Blackwell Science, LtdOxford, UKOBRobesity reviews1467-78812004 The International Association for the Study of Obesity. 693111Review ArticleAdverse events of herbal food supplements M. H. Pit- obesity reviews Adverse events of herbal food supplements for body weight reduction: systematic review* Complementary Medicine, Peninsula Medical School, Universities of Exeter and Plymouth, Herbal
Physician participation in clinical studies and subsequent prescribing of new drugsPhysician Participation in Clinical Studies
and Subsequent Prescribing of New Drugs
Mark H. Corrigan, MD, and Harold E. Glass, PhD
practice characteristics, and prescribing patterns of physi- Using a matched case–control design, we conducted an cians of any type who adopt drugs early on. The few studies analysis to determine whether U.S.-based physicians who had that have been published suggest that the following categories been investigators in a phase 3 clinical study were more likely of physicians tend to be early adopters of new drugs:4,6,18,19 to prescribe the study drug than were matched control (non-investigator) physicians over the 18 months following the product’s launch. We found that for all therapeutic indications, investigator-physicians made the study drug available to patients sooner after product launch and more frequently than did control physicians. Clinical investigators, as practicingphysicians, may have a major influence on how other doctors Other studies have examined the influences on decision- making within the general physician populations in prescrib-ing new drugs. Despite physicians’ perceptions, the research Key Words: prescribing, clinical trials, pharmaceuticals
highlighted the fact that peer-reviewed academic sources ofinformation about new drugs are of limited importance in Phase 3 clinical trials are critical to new drug development their prescribing behaviors.7–10 Physicians also appear to place because they help to secure regulatory approval for the new limited value on pharmaceutical marketing and Continuing drugs and add to the general medical knowledge base. The Medical Education (CME) as important sources of informa- published results from these clinical trials are important in the tion.11 More recent studies, however, have asserted that phar- adoption of new drugs; key results relating to a new drug’s maceutical information sources, such as sales forces and CME safety and efficacy appear in scholarly journals and are pre- programs, may have a greater influence on actual prescribing sented at professional meetings. Scant quantitative research behavior than many physicians wish to acknowledge.9,12 exists, however, on how physician participation in new drug For more than 50 years, the literature emphasized the func- clinical studies may influence their subsequent prescribing tion of medical peers in influencing how physicians prescribe behavior. This research examines whether physicians who drugs.13 The newer literature highlights the role that clinical take part in phase 3 clinical trials of a new drug are more trial investigators in particular have on a local level in terms likely to prescribe that drug after it comes to market.
of the patterns of practice of other physicians as follows:14–17 Few papers have measured how participation in clinical tri- als of any type may affect study drug prescribing. In one such . . . physicians involved in clinical trials tend to be local opinion case, a Dutch study of one hospital determined that the hos- leaders who can influence local practice styles because their pital’s participation in either a drug’s phase 2 or phase 3 clin- patterns of practice are adopted quickly by other physician- ical trial was associated with a 100% adoption rate of the stud- ied drugs by the hospital’s formulary.1 In another study, the useof lansoprazole (Prevacid®, TAP) was greater in a teaching These peer opinion leaders might not be internationally rec- hospital that had undertaken clinical trials of this agent.2 ognized experts but, rather, practicing physicians whose opin- More recently, Majumdar and colleagues attempted to ions are valued by other practicing physicians.
