Pt_0411_rosenthal.3_pd

Demographics, Practices, and Prescribing
Characteristics of Physicians Who Are
Early Adopters of New Drugs
Harold E. Glass, PhD, and Bruce Rosenthal, MBA
ABSTRACT
portant factors in influencing doctors’ adoption of a new drug, We conducted an analysis to determine the factors that even if physicians sometimes minimize the importance they influence physicians’ decisions to adopt a new drug by exam- place on pharmaceutical salespeople.3,8,9 ining their demographics, practices, and prescribing charac- Many research studies highlight the role that medical bul- teristics. In addition to the level of marketing support letins and journals play as sources of information on new expended by a company for its newly launched drugs, several drugs. Although some researchers have debated the value of physician-specific variables (e.g., age, specialization, and pre- peer-reviewed journals as a source of information on new scribing practice) are critical in quantifying the likelihood of drugs,10 other data indicate that medical bulletins and journal articles do represent an important channel of informationabout both old and new drugs.1–3 INTRODUCTION
Specialist meetings, presentations, conferences, and sym- As total expenditures related to health care grow each year, posia form a communication channel that appears to be par- pharmaceuticals play an ever-increasing role in medical care.
ticularly important for disseminating information about new Prescription drugs now account for a larger percentage of drugs. The literature indicates that these forums provide a health care costs than ever before.1,2 Physicians remain the highly valued source of information and facilitate interaction new drug.3 Although a fairly large amount of research has focused on factors that in- fluence a physician’s decision to adopt a although it is allowed only in the U.S. and scribing characteristics of physicians who are the first to adopt new drugs. This ar- Harold E. Glass,
Bruce Rosenthal,
cerning new drugs and the factors that in- vertising is actually effective in getting fluence an individual physician’s decision to prescribe.
physicians to write prescriptions, however, is still being de- Sometimes the research refers to a new drug on the market bated. Even though many pharmaceutical companies have or to an existing drug first prescribed by the physician. A wide dramatically increased their spending on DTC advertising, variety of research has shown that interpersonal communica- visits by sales representatives who bring free samples still tion between opinion leader–physicians and their peers can be constitute a much larger percentage of U.S. pharmaceutical (and many times has been) the critical factor in the rapid, company promotional spending and has proved to be more ef- wide-scale acceptance of innovative drugs.4–6 A report in the Journal of the American Medical Association The literature is scant in covering the practices and pat- from 2000 estimated that of the $11 billion spent each year by terns of physicians who adopt new drugs early on. The limited pharmaceutical companies for promotion and marketing, research that has been done suggests that the following cate- almost 50% of that went to sales representatives.7 The same gories of physicians tend to be early adopters of new study found that interactions with drug salespeople had a very strong impact on preference and rapid prescribing of newdrugs. Many other studies have echoed these findings, listing • young physicians—or at least those who have been prac- the pharmaceutical sales representative as one of the most im- • male physicians, when compared with female physicians13 Dr. Glass is a Research Fellow at the Center for Evidence-Based Policy, Queen Mary, University of London in the United Kingdom, • physician graduates from the most recently established and President of TTC, LLC, in Philadelphia, Pennsylvania. Mr. Rosenthal is Assistant Director of the Graduate Program in Phar-maceutical Business at the University of the Sciences in Philadel- Drawing upon multivariate models, this paper describes the characteristics of U.S. physicians who are the first to adopt new 2 P&T® • November 2004 • Vol. 29 No. 11
Physicians and Early Adoption of New Drugs
drugs in their prescribing practices. The research focuses on distributed among all of the 32 drugs. The drugs were indicated those characteristics as they relate to the adoption of both ther- for the outpatient treatments of asthma and allergic rhinitis, apeutically novel drugs (i.e., first-in-class drugs) and later fol- hypertension, osteoarthritis and rheumatoid arthritis, low-on drugs in drug classes already established.
depression, pneumonia, hypercholesterolemia, and diabetes.
