SK. .sci. .MPd Vol. 31. No. 2. pp. 153-157. 1990 Primed in Great Bntain. All rights reserved

0277-9536’90 53.00 + 0.00 Copyright $ 1990 Pergamon Press plc

RESPONSIVENESS TO PHYSICIANS’ REQUESTS FOR INFORMATION CONCERNING DRUG INTERACTIONS: A COMPARISON OF BRAND AND GENERIC COMPANIES

MARTIN THOMAS’* and JOEL LEXCHIN?

‘Department of Political Science. York University, 4700 Keele St, Downsview, Ontario. Canada M3J IP3 and ‘121 Walmer Rd, Toronto, Ontario, Canada M5R 2X8

Abstract-Research-based pharmaceutical companies maintain that there are important differences between themselves and their generic competitors. Prominent among them is an alleged greater ability to provide accurate and rapid responses to requests from physicians for information about drug products. This study evaluates pharmaceutical company behavior with regard to these issues.

Two drug-drug interactions were identified, along with all of the companies in Canada marketing any of the four drugs involved. Each company received a letter describing symptoms suggestive of an interaction in a patient taking its particular product and the relevant second drug. The companies were asked if they were aware of any evidence of an interaction involving the two drugs. They were also asked to provide references regarding the interaction. Responses were received from all companies contacted except one.

There were no significant differences (in the hypothesized direction) between the generic and brand companies with regard to either the accuracy or promptness of the response, or the usefulness of the references cited. On the contrary, generic firms were markedly quicker to respond than were brand manufacturers. The latter were slightly more likely to acknowledge evidence of an adverse drug interaction, and to provide useful references to relevant published research.

Key rcsords-pharmaceutical manufacturers, drug information, Canada

INTRODUCTION

Research-based pharmaceutical companies routinely claim that they are able to provide comprehensive and accurate information about their products to physicians and other interested health care providers.

They also often contend that they are superior to their generic competitors in that regard.

The assertions of research-based (or name- brand) drug companies in Canada are illustrative. The Pharmaceutical Manufacturers Association of Canada (PMAC), which represents them, contends that its members maintain “a complete scientific information service on all of the company’s products in order. . . to be able to provide factual, current information to members of the health professions” [I, p. 1081. The claims of the PMAC concern not only the competence and commitment of its mem- bers, but also their superiority to generic competitors: “the source of the most complete information on a new drug, its advantages and disadvantages, its side effects and its results in worldwide use in medical practice is obviously its originator” (I, p. 1081. More pointedly, generic firms are accused of having “little interest in investing in information and promotion support to keep the health professions informed on the current safety/efficacy profile of the generic brand” [I, p. 1091. Finally, the compari- sons with generic companies refer not only to the quality of responses, but also to their alacrity. The PMAC alleges its members “can make this

*Address correspondence to: Martin Thomas, 49 Melgund Rd, Toronto, Ontario, Canada MSR 2AI

information available most promptly. . . to the pro- fessions” [I, p. 108).

The implications of these kinds of claims on behalf of research-based manufacturers are that they bear a significant cost in maintaining large, highly skilled professional staffs to respond to questions about their products, that generic companies do not have comparable staff expertise or expenses and that costs incurred in attempting to serve the community ultimately must be reflected in the prices of the products marketed.

This research examines the responses of name- brand and generic drug manufacturers to physicians’ requests for information concerning an adverse drug interaction. The research focuses on companies in Canada, where the topic is particularly relevant due to the prominent role generic companies play in providing drugs through the provincially adminis- tered drug plans.

The following hypotheses are tested.

The responses of research-based drug manufacturers to physicians’ requests for information about drug interactions are more likely than those of generic companies to provide accurate information. Requests for information about drug inter- actions tend to be answered more rapidly by research-based manufacturers than by generic drug producers. The responses of research-based companies are more helpful than those of generic firms in citing the relevant professional litera- ture.

153

154 MARTIN THOMAS and JOEL LESCHIS

LITERATURE REVIEW

There is not a great deal of empirical evidence to either support or refute the contentions of the re- search-based manufacturers, and to date, none of the studies comparing the responsiveness of different kinds of pharmaceutical companies has examined the behavior of firms in Canada.

