Opioid prescribing habits differ between Denmark, Sweden and Norway – and they change over time

1 Introduction

The medical use of opioids in different countries is often subject to public debate and concern, both in the individual countries and in comparisons between countries. The discussions often reflect differing views for or against [1], [2] the use of opioids in pain relief in different patient populations or different pain settings while referring to the same bottom-line results from drug use statistics. Bearing in mind the potential for serious side effects and risks, but also the favorable effects and gains of using opioids, it is important that discussions concerning the use of opioids and policy making are based on the best possible evidence. Changes in opioid use statistics can appear more or less dramatic depending on the time periods and/or the types of opioids chosen for presentation. Latest, the “opioid epidemic” or “opioid crisis” in USA are intensely debated on the basis of aggregated prescription data of changes in opioid use in USA linked with opioid misuse, increased numbers of deaths among drug addicts [3], [4] and with opioids used in the treatment of chronic non-malignant pain [5]. This has occasionally led to scare campaigns against the use of opioids and stigmatizing of opioid prescribing doctors [6], and harmful tapering of patients in stable opioid use [7]. According to the International Narcotics Control Board (INCB) [8] USA has a leading position for consumption of narcotic drugs per million inhabitants, but the Scandinavian countries are close behind, with Denmark as no. 5, Norway as no. 15 and Sweden as no. 19 in the world. Discussions of how to prescribe and use opioids based on statistics like INCB are undesirable [9]. However, important knowledge on the opioid prescribing behavior can still be drawn from detailed and public assessable information enclosed in other national prescription-databases. When aggregated opioid prescription data are used for comparisons between countries or among selected patient populations it is important to consider at least two factors in order to reduce the risk of serious flaws in the interpretations and conclusions (the risk of the ecological fallacy [10]). These factors are; (1) the potencies of the different prescribed opioid types, and (2) differences in the doctors’ preferences with regard to choices of opioid types. The underlying reason for this is the pharmacology of the chemically different opioids, which all are defined by their μ-receptor agonist action but with different potencies and pharmacokinetics, resulting in large differences in equianalgesic doses between different opioid types.

This study explores national trends and changes in prescribing behavior in Denmark, Sweden and Norway by use of public assessable data on dispensed opioids. The aim is to demonstrate how these data can qualify and provide a deeper understanding of the mechanisms behind opioid prescribing and opioid use.

This study aims to:

1. Compare the amounts of dispensed opioids in Denmark, Sweden and Norway during 2006–2014, using public assessable information from each nation’s annual drug-use statistics.

2. Compare the opioid prescribing behavior in the three Scandinavian countries, and whether it has changed during 2006–2014

2 Methods

Each of the three Scandinavian countries has nationwide and complete prescription databases containing electronically submitted information of dispensed prescriptions [11]. Each country’s prescription database can be assessed on the internet, to obtain more detailed information on the annual amount of dispensed opioids, both with regard to opioid types and number of users. In this study, the amounts of dispensed opioids are used as proxy for the amounts of opioids used. Knowledge of the exact amounts of opioids prescribed and/or administered to the single users are not available in such databases. Information of the dispensed opioids from 2006 to 2014 was downloaded via the internet from the three databases [12], [13], [14]. The period was chosen to begin in 2006, because complete data were first assessable from July 2005 in the Swedish database [11]. Only data on opioids sold in the pharmacies were included, because this use can be linked to the number of users. The amount of opioids used in hospitals, which includes less than 5% of the total consumption, cannot be related to the number of users.

The different opioid types were identified in the prescription databases by their 7-digit ATC-code (Table 1). Opioids indicated for analgesia all start with N02A. Few other opioids are also relevant as analgesics – like methadone and codeine (ATC-code: R05DA04). These drugs are not routinely included in the usual opioid use statistics, because they are indicated for other conditions than analgesia. In this study, we only used data for N02A drugs, except in the presentation of the different opioid types’ proportions of the total volume, where R05DA04 – codeine was is included, because it is commonly used as an analgesic in Denmark, and to some extent in Norway. Opioids are classified in two groups – the weak and the strong opioids. Weak opioids are codeine, tramadol and dextropropoxyphene (blue color codes in Table 1), the remainders are strong opioids.

Table 1:

Opioids (N02A) registered in the three Nordic countries, with DDDs, equianalgesic potencies and conversion factors between DDDs and mg omeqs.