The list of drugs appears in Table 1.
Although hospital pharmacies are a major source of pre- We examined the prescribing behavior of 3,646 physicians scription fulfillment, it is often difficult to link an individual pre- in relation to the introduction of 32 new drugs on the market scription to a particular physician. Therefore, by concentrat- from 1997 through 2000. The data reflected writing prescrip- ing on outpatient indications, we were able to better ensure the tions for one drug per physician, with the physicians randomly comprehensiveness of the prescribing data for each physician Table 1 Study Compounds Introduced to the Market
We originally designed the sample to test the from 1997 Through 2000
relationship between participation in a clinical trialand subsequent prescribing of the study drug. Theoriginal analysis compared physicians who had par- Beclomethasone dipropionate (Qvar®, Ivax) ticipated in a clinical trial with a matched set of (con- Beclomethasone dipropionate inhaler (Vanceril®, Key) trol) physicians who had not participated in clinical Budesonide (Rhinocort® Aqua, AstraZeneca) trials of any sort in the previous five years. An analy- Budesonide inhalation powder (Pulmicort® Turbuhaler, AstraZeneca) sis of both the clinical and the control physicians Budesonide inhalation suspension (Pulmicort® Respules, AstraZeneca) demonstrated that clinical trial physicians were more Candesartan cilexetil (Atacand®, AstraZeneca) likely to prescribe a study drug after it had been on the market for a period of at least 18 months.15 Approximately 50% of the study’s physicians had served as clinical investigators, and 50% constituted Diclofenac/misoprostol (Arthrotec® 75, Pfizer) the matched control set. As with the general phy- sician population in the U.S., most clinical trial sites are office-based, not hospital-based. One particular type of hospital—the major academic medical cen- ter—sometimes receives extensive press coverage for its work in clinical research. However, these cen- Irbesartan (Avapro®, Bristol-Myers Squibb) ters constitute a decreasing proportion of all phase 3 clinical trial sites, and they perform a minority of all phase 3 studies. Most clinical investigators see patients in their office-based practice and enroll patients for clinical studies from these practices.16 We conducted a comparative analysis (tests of Mometasone (Nasonex® Nasal Spray, Schering) statistical independence) between the two investi- Mometasone furoate inhalation powder (Asmanex®, Schering-Plough) gator and control groups of physicians and analyzed other variables that might have affected new drug- prescribing behavior. We found that although physi- cians who had worked as clinical investigators were more likely to prescribe the study drug when it Quinupristin/dalfopristin (Synercid®, Monarch) arrived on the market, the relationship of the other demographic, practice, and prescribing variables in explaining new drug prescribing did not differ in any meaningful theoretical or statistical way. Hence, we decided to combine both sets of physicians into one Telmisartan (Micardis®, Boehringer Ingelheim) data set, and, during subsequent analyses, to statis- tically control for the impact on new drug prescrib- ing levels of a physician’s participation in at least one phase 3 clinical trial for the new drug. The analysis always tested for the appearance of statistical inter- action between the two sets of physicians, the in- dependent variables, and the likelihood of a physi- * Four of study drugs were later withdrawn from the market: mibefradil cian’s being an early new drug adopter.
(Posicor®), cerivastatin (Baycol®), troglitazone (Rezulin®), and rofecoxib(Vioxx®). Eighteen-month prescribing data were available for cerivastatin and Study Population
troglitazone; six-month prescribing data were available for mibefradil.
We obtained the names of physicians from the IMS Health, Inc., database of active prescribing U.S.
Vol. 29 No. 11 • November 2004 • P&T® 3
Physicians and Early Adoption of New Drugs
egory, and USC5 is the most detailed category, allowing for Table 2 Study Population and Physician
more specificity within a category. The study used the USC5 Demographics in the U.S.