Two research projects which did not distinguish between types of manufacturers are nevertheless relevant. Russell and Dahlin [2] asked 20 multi- national companies for the same information that physicians routinely receive with each of the com- pany’s products. Only 9 of the firms bothered to respond, and of those, only one complied with the request. In contrast, Tse and Mansour [3] received responses to all 85 of their requests for information over an 8 month period. They deemed 60% of the responses ‘complete,’ another 34% ‘partial’ and the rest ‘unsatisfactory.’

Bainbridge rt al. [4], employees of a name-brand company, surveyed a variety of health-care providers who had requested information about some of the company’s drug products. Although 60% of the responses pertained to only two specific drugs, and although the respondents could not have regarded the researchers as impartial, the findings are still interest- ing. Only 52% of the respondents felt that their question had been answered completely. Further- more, more than half of the respondents reported that they did not receive a response within 2 weeks.

Tse [5] compared the responsiveness of relatively small manufacturers (those producing 25 or fewer products) and large companies. He found that the two groups did not differ with respect to response rate. but did with respect to response quality. The larger firms were more likely to give unsatisfactory responses. Although research-based manufacturers tend to be larger, Tse did not indicate which, if any, of the firms he surveyed were research-based.

Knight and Baumgartner [6] compared research- based and generic companies with respect to the provision of quality control information, but found no appreciable intergroup differences.

Finally, Generali and Hogan [7] telephoned 45 companies which marketed a popular antibiotic, re- questing information about a possible interaction between it and another drug. The respondents in- cluded 18 name-brand and 22 generic companies. Seven of the former and 4 of the latter provided information on the interaction. Only 5, all name- brand manufacturers, provided documentation de- scribing the interaction. Although neither group’s responses are exemplary, this study does provide some support for the PMAC claim that research- based manufacturers provide superior information when asked. Generali and Hogan gave no informa- tion on the timeliness of the responses.

METHOD

Ocerrieu

All manufacturers of specific selected drugs were sent a letter on physician’s stationery. It purported to describe a real patient (who was, in fact, fictitious) exhibiting symptoms of an adverse drug interaction

involving one of the company’s products and another drug. The letter asked about the possibility of such an interaction. and requested references to, or photo- copies of. relevant publications.

Sample

Since the intent is not to draw inferences about the groups of drug companies generally. but about their responses to typical requests for information. the sample was selected not from a population of compa- nies, but from a population of drug interactions. The optimal research design is one in which the two groups of companies experience identical treatments. The sampling method, therefore, is a form of cluster sampling. in which each drug selected identifies a group of firms. Their responses are the units of analysis.

The drug interactions selected met two criteria. They had been well-reported in the professional literature, and were manufactured by a large number of firms, including both research-based and generic types. Two interactions satisfying these criteria were identified from The Medical Letter Handbook of Drug Interactions (81: erythromycin with aminophylline (interaction I), and hydrochlorthiazide with allo- purinol (interaction 2). Research has consistently shown interaction I to be a true adverse drug inter- action. Interaction 2 is not as well understood, and may be either a true interaction or the result of pre-existing renal disease.

All companies marketing any of these four drugs in Canada, whether as single agents or in combination products. were identified, using the Compendium of Pharmaceuticals and Specialties (CPS) [9]. Companies marketing more than one of the drugs were contacted only once, to avoid arousing suspicion concerning the legitimacy of the request. Similarly, if selected compa- nies were known to be organizationally related, only one of them was contacted. In both of these kinds of cases, the choice of a specific company was made to equalize the number of companies contacted concern- ing each of the interactions.

Letters were sent to 35 companies. 21 of which were research-based. Three companies indicated that they no longer manufactured the drug, or never had. Another company apparently had gone out of business. The working sample consisted of responses from 31 drug companies, 20 of which were research- based.

Significance tests

Because the hypotheses are stated directionally, statistical tests are one-tailed. Probabilities are re- ported only if the relationship is in the hypothesized direction.

Operational definitions

PMAC membership. Name-brand companies were defined as those on the PMAC’s 1988 membership list, or subsidiaries of those on the list. XII others are defined as generic companies.

Response time. The rapidity of response was oper- ationally defined in two ways. One variable was dichotomous. reflecting whether the company re- sponded prior to receiving a follow-up letter, mailed 30 days after the original request had been sent. The

Responsiveness of brand and generic companies 155

other variable represents the number of days between the original request. and the date of the response postmark, or the phoned response, whichever was earlier. A dated ‘receipt notification’ was not used because of the possibility that it would create sus- picion concerning the legitimacy of the request.