assess the subsequent prescribing impact of participation in Rogers’ seminal work on the “diffusion of innovations” clinical trials.3 They concluded that the effect, if it was present theory highlighted, among other factors, the role of peer opin- at all, was small. However, they identified the major limitations ion leaders in the acceptance of new ideas in a wide variety of to their study. The results applied to the hospital level, not to settings.18 Regardless of whatever information is broadcast, the individual physician level, and they were reported for all published, or otherwise disseminated among a large pop- prescriptions within a drug class because drug-specific pre- ulation, most people look to the ideas of peers whose opinions are widely respected by others in the community before they The literature is also limited in terms of the demographics, accept the new idea. The same dynamic may well be at workin the acceptance of new drugs, regardless of the individual Dr. Corrigan is Executive Vice President of Research and Develop- ment at Sepracor, Inc., in Marlborough, Massachusetts. Dr. Glass Using individual physician drug-specific prescribing data, is a Research Fellow at the Center for Evidence-Based Policy, Queen the study addresses, in a matched case–control design, Mary, University of London in the United Kingdom; a Professor with whether physicians’ involvement in phase 3 clinical studies the Graduate Program in Pharmaceutical Business at the Univer- affects their subsequent prescribing behavior of the study sity of the Sciences in Philadelphia, Pennsylvania; and President of drug, particularly if that study involves a first-in-class drug. We compared the prescribing behaviors of the two groups of 60 P&T® • January 2005 • Vol. 30 No. 1
Physician-Investigators and Prescribing of New Drugs
physicians—phase 3 clinical investigators and a matched These data came from the AMA’s Annual Survey of Practicing set of control (non-investigator) physicians—three months Physicians. These four variables were not used to match the before the start of their participation in a specific phase 3 case (investigator) and control groups.
clinical trial, and at three, six, and 18 months after the launch Investigator and control physicians differed statistically on three of these additional descriptive variables—namely, sex, METHODS
Table 1 Study Compounds Introduced into the
To answer the study question, we randomly selected 2,187 Market Between 1997 and 2000
physicians who were involved with at least one phase 3 clini-cal study as principal investigators at 2,897 clinical sites from First-in-Class Drugs
January 1, 1995, to December 31, 1996. (Some physicians con- ducted studies for more than one indication.) For physician Diclofenac/misoprostol (Arthrotec® 75, Pfizer) selection, we used a database that tracked data on clinical study investigator grant payments and protocol design infor- mation from more than 40 pharmaceutical companies that were conducting phase 3 clinical studies in the U.S.21,22 The2,187 physicians were randomly selected within each indica- Follow-on (“Me Too”) Drugs
tion group and represented about one third of the physicians Beclomethasone dipropionate (Qvar®, Ivax) in each of these groups. We could not use a complete census Beclomethasone dipropionate inhaler (Vanceril®, Key) of physicians in each indication group because of the expense Budesonide (Rhinocort® Aqua, AstraZeneca) involved in matching individual prescribing level data from Budesonide inhalation powder (Pulmicort® Turbuhaler, We matched several characteristics of each investigator- Budesonide inhalation suspension (Pulmicort® Respules, physician in the IMS Health database of practicing physicians with those of a physician who had not been involved with the Candesartan cilexetil (Atacand®, AstraZeneca) conduct of a clinical study. Although a larger control group (e.g., four times the number of cases) would have been an opti- mal design, we had to use a research design with a one-to-one matching ratio because of the financial constraints of obtain- ing these additional physician data points.
The matching variables for the two physician groups were (1) the physicians’ geographic location; (2) their specialty, as classified by the American Medical Association (AMA); and Irbesartan (Avapro®, Bristol-Myers Squibb) (3) their prescribing behavior three months before their par- ticipation in the clinical study, including: Levofloxacin (Levaquin®, Ortho-McNeil)Loratidine (Claritin®, Schering) • the total number of prescriptions written.
• the total number of sponsored company prescriptions as a percentage of all prescriptions written by the physician.
Mometasone (Nasonex® Nasal Spray, Schering) • the number of prescriptions written by the physician for • the number of prescriptions written by the physician in the Olanzapine (Zyprexa®, Lilly)Quinupristin/dalfopristin (Synercid®, Monarch) We matched geographic areas by using U.S. Post Office Zip Codes. There were no statistically significant differences between the two sets of physicians on the matched variables.