New drug prescribing data were available for all the physicians Independent Variables
The independent variables can be divided into several cate- • physician demographics• the physician’s practice and prescribing behavior physicians (Table 2).17 Compared with known parameters of the U.S. prescribing physicians, the study population was slighter older and more likely to be based in an office ratherthan in a hospital.16 The data on the physician’s sex and age; whether the physi- Physicians were enrolled from all 50 states, corresponding cian is based in a hospital or in an office; the physician’s spe- with the distribution of active physicians within the U.S. A cialty; and the physician’s board certification were derived Spearman Rank Order correlation of .932 between the num- from the AMA’s annual survey of practicing physicians.
ber of physicians in the study and the number of physicians The IMS Health database provides the information for the active within each state (as indicated in the American Medical prescribing variables used as independent variables (Table 3): Association’s annual survey of practicing physicians main-tained by IMS Health, Inc.) indicated a strong rank order • Total Pre-product Launch Prescribing Volume • Total Pre-product Launch Drug Class Prescribing• Pre-product Launch Company Prescribing Loyalty Dependent Variable
The dependent variable is dichotomous. Physicians were Prescribing data were available only at the Total Volume characterized as either new drug adopters or not new drug level for these variables, not for the individual drugs within adopters. The new drug adopters prescribed the new drug at each of the prescribing categories. A drug was classified as some time during the first six months after its launch and con- either “first in class” or as “follow-on,” according to its tinued to prescribe it for the next 12 months. We chose an ini- respective order of appearance on the market in accordance tial timeline of six months (a commonly used industry con- with the IMS Health, Inc., USC coding scheme.
vention). The date of the product launch was calculated as the IMS Health, Inc., also provided the rank order of spending date of the drug’s first prescription, as recorded by IMS Health, data used for the variable Pharmaceutical Marketing Support.
Although this variable was not a physician demographic or a All other physicians were characterized as not being new practice characteristic, we included it in a control function. The drug adopters. Physicians who prescribed the drug during the various drugs in our analysis came from companies in the first six months—but who did not continue to prescribe the largest revenue category to those with sales under $1 billion.
drug over the next 12 months—were not considered to be The use of this variable helped to control for the role of dif- ferential marketing expenditures in understanding new drug The term “early adopters” is derived from Rogers,18 who adoption and thus helped to isolate the explanatory impor- stated that “innovators” and “early adopters” made up 16% of tance of physician demographic and practice characteristics.
individuals overall. In our study, the new adopters constituted We obtained information about a physician’s participation in 22% of the total number of physicians in the study.
a drug’s phase 3 clinical trials from a pharmaceutical industry Drugs are designated by their Uniform System of Classifi- database of clinical trials and the U.S. Food and Drug Admin- cation (USC) code. IMS Health, Inc., and a majority of phar- istration’s (FDA’s) database of 1572 Forms, filed as part of new maceutical manufacturers created the USC in 1975. The sys- drug clinical trial activity.19 Missing data never exceeded 2% tem uses five digits to standardize and categorize all pharmaceuticals in the U.S. on the basis of product type.
USCs are used in the U.S. and Canada. In Europe, the equiv- Logistic Regression
alent classification is called an Anatomical Therapeutic Chem- The analysis employed binomial (binary) logistic regression, which is used when the dependent variable is categorical and USCs have four levels of hierarchy. USC2 is the broadest cat- the independent variables are either categorical or interval. Lo- 4 P&T® • November 2004 • Vol. 29 No. 11
Physicians and Early Adoption of New Drugs
gistic regression is considered to have overcome many of the used significance test for a logistic model. A well-fitting model restrictive assumptions of ordinary least squares (OLS) has a P value (Sig.) of .05 or lower; that is, we want model chi- regression. It does not require the assumption of a linear rela- square to be significant at the .05 level or better.