Response yualit_r. Two measures of response quality are used, both operationalized in terms of the usefulness of the information to a typical physician. The first measures the accuracy of the written re- sponses and accompanying material. Three physi- cians independently evaluated the responses after being provided with a written description of the current state of knowledge concerning the drug inter- actions. The physicians were asked to evaluate the responses on a three-point scale defined numerically (0, 1, 2). and verbally (‘does not support the consen- sus in the medical literature,’ ‘weakly supports the .‘. and ‘strongly supports the . . .‘). This can be viewed as an interval scale. but is conservatively treated as ordinal for the sake of the statistical test.

When both a telephone and letter response were received. only the latter was scored. In each of those cases, it was obvious that the letter was a response to the original request for information, and not simply a follow-up to the phone call. In the two cases in which only a phone response was received, the score was assigned after consultation with the physician who had received the call.

The second ‘response quality’ variable is somewhat less relevant, measuring the usefulness of publications cited. Although no attempt was made to grade indi- vidual references with regard to either their quality or the nature of their conclusions, they were valued differentially according to type: textbooks or litera- ture reviews, professional journal articles, and com- mercial or trade publications. The index, described more completely in Table 1, is based on both the number and type of citation.

RESULTS

Because the two drug interactions seemingly differed with regard to both obscurity and complex- ity, the possibility existed that observed associations between type of company and the measures of re- sponsiveness would be spurious, caused by factors related to the drug interaction. The data indicate no statistical relationship between interaction type and company type, with 65% of research-based, and 64% of generic companies producing the products relevant to interaction 1.

Response time

All companies responded to the request for infor- mation, with the exceptions noted previously. Of the 20 research-based corporations, 19 responded without receiving a follow-up letter. The other one ultimately responded almost 2 months after the follow-up letter was mailed, attributing the delay to ‘a reorganization.’ All firms marketing generic products responded prior to receiving the follow-up letter.

Although the precise amount of time it took for the original letters to be delivered is not known, most of the companies apparently attempted to respond promptly. Some of them received the request within 2 or 3 days of its mailing, but, with the intercession of a weekend, some may not have received it until 6 days after its mailing. Within 5 working days of the mailing, a majority of the companies had mailed their responses.

The responses of research-based and generic companies differ markedly with regard to response time, although not in the direction hypothesized. The mean number of days which the generic firms took to respond is 7.8; for name-brand producers, the figure is 14.6. And the difference is not attributable solely to the single late respondent alluded to above. Even with that case excluded, name-brand companies averaged more than 11 days to respond.

The median response time of the name-brand group was 13 days; only one of the generic companies took that long to respond. The modes are 18 and 10 days, respectively.

Because geographical location could have accounted for part of the observed difference in measured response time, as a result of variations in mail delivery speed, the locations of name-brand and generic companies were compared. There is a geo- graphical bias, but it apparently worked to the benefit of the former group. The Toronto metropolitan area, from which all of the requests were mailed, is home to 55% of the PMAC companies in our sample, and the rest are headquartered elsewhere in the Province of Ontario or contiguous Quebec. Only 27% of the generic firms are located in the Toronto region. Most, but not all of the rest are in Ontario and Quebec. Furthermore, the data reveal that proximity is posi- tively correlated with response time. Therefore, the relatively rapid responses of the generic companies occurred in spite of the fact that they probably received the requests for information later than the name-brand manufacturers did.

Table I. Index of usefulness of references tiled

Tvoe of reference

Number of references Textbooks and Professional Commercial and of each type literature reviews journal articles trade publications

0 0 0 0 I IO 5 3 2 II 8 3 3 I2 IO 3 4 I2 II 3

5 or more I? I2 3

Points are assigned separately for each category of reference, then cumulated. When more than one kind of reference is provided. the number of points in categories other than the primary one is halved (e.g. a response citing 5 journal articles. 2 texts, and 2 trade publications. would obtain the score; l2+(11+3)2=19).