We confirmed the clinical investigators from information that Telmisartan (Micardis®, Boehringer Ingelheim) had been supplied to the Food and Drug Administration (FDA). Before an investigational new drug is allowed to be Valsartan hydrochlorothiazide (Diovan® HCT, Novartis) used with enrolled patients at a site, the FDA’s 1572 Form must be completed, identifying the principal investigators and the sub-investigators. All of the investigators’ names appeared inthe database. None of the control physicians’ names appeared * Four of the study drugs were later withdrawn from the market: in that FDA database from 1988 to 1997.
mibefradil (Posicor®), cerivastatin (Baycol®), rofecoxib (Vioxx®), Additional descriptive information was available from and troglitazone (Rezulin®). Eighteen-month prescribing data wereavailable for cerivastatin, rofecoxib, and troglitazone; six-month pre- another database about each investigator and control physi- scribing data were available for mibefradil.
cian, such as age, sex, board certification, and practice type.
Vol. 30 No. 1 • January 2005 • P&T® 61
Physician-Investigators and Prescribing of New Drugs
board certification, and practice type. Becausethe literature had pointed to a possible relation- Table 2 Comparison of Physician Characteristics in
ship between age, sex, board certification, and Prescribing New Drugs
new drug adoption, we included these three vari-ables, along with practice type, in subsequent Investigator-
(n = 2,108)
(n = 2,108)
The drugs, or compounds, in clinical trials are No. (%), Mean ± SD (Range),
classified by an indication based on the Inter- or Median
national Classification of Diseases, Ninth Revi- sion (ICD-9) coding system. All of the drugs • Total of sponsor company prescriptions that we examined in our study represented new molecular entities (NMEs), or substantial re- formulations of an existing drug. We required the re-formulations (e.g., extended-release) to be complete, original New Dr ug Applications (NDAs) that had been filed with the FDA. The compounds in the study came to market between 1997 and 2000. Several of the study compounds had never been submitted by the sponsoring com- pany for FDA approval or had not received FDA approval by 2000. These compounds could not be included in the analysis. A complete list of study compounds is presented in Table 1.
The clinical studies selected for use in the analy- sis included investigation of 38 compounds in phase 3 clinical trials that were investigated for the outpatient treatments of asthma and allergic rhini- tis, hypertension, osteoarthritis and rheumatoid arthritis, depression, pneumonia, hypercholes- terolemia, and diabetes. These indications were • They are outpatient indications, and the pre- • They continue to receive a substantial amount of funds for research and develop- Dependent Variables
Study drug prescribing was reported at three, six, and 18 months after product launch, as deter- mined by the date of the first filled prescription recorded by IMS Health, Inc. The IMS Health USC = Uniform System of Classification.
prescribing database tracks the retail fulfillment prescribing behavior of virtually every practicingAmerican physician, recording more than 70% ofall prescriptions filled in the U.S.24 The database does not manufacturers created the USC in 1975. The system uses five include prescriptions filled by hospital pharmacies or hospital digits to standardize and categorize all U.S. pharmaceuticals systems such as the Veterans Health Administration.
according to product type. USCs, which are used in the U.S.
The prescribing share percentage for an individual drug by and Canada, have four levels of hierarchy. USC2 is the broad- an individual physician is measured in a Uniform System of est category, and USC5 is the most detailed category, allow- Classification (USC) share. The USC share is the percentage ing for more specificity within a category. Our study used the of the physician’s prescriptions in the corresponding drug USC5 level of specificity. (For more information on USC lev- class that are represented by that physician’s actual number els, see page 702 of the November 2004 issue of P&T.) IMS of prescriptions for the study drug in question.
Health also indicates whether a new drug represents a new IMS Health, Inc., and virtually all of the pharmaceutical drug class, based on a new mode of action.
62 P&T® • January 2005 • Vol. 30 No. 1
Physician-Investigators and Prescribing of New Drugs
Missing data never exceeded 2% of any variable Statistical Methodology
Table 2 contains both means and shows stan- dard deviation (SD) and percentages. Figures 1 and 2 and Table 3 contain percentages, with levelsof statistical significance derived from an analysis- of-covariance model. An illustrative analysis-of- covariance model, addressing new drug prescrib-ing three months after product launch, is presented We used SPSS for Windows, Version 12 (SPSS Inc., Chicago, IL), to perform the analyses.