tionship between the independent and dependent variables, The significance of individual parameters can also be esti- and it is not necessary for the dependent variable to be nor- mated. Logit coefficients, also called unstandardized logistic mally distributed. There is no assumption of homogeneity of regression coefficients, are similar in interpretation to the b variance, one does not need to assume normally distributed (unstandardized regression) coefficients in OLS regression.
error terms, and the independent variables do not need to be Logits are simply the natural log of the odds; they are used in a logistic regression equation to estimate the log odds that the Logistic regression applies a maximum likelihood estimation dependent equals 1 in a binomial logistic regression.
after transforming the dependent into a logit variable, or the nat- Partial R is a method of assessing the relative importance ural log of the odds of the dependent occurring or not. Logis- of the independent variables, similar to beta weights or stan- tic regression estimates the probability of a certain event occur- dardized partial regression coefficients in OLS regression.
ring in the dependent variable; it calculates changes in the log The odds ratio is another method of determining the relative odds of the dependent variable, not the changes in the importance of the independent variables. It avoids some of the dependent variable itself, as OLS regression does. It is the like- interpretative difficulties involved with the Wald statistic, par- lihood, or probability, that the observed values of the depen- ticularly the greater possibility of type II errors. In this study, dent variable may be predicted from the observed values of the the odds ratio is used to help readers understand the relative independent variables. The likelihood probability, like any importance of each independent variable.
probability, varies from zero (0) to one (1).
Hosmer and Lemeshow20 and Menard21 present a more The log likelihood (LL) is its log, and it varies from 0 to extensive discussion of logistic regression.
minus infinity; it is negative because the log of any number lessthan 1 is negative. LL is calculated through iterations, making use of the maximum likelihood estimation (MLE).
We divided new drug introductions into two steps: (1) first- The log-likelihood test of a model can be used to estimate in-class drugs to reach the market and (2) later follow-on drugs the statistical significance of the entire model. Frequently in established therapeutic categories (i.e., with previously called the “model chi-square test” or the “likelihood ratio test,” existing USC codes). In this research, our analysis used two it is based on –2LL (deviance). It is an alternative to the Wald logistic regression models: one for first-in-class drugs and a statistic. The model chi-square provides the most frequently second for follow-on drugs. First-in-class drugs represent a Table 3 Independent Variables in the Early Drug Adoption Logistic and Ordinary Least Squares Models
Name of Variable
Description of Variable
• Physician office-based or hospital-based • Total Pre-product Launch Prescribing Volume • Physician’s total pre-product launch (3 months) prescribing volume • Total Pre-product Launch Drug Class Prescribing • Physician’s total pre-product launch (3 months) prescribing USC • Pre-product Launch Company-Prescribing Loyalty • Physician’s total pre-product launch (three-month) prescribing USC share of all the respective drug’s company products as a percentageof all the prescriptions written by the physician • The rank order of the pharmaceutical company’s total spending on the study drug during the pre-launch and first six months after launchof the study drug • Physician’s classification as a specialist in the respective specialty of the study drug or as a generalist (e.g., internist, family practitioner) • Physician’s participation as a clinical trial investigator for the study USC = Uniform System of Classification.
Vol. 29 No. 11 • November 2004 • P&T® 5
Physicians and Early Adoption of New Drugs
Table 4 “Model 1”: Early Adoption of First-in-Class Drugs
three months before a productlaunch (i.e., Total Pre-product Variable
Launch Prescribing Volume). Thegreater the number of total pre- B = unstandardized regression coefficient; df = degrees of freedom; SE = standard error; Sig = sig- adopter, although the relationshipwas not completely linear. Theyoungest doctors, those under 36 Table 5 “Model 2”: Early Adoption of Follow-on Drugs
years of age, and the oldest doc-tors, those over 65, were the least Variable
likely to be new adopters. How-ever, these age groups repre- ceutical Marketing Support put be- Clinical Investigator Experience –1.0418 B = unstandardized regression coefficient; df = degrees of freedom; SE = standard error; Sig = sig- nificance (P value); Wald = Wald statistic.
ties as increased product detailingand free samples provided by com- potentially greater innovation than follow-on drugs, which pany sales representatives, larger DTC advertising, and more enter the market at a later date in an existing drug class.
advertising in professional publications. The relative role ofeach could not be assigned in this study.