Color code (Fig. 3)	Opioid-type (N02A)	mg/DDDa	Equianalgesic potency in relation to oral morphineb	Conversion factorc
	N02AX02 Tramadol	300	0.2	60
	N02AX52 (375 mg tramadol+325 mg paracetamol)	150	0.2	30
	N02AA59 Codeine comb.	100	0.1	10
	N02AA08 Dihydrocodeine	150	0.1	15
	N02AC04 Dextropropoxiphene	300	0.15	45
	N02AC54 Dextropropoxyphene comb.	300	0.15	45
	N02AA01 Morphine	100	1	100
	N02AG01 Morphine, comb (only for injection)	10	3	30
	N02AA05 Oxycodone	75	1.5	112.5
	N02AA55 Oxycodone comb.	75	1.5	112.5
	N02AB03 Fentanyl	0.6	100	60
	N02AE01 Buprenorphine	1.2	75	90
	N02AB01 Ketobemidone	50	1	50
	N02AG02 Ketobemidone, comb.	25	1	25
	N02AA03 Hydromorphone	4	7.5	30
	N02AA04 Nicomorphine	30	1	30
	N02AB02 Pethidine	400	0.1	40
	N02AD01 pentazocine	200	0.17	34
	N02AX06 Tapentadol	400	0.33	132

1. ^amg pr DDD for each opioid are indexed in: http://www.whocc.no/atc_ddd_index/?code=N02A.

2. ^bEquianalgesic potencies between oral morphine and oral opioids, except for fentanyl and buprenorphine, where the potencies refer to the transdermal route and the sublingual route, respectively.

3. ^cThe conversion factors are the factors used for multiplication with the sold amounts of DDDs to estimate the sold amounts of mg oral morphine equivalents (see also Ref. [15]).

4. The color codes refer to Fig. 3 where the blue colors designate the weak opioids.

Annual values for; the number of opioid users per 1,000 inhabitants, the consumption of opioids per 1,000 inhabitants, and the average amount of opioids dispensed per user per day, were calculated. The size of the populations in the three countries were downloaded from public assessable demographic databases and shown in Table 2.

Consumption of opioids can be measured in defined daily doses (DDDs) or in mg oral morphine equivalents (mg omeq). The DDD for a specific drug is a predefined, international standard amount of a generic substance designed for drug use statistics assigned by WHO [16]. For many drugs one DDD will reflect the usual daily dose prescribed in the clinic. However, when it comes to pain treatment opioid doses vary between individuals to achieve the most optimal balance between effect and side effects. Therefore, DDDs for opioids may not necessarily reflect the doses used in different clinical situations. Furthermore, the assigned DDDs do not always mirror the analgesic potencies of the different opioids. For example, 1 DDD of codeine is 100 mg codeine and 1 DDD of morphine is 100 mg morphine. Morphine is around 10 times as potent as codeine, which implies that 10 DDDs codeine is considered equipotent to 1 DDD of morphine. This explains why different opioid prescribing habits in different populations can give rise to skewed pictures, if DDDs are used as units for consumption with regard to the clinical value of the opioid consumption [17]. For this reason, DDDs of opioids were converted to mg omeqs, using the same conversion factors as used in a recent paper on cancer patients’ consumption of opioids [15]. The conversion factors are the factors used for multiplication with the dispensed amounts of DDDs to estimate the dispensed amounts of mg oral morphine equivalents. Most of the conversion factors shown in the right column in Table 1 were calculated based on the DDDs assigned by WHO and the equianalgesic potencies between different orally administered opioid types and oral morphine. Two exceptions were fentanyl and buprenorphine, where the conversion factors were calculated on the basis of equianalgesic potencies for the transdermal route and the sublingual route, respectively. Apart from the two exceptions, the decisions to choose the equianalgesic potencies for orally administered opioids were; 1: the large amount of dispensed opioids in the pharmacies is for oral use, and 2: it is not possible to discriminate between DDDs sold for different administration routes in the public assessable data from the prescription registries.

3 Results

3.1 Opioid users and opioid consumption per 1,000 inhabitants

The numbers of opioid users per 1,000 inhabitants are shown in Fig. 1. Norway had the highest number of opioid users in the population and Denmark the lowest. The number of users per 1,000 inhabitants increased slightly over the period in both Denmark and Norway, but seemed to have stabilized in Denmark since 2012. Sweden appeared stable throughout the period. The absolute increase in number of users in each country throughout the period was 17% in Denmark (from 357,653 to 419,261 users), 7% in Sweden (from 714,354 to 760,957 users) and 18% in Norway (456,189–536,894 users).

Figure 2 shows the consumption of opioid per 1,000 inhabitants using two different units for measuring; the upper part using DDDs, and the lower using mg omeqs. When measured in DDDs, the three countries opioid consumption appeared rather similar, except for a small reduction in Sweden during the period (Fig. 2, upper panel). However, large differences appeared when the consumption was calculated in mg omeqs per 1,000 inhabitants (Fig. 2, lower panel). Denmark had a significantly larger and increasing consumption throughout the period compared to Sweden and Norway. Norway had the lowest consumption, but with an increasing tendency over the period. In Sweden, the decrease in consumption when measured in DDDs almost disappeared when the consumption was converted to mg omeqs.

