Generic substitution, financial interests, and imperfect agency

PD is the counterpart of prescribing pharmacists, who exist e.g. in the case of refills in the United States, Canada, the United Kingdom, and New Zealand (Emmerton et al. 2005). In both cases, the prescriber and the dispenser is one and the same person or institution, respectively.

The prices for brand-name drugs are also higher in Switzerland, but the markups for physicians are smaller (see Sect. 3.2).

The difference between margin optimization and maximization is that in the first case, PD providers prescribe several small packages instead of one large package while in the second case, they prescribe a higher quantity to maximize their income.

A small number of physicians works in managed-care type arrangements, where other modes of payment are possible.

This may be true although dummy variables for cantons and community types are included in the estimation in order to control for differences between regions (see Sect. 7).

The regulation of reimbursement allows generic producers to charge higher prices in Switzerland than elsewhere in Europe. In Switzerland, generic drugs have to be at least 40 % cheaper than the brand-name drugs. Thereafter, insurers are obliged to reimburse each price that is set by generic producers. By way of contrast, many European countries install reimbursement ceilings that are oriented towards the cheapest prices that are offered in the market (internal reference pricing, Vogler et al. 2008). If the price exceeds the ceiling, the difference needs to be paid by the insured, which is avoided by most generic producers.

The Swiss health insurers’ association (Santesuisse) scrutinizes physicians who exhibit inexplicably high cost of treatment compared to their peers and occasionally sues them.

In fact, non-dispensing physicians get a fee (TARMED) for prescribing a drug, which however does not differ between brand-name and generic drugs. This fee is therefore irrelevant to our analysis.

The mixed-effects model did not converge using Stata 10.

ATC-code: omeprazole (A02BC01), amlodipine (C08CA01), ciprofloxacin (J01MA02). For more details about the investigated agents see www.​drugbank.​ca/​drugs.

Number of generics available on the Swiss market (2005–2007): omeprazole (11), amlodipine (12), ciprofloxacin (11).

This is regulated by national law (specifically paragraph Art.38a KLV).

The concept of odds ratios and their calculation in the presence of interaction terms can be found in Hosmer and Lemeshow (2000).

The 95 % confidence interval is calculated according to \(CI=exp(\hat{\beta } \pm 1.96\cdot \widehat{SE}(\hat{\beta }))\), where \(\hat{\beta }\) is the logit coefficient. Because Tables 5 show ORs, the reader can calculate the necessary quantities according to \(\hat{\beta } = ln(\widehat{OR})\) and \(\widehat{SE}(\hat{\beta })=\widehat{SE}(\widehat{OR})/\widehat{OR}\) using the values from the table.

We additionally compared the prescribing behavior of non-dispensing physicians from regions where PD is prohibited with those located in communities that allow PD. While the first type of physicians might want to sell drugs but lack authorization, the latter type is free to dispense but does not size the opportunity to do so. Hence, the second type serves as a reference group, comprised of physicians that are not (or at least comparatively less) prone to financial incentives, because they let go on profits from drug selling. We found no empirical evidence for a difference in prescribing patterns between non-PD across cantons for the two substances omeprazole and amlodipine. For the third substance (ciprofloxacin), the likelihood of generic substitution is significantly lower in cantons with a less restrictive stance with regard to PD.

The theoretical model focuses on the first prescription, neglecting decisions with regard to follow-up prescriptions. In an attempt to make the econometrics match theory more closely, an estimation using only the first observation per patient was performed as well. The odds-ratios are 2.84 (0.38) for omeprazole, 1.90 (0.16) [specialists] and 6.84 (0.81) [GPs] for amlodipine, and 1.04 (0.11) [specialists] and 6.77 (0.72) [GPs] for ciprofloxacin, respectively. Therefore, the conclusions based on Table 5 seems to hold regardless of whether physicians decide about a first or a follow-up prescription.

Cost containment through cheaper generic drugs could be (over) compensated through unnecessarily prescribed drugs.

To facilitate the calculation, all models are estimated without an interaction of PD and GP. Further, one has to keep in mind that the brand-name drug went off patent in the first month of the study period. This contributes to a strong effect of PD because physicians with dispensing rights are targeted by sales activities and are immediately informed about market entry of new generic drugs.