A rational approach to the treatment of pain is to combine treatments that act on distinct pain mechanisms in order to improve analgesia and, hopefully, to reduce the incidence of adverse events [1]. This is the concept of multimodal analgesia. The World Health Organization (WHO) three-step Analgesic Ladder, proposed for cancer pain in 1986, is a stepwise approach to analgesic management, where a patient’s pain severity determines the level of analgesics [2]. In this ladder, when pain is not relieved by WHO step 1 analgesics such as paracetamol (PC) or non-steroidal anti-inflammatory drugs (NSAIDs), a WHO step 2 analgesic, i.e. one of three weak opioids – codeine, tramadol or dextropropoxyphene (DXP) – is recommended, generally in combination with PC.

However, DXP-related hepatotoxicity and its frequent use for suicidal poisoning reported in North Europe, United States and Australia, led to its progressive withdrawal. For instance, in the United Kingdom (UK), co-proxamol was withdrawn in 2007 [3]. In 2009, the European Medicines Agency (EMA) recommended the withdrawal of DXP/PC throughout the European Union [4]. Two years later, and despite objections from the French health authorities based on the benefit/risk ratio considered to be locally acceptable, DXP/PC was totally withdrawn from the French market, in March 2011 [5].

DXP/PC was widely used in France with more than 70 million of DXP/PC packs sold per year [6]. Until market withdrawal in France, DXP/PC was the second most prescribed analgesic drug after PC [7], and there were 41 different medications containing DXP/PC.

To accompany the withdrawal, the French Regulators, in collaboration with scientific societies, provided recommendations regarding therapeutic alternatives to DXP/PC. It was suggested to replace DXP/PC by another step 2 analgesic, e.g. tramadol or codeine, and by PC for weaker pain level [8].

While the consequences of DXP/PC withdrawal on suicidal deaths have been studied [9–15], its effect on overall analgesic prescriptions have been little investigated [3]. Due to their specific benefit/risk ratios and their extensive use, drugs replacing DXP/PC have impacted the management of pain, and the quality of care. These changes need to be detailed, first of all in terms of use of analgesics.

To provide a first set of data, an analysis of the Rhône-Alpes URCAM (Regional Union of Health Insurance Fund) database with 5 million inhabitants covered by the general health insurance scheme was performed, to describe analgesics dispensation between 2009 and 2012, around the time of DXP/PC withdrawal.

In France, the withdrawal of DXP/PC took place in two phases, with a first decrease in 2009 following an EMA opinion, and a second, final decrease in 2011 due to national regulatory decisions. The global dispensation data of prescribed analgesics suggest that the DXP/PC withdrawal had a small impact on the overall use of analgesics in France, as the total dispensations of these drugs decreased by 14% over the four years considered. However, over the same period, there was an increased use of PC, a step 1 analgesic, and to a lesser extent, of step 2 analgesics, codeine, opium, and tramadol, mostly combined with PC.

The data suggest that physicians replaced DXP/PC quickly after withdrawal from the French market. This was not straightforward, as this medication had been widely prescribed since 1964 – e.g., figures from 2009 show a use of 29 DDDs/1000 inhabitants/day – in a large set of indications. Part of the explanation for this quick replacement could be the fact that the benefit-risk ratio of DXP/PC was considered – by EMA, health care professionals (HCPs) and scientific societies – to be disputable. For instance, in 2008, a consensus conference on postoperative pain care concluded that DXP/PC should not be prescribed for this indication [18]. Surveys conducted among HCPs also revealed concerns regarding the safety profile or the limited efficacy of DXP/PC [19].

However, the use of analgesics after DXP withdrawal had not been predicted. Before withdrawal, step 2 analgesics, particularly tramadol/PC and codeine/PC, were expected to be used much more frequently, but our data show that after DXP withdrawal, the use of PC increased more than the use of step 2 analgesics. The reasons for this limited increase of step 2 analgesics could be related to safety concerns as tramadol is known for its poor tolerability, while codeine is under surveillance for its respiratory effects [20–22]. Also, opium-containing drug are seldom prescribed, except for elderly patients. In that context, for the prior indications of DXP/PC, prescribers probably chose PC, i.e. a less effective, but safe alternative to step 2 analgesics.

Of interest, the results of this study differ from the results of a survey performed among Pain specialists asked to describe alternatives to DXP in France [23]. HCPs declared tramadol combined with PC to be the substitutive analgesic of choice, while only 24% of considered PC alone as a substitute.

By contrast, another study conducted in a teaching hospital in 1997, i.e. long before withdrawal, suggested that DXP/PC should be predominantly replaced by PC alone, in agreement with our findings [7]. Also in line with our data, a study conducted after withdrawal among community-dwelling elderly suffering from chronic pain and previously treated with DXP/PC, showed that a majority of patients remained treated with step 2 analgesics, mainly tramadol, but that 40% were switched to step 1 drugs [24]. Altogether, the available data suggest that the choice of replacement analgesics depended on physician specialty, setting – e.g. primary vs. secondary care – indication, patients’ comorbidities and age.

The effects of DXP withdrawal have also been investigated in other countries, notably in the UK, where withdrawal was effective in 2008. In the UK, a 23%-increase in codeine/PC, a 19%-increase in tramadol and a 16%-increase in PC prescriptions were reported [25], confirming international differences in pain management.

Our findings had some limitations. This study relied on the use of aggregated data, i.e. monthly dispensations delivered to a population of five million people after analgesic prescribing by regional physicians. As such, it was not possible to distinguish successive episodes of use of analgesics in individuals, to identify analgesic therapy prescribed after DXP/PC withdrawal in chronic or repeated users. It was also not possible to assess the impact of therapeutic changes on the effectiveness of pain therapy, in the absence of patient-reported data. Access to individual drug histories would have allowed exploring differences in patients’ characteristics, such as gender, age or comorbidities, and differences in prescribers’ specialties, in addition to providing some markers of treatment effectiveness.

Also, our data did not allow to verify the occurrence of a storage phenomenon that was shown to delay DXP replacement in the UK, where 30% of patients were still using DXP/PC one year following its withdrawal [25].

A last limitation refers to the absence of over-the-counter data, since claims data include only information on drugs that were both prescribed and dispensed. However, prior research on this issue support the validity of the results obtained with claims data, as in France, PC is mostly used as prescribed therapy, while step 2 and step 3 analgesics are Prescription-Only-Medicines [26].

In conclusion, after DXP withdrawal, large increases in PC use were observed, while the use of other step 2 analgesics increased to a lesser extent. Detailed analyses of individual longitudinal drug histories would be helpful to confirm the nature and the effectiveness of replacement therapy according to patients’ and prescribers’ characteristics, and the kinetics of this process, besides allowing investigations of quality of care in analgesia, and much-needed international comparisons.