcohol.14 Alcohol can also enhance the sedative effects of some medications, such as muscle relaxants and benzodiazepines, which can lead to falls and even death.15 Despite the increased likelihood of being prescribed multiple drugs the RCP7 8 state that older people are often given confusing and conflicting advice on how much they can drink or no advice at all.

To date, there have been no studies in the UK which have looked at ‘at-risk’ (AR) drinking using the limits recommended by the RCP7 and compared them to the new Department of Health guidelines.4 This study uses data collected from the HSE16 which is regarded as a gold-standard general population survey and data are quoted from it in National Health Service Statistics.17

The aim of the study is to examine the concurrent use of AR drinking (>10 units of alcohol per week) and prescription medications, while controlling for sociodemographic, and health-related factors, among older adults (aged 65–89 years).

Methods

The Health Survey of England

The study uses data collected from the HSE which has been conducted annually since 1991 on adults aged 16 years or more. The data used in the current study is from HSE 2013. The core modules that are repeated regularly are demographic characteristics, general health, long-standing illness, doctor-diagnosed hypertension, diabetes, smoking and alcohol consumption. Prescription medication was especially collected for the HSE 2013 and is not routinely collected.

Assessment of AR drinking

The current study uses the cut-off points for AR drinking as those who drink >10 units of alcohol per week. The 10.5 (not including two alcohol-free days) unit limit is regarded as the cut-off for AR drinking in older adults in accordance with the recommendations of the RCP.7 The nearest category to the RCP cut-off point used by the HSE is 10 units per week, which is used in the current study. The 14 units of alcohol per week, limit is the current Department of Health AR drinking guidelines4 which is a cut-off also used by the HSE.

Data collection

The HSE sample was selected using a stratified random probability sample of private households in England. In 2013, 8795 adults aged 16 years and over and 2185 children ages 2–15 years were interviewed at households from the selected addresses and 6183 adults and 1455 children had a nurse visit when measurements such as blood samples were taken and data were recorded concerning prescription drug use. Thus, our sample (aged 65–89 years) consisted of 2169 (24.6%) participants who had the interview, and 1607 (25.9%) who received the nurse visit. Ethical approval for this wave of the study was provided by the Oxford A Research Ethics Committee (reference number: 12/SC/0317).18

Patient and public involvement

The data are derived from a secondary dataset and there is no direct patient involvement in this study. Details of arrangements for patient and public involvement can be obtained through University College London.16

Socio demographic measures

In this study, the variables age, gender, education level, employment status, ethnicity, religion, marital status, household type, rurality of dwelling, Index of Multiple Deprivation (IMD)19 and social class were used. The IMD was calculated by assigning a number of indices of deprivation/affluence to a specific locality. They are divided into quintiles and higher quintiles equate to greater deprivation. Social class categorisations were assigned with reference to the Office of Population Surveys20 and occupations were coded according to Office of National Statistics guidance.21

Indicators of physical and mental health

Measures of height (cm), weight (kg), BMI, blood cholesterol levels and blood pressure were collected. Diet was also assessed regarding the amount of fruit and vegetables consumed. Each adult was asked to rate their current general health on a scale of very good, good, fair, bad and very bad. This was converted into a score of 1–5 (higher scores indicated better health). Level of physical activity was assessed by the International Physical Activity Questionnaire.22 Individuals record the amount of time they spent carrying out physical activities such as walking. The Warwick and Edinburgh Mental Well-being Scale (WEMWBS)23 was used which assesses mental health well-being across the general population and its validity and reliability is well established where higher scores indicate better mental health status.24

Cigarette and alcohol consumption

Participants were asked how many cigarettes they typically smoked daily and to compile a drinking diary concerning the amount of alcohol consumed in the last 7 days. This was then subsequently converted into units of alcohol consumed per week. In the UK, a unit of alcohol is 8 g of pure alcohol, the equivalent of half a pint of normal strength beer.25

Prescription medications

Participants were asked whether they had been prescribed any of the following medications in the last 7 days; antidepressants, analgesia, antihypertensives, cardiovascular, antidiabetic medications, proton-pump inhibitors, antiplatelet, lipid-lowering, antiasthma/Chronic Obstructive Pulmonary Disease (COPD) and antibacterial medications. The use of over the counter (OTC), medication purchased at a pharmacy, was not recorded in the survey.

