Determinants of drug expenditure in the Swiss healthcare market in 2006

In 2018, drugs were sold for CHF 9.5 billion (public prices) in Switzerland [1], which represented 11.8% of total health costs (CHF 80.2 billion) and CHF 1,108 per inhabitant. In terms of ex-factory prices, the annual turnover of drugs rose from CHF 3.3 to CHF 5.0 billion between 2006 to 2018 [2], corresponding to an increase in ex-factory cost of drugs per capita of 33.5%, while the inflation was 2.5% during the same period [3] (Table 1). A part of this increase was related to a growth in the number of prescribed packaging per capita (+ 7.4%), but a higher part was due to the ex-factory price increases (+ 24.3%). This price driven contribution to drugs expenditure was also observed in other countries [4].

Several measures to curb drug expenditures are in force in Switzerland since 2001. First, patients have to pay the full cost until they reach a certain deductible, set by contract with their health insurer [5]. When they reach their deductible, patients still have to pay a 10% share up to a certain ceiling. Second, pharmacists receive an administrative tax for delivering drugs to reduce the perverse incentive of a remuneration proportional to the net sales [6]. Third, pharmacists are allowed to substitute medical prescriptions for the cheapest products if there are generic drugs [7]. Fourth, the prices of drugs are set administratively by the Confederation (Federal Office of Public Health, FOPH) on the basis of legal criteria (i.e. economics, efficiency, adequacy of the prescriptions according to scientific criteria), after a notice from the Swiss Medicines Commission [8].

The effectiveness of these measures was however criticized [9], and given the strong and continuing rise in the drug expenditure in Switzerland, politicians are willing to understand determinants of this rise in order to design appropriate measures that would limit it.

In 2017, the Swiss Federal Council commissioned a panel of expert to tackle the issue of costs increase in compulsory health insurance [10]. Among the 37 measures proposed in their report, four might help reducing drug expenditure. In August 2019, the Federal Council proposed to the Swiss Parliament to implement a first set of nine measures, including one related to drug pricing. Such a measure is applied in Europe [11], and consists in introducing a reference price for three kinds of drugs: generic drugs (i.e. same active substance as the original drug available when the patent has expired), "me-too" (i.e. different active substances with the same mechanism of action as the original drug), and some patent-protected drugs that are not very innovative. For drugs whose price is higher than the reference price, the delta would be at the expense of the patient. However, the motion currently (2022) discussed by the Swiss Parliament is limited to generics. In order to determine whether imposing a reference price to other substances (i.e. me-too and patent-protected drugs) is necessary, one might be willing to look at the impact of the age of drugs (i.e. the time since they have been introduced on the Swiss reimbursement list) on their prices. Indeed, if products introduced recently, by stimulating competition, reduces drug prices, extending reference prices to other substances (“me-too”) is not necessary. On the opposite, it might be wise to recommend this extension if medicinal products introduced recently tends to increase expenditure.

From a theoretical economic point of view, the first assertion should be observed in a perfect competition market [12]. Nevertheless, there is some evidence that the drug market is imperfect [13]. It is thus important to analyze empirically the determinants of drug retail expenditure, which we did in this study. Our objective was to analyze the possible impact of several variables on different levels: market (number of treated patients, competition for instance), physician (specialist or not), patients (age, gender, deductible, co-morbidities), regions (cantons), and the treatments (generics for instance), with, for every level, a special attention paid to the recentness of drugs (i.e. minus the age of drugs).

A recent literature review has shown a significant effect of the expiry date of patents, the access to the generic market, and it emphasized the importance of tendering [14]. However, this review paper also indicates that scientific knowledge on the determinants of drugs expenses is lacking, largely related to a lack of transparency on this particular market. What is more, in Switzerland it is difficult to obtain representative drug data from insurers because such data could reveal the risk structure of insured and harm their risk financial compensation interests. Our study is based on 2006 data that four major insurance companies accepted to provide, anonymously, thanks to the support of the Federal Office of Public Health [15]. Although these data are arguably old, they provide a valuable and very rare source to understand the functioning of the Swiss drugs’ market, while presenting some important advantages: they are highly representative of the Swiss population and they describe a drug market offering a significant choice among multiple products (18 per specific market in average) at a time when the generics were significantly gaining importance.