Table 2 compares the characteristics of investi- In the three months prior to the drug’s launch, both investigator and control physicians wrote a mean of 2.2% of all their prescriptions from thepharmaceutical company launching the new drug Figure 1 Uniform System of Classification share after product launch (SD ± 6.0 for the investigators, SD 4.3 for the con- for investigator and control physicians (P = .001).
trol physicians). During this time, investigatorswrote a mean total of 677 prescriptions (SD ± 758); the control Most of the professional activity for both investigators (77%) physicians wrote 653 prescriptions (SD ± 715). In addition, the and control physicians (82%) was office-based; however, more investigators wrote a mean of 24 prescriptions (SD ± 64) for the investigators (17%) were based in medical schools, conducted drug class of the new drug; control physicians wrote 22 pre- medical research outside of medical schools, or concentrated on teaching, compared with the control physicians (5%).
The average age of both types of physicians was 53 years.
For all indications in the database, investigators prescribed The SD was ± 8.5 for the investigators and 10.9 for the control the study drug more often than control physicians did. Three months after product launch of the new drug, the number of Because geography was a sample control variable, there was prescriptions for the clinical trial drug written by the investi- no statistically significant difference in the geographic distri- gators accounted for a statistically significant higher USC bution of the two sets of physicians.
share than did the number of prescriptions written by controls Roughly three-quarters of both groups of physicians were (26% vs. 16%; P ≤ .001), even with controlling for the additional specialists (76% of the investigators and 74% of the control variables used in the multiple analysis-of-covariance model reported in Table 4. There were no statistically significant sec- A higher percentage of investigators (92%) were men, com- ond-order interactions. This difference in prescribing behav- pared with the control group, in which 83% were men. Similarly, ior between the case and control physicians remained virtually more investigators were board-certified (91%) compared with unchanged over the next 15 months (see Figure 1).
Prescribing behaviors did vary, however, according to the Table 3 Uniform System of Classification (USC): Share by Practice Type After Product Launch*
USC Share (%) at Time After Product Launch
Vol. 30 No. 1 • January 2005 • P&T® 63
Physician-Investigators and Prescribing of New Drugs
Table 4 Analysis-of-Covariance Model, with Uniform System of Classification (USC) Share 3 as the
Type of Doctor (Trialist or Control) df = degrees of freedom; Rx = prescription; Sig. = significance (P value).
nature and type of the physicians’ practice. At three months after product launch, in every comparison except for research-based or medical teaching- based physicians, the investigators’ prescriptionsmade up a larger USC share than the controls’ prescriptions did. Overall, prescribing behaviors barely changed over the 18-month analysis period for physicians based in offices or hospitals (seeTable 3).
At three months, the number of research-based or medical teaching-based investigators’ pre- scriptions was similar to that of the controls. How-ever, the difference between the groups became apparent at six months after product launch and increased over the next year, with prescriptions of investigators accounting for 23% of the USC shareat 18 months and those of the controls decreasing Investigators were more likely than controls to prescribe a first-in-class new drug. The average Figure 2 Uniform System of Classification share at three, six, and 18 number of drugs within a new class was two dur- months after product launch for investigator-physicians and controls ing the 18 months of the study data. None of the for “me too” (follow-on) drugs (P = .01).
first-in-class drugs entered existing classes, and an 64 P&T® • January 2005 • Vol. 30 No. 1
Physician-Investigators and Prescribing of New Drugs
average of six other drugs were present in previously existing At all times during this study, and for nearly every compar- drug classes. The absolute USC share figures for “me too” (fol- ison except for research-based or medical teaching-based low-on) drugs were substantially lower than for first-in-class physicians at three months after product launch, investigators drugs simply because there were more drugs in the class.
