Social inequalities in use of potentially addictive drugs in Norway – use among disability pensioners

Objectives: The Norwegian Government urges that actions are needed to stimulate the working capacity in disability pensioners (DPs) with such a potential. Information on factors that may impair rehabilitation efforts, including use of potentially addictive drugs, may be useful in this context. Thus, the aim was to study the association between DP on initiation as well as long-term use of benzodiazepines (BZDs), and to describe aspects of problematic use of BZDs in terms of: long-term use pattern, including escalation of dose over time, and use of other potentially addictive drugs. Methods: We followed a cohort of 8,942 men and 10,578 women aged 40, 45, 60 years (non-users of BZDs at baseline), who participated in health surveys in 2000-01 in three Norwegian counties, with respect to use of BZDs, and other potentially addictive drugs, by linkage to the Norwegian Prescription Database (NorPD) for 2004-2007. Information on DP status was retrieved from Statistics Norway. Results: Incident BZD use was highest among female DPs; 18-20% compared to 5-8% of the non-DPs. Multivariable analyses revealed an independent effect of DP on incident (OR 1.6 (95% CI 1.4-2.0)) and long-term use (OR 2.47 (95% CI 1.90-3.20)) of BZDs. Among incident users, 51-60% of the DPs retrieved BZDs throughout the period 2004-07, as compared to 32-33% of the non-DPs. The annual median defined daily doses (DDDs) of BZDs among long-term users increased throughout the period 2004-07, most pronounced in the youngest DPs; from 50 (interquartile range (IQR) 14,140) DDD to 205 (IQR 25,352) DDD. Conclusions: The chance of being prescribed BZDs as well as becoming a long-term user was higher among DPs. High continuation rates, with a steadily increasing annual amount of use among the long-term users may reflect an unfavourable use pattern of potentially addictive drugs among DPs, most worrisome among the youngest.


INTRODUCTION
Socioeconomic inequalities of health have become an important part of the political agenda in Norway and other countries 1 .Commonly used measures of socioeconomic status are level of education, occupation and level of income 2 .In addition; other measures of sosioeconomic status are used, such as disability pension which is associated with a lower educational level 3 .Equal access to and quality of health care is supposed to be a fundamental part of the Norwegian welfaresystem 1 .In this context; drugs should be prescribed according to medical need, irrespective of socioeconomic circumstances, and problematic use of potentially addictive drugs should be equally avoided.
Studies on socioeconomic inequalities in drug use are relatively scarce in the literature.One area highlighted, however, concerns socioeconomic inequalities in the use of statins.Statins have a well documented beneficial effect on cardiovascular morbidity and mortality 4 .Thus, an inequality in statin use, due to sosioeconomic factors and not need, would imply inquality in the prevention of one of the leading causes of death in Norway.Swedish and Danish cross-sectional studies have shown a socioeconomic gradient in use of statins [5][6][7] .However, a recent Norwegian study showed that this gradient could be explained by different cardiovascular disease (CVD) risk factor profiles (medical need) prior to the initiation of therapy, at least for primary CVD prevention 8 .In the secondary prevention group, however, there was a tendency to a higher statin use among the highly educated compared with people of lower educational level, after adjustment for other CVD risk factors, particularly in women.It may reflect that highly educated women may be more aware of their own health and new treatments than women of low educational level, which in turn may add up as a factor contributing to the observed social gradient in cardiovascular mortality.
The proportion of Norwegians on disability pensions has doubled since the 1980s 9 .About 11 per cent of the working population (18-67 years) now receives disability benefit in Norway which constitutes both social and economical challenges.The Norwegian Government has urged action to arrest the upward trend of people on disability pension.Information on factors that may impair rehabilitation efforts, including unfavourable use of potentially addictive drugs, may be useful in this context.
In Norway, and other OECD countries, mental health disorders and musculoskeletal problems are the main reasons for claiming disability benefit 9 .Benzodiazepines (BZDs) have anxiolytic, sedative, anticonvulsant and muscle-relaxant effects, so they are likely to be used by disability pensioners to deal with these problems 10 .However, there are concerns about benzodiazepines because of their potential adverse effects, which include impaired cognitive function and psychomotor skills, unwanted sedation, as well as dependence associated with long-term use 10,11 .Because of this, benzodiazepines are only recommended for short-term use over a few weeks 12 , and guidelines emphasize that combined use with other potentially addictive drugs, such as opioids and carisoprodol, is especially problematic 13 .Unfavorable use may be of particular concern in disability pensioners, as it may add yet another potential problem to the deterioration of people's daily functioning.
To our knowledge, information on the use of potentially addictive drugs according to disability pension status is scarce in the literature.However, recent studies evaluating use of potentially addictive drugs over a 20 year period (from 40 to 60 years of age), suggest that such drugs are extensively used among disability pensioners, and that disability pension was an independent predictor of initiation as well as continued use of BZDs 14,15 .
These studies had a special focus on long-term use, with a 20 year period between the two measuring points.Thus, a limitation of these studies, was the description of use of benzodiazepines according to selfreported disability status 20 years in the past.The proportion on a disability pension increases with age; e.g. in 2006 about 25% of all men and 37% of all women aged 60 years in Norway were receivers of a disability pension, as compared to 1-2% of the 40-42 year old participants in the previous study 9 .Thus, an unknown number of the initially non-disabled 40-42 years old will have changed their disability status during this period.In this situation, the OR estimates presented will be underestimated, and biased towards the null.
Also, information on long-term use was collected in a limited cohort of individuals, 40-42 year olds only.We do not know whether these findings are valid for other age-groups.And last, the study population constituted individuals living in regions in Norway with above average consumption of anxiolytics and hypnotics over the past years, compared to other Norwegian regions 16 .Thus the trends shown in the previous studies may not be representative to other parts of Norway.
In this current study, we wanted to extend the knowledge of aspects of use of potentially addictive drugs to other age-groups, based on more recent data from regions with average overall consumption of these drugs.The aims of our study were to evaluate disability pension as a predictor of initiation and longterm use of benzodiazepine use among individuals reporting to be non-users at baseline, adjusting for several other factors that may influence use of such drugs.
Further, other aspects of problematic use of benzodiazepines studied were: long-term use pattern in terms of percentage of incident users becoming long-term users and escalation of dose over time, and prevalence of medication with other potentially addictive drugs (zhypnotics, opioids and carisoprodol).

Health surveys -information on independent variables
Data from Cohort of Norway (CONOR) was included in this study, which is a collection of health data and blood samples from several Norwegian health surveys described elsewhere 17,18 .CONOR data from population based surveys conducted in 2000-01 in the three Norwegian counties Oslo (HUBRO and the Oslo II study), Hedmark and Oppland (The OPPHED study) are included in this study, which cover both rural and urban regions.Our analyses are restricted to data on participants aged 61 years and younger, thus including people below the early age of retirement (contractual pension) only.At the time of investigation, the subjects were aged 40-41, 45-47 and 59-61 years (from now on referred to as two age-groups: 40+45 and 60).All subjects in selected age cohorts were invited to participate in the respective surveys.In these age cohorts 18,612 men and 18,273 women were invited in the HUBRO and OPPHED study.Of these, 9,363 men (50.3%) and 11,236 women (61.5%) participated and agreed to the storage of blood samples and data for research purposes and agreed to have their data linked to other health registers (Figure 1).In addition, our study-population included 46 male participants aged 61 years from the Oslo II health study, in which the overall response rate was 48.7% 17 .129 men and 124 women who died or emigrated before January 1 2004 were excluded from analyses.In a second step we removed all those whose drugs were reimbursed for treatment of cancer (Figure 1).
As our purpose was to study incident use of BZDs, BZD users at baseline were excluded.The drug use questions included an open question on drug trade names.Those who wrote trade names of benzodiazepines, were excluded from our study (245 men and 442 women).And last, persons with missing information in the administrative register on disability pension status (14 men and 6 women) were excluded resulting in a study-population of 19,520 people (8,942 men and 10,578 women).
The participants completed a self-administered questionnaire covering drug use questions, history of diseases and risk factors, and different lifestyle habits such as alcohol use, smoking and physical activity and others.See Table 1 for details on questions and variable definition.The English version of the questionnaire, as well as further details of the surveys, is available at the home page of the Norwegian Institute of Public Health 18 .Table 1.Questions with answering categories and variable definition from the health surveys in Oslo, Oppland and Hedmark 2000-2001.

Physical activity
Describe the physical activity in your spare-time, Description of four levels of physical activity, seldom/never physical active =1; active =2-4 on average during the last year.

Information on disability pension status
The variable "current activity status" from the nationwide population and housing census in Norway 2001 (Statistics Norway) was used to categorize individuals according to disability pension status 19 .This variable is based on information from several administrative registers, and contains information on all citizens living in Norway.The variable categorizes Norwegian individuals according to their labour force participation; employed, unemployed, in education, or receipt of national insurance benefit (disability-, retirement-or survivor pension).The study population was categorized as either receivers or non-receivers of a disability pension.

Norwegian Prescription Database (NorPD)information on dependent variables
Prescription data about benzodiazepines in 2004-2007 were drawn from the Norwegian Prescription Database (NorPD) which covers the entire nation (4.7 million inhabitants) 20,21 .From 1 January 2004 all the pharmacies in Norway became legally obliged to send in all electronic data on prescriptions.These returns are sent to the Norwegian Institute of Public Health.NorPD contains information of all individuals living outside institutions who have received prescription drugs dispensed at pharmacies.All prescriptions reimbursed or not, are stored in the database.The drugs are classified according to the Anatomical Therapeutic Chemical (ATC) classification system 22 .The data collected for our study were patients unique identifying number (encrypted), sex, age, the date of dispensing, and drug information (brand name, package size, number of packages, ATC-code, Defined Daily Dose (DDDs)).
The code of reimbursement is also recorded and this may function as a proxy of diagnosis.Code §9.9 is dedicated to cancer diseases.Benzodiazepines were defined by the ATC-codes N05BA (anxiolytics), N05CD (hypnotics) and N03AE01 (clonazepam), benzodiazepine-like drugs (z-hypnotics; zolpidem and zopiclone) by the ATCcode N05CF, opioids by the ATC-code N02A and carisoprodol by the ATC-code M01BA02.Incident use was defined when an individual (non-user at baseline) was dispensed at least one prescription of a benzodiazepine during the period 1 January 2004 -31 December 2004.Long-term use was defined when an individual (non-user at baseline) was dispensed at least one prescription of a benzodiazepine each of the years 2004 to 2007.Among BZD users in 2004, the proportion of long-term users throughout the period 2004-07 was calculated, according to disability pension status and age.
In addition, among the long-term benzodiazepine users, the annual amount of BZDs dispensed in the period 2004-2007 was calculated in terms of total defined daily doses (DDDs).
A DDD in clinical practice is defined as the assumed average maintenance dose per day for a drug used on its main indication in adults 23 .The DDDs for the benzodiazepines registered in Norway, and included in our analysis, are as follows; diazepam 10 mg, oxazepam 50 mg, alprazolam 1 mg, nitrazepam 5 mg, flunitrazepam 1 mg, midazolam 15 mg, clonazepam 8 mg.Data from the health surveys, administrative register and NorPD were linked based on the unique encrypted identification number.

Statistics
Chi-square test was used to assess equality of proportions across the groups of drug use.Mann-Whitney test was used for variables with a skewed distribution (DDDs, presented as median and interquartile ranges (IQR)).Association between disability pension status and others factors with incident BZD use in 2004 and long-term BZD use (2004-07) was estimated by logistic regression, and presented as crude and adjusted odds ratios (ORs) with 95% confidence intervals (CI).All analyses were done using SPSS 17.0 for Windows.Level of significance was set to p<0.05.

Ethical considerations
The study protocol was assessed by the Regional Committee for Medical Research Ethics and approved by the Norwegian Data Inspectorate.

Baseline characteristics (non-users of benzodiazepines)
Characteristics of the study population, non-users of benzodiazepines at baseline, are shown in Table 2.There was a higher prevalence of teetotallers and current daily smokers among disability pensioners.Further, there was a higher proportion of unmarried people (except in 60 year old women) and physically inactive persons (except in 60 year old men) among receivers of a disability pension.
About 2/3 of the disability pensioners reported poor self-perceived health at baseline, irrespective of age and gender.Also, the prevalences of mental health problems, fibromyalgia/chronic pain syndrome and cardiovascular disease were significantly higher among receivers of a disability pension.
Self-reported use of z-hypnotics, opioids and carisoprodol at baseline were higher among disability pensioners, except for z-hypnotics in 60-year old men.

Retrieval of prescription drugs in 2004 and 2004-2007
In both genders, and irrespective of age, incident BZD use 3-4 years later was significantly higher among disability pensioners (Table 2).Highest initiation rates were observed among female disability pensioners, of whom 18-20% had started BZDs, as compared to 5-8% of the female non-receivers of a disability pension at baseline.Prevalence of use of all other potentially addictive substances, such as z-hypnotics, opioids and carisoprodol had increased 3-4 years later, irrespective of age, gender and disability pension status at baseline.E.g. in 2004, 18-27% of disability pensioners retrieved a prescription on an opiod, about twice the level of retrieval among the non-receivers of a disability pension.
A significantly higher proportion of the disability pensioners became long-term users of benzodiazepines as compared to the non-receivers of a disability pension, 5-6% of the male and 10-11% of the female disability pensioners.