156 MARTISTHOMAS and JOEL LEXCHIN

One way in which brand manufacturers were more responsive was a greater tendency to respond by telephone, possibly reflecting a greater desire to provide effective or rapid service. Of the 20 brand- name respondents, 5 phoned an answer, while only 1 generic company did so. Unfortunately, the data are not easily interpreted. Two of the 5 who phoned made their calls more than 2 weeks after the request was sent, and a third phoned, but never wrote. These actions may reflect a decision based as much on convenience as a desire to provide information promptly. Furthermore, at least part of the relation- ship between ‘response method’ and ‘type of manu- facturer’ appears to be accounted for by geographical location. Companies in the Toronto metropolitan area were much more likely to call than were those for whom it would have been a long distance call, and, as noted above, a relatively large proportion of name-brand manufacturers are located in Toronto.

Finally, the two sets of responses revealed a striking and unexpected difference in response time variability, with the generic companies showing much greater consistency. The standard deviations are 2.4 and 15.9 days for the two groups.

Response qualit?

Accuracy. The overall quality of the responses was reasonably good. The possible numeric values assigned to the companies’ responses were 0, 1, or 2, with 2 corresponding to correct, accurate responses. No company received a zero from all three referees. Only two name-brand companies (of 21) and one generic company (of 11) had an average score less than 1.0. At the other extreme, the reponses of four generic and ten name-brand companies were scored 2.0. The average scores for the name-brand firms was 1.65: for the generics, the mean was 1.55.

Treating the data as ordinal, and using the Mann-Whitney test, the probability that the inter- group differences occurred by chance is less than or equal to 0.44. (If the data are considered interval, P < 0.29.) The null hypothesis is not rejected.

It was expected that interaction 2, which had not been consistently described in the literature, would more frequently result in inaccurate advice being given. However, the type of interaction was not significantly related to response quality.

References. Research-based manufacturers were slightly more helpful than generic companies in citing relevant publications, with mean response scores of 10.7 and 10. I respectively. This difference is not statistically significant (P < 0.37), and the null hypothesis is not rejected. The medians are II and IO respectively.

Two firms, both research-based, were exceptionally helpful. They provided many citations, and were more likely to enclose photocopies or abstracts. One also reviewed the literature in its letter. Although the group of name-brand companies included two whose responses were outstanding, it also included the only respondent which provided no citations at all. Once again, these firms evidenced greater variability in the quality of their responses than the generic companies did.

There was no significant difference between the two groups’ reliance on trade publications. Surprisingly

few respondents cited them. with only 2Soi0 of re- search-based companies and 18% of generic firms doing so.

Similarly. there was no significant intergroup difference in citations of texts, compendia. or reviews of the literature. They were frequently cited. however, with 64% of generic. and 60% of PMAC affiliated companies alluding to them. When only one publi- cation was cited, it was most likely to be a text or review of the literature. Perhaps recognizing the redundancy. no respondent mentioned more than two publications of this type.

It is with regard to the proportion of companies citing research articles, and the number of such publications cited, that research-based manufacturers clearly outdo generic producers. Relevant research is cited by 70% of the former, but only 36% of the latter. The mean numbers of citations are 3.4 and I.0 respectively. although these figures are somewhat misleading. since the efforts of the two companies mentioned earlier badly skew the results. One of them provided 21 references; the other cited 14. No one else mentioned more than four journal articles: companies in both groups cited that many.

Of course. a simple count of the number of refer- ences provided is not a valid index of the usefulness of the response, nor even a valid measure of the companies’ efforts in preparing the response. And, in this case. the skewness of the data compel caution in its interpretation. Consider that the median number of citations in the responses of both groups was identical: only 2.

Because research-based manufacturers, who tended to be considerably slower in responding. also tended to be slightly more helpful in citing relevant publica- tions. we considered whether their tardiness might be attributable, in part. to greater diligence in seeking out, collecting, or photocopying material. The data do not support that view. The measures of response time and usefulness of publications cited are not correlated. This finding is similar to the one drawn by Tse and Mansour [3], who found that the time taken to respond to their request for information was not correlated with ‘completeness’ of responses.

CONCLCSlON

Our findings do not support the contentions of research-based pharmaceutical manufacturers that they are more thorough than generic companies in providing information to physicians, and do so more rapidly. Although the research of Generali and Hogan [7] is methodologically similar to the current project, we failed to find the significant differences between brand and generic companies which they did. Three plausible explanations for the disparity come to mind.

Prior to the United States’ Patent Term Restora- tion Act of 1984, generic drug companies faced severe restrictions on their ability to market products in that country. It is possible that the relatively limited resources of generic companies at the time Generali and Hogan conducted their survey affected those companies’ abilities to be useful sources of informa- tion. Yet the research of Knight and Baumgartner [6], which also preceded 1984 and focused on American

Responsiveness of brand and generic companies 157

companies, described no significant difference in responsiveness between the two types of firms.

A more plausible explanation concerns the type of drug interaction. The interactions which we asked about had been reported in the literature over a number of years. Generali and Hogan’s interaction had only recently been identified, and relevant infor- mation might have been difficult to come by. This suggests that research-based companies may be better able to provide information about very recent discoveries (or. at least that that was the case prior to 1984).

Most importantly, Generali and Hogan’s infer- ences were based on an extremely small sample. They contacted 45 companies, only I I of which were able to provide any information, and only five of which gave complete answers. These numbers are too small to be conclusive.

Future research might evaluate whether the com- plexity of information requested affects responsive- ness. Some research has suggested that response rapidity varies with type of information requested, but it did not indicate whether name-brand and generic manufacturers were similar in that regard [3].

One surprising finding is the dramatic difference in homogeneity between the two corporate groups. For every variable considered, the generic companies were fairly similar in their responses. Among name-brand companies, the variation was greater, occasionally markedly so. No theoretical explanation for these differences between types of firms is immediately apparent. and our data offer no explanation. The age of the companies may be relevant. Those formed recently, which are more likely to be generic, may tend to adhere to a similar standard, while the diversity of those which are older may result from temporal changes in perceptions of corporate responsibility, and related idiosyncratic differences in management style and values.

Notwithstanding the efforts of governments to establish registries of adverse drug interactions, physicians and other health care providers will continue to rely on drug manufacturers for infor- mation about their products. Although some of the companies we contacted responded both promptly and accurately, many did not, indicating that phar- maceutical manufacturers generally are not an en- tirely reliable source of information for health care providers.

More specifically, the great variability in the com- mitment or ability of research-based manufacturers

to provide information accurately and quickly belies some of the claims of the Pharmaceutical Manufac- turers Association of Canada. The current research indicates that there is no widely held standard of quality among the research-based companies. The dissimilarities among them are so great, that any generalized claims on their behalf are suspect. This is quite apart from the fact that their specific claims of superiority considered here are not supported by the empirical evidence.

Ackno,cled~~menrs-Obviously. it is with reluctance that one requests information on behalf of fictitious patients. The authors sincerely commend the medical directors and other staff of the pharmaceutical companies contacted for their efforts. We trust they will forgive the innocuous deceit, and consider the substantive findings informative.

We gratefully acknowledge the assistance of MS Suzanne Francis for her assistance related to the masking of com- pany names on correspondence.

We also appreciate the unselfish assistance of three physicians who evaluated the pharmaceutical companies’ responses. They prefer to remain anonymous.

I.

2.

3.

4.

5.

6.

7.

8.

9.

REFERENCES

Brief to the Commission of Inquiry into the Pharma- ceutical Industry. Pharmaceutical Manufacturers Asso- ciation of Canada. 1984. Russell W. L. and Dahlin P. A. Industry’s interest (letter). Drug Intell. Clin. Phurmuc. 11, 493494. 1977. Tse C. S. T. and Mansour J. M. Drug industry response time (letter). Drug In/eN. C/in. Pharmuc. 18, 325, 1984. Bainbridge C. V.. D’Ambrosio G. G.. Petrick R. J. and Weiss E. M. Pharmaceutical manufacturer’s responses to drug information inquiries. Am. J. Hosp. Pharmuc. 39, 1532-1534. 1982. Tse C. S. T. Disclosure of product information by manufacturers. .-lni. J. Hosp. Phurmuc. 39, 645446. 1982. Knight J. L. and Baumgartner R. P. J. Quality control information: its availability from pharma- ceutical manufacturers. Hosp. Formul. Mgmt. 9, 35-52. 1974. Generali J. A. and Hogan L. A comparison of pharma- ceutical manufacturers as a source of drug information to a telephone inquiry: generic vs. brand. Drug Inform. 17, 195-204, 1983. Rizack M. A. Tlie .Wedicol Letter Handbook of Adterse Drug Interactions. The Medical Letter. New Rochelle, N.Y.. 1989. Krogh C. M. E. (Ed.) Compendium of Phurmaceutica1.r and Specialries 1987Canadian Pharmaceutical Associa- tion, Ottawa. 1987.