In 2016, drug expenditure represented 11.8% of the total health costs in Switzerland. While several measures are being proposed to curb these ever-growing costs, we found it interesting to propose an overview of the likely determinants of drug expenditure and to assess their relative impact. We started by partitioning the total variance in drug expenditure across the five levels: treatment, patient, physician, market, and canton. Then, we included explanatory variables at each level and determined the contribution of these variables to explain the variance measured at each level. To assess the impact of each statistically significant variable on drug expenditure on an intuitive scale, we computed contrasts.

We found that more than two third of the drug expenditure variance was determined at the market level, and the variable having the greatest impact on drug expenditure at this level was market size (average expenditure decreased by 108 CHF in market with 10,000 vs 100,000 insured). Notice, unfortunately, that market size is not controllable per se as it depends mainly on the number of people with a specific illness and may not be used as tool to control drug expenditure. Notwithstanding this, Swiss drug authorities might still strive to negotiate lower prices with their providers together with other countries. Regarding market recentness, we found a sharp contrast between markets with new drugs and markets with older drugs (there was a difference of 104.8 CHF when the average age of drugs differed by 15 years). However, this contrast describes markets selling different drugs which may not be substitutable. Nevertheless, some markets may be substitutable and fostering the selling of old drugs proven to be efficacious may be one way to reduce expenditure. Of the four variables (HHI, number of mee-too, number of drugs, and number of brands) used in turn in different models to measure competition on the drug market none turned out to be significantly associated with expenditure. This is somewhat unexpected given economic theory claiming that in a free-market competition will lower prices.

The treatment level also had a significant influence on drug expenditure (about 20% of the expenditure variability was explained at this level). The number of prescribers had the largest impact (drugs prescribed by two physicians were 217 CHF more expensive than those prescribed by one single physician). This should not be over interpreted, however, as in practice most insured had (on average) between 1.0 to 1.1 prescribers and the effect of this variable would thus be much weaker. This effect might be explained by different mechanisms. First, we might assume that having multiple prescribers is a proxy for drugs for chronic conditions, which naturally require lengthy and more costly treatments. In this case, unfortunately, prohibiting consulting several physicians would not produce any economies. Alternatively, it also possible that patients solicit other physicians to lengthen the prescription duration that was refused by the first physician (e.g. sleeping pills); gatekeeping mechanisms might be effective to reduce costs in such cases. We found that the use of generics (-58.6 CHF) and older drugs (-36.2 CHF) were associated with lower expenditure. Generics are known to be cheaper than originals and incentives to use generics might contribute to cost reduction.

We found that older, substitutable, drugs were cheaper than newly released treatments, raising the question of the added value of drugs’ recentness. Between 2012 and 2014, 68.9% of 270 newly registered medications corresponded indeed to already known active substances [29]. This shows that most of the newly released medications consisted essentially in a modification of the presentation of the drugs (administration mode, dosage, number of pills, co-marketing of same substance) or in “me-too” drugs without great differences in efficacy. One option to curb this tendency could be to adopt a same fixed price per defined daily dose (DDD) for substitutable drugs among a same ATC 4^th level (generics and me-too drugs). This would provide incentives to convince physicians and patients to choose the most effective and comfortable drugs (substance, administration way, packaging, etc.) without pernicious effects on expenditure. This strategy makes sense given that physicians are often unaware of the actual cost of medications [30]. Another solution would consist in fixing prices per defined daily dose (DDD) for all products within a same market, which would not change the prices at short term but would at least stabilize them. Of course, innovative drugs (i.e. new substances) might increase expenses ([31] for instance), but this was not studied here, because our results were adjusted by markets (ATC4^th level). Finally, although Rx prescription should not depend on expenses’ aspects, the fact that drugs requiring a physician’s prescription tended to be associated with higher expenditure (+ 92 CHF) might suggest that marketing efforts were effective [32].