Statistical analysis

SPSS V.23 was used for all statistical analyses. A two-sided p value of <0.05 was considered statistically significant.

In stage 1 analysis (of all participants), χ2 test, one-way Analysis of Variance (ANOVA) and Kruskal-Wallis test were used.

In stage 2 analysis, only the individuals who consumed any alcohol in the last 7 days were included. Logistic regression model was used, with the dependent variable as AR versus not-AR drinkers. A cut-off using 10 units of alcohol per week as cut-off was used as the primary outcome measure for AR drinking, and a secondary outcome measure used a cut-off was 14 units of alcohol per week. All variables which were significant in the bivariate analysis in stage 1, were used as predictors in the logistic regression model of stage 2. The variable, total number of prescription drugs taken, included the 10 drugs: analgesics, antibacterial, antidepressants, antidiabetic, antihypertensives, antiplatelet, asthma/COPD, cardiovascular, lipid-lowering and proton-pump inhibitor. A further analysis using only nine of these drugs (removing antidepressants) was carried out, as it was the only psychotropic medication compared with the others. A comparison was also made, to assess the representativeness, of the subset of individuals who provided prescription drug use data, with all participants in the full sample.

Results

Sociodemographic, cigarette and alcohol consumption data for the study sample are shown in table 1. There were 2169 (52.2% women) participants in the sample aged 65–89 years. Mean age in the sample was 73.7 years (SD 6.5). There was no difference in the mean age between men and women (p=0.402). Over 92% of individuals was White British and the majority were married (59.8%), affluent (IMD quintiles 1 and 2: 47.5%) and lived in urban areas (75.2%). Most had retired (87.7%), but the most common social class categories (based on employment when working) were II and IIIN.

Table 1

Percentage of sociodemographic, smoking and drinking categories in the sample

Variable (n)	Category	%
Gender (2169)	Male	47.8
	Female	52.2
Ethnicity (2158)	White British	92.2
	Other	7.8
Marital status (2169)	Single	5.6
	Married	59.8
	Divorced/separated	10.4
	Widowed	24
	Other	0.2
Rurality of dwelling (2169)	Urban	75.2
	Town and fringe areas	11.8
	Rural	13
Deprivation index, IMD (2169)	Quintile 1 (least)	23.7
	Quintile 2	23.8
	Quintile 3	22.4
	Quintile 4	15.8
	Quintile 5	14.2
Social class21 22 (2087)	I professional	4.9
	II managerial/technical	28.6
	IIIN skilled non-manual	24.5
	IIIM skilled manual	20
	IV semiskilled manual	16.2
	V unskilled manual	5.8
Cigarettes per day (2162)	Non-smoker	90.6
	Light (<10)	3.1
	Moderate (10–20)	4.1
	Heavy (>20)	2.2
Alcohol consumption, last 7 days (2119)	None (last 12 months)	21.7
	<1 unit	17
	1–7 units	27.7
	>7–10 units	6.7
	>10–14 units	6.2
	>14 – 21 units	7.2
	>21–28 units	4.9
	>28–35 units	3.3
	>35–50 units	3
	>50 units	1.9

IMD, Index of Multiple Deprivation.

Over 90% of individuals were non-smokers. Nearly 40% (n=822) were either non-drinkers or typically drank <1 unit per week. In total, 26.8% (n=568) individuals were drinking >10 units of alcohol per week and 20.5% (n=435) drinking >14 units per week. The most commonly prescribed drug was cardiovascular medications (n=927), followed by lipid-lowering medications (n=729) and antihypertensives (n=665). The total number of drugs taken by individuals, treated as a continuous variable gave a mean of 2.5 drugs per individual (SD=1.9, range=0–8, n=1638), with 79.4% taking at least one of the prescribed medications.

Sixty-three per cent (n=1373) of participants had their BMI measured, and 26.5% were normal/underweight (BMI <25), 43.3% overweight (BMI 25–29), 20.8% obese 1 (BMI) 30–34) and 9.2% obese 2 (BMI >34). Fifty-five per cent (n=1197) of participants provided cholesterol data, and of these 49.9% had total cholesterol levels >5 mmol/L. Nearly 75% (n=1459) of individuals had their blood pressure measured. Taking the cut-off as a systolic reading of 140 or more,26 35.9% (n=524) could be classified as having hypertension. Only 28.4% of participants ate five or more portions of fruit and vegetables daily. The results for self-rating of general health was as follows; very good/good 57.2% (n=1242), fair 29.9% (n=650) and poor 12.9% (n=277). Over 80% (n=1740) answered the question concerning physical exercise (time spent walking and no of steps, per day) and 51.8% were classified low (<30 min, <2500 steps), 23.7% medium (30–60 min, 2500–5000 steps) and 24.3% high (>60 min, >5000 steps).22 The mental health and well-being scores (WEMWBS)23 indicated that most participants had good mental health and well-being (n=1846, mean=52.0, SD=8.5).

The Office of Population Censuses Surveys (1991)20 categorisations were used to assign social class, and occupations were coded according to guidance provided (Office of National Statistics 2000).21

Stage 1 analysis: all participants

This analysis included all participants aged 65–89 years. The association between alcohol consumption, and sociodemographic, health-related and prescribed drugs variables, was explored (table 2). The sociodemographic variables that were associated with alcohol consumption were gender (males drinking more, χ2(4)=156.710, p<0.0001), social class (I and II, drinking more, χ2(20)=93.436, p<0.0001), marital status (married drinking more, χ2(12)=62.320, p<0.0001), rurality of dwelling (urbanites drinking less, χ2(8)=24.334, p=0.002) and deprivation index IMD (most deprived drinking less, χ2(16)=73.994, p<0.0001).

Table 2

Sociodemographic, health-related, prescription drug factors with alcohol consumption during last 7 days

Variable (n)	Non-drinkers	1–10 units	>10–21 units	>21–35 units	>36 units	P value
	n (row %)	n (row %)	n (row %)	n (row %)	n (row %)
Gender (2119)						<0.0001
Male	285 (28.3)	347 (34.2)	167 (16.5)	128 (12.6)	85 (8.4)
Female	535 (48.4)	382 (34.6)	120 (10.9)	47 (4.3)	21 (1.9)
Social class (2087)						<0.0001
I professional	23 (23.0)	43 (43.0)	14 (14.0)	15 (15.0)	5 (5)
II managerial/technical	174 (29.7)	216 (36.9)	93 (15.9)	62 (10.6)	41 (7.0)
IIIN skilled non-manual	193 (38.6)	190 (38.0)	71 (14.2)	34 (6.8)	12 (2.4)
IIIM skilled manual	149 (36.8)	127 (31.4)	64 (15.8)	38 (9.4)	27 (6.7)
IV semiskilled manual	165 (49.1)	114 (33.9)	31 (9.2)	12 (3.6)	14 (4.2)
V unskilled manual	65 (55.6)	30 (25.6)	8 (6.8)	11 (9.4)	3 (2.6)
Marital status (2169)						<0.0001
Single	49 (42.6)	35 (30.4)	11 (9.6)	14 (12.2)	6 (5.2)
Married	426 (33.5)	456 (35.8)	202 (15.9)	117 (9.2)	72 (5.7)
Divorced/separated	87 (40.1)	78 (35.9)	20 (9.2)	17 (7.8)	15 (6.9)
Widowed	260 (50.6)	160 (31.1)	54 (10.5)	27 (8.3)	13 (2.5)
Rurality of dwelling (2119)						0.002
Urban	649 (40.8)	549 (34.5)	204 (12.8)	112 (7.0)	76 (4.8)
Town and fringe areas	84 (33.2)	91 (36.0)	40 (15.8)	27 (10.7)	11 (4.3)
Rural	89 (32.2)	89 (32.2)	43 (15.6)	36 (13.0)	19 (6.9)
Deprivation index, IMD (2119)						<0.0001
Quintile 1 (least)	143 (28.1)	197 (38.8)	90 (17.7)	50 (9.8)	28 (5.5)
Quintile 2	169 (33.3)	189 (37.3)	70 (13.8)	48 (9.5)	31 (6.1)
Quintile 3	199 (42.0)	165 (34.8)	57 (12.0)	33 (7.0)	20 (4.2)
Quintile 4	149 (45.2)	102 (30.9)	44 (13.3)	22 (6.7)	13 (3.9)
Quintile 5 (most)	162 (54.0)	76 (25.3)	26 (8.7)	22 (7.3)	14 (4.7)
Cigarettes/day (2117)						0.027
Non-smoker	732 (38.2)	678 (35.4)	262 (13.7)	152 (7.9)	106 (5.0)
Light (<10)	35 (52.2)	15 (22.4)	7 (10.4)	5 (7.5)	5 (7.5)
Moderate (10–20)	30 (34.5)	26 (29.9)	15 (17.2)	11 (12.6)	5 (5.7)
Heavy (>20)	23 (47.9)	10 (20.8)	3 (6.3)	7 (14.6)	5 (10.4)
General health (2118)						<0.0001
Very good/good	366 (30.0)	461 (37.7)	199 (16.3)	130 (10.6)	66 (5.4)
Fair	293 (46.7)	201 (32.0)	72 (11.5)	32 (5.1)	30 (4.8)
Bad/very bad	163 (60.8)	66 (24.6)	16 (6.0)	13 (4.9)	10 (3.7)
Exercise level (1708)						<0.0001
Low	388 (43.8)	280 (31.6)	95 (10.7)	70 (7.9)	53 (6.0)
Medium	122 (29.8)	165 (40.3)	63 (15.4)	41 (10.0)	18 (4.4)
High	104 (25.2)	162 (39.2)	83 (20.1)	47 (11.4)	17 (4.1)
BMI (1349)						0.035
Normal/underweight (<25)	120 (33.3)	144 (40.0)	50 (13.9)	32 (8.9)	14 (3.9)
Overweight (25 to <30)	181 (31.2)	229 (39.4)	87 (15.0)	51 8.8)	33 (5.7)
Obese I (30 to <35)	97 (34.4)	104 (36.9)	42 (14.9)	24 (8.5)	15 (5.3)
Obese II and III (≥35)	65 (51.6)	34 (27.0)	11 (8.7)	10 (7.9)	6 (4.8)
Prescription drugs, taken last 7 days (1607)
Antidepressants, yes	77 (51.3)	43 (28.7)	14 (9.3)	12 (8.0)	4 (2.7)	0.004
No	(35.6)	(37.1)	(13.5)	(8.6)	(5.2)
Analgesics, yes	171 (47.5)	118 (32.8)	43 (11.9)	20 (5.6)	8 (2.2)	< 0.0001
No	(34.1)	(37.3)	(13.5)	(9.4)	(5.8)
Antihypertensives, yes	272 (55.9)	234 (35.2)	80 (12.0)	47 (7.1)	32 (4.8)	0.067
No	(34.4)	(37.0)	(13.9)	(9.6)	(5.1)
Cardiovascular, yes	381 (41.1)	321 (34.6)	113 (12.2)	64 (6.9)	48 (5.2)	0.001
No	(31.6)	(38.5)	(14.4)	(10.7)	(4.7)
Antidiabetic, yes	91 (47.2)	64 (33.2)	18 (9.3)	13 (6.7)	7 (3.6)	0.031
No	(35.7)	(36.7)	(13.6)	(8.8)	(5.2)
Proton-pump inhibitor, yes	176 (46.0)	126 (32.9)	39 (10.2)	26 (6.8)	16 (4.2)	0.001
No	(34.3)	(37.3)	(14.1)	(9.1)	(5.2)
Antiplatelet, yes	168 (42.4)	128 (32.3)	53 (13.4)	24 (6.1)	23 (5.8)	0.029
No	(35.3)	(37.6)	(13)	(9.3)	(4.7)
Lipid-lowering, yes	284 (39.0)	252 (34.6)	89 (12.2)	60 (8.2)	44 (6.0)	0.176
No	(35.5)	(37.7)	(13.9)	(8.8)	(4.1)
Asthma/COPD, yes	82 (42.9)	72 (37.7)	16 (8.4)	11 (5.8)	10 (5.2)	0.098
No	(36.3)	(36.1)	(13.8)	(8.9)	(4.9)
Antibacterial, yes	12 (32.4)	19 (51.4)	3 (8.1)	3 (8.1)	0 (0.0)	0.266
No	(37.2)	(35.9)	(13.2)	(8.5)	(5.1)

IMD, Index of Multiple Deprivation.

The health-related factors that were associated with alcohol consumption were smoking (moderate/heavy smokers drinking more, χ2(12)=23.037, p=0.027), self-reported general health (decreased use of alcohol with poorer health, χ2(8)=121.740, p<0.0001), exercise levels (higher levels of exercise with less drinking, χ2(8)=65.549, p<0.0001) and BMI (decreased use of alcohol associated with higher BMI, χ2(12)=22.185, p=0.035). Blood Pressure (χ2(8, n=1434)=5.848, p=0.664) and cholesterol levels (F(4,1177)=1.420, p=0.225) were not associated with alcohol consumption.

One thousand six hundred and seven participants (75.8%) provided data on prescription drug use in the last 7 days and of these (n=1301, 790.4%) were taking at least one prescription drug. As only 75.8% of the participants had provided data on prescription drugs, a check on the representativeness of this subset of individuals, to the full sample, was made in terms of their age, gender, ethnicity, social class and alcohol consumption, which showed no difference in these factors (table 3).

Table 3

Sociodemographic, and alcohol consumption, comparison of total sample with prescription drug users only

	Total sample (n=2169)	Prescription drug users only (n=1301)
	%	%
Gender (n=2169)
Male	47.8	48.6
Female	52.2	51.4
Ethnicity (n=2158)
White British	92.2	93.6
Other	7.8	6.4
Social class (n=2087)
I: professional	4.9	4.1
II: managerial/technical	28.6	28.6
IIIN: skilled non-manual	44.5	44.4
IIIM: skilled manual	20	20.1
IV: semiskilled manual	16.2	17.1
V: unskilled manual	5.8	5.9
Alcohol consumption, last 7 days (n=2119)
None in last 12 months, or <1 unit	38.7	38.9
1–10 units	34.4	34.9
>10–21 units	13.4	12.1
>21–28 units	4.9	4.6
>28–50 units	6.3	5.8
>50 units	1.9	1.8
Age, years (n=2169)	Mean=73.7 (SD=6.5)	Mean=73.9 (SD=6.4)

The prescription drugs associated to alcohol consumption in the last 7 days, were antidepressants (χ2(4)=15.341, p=0.004), analgesics (χ2(4)=27.503, p<0.0001), cardiovascular medications (χ2(4)=19.561, p=0.001) and antidiabetic drugs (χ2(4)=10.621, p=0.031), proton-pump inhibitors (χ2(4)=17.993, p=0.001) and antiplatelets (χ2(4)=10.798, p=0.029) (table 2). Higher prescription drug use was associated with less alcohol consumption or abstinence. Antihypertensive (χ2(4)=8.758, p=0.067), lipid-lowering (χ2(4)=6.330, p=0.176), antiasthma/COPD (χ2(4)=7.820, p=0.098) and antibacterial medication (χ2(4)=5.213, p=0.266) was not associated with alcohol consumption levels. Three continuous variables, also associated with alcohol consumption, were age (OR −0.279, 95% CI −0.365 to −0.193, p<0.001), WEMWBS scores (OR 0.101, 95% CI 0.028 to 0.173, p=0.006) and number of prescription drugs (OR −0.011, 95% CI −0.018 to −0.004, p=0.001).