were more likely than controls to prescribe the new drugs. For Investigators also tended to prescribe new “me too” drugs physicians in research or teaching institutions, their more (Figure 2). The percentage difference between the investiga- restrictive outpatient formularies may be slower in admitting tor and control physician USC share was greater for first-in- new drugs. Teaching hospitals often have a relatively higher class drugs than for “me too” drugs. The difference never ex- percentage of patients whose pharmaceutical and medical ceeded nine share percentage points for “me too” drugs and care is compensated by government sources. These investi- never dropped below 16 share percentage points for first-in- gators may have initially been inhibited in prescribing new products because of the more restrictive formularies in theirworkplaces.
LIMITATIONS OF THE STUDY
Eventually, however, the research-based and teaching-based The data were limited to U.S. results in a specific number of investigators prescribed new drugs in a manner similar to indications, but the results may differ for other indications or that of other investigators, perhaps because their institutions in other countries. As with any case–control study, there is the began adding the new drugs to the formulary. Three months potential for confounding variables not matched or included in after product launch, the investigators’ prescriptions for the study drug accounted for a statistically significant higher USCshare than did the controls’ prescriptions. This difference DISCUSSION
remained constant over the next 15 months.
Clinical investigators are integral to the drug-development Phase 3 investigators were more likely to prescribe new process. They often play an important additional role in the drugs for all indications studied, thereby making these drugs adoption of new drugs, particularly for first-in-class agents. In available to their patients sooner than control physicians did.
the U.S., most investigators are practicing physicians in office Clinical investigators often have the longest and most exten- or hospital settings, with only a portion of them based in med- sive exposure to new drugs, and their patients may benefit ical schools and teaching hospitals.25,26 These practicing in- from this experience during the clinical trial and after the vestigators, although not necessarily authors of major profes- product is introduced into the market.
sional papers on drug development, may be a useful source of Clinical studies are an essential step in the drug-develop- information for physicians who were not participants in clini- ment process. Investigators often play a central—and here- tofore underappreciated—role in the way that new drug adop- Investigator-physicians are the first to work with drugs and tion is actually dif fused into the general population of thus have the greatest experience with them. Their behavior physicians. This might be especially true for first-in-class after the launch of drugs may well influence other physicians.
Participation in phase 3 studies provides a mechanism bywhich investigators can gain access to new drugs under de- REFERENCES
1. Denig P, Haaijer-Ruskamp FM, Wesseling H, Versluis A. Impact velopment nearly three years ahead of the general population of clinical trials on the adoption of new drugs within a university hospital. Eur J Clin Pharmacol 1991;41:325–328.
By and large, clinical investigators are likely to be early 2. Jones MI, Greenfield SM, Jowett S, et al. Proton pump inhibitors: adopters of new medications and hence may be more willing A study of GPs’ prescribing. Fam Pract 2001;18:333–338.
to participate in clinical trials. We have no way of measuring 3. Majumdar SR, Chang WC, Armstrong PW. Do the investigative sites that take part in a positive clinical trial translate that evidence this general mindset regarding early drug adoption in this into practice? Am J Med 2002;113:140–145.
study. We do not have psychographic data on the investigators, 4. Steffensen FH, Sorensen HT, Olesen F. Diffusion of new drugs and we have no information about how they prescribe other in Danish general practice. Fam Pract 1999;16:407–413. new drugs when they have not participated in these other 5. Anderson GF, Hurst J. Hussey PS, Jee-Hughes M. Health spend- ing and outcomes: Trends in OECD [Organisation for Economic clinical trials of new drugs. We do know from the data in this Co-operation and Development] Countries, 1960–1998. Health study, however, that investigators, more than other physicians in the study, do tend to adopt first-in-class drugs.
6. Tamblyn R, McLeod P, Hanley JA, et al. Physician and practice The investigators had earlier access to the study drugs, and characteristics associated with the early utilization of new pre- they have seen new drugs being carefully tested in their own scription drugs. Med Care 2003;41:895–908.