Predictors of BZD use
Univariate and adjusted ORs for incident retrieval of benzodiazepine prescription were higher among disability pensioners, as compared with those registered as non-receivers of a disability pension 3-4 years in the past (Table 3).
Further, univariate OR for long-term use of benzo-diazepines was above six times higher among the receivers compared to the non-receivers of a disability pensioners (Table 3).After adjustment for age, sex, lifestyle, physical and mental health, and socioeconomic variables the OR was lowered to 2.5 (1.9-3.2).Other independent predictors of long-term BZD use were; increasing age, female gender, an unmarried status, previous smoking, physical inactivity, poor self-perceived health, fibromyalgia/chronic pain syndrome and a CVD history.The strongest predictor of long-term BZD use was self-reported mental health problems at baseline.

Patterns of long-term BZD use, and amount of use among the long-term BZD users
The proportion of incident BZD users in 2004 who continued to retrieve BZD prescriptions throughout the period 2004-07 was higher among receivers compared to the non-receivers of a disability pension at baseline (Figure 2).Among BZD users in 2004, half of the   youngest (40+45 years at baseline) and 60% of the oldest (60 years at baseline) continued to retrieve benzodiazepines throughout the period 2004-07, as compared to a third of the non-receivers of a disability pension (Figure 2).Among long-term users, the median amount of BZDs retrieved among disability pensioners in 2004 was 50 DDD, as compared to 18-20 DDD among the non-receivers of a disability pension (Figure 3).The annual median DDD of benzodiazepines among continued users increased steadily throughout the period, most pronounced among the youngest disability pensioners.

DISCUSSION
In a study-population reporting to be non-users of BZDs at baseline; the proportion of disability pensioners that had started benzodiazepines 3-4 years later were about fourfold among 40+45 year old and about doubled in the 60 year old compared to those registered as non-receivers of a disability pension at baseline.Further, once started BZD use in 2004, continuation rates were high, with a steadily increasing amount of BZDs (DDDs) throughout the period 2004-2007.
A higher use of benzodiazepine use among those registered as disability pensioners may partly be explai-   ned by a higher prevalence of mental health symptoms in this group at baseline.Mental health problems are the main reason for being granted a disability pension in Norway, in young people in particular 24 .Thus, mental health problems were the strongest predictor of incident and long-term BZD use, and almost half of all 40+45 year old disability pensioners in our study were registered as receivers of a disability pension.Still, however, adjustment for various confounders, including self-perceived health and self-reported presence of mental health problems at baseline, revealed an independent effect of disability pension on the initiation and thereby long-term use of benzodiazepines 3-4 years later.Our study supports that initiation and longterm use of BZDs may be attributed to the DP situation which itself may bring about a number of psychological problems as a result of exclusion from the social and personal advantages associated with being a part of the workforce 25 .Drug treatment with benzodiazepines may be initiated to relieve these symptoms.Our evaluation of disability pension as a risk-factor of benzodiazepine use must, however, be interpreted with caution.Even if many potential confounders associated with prescription of benzodiazepines were registered and adjusted for in our analysis, all possible confounders may not have been taken into account.Our observations of an independent effect of disability pension on the initiation and long-term use of BZDs, however, are in agreement with a previous study evaluating incident benzodiazepine-taking behaviour over a 20 year span 14,15 .
Because of concerns related to the potential sideeffects, such as impaired cognitive function and psychomotor skills, and addiction problems 10,11 , benzodiazepines are only recommended for short-term use over a few weeks 12 .Still 50-60 per cent of those disability pensioners who retrieved a BZD prescription in 2004 continued to retrieve BZD prescriptions throughout the period 2004-2007.Parallel continuation rates were observed over a 20-year period in a previous study; more than half of those on disability pensions continued to receive benzodiazepine prescriptions 20 years later, a span covering a large part of the potential active workforce period 14 .Our findings add to observations in other studies on long-term use; having once entered a cohort of benzodiazepine users, the probability of still being part of that cohort in the following years is consistently high in different populations [26][27][28] .Thus, in this context, high initiation rates among the youngest disability pensioners observed in our study may perhaps be most worrisome.Once started, larger proportions will most likely continue to use BZDs throughout their potential work-force period.
One could argue that one annual prescription over a 4 year period is neither an outcome for long-term use, nor an addictive use pattern.However, the observed annual increase in amount of benzodiazepines dispensed over this 4-year period indicates more than sporadic use, among the youngest disability pensioners (40+45 years) in particular.In 2004 half of the youngest disability pensioners were dispensed an annual amount of benzodiazepines corresponding to the use of a daily dose every week (median 50 daily doses), increasing to retrieval of a daily dose every second day in 2007 (median 205 daily doses).A corresponding amount, a dose every second day, was observed in the previous study evaluating continued use of benzodiazepines over a 20-year period 14 .In addition, and relevant to the discussion of unfavourable use of potentially addictive drugs, baseline level (2000-2001) of use of all other potentially addictive substances, such as z-hypnotics, opioids and carisoprodol had increased 3-4 years later in disability pensioners, irrespective of age and gender.Parallel, as revealed in our previous study, the majority of those who were still dispensed benzodiazepines 20 years later, were also dispensed opioids; half of all men and 3 out of four women.15-20% retrieved benzodiazepines combined with both opioids and carisoprodol 14 .Guidelines emphasize that combined use with other potentially addictive drugs, such as opioids and carisoprodol, is especially problematic 12,13 .These recommendations are being violated when continued and combined use of benzodiazepines exists to the extent observed in our present and previous studies.Unfavourable use may be of particular concern in disability pensioners, as it may add yet another potential problem to the deterioration of people's daily functioning, help consolidate the disability situation.
As in previous studies, in the present study we chose to focus on use of traditional benzodiazepines, with its well known potential of causing addiction problems and other negative effects interfering with people's daily functioning 10,11 .Thus, we excluded prescriptions on benzodiazepine-like hypnotics from our analysis.In Norway benzodiazepine-like hypnotics, zhypnotics, are now recommended as first-choice hypnotics for short-term use in severe insomnia.These substances, zopiclone in particular, are the most commonly used hypnotics in Norway.Hence, 85-90% of all individuals who retrieved at least one prescription on a hypnotic in Oslo, Hedmark and Oppland were users of benzodiazepine-like hypnotics in 2007 21 .Recent literature, however, adds to the growing evidence that these substances have a similar potential for negative effects as traditional benzodiazepines [29][30][31] .Still we wanted to restrict our analysis to the prescription of the traditional benzodiazepines, which constitutes a better documented marker of an addictive use pattern 29,32,33 .
This study is prospective with information from nationwide registers on exposure (disability pension status) and outcome (BZDs) variables.However, our study-population is attendants in health surveys, in which attendance rates at 50-60% may imply a risk of selection bias.Unfortunately we do not have information on all non-attendants in our study.However, an evaluation of the non-attendees in the Oslo Health Survey (HUBRO) was performed; receivers of a dis-ability pension were over-represented among the nonattendees, while some health indicators, including mental health (HSCL), probably did not differ 34 .Low attendance may bias the representativity of baseline characteristics or prevalences, but is less likely to influence estimates of associations such as odds ratios.Further, self-reported use of BZD at baseline may be underreported.In general, anxiolytics and hypnotics/ sedatives are prone to be underreported, possibly due to the stigma of mental health problems 35 .Thus, there may be an unknown number of BZD users at baseline included in our study-population.
In conclusion, compared to previous studies which had a special focus on incident and long-term use of benzodiazepines over a 20-year period, this study adds complementary and new information of aspects of use of potentially addictive drugs among disability pensioners in Norway.Our study suggests that BZDs, and other potentially addictive drugs, are extensively used among disability pensioners, and that disability pen-sion may have an independent effect on initiation followed by long-term use of BZDs.Once started, rates of continued retrieval of BZDs are high, with a steadily increasing annual amount of use, most pronounced among the youngest disability pensioners.Efforts to bring those with potential to work back into employment have been a part of the political agenda for the last decade.Improved management of BZD use may be one part of this effort.

Figure 2 .
Figure 2. Proportion of long-term use of benzodiazepines among users (men and women) in 2004, throughout the period 2004-07, according to disability pension status and age.DP = disability pensioners.

Figure 3 .
Figure 3. Median amount (DDDs with interquartile range (IQR)) of benzodiazepines retrieved among long-term users (men and women) throughout the period 2004-2007, according to age and disability pension status.DP= disability pensioners.

Table 2 .
Prevalence (n (%)) of baseline characteristics among men and women who were nonusers of benzodiazepines when surveyed in 2000-01.Use of prescription drugs retrieved from the Norwegian Prescription Database in 2004-2007 according to baseline characteristics.
a Sought help for mental health problems b use was defined as retrieveal of at least 1 prescription in 2004 c long-term use was defined as retrieveal of at least 1 prescription in2004, 2005, 2006 and 2007