Variables not statistically significantly associated with alcohol consumption were BP, cholesterol level and prescription medication of antihypertensive, lipid-lowering, antiasthma/COPD and antibacterial medication.

Stage 2 analysis: alcohol drinkers

All variables which were associated to alcohol consumption in stage 1 analysis, were used as predictors in logistic regressions using AR versus not-AR drinking (>10 units of alcohol in the last 7 days as cut-off). These were gender, age, social class, marital status, rurality of dwelling, deprivation index, general health, cigarette smoking, BMI, exercise level and WEMWBS scores, number of prescription drugs.

Individuals who had not drunk any alcohol in the previous 12 months were excluded. Table 4A shows that men were 3.44 (95% CI 2.59 to 4.57, p<0.0001) times more likely to be AR drinkers than women, after controlling for age, use of prescription medication and rurality of dwelling. Other variables were not significant predictors. Higher age was associated with a lower probability of AR drinking, by a factor of 0.95 (95% CI 0.93 to 0.98, p<0.0001) for each year older the individual is. Using a prescription drug reduces AR drinking by a factor of 0.92 (95% CI 0.85 to 0.93, p=0.033). A similar analysis was conducted excluding antidepressants from the total number of drugs taken which showed that there was little difference in the odds ratio (OR 0.92, 95% CI 0.86 to 0.99, p=0.020), compared with the original model in table 4A.

Table 4A

Logistic regression model: AR drinkers (>10 units per week, n=566, 34.2%) versus not-AR drinkers (n=1093, 65.8%)

Variables	OR (95% CI)	P value
Gender (reference category: female)	3.44 (2.59 to 4.57)	<0.0001
Age	0.95 (0.93 to 0.98)	<0.0001
Total no of prescription drugs taken	0.92 (0.85 to 0.93)	0.033
Rurality of dwelling (reference category: rural)		0.046
Urban	0.74 (0.50 to 1.10)	0.136
Town and fringe areas	1.16 (0.70 to 1.92)	0.568

A further logistic regression analysis was conducted using the higher cut-off point for AR drinking of >14 units per week (table 4B), which showed similar results to the lower cut-off point logistic regression model. Again, if we excluded antidepressants from the total number of drugs taken, there was little difference in the odds ratio (OR 0.90, 95% CI 0.83 to 0.97, p=0.005) compared with the model in table 4B.

Table 4B

Logistic regression model: AR drinkers (>14 units per week, n=330, 25.6%) versus not-AR drinkers (n=960, 74.4%)

Variables	OR (95% CI)	P value
Gender (reference category: female)	3.27 (2.49 to 4.31)	<0.0001
Age	0.97 (0.95 to 0.99)	0.004
Total no of prescription drugs taken	0.91 (0.84 to 0.98)	0.008
Rurality of dwelling (reference category: rural)		0.088
Urban	0.67 (0.47 to 0.96)	0.028
Town and fringe areas	0.75 (0.47 to 1.21)	0.239

Discussion

Summary

The main finding from this study shows that there is an association of decreased AR drinking as individuals age, and using prescription drugs is also associated with a reduction in the probability of AR drinking. A recent BMJ editorial3 has highlighted that substance misuse (including illicit drugs but predominantly alcohol) by a group they term baby boomers (people born 1946–1964) is likely to treble in the USA and double in Europe by 2020.27 The term baby boomers usually refer to those aged 50 years and above. In the current study, we have only used data collected on individuals 65 years or above and have assessed AR drinking by using criteria suggested for those aged 65 years or above others by the RCP7 and the current Department of Health guidelines.4

Not-AR drinking was associated with high prescription medication levels, and men were more likely to be AR drinkers than women. There was also an association between rurality of dwelling and AR drinking showing that living in rural areas was associated with a higher level of AR drinking of >10 units per week. Greater age, and being on prescribed drugs, was associated with a lower level of AR drinking. This suggests that diminished health status as indicated by being prescribed drugs is associated with less AR drinking in individuals aged 65–89 years. Similar trends were evident when the cut-off point for AR drinking was taken as >14 units per week.