The three remaining levels were responsible for only 5% of the drug expenditure variability, showing that measures based on patients, physicians or cantons should not have a great impact on costs reduction. At the patient level, we found that elderly and patients with multi-morbid conditions might receive more costly drugs to avoid interaction for instance, confirming the results of previous studies [33]. The contrasts observed for the other patient-level variables (deductible and gender) were small. These results confirm the weak role of patients in choosing treatments, even if a greater involvement of them might be desirable [34]. Although the physician level was associated to less than 1.7% of the drug expenditure variance, variables that were introduced in our model at this level (i.e. type of physician and their propensity of using recently introduced drugs) explained half of it. Finally, it is interesting to see that regional disparities were relatively slight, contrarily to other geographic comparison of drug consumption in Switzerland [35].

The main finding of this study is that drug recentness was associated with an increase in drug expenditure, confirming former studies stating that pharmaceuticals recently introduced to the Swiss market generated higher costs [36]. In most economic sectors, new products enables the competitors to keep market shares, in a context of stable or decreasing prices. But, in the drug sector, it is a way to increase expenditure. Moreover, we found that several other aspects of the economic theory of perfect competition did not apply (e.g. more concurrence and higher deductibles were not associated with a decrease in drug expenditure). Such results are congruent with former researches on the absence of effect of the co-payment on rational use of medicines [37], and on the weak impact of financial incentives on prescribing medicines [38, 39]. They also provide evidence that the drug market cannot be considered as a “perfect market”, knowing that the patient is the customer, the decision is medical, the payment is not made by the insurance, the pharmaceutical industry sometimes has a dominant position, the entry of a new supplier into the market is not easy, and the substitution among products is not always possible [13].

Our study had some limitations. First, we assumed that physicians might substitute drugs in each market without damaging the efficacy of treatment. Even if this assumption should hold in most of the cases, physicians might have good reasons to choose a more costly drug, e.g. to provide a better comfort to the patient (administration mode for instance), or to avoid some side effects or drugs’ interactions [40], especially for multi-morbid patients [41]. However, we adjusted our results for age and co-morbidity index and only substitution among the same or similar substances were taken into account.

Second, our dataset was relatively old (i.e. 2006). There is no guarantee that the results obtained that year would be identical today. For this reason, this analysis should be seen as historical (or even pioneer), illustrating which kind of conclusions might be obtained with similar data and model. For instance, asserting that the variables linked to market and treatments (generic, etc.) explain respectively 76% and 20% of the variance of annual treatment costs provides an information that is crucial to know to take sensible decisions. Whereas it is well known that superficial innovations (co-marketing, packaging, etc.) might help imposing higher prices, it is also important to confirm it empirically and, eventually, to measure the magnitude of this effect. If our results were to be confirmed with updated data, it would indicate that it is probably more effective to act on the structure of the market than to target patients or doctors. An analysis with more updated data would also enable to take into account new aspects such as the emergence of healthcare networks or biosimilar drugs [42, 43].

Some drivers of drugs’ expenditures in 2006 Switzerland’s market might well still play an important role in today’s drug market, both markets displaying several similarities (e.g. many substitutable drugs, new products regularly introduced in the reimbursement list of drugs). However, we must also mention some changes that have occurred in 16 years, such as the apparition of new treatments against cancer and autoimmune diseases, an increased pressure of Federal authorities on prices since 2012, and a growing market share for generics. We could thus only encourage further research on this subject with updated data to see whether and to which extent results obtained here still apply, in Switzerland or in other countries, and we hope that this study will motivate stakeholders to make the required datasets available in the near future.