Fig. 1:

Annual number of N02A opioid users per 1,000 inhabitants – during 2006–2014.

Fig. 2:

Average daily consumption of opioids per 1,000 inhabitants per day – during 2006–2014. DDD is used as unit in the upper panel, and mg omeq is used in the lower panel.

3.2 Dispensed opioids per user

In 2006, the annual average amount of opioids dispensed per user was 1979, 3615 and 6025 mg omeq in Norway, Sweden and Denmark, respectively, and in 2014 the corresponding figures were; 2426, 3473 and 6361 mg omeq. The average daily consumption is shown in Fig. 3, where Denmark had the highest consumption of opioids per user, and Norway the lowest.

Fig. 3:

Average daily opioid consumption per user – during 2006–2014.

In 2006 and 2014, the average daily consumption/user in Denmark was 3.0 and 2.6 times higher than in Norway. Norway showed a slight increase in average daily consumption/user over the period from 5.4 mg omeqs/day to 6.6 mg omeqs/day (equals a 22.6% increase). The consumption was rather stable in Sweden and Denmark over the period; in Sweden the consumption was 9.9 mg and 9.5 mg omeq/user/day in 2006 and 2014 (3.9% decrease), while in Denmark the consumption was 16.5 mg and 17.4 mg omeq/user/day in 2006 and 2014 (5.6% increase).

3.3 Choice of opioids

The preferred types of prescribed opioids in Denmark, Sweden and Norway in 2006 and 2014 are shown in Fig. 4. The figure shows each opioid’s proportion in the total annual opioid consumption in the countries in the 2 years when measured in mg omeqs.

The weak opioids are represented by the blue colors in the discs, and R05DA04 (light purple) is only represented in the Danish consumption with 3% and 2% in 2006 and 2014. In 2006 in Sweden, almost three quarters of the opioid consumption came from weak opioids. Gradually (annual data not shown) over the period, the preference to use strong opioids in both Sweden and Norway increased. In 2014 a little more than half of the opioid consumption came from the use of strong opioids in all three countries. In 2014, in Denmark morphine was the most frequently used strong opioid, while in both Sweden and Norway oxycodone was used more frequently than morphine. The relative consumption of oxycodone increased substantially over the period, in Sweden from 10% to 23%, and in Norway from 17% to 26%, while in Denmark there was a decrease from 22% in 2008–2010 to 16% in 2014. The relative consumption of tramadol (light blue) increased from 20% to 32% in Norway, from 42% to 47% in Denmark and decreased from 49% to 37% in Sweden. The use of dextropropoxyphene had disappeared in all three countries in 2014.

Fig. 4:

The different opioids’ share in the total consumption of opioids in mg omeq/year in 2006 and 2014 – in Denmark, Sweden and Norway. The blue colors represent weak opioids. **Ketobemidone combined with antispasmodics.

In 2014, the total consumption of codeine, R05DA04 plus N02AA59, in Denmark, Sweden and Norway was around 17,000, 27,000 and 35,000 mg omeq/1,000 inhabitants, respectively (Fig. 5). Codeine combined with paracetamol, was used much more frequently in Sweden and Norway compared to Denmark, where most of the codeine was prescribed as pure codeine (ATC code; R05DA04). In Denmark, 60% of the codeine consumption came from R05DA04 prescriptions, while in Sweden and Norway less than 0.2% and 0.1%, respectively, of the total codeine consumption came from R05DA04-prescriptions.

Fig. 5:

The average annual consumption of codeine (N02AA59+R05DA04) per 1,000 inhabitants in mg omeq – during 2006–2014.

4 Discussion and conclusion

Relatively large differences in opioid prescribing habits between the three Scandinavian countries were demonstrated. While Denmark had the fewest number of opioid users compared to Sweden and Norway, the average consumption of opioids per user or per 1,000 inhabitants was clearly higher when omeqs was used as unit for consumption. Figure 2 shows the implications of using either DDDs or omeqs as units for consumption, which was pointed out in a recent paper from Norway [17]. Several explanations for the differences between the countries average consumption per user can be considered; treatment with higher equianalgesic doses per user, longer treatment periods and/or a larger proportion of long-term users. Differences in the proportions of 65+-aged people in the populations (Table 2) could also play a role in the differences and changing patterns of opioid use. However, Sweden had the largest proportion of older people, while Denmark had the largest increase in this proportion since 2006. So, the proportion of older people does not seem to be a major explanation for the changing patterns of opioid consumption per user or per 1,000 inhabitants. The absolute increase in the number of users in Denmark and Norway was equal, and considerably higher than Sweden, while the changes in number of users per 1,000 inhabitants were higher in Norway compared to Denmark and Sweden. These differences are most likely explained by different changes in prescribing habits and to a lesser degree changes in the demography. In Denmark, a discussion of the risks of using NSAIDs has been very intense during 2009 and 2010, which might suggest that a proportion of NSAID-users had been switched to opioids for pain treatment. The differences between the three Scandinavian countries are unlikely to be explained by large differences in the prevalence of pain in the populations [18], [19]. In 2013 in Denmark, the proportion of long-term users (defined as continuous use over 6 months) constituted 35% of all opioid users (169,750 users) or 3% of the population, and as a consequence of this, most of the annual opioid consumption in Denmark was related to chronic use [20]. The duration of opioid use surpassed 1 year for around 65% of the Danish long-term users [20], which equals 2% of the population, while in Norway a recent study [21] showed that 1.3% of the population met the criteria for being regarded as persistent opioid users. No figures or estimates for the proportion of Swedish long-term users could be identified. The average consumption of opioids per user in Denmark and Sweden appeared rather stable during the 9 year period. There was a 22% increase in Norway throughout the period in this parameter. This relatively high increase reflected the increase from a very low to a less low consumption per user in Norway compared to Denmark and Sweden. Throughout the period Norway continued to have the lowest consumption per user of the three countries. The results showed that prescribing habits changed over time. The preferences for certain opioid types differed between all three countries, especially in the beginning of the period of investigation. In Sweden, the preference for using strong opioids increased over the period. In both Norway and Sweden oxycodone seemed to be preferred over morphine. In Denmark the use of oxycodone also increased until 2009 [22], but recommendations for morphine and restrictions for oxycodone as first choice of a strong opioid appear to have changed this behavior. The preferences for weak opioids also differed between the countries. In Denmark, codeine was used to a much lesser extent than in Sweden and Norway. In Denmark pure codeine seemed to be preferred over codeine in combination with paracetamol, in contrast with both Sweden and Norway where codeine combined with paracetamol was preferred.

The rather stable opioid consumption in all three countries can be interpreted as conservatism among prescribers within each country, with regard to indications for prescribing the opioids and with regard to the dosages used, measured in equianalgesic “currency”. On the other hand, the data have indicated a willingness among the prescribers to change their choice of opioid types in all three countries. Changes that may be explained both by regulatory and legislative demands from the authorities, guidelines and/or public debate.

The weaknesses and strengths in the study are mostly coherent with the way the prescription databases are built and administered [11]. The study was weakened from the facts that the results were restricted to opioids used in the primary care sector, and that the conversion factors used to change DDDs to omeqs were mainly based on orally administered opioids, as explained in the method’s section. The decision for the latter was the assumption that the overwhelming amount of opioids sold in the pharmacies was for oral use. The public assessable data on the amounts of DDDs sold from the pharmacies provided no knowledge on the distribution of opioids for different administration routes. Furthermore, the data provided no knowledge of indications, effects or side effects related to the dispensed prescriptions or whether people, who have redeemed the prescriptions, actually used the opioids (compliance). However, to the extent people do not consume the opioids they have redeemed from the pharmacies, the consumption per user is equivalently overestimated. A major strength of the study is the high accuracy and completeness in the registration of dispensed prescriptions, where the whole population in the country is the study cohort. The conversion of DDDs to mg omeqs suffers from the weaknesses caused by the underlying uncertainty in equianalgesic dose ratios [23], [24], so the amounts in mg omeqs must be considered as estimates and not exact amounts.

Apart from population based drug use statistics, record linkage between prescription registries and other population based databases hold many possibilities to achieve a more detailed understanding of the advantages or disadvantages of using opioids, and the data in the registries may also be useful in pharmaco-epidemiological follow up studies after pain treatment interventions.

5 Conclusion

Statistics based on dispensed opioid prescriptions reflect differences in medical culture and attitudes toward pharmacological pain treatment. High quality pharmacological pain treatment – i.e. using the right drug, initiated at the right time, provided in the right doses and for the right duration, to the right patient, will appear in the prescription databases’ aggregated data – and so will prescriptions from poor pain treatment behavior. The quality of pharmacological pain treatment cannot be appraised only from looking in drug use statistics. Prescription data combined with other sources of research based information can provide a clearer picture of “what is going on” with regard to the quality of the pain treatment on the larger scale. In this study it remains speculative, which of the three Scandinavian countries are closest to the optimal prescribing behavior, when it comes to treatment of opioid responsive pain? However, judged by the drug statistics used in this study, no opioid epidemic appears to be knocking on the doors in Denmark, Sweden or Norway.