Comparison with existing literature

Our study is consistent with findings that alcohol consumption is likely to reduce because of declining health status,28 in particular, being on prescription drugs. The finding that community-based men over 65 years are more likely to be AR drinkers than women is similar to data from the American National Survey on Drug Use and Health.29 There are indications that living in a rural area or village is associated with a greater likelihood of AR drinking. This is an under-researched area but Li et al 30 found that Norwegian older adults living in urban areas were more likely to drink alcohol and this was associated with positive health outcomes. In contrast poor health outcomes including greater alcohol consumption were reported in the same paper in a Chinese sample of men living in rural areas.30 The association with the diminished likelihood of AR drinking as an individual gets older has been confirmed by international research.31 An American survey (n=83 321) of alcohol use and prescription drug in older people32 found that nearly twenty percent of the sample reported being prescribed an alcohol interactive medication and continued to drink. This is consistent with our finding that there are a proportion of older people who consume alcohol in combination with those medications known to cause an adverse reaction and are likely to be unaware of the risks involved.7 In this context it is noteworthy that one third, of all men and women aged 65 years and over are prescribed four or more types of prescription drug daily33 notwithstanding OTC drug consumption. There was no relationship between social class and AR drinking. The English Longitudinal Study of Ageing (ELSA)34 suggests that affluence, higher social class and alcohol consumption is associated with ‘successful’ ageing.35 The RCP8 acknowledged, that alcohol abuse in older people now needs to be considered in the context that alcohol plays in encouraging ‘social cohesion’. Important insights are provided by a 10-year longitudinal study36 which confirmed that higher social class was associated with greater alcohol consumption, and equally poor self-rated health was associated less alcohol consumption over time. The authors36 suggest that as successful ageing could be associated with a level of drinking, as it provides greater opportunities for social interaction, and as health declines there are fewer opportunities to facilitate this. However, it should be noted that ELSA includes individuals who are aged 50 years or more, unlike the current study which only considers those 65 years or more. The finding that being prescribed medication is associated with less AR drinking but not social class suggests that as some individuals age the positive impact of drinking and social class lessens, as their health status diminishes.

Strengths and limitations

The main strength of this study is that it is based on data taken from an established national dataset using a complex random probability sample derived from post codes.16 However, prescription drugs use is reliant on self-report and limited to those included in the survey. Thus, it was not possible to investigate the role of, for example, prescribed psychotropic medication other than antidepressants. The role of OTC medication cannot be assessed as these data were not collected in the HSE. US evidence indicates that older people living in the community, regularly use OTC medication though this was based on self-report data,37 and a study across 15 European countries found that in a sample of 46 394 adults with chronic pain over half of the sample used OTC analgesic medication.38 Alcohol consumption levels reported, may also have been under-estimated. It is likely that heavy drinkers would not participate in a general population survey39 this is coupled with a tendency towards underestimation of alcohol consumption in the general population40 and into what constitutes a unit of alcohol.41 In addition, estimates of reported wine consumption have underestimated the impact of the strength of drinks consumed and the size of glasses in which they are either served or poured.42

Implications for research and/or practice

We suggest that future waves of the HSE continue to survey prescription drug use, but expand the types of prescription drugs being assessed, and consider how to integrate assessing use of OTC medications. Thus, we recommend that public health professionals should highlight the need for GPs, pharmacists and other professionals to address the possible interaction of any prescribed or OTC medications with alcohol in older people. In addition, it would be beneficial to design information leaflets that can be distributed in health centres and pharmacies providing information concerning the interaction between certain prescription drugs/OTC medicines and alcohol. Finally, we also suggest that professional medical training bodies such as the General Medical council (GMC) (medicine), Nursing and Midwifery Council (NMC) (nursing) and General Pharmaceutical Council (GPhC) (pharmacy) address their current curricula to take steps to ascertain current practice and thus ensure that medication interactions with alcohol are addressed.