First-in-Class Drugs
The nature of the practice (Practice Type) was critical as well.
The first-in-class logistic model has a significant model chi Office-based physicians were more likely to adopt a first-in- square of .0000 and correctly predicts a robust 80% of the cases class drug than were hospital-based doctors, who might work (Table 4). In the final logistic regression model, seven variables with more restrictive formularies. For example, hospital-based were statistically significant. In order of relative importance, as physicians may see a higher percentage of patients who are tak- measured by the log odds, these variables were: (1) Pre-prod- ing a product on some type of public formulary. Even if these uct Launch Company Prescribing Loyalty, (2) Total Pre-prod- physicians wished to prescribe the new drug, they might be uct Launch Prescribing Volume, (3) Age, (4) Pharmaceutical unable to do so early in the new drug–adoption process.
Marketing Support, (5) Practice Type, (6) Clinical Investiga- Physicians who participated in at least one of the phase 3 clin- ical trials of the new drug had a greater chance of being early Early adopters of first-in-class drugs tended to write a greater adopters (Clinical Investigator Experience). They were famil- percentage of their prescriptions, three months before prod- iar with the drug for some time, and they had the most exten- uct launch, for drugs from the pharmaceutical company mar- sive experience using the drug in clinical settings.
keting the new drug than were physicians who were not early Consistent with findings in other literature, specialists in the adopters. Pre-product Launch Company Prescribing Loyalty was drug’s therapeutic area tended to be early adopters more often the most significant variable in the model explaining adoption than generalists or other specialists (Specialty in Tables 3, 4, Second in importance was the absolute number of pre- Neither the physician’s Board Certification nor Sex appeared 6 P&T® • November 2004 • Vol. 29 No. 11
Physicians and Early Adoption of New Drugs
in the model. Board-certified physicians in this study popula- sent in the follow-on drug model, yet it was the most important tion were actually more likely to adopt a first-in-class drug; how- variable in the first-in-class model. In this case, the higher the ever, the variable was no longer significant as part of a multi- percentage of drugs represented by the launch company in a physician’s total prescribing, the more likely that physician wasto be an early adopter of a drug from that company. It may well Follow-on Drugs
be that the variable reflected increased detailing by a phar- The follow-on drug logistic model also had a highly sig- maceutical company to that physician. However, the variable nificant model chi square of .0000 and correctly predicted a might also indicate a degree of confidence and trust in that strong 75% of the cases (see Table 5). Some variables in the sec- company, or in that company’s sales representatives, by the ond model were similar to those in the first-in-class model, and physician prescribing a therapeutically novel new drug from several had distinctly different explanatory roles.
The most important variable in the second model, Total Because a novel drug represents a new class of drug, com- Pre-product Launch Drug Class Prescribing, was able to be pany trust may be a factor in a physician’s decision to be an present only in a model that examined drugs for which an early adopter of a novel drug from that company. If increased established therapeutic category existed at the time of the detailing were the only factor at work in explaining the impor- new drug launch. The number of prescriptions written in the tance of this variable, we would expect to see the variable in new drug’s therapeutic class was a key explanatory variable for both models; however, it was present only in the first-in-class understanding the adoption of another drug in that therapeutic class. The more prescriptions written in a drug class by a The explanatory importance of trust was further supported physician, the greater the likelihood that the physician would by an examination of the subset of physicians who tried the adopt a new drug in that therapeutic class.
first-in-class drug within the first six months of a product’s Physicians who did not write any prescriptions at all in the launch but who later stopped prescribing that drug. These drug class were unlikely to prescribe a new drug in that class.
physicians were statistically similar in virtually every respect New adopters in this drug class may be prescribing these to the early adopters (physicians who continued to prescribe new, but non-novel, drugs for patients who have not responded the new drug after its adoption within the first six months).
well to older drugs already on the market in that drug class.