7. McKinney WP, Schiedermayer DL, Lurie N, et al. Attitudes of practices. Investigator-physicians who are new drug pre- internal medicine faculty and residents towards professional scribers may be more numerous, simply because they are interactions with pharmaceutical sales representatives. JAMA often comfortable prescribing drugs with which they have become familiar. Their involvement with clinical trials affords 8. Lichstein PR, Turner RC, O’Brien K. Impact of pharmaceutical company representatives on internal medicine residency pro- them this added experience with the drugs.
grams: A survey of residency program directors. Arch Intern Whether or not investigators are more innovative in their general approach to medicine, or whether they simply tend to 9. Wazana A. Physicians and the pharmaceutical industry: Is a gift prescribe what they know, they are more likely to prescribe ever just a gift? JAMA 2000;283:373–380.
clinical study drugs when they have participated in the clini- 10. Omoigui NA, Silver MJ, Rybicki LA, et al. Influence of a ran- domized clinical trial on practice by participating investigators: Lessons from the coronary angioplasty versus excisional atherec- Vol. 30 No. 1 • January 2005 • P&T® 65
Physician-Investigators and Prescribing of New Drugs
tomy trial (CAVEAT). J Am Coll Cardiol 1998;31:265–275.
18. Rogers EM. Diffusion of Innovations, 4th ed. New York: The Free 11. Lexchin J. Interactions between physicians and the pharma- ceutical industry: What does the literature say? Can Med Assoc J 19. Peay MY, Peay ER. The role of commercial sources in the adop- tion of a new drug. Soc Sci Med 1988;26:1183–1189.
12. Avorn J, Chen M, Hartley R. Scientific versus commercial sources 20. Miller RR. Prescribing habits of physicians: A review of studies of influence on the prescribing behavior of physicians. Am J Med on prescribing of drugs (Parts I–III). Drug Intelligence Clin Phar- 13. Guldal D, Semin S. The influences of drug companies’ advertis- 21. PICAS® [database]. Fort Washington, PA: Data Edge, LLC; 1992.
ing programs on physicians. Int J Health Serv 2000;3:585–595.
(No longer commercially available; rights sold to IMS Health in 14. Soumerai SB, McLaughlin TJ, Gurwitz JH, et al. Effect of local medical opinion leaders on quality of care for acute myocardial 22. Glass HE, Sullivan JM. Managing clinical grant costs through the infarction: A randomized controlled trial. JAMA 1998;279: use of unaffiliated investigators. Drug Information J 1995;29.
23. R&D Insight [database online]. Yardley, PA: Wolters Kluwer 15. Soumerai SB, Majumdar SR, Lipton HL. Evaluating and improv- Health, Pharma Solutions Division; 1995. Updated daily.
ing physician prescribing. In: Strom B, ed. Pharmacoepidemiol- 24. Xponent® [database]. Plymouth Meeting, PA: IMS Health, Inc.; ogy, 3rd ed. Toronto, Canada: John Wiley & Sons; 2000:483–503.
16. Greer AL. The state of art versus the state of science: The dif- 25. Hovde M, Seskin R. Selecting U.S. clinical investigators. Appl Clin fusion of new medical technologies into practice. Int J Technol Assess Health Care 1988;4:5–26.
26. Glass HE, Kane RA. Higher payments to investigators don’t speed 17. Naylor CD. Better care and better outcomes: The continuing study completion. Appl Clin Trials 1995;9:46–54.
challenge. JAMA 1998;279:1392–1394.
66 P&T® • January 2005 • Vol. 30 No. 1
Africa in Motion Special Focus – Voices of the Bushmen Al members of the 'Voices of the Bushman' panel discussion are available for interview or to provide feature articles. For more information, please contact Kirsty Dickson, Press Officer, on 07826 850763 or ---AiM is pleased to present a spotlight on the Bushmen of Southern Africa. As part of this focus, there wil be an exhibition of Bu