This similarity included demographic variables such as age and Non-prescribers in that drug class may not have patients who sex as well as the practice and prescribing variables.
are appropriate candidates for drugs within that class, or they The one important exception was Pre-product Launch Com- might simply not be convinced of that drug class’ medical pany Prescribing Loyalty. Early adopters demonstrated a value and thus might not be disposed to prescribing any new statistically significantly higher percentage (.0001) of their drugs from the pharmaceutical company marketing the new Total Pre-product Launch Prescribing Volume constituted drug than did physicians who first prescribed the drug but who the second most important variable in predicting whether a eventually stopped prescribing it. The willingness of some doctor would be an early adopter of a new drug, as was the case physicians to be early adopters of a first-in-class drug was with first-in-class drugs. The more prescriptions a physician highly related to their total level of prescribing drugs from the has written, the greater the chance that he or she will become company bringing the new novel drug to market, and it prob- a new drug adopter for follow-on drugs as well.
ably reflected a degree of trust by the physicians in that com- The relative amount of money spent in support of the new pany and their representatives to provide a safe and efficacious drug by the pharmaceutical company, the Pharmaceutical Marketing Support, was the third most significant explanatory After reviewing the results by indication as part of our over- all analysis, we found no significant, systematic differences in Board Certification was a statistically significant variable in the overall pattern of results between the chronic and acute this model, although it was not in the first model.
Office-based physicians, specialists, and those who partici- pated in a drug’s phase 3 clinical trials tended to be early LIMITATIONS
adopters of follow-on drugs in this study (Practice Type, Spe- The study design had several limitations: cialty, and Clinical Investigator Experience in Table 3).
In stark contrast to the first model, Pre-product Launch Com- 1. The data were restricted to drugs for selected outpatient pany Prescribing Loyalty was not a significant variable. Age was indications. The dynamics might differ for in-patient or not an important explanatory variable in the follow-drug model.
The Sex of a physician was not a significant factor in either 2. Although the study population covered a broad range of model. Female physicians were less likely to be new drug physicians, it was not a statistically projectable sample to adopters, but the difference was not statistically significant.
the entire U.S. physician population.
Several variables were common to the adoption models for 3. The study covered U.S. data only. The U.S. is the only both types of drugs, including (1) Total Pre-product Launch Pre- major pharmaceutical market that currently allows data on scribing Volume, (2) Clinical Investigator Experience, and (3) individual physician prescribing patterns to be tracked Specialty and Office-Based Practices. The most striking differ- and sold without explicit physician approval. Most coun- ence, though, between the two models was the role played by tries prohibit the selling of these data under any circum- Pre-product Launch Company Prescribing Loyalty. It was not pre- stances, and a few countries allow the data to be sold with Vol. 29 No. 11 • November 2004 • P&T® 7
Physicians Who Are Early Adopters of New Drugs
the express agreement of the physician. It is probably physicians were also more likely to adopt a new follow-on impossible at present to replicate this study outside the U.S. because of the absence of widespread individualphysician-prescribing data.
REFERENCES
1. National Institute for Health Care Management. Prescription DISCUSSION
drug expenditures in 2000: The upward trend continues. Availableat: www.nihcm.org/spending2000.pdf. Accessed January 14, 2004.
Prescription drugs represent a growing cost in the provision 2. Anderson GF, Hurst J, Hussey PS, Jee-Hughes M. Health spend- of health care. New drugs are now used in treatments for ing and outcomes: Trends in OECD [Organisation for Economic which older drugs were not developed. In addition, existing Co-operation and Development] Countries, 1960–1998. Health drug treatments are being replaced by newer, sometimes more 3. Glass HE. Direct-to-consumer advertising. PMT 2002;13:36–39.
4. Coleman JS, Katz E, Menzel H. Medical Innovation: A Diffusion Much has been written about the influences on physicians Study. New York, NY: Bobbs-Merrill; 1966.
to prescribe newer drugs. These pressures originate from 5. Weiss R, Charney E, Baumgardner RA, et al. Changing patient pharmaceutical companies, through their efforts to market management: What influences the practicing pediatrician? Pedi- and detail newer drugs; the medical literature; professional 6. McGettigan P, Golden J, Fryer J, et al. Prescribers prefer people: meetings; patients themselves; and the increasing use of DTC The sources of information used by doctors for prescribing sug- advertising for some medical indications.
gest that the medium is more important than the message. Br J In contrast, the literature on the characteristics of new drug Clin Pharmacol 2001;51:184–189. adopters is sparse. In most of these studies, interviews and 7. Wazana A. Physicians and the pharmaceutical industry: Is a gift ever just a gift? JAMA 2000;283:373–380.
reported behavioral data have been the sources of the infor- 8. Prosser H, Almond S, Walley T. Influences on GPs’ decision to pre- mation. The few studies that have used prescription data have scribe new drugs: The importance of who says what. Fam Pract not been able to link individual physicians with their specific 9. Peay MY, Peay ER. The role of commercial sources in the adop- Similarly, little use has been made of multivariate models to tion of a new drug. Soc Sci Med 1988;26:1183–1189.
10. Groves KEM, Flanagan PS, MacKinnon NJ. Why physicians start examine new drug adoption. Our study examined individual or stop prescribing a drug: Literature review and formulary impli- physicians and their actual prescribing behavior from this cations. Formulary 2002;37:186–194.
type of multivariate view. It also examined new drug adoption 11. Jones MI, Greenfield SM, Bradley CP. Prescribing new drugs: by new drug novelty (first-in-class) and follow-on drugs (those Qualitative study of influences on consultants and general prac-titioners. BMJ 2001;323:378–381.
12. Miller RR. Prescribing habits of physicians: A review of studies The level of marketing support expended by a company for on prescribing of drugs (Parts I–III). Drug Intelligence Clin Phar- its newly launched drug is an important explanatory variable in understanding new drug adoption; in our analysis, it was 13. Steffensen FH, Sorensen HT, Olesen F. Diffusion of new drugs in Danish general practice. Fam Pract 1999;16:407–413.
14. Tamblyn R, McLeod P, Hanley JA, et al. Physician and practice characteristics associated with the early utilization of new pre- CONCLUSION
scription drugs. Med Care 2003;41:895–908.
Several physician-specific variables are critical in quantify- 15. Glass HE, Corrigan MH. Physician participation in clinical stud- ing the likelihood of new drug adoption. A physician’s Total ies and subsequent prescribing of the new drug. Submitted forpublication.
Prescribing Volume is essential in understanding the likeli- 16. Hovde M, Seskin R. Selecting U.S. clinical investigators. Appl Clin hood that the physician will adopt a new drug, whether or not that drug is first in its therapeutic class. Physicians who par- 17. Xponent® [database]. Plymouth Meeting, PA: IMS Health, Inc.; ticipated in a drug’s phase 3 clinical trials or who were spe- cialists in the drug’s therapeutic area were more likely to adopt 18. Rogers E. Diffusion of Innovation, 4th ed. New York: The Free a new drug, whether or not it was therapeutically novel.
19. Glass HE. Higher payments to investigators don’t speed study Younger physicians were more likely to adopt a novel new completion. Appl Clin Trials 1995;4:40–44.
drug, as were physicians who already prescribed a higher per- 20. Hosmer DW, Lemeshow SL. Applied Logistic Regression, 2nd ed.
centage of all their drugs from the company launching the 21. Menard S. Applied Logistic Regression Analysis. Series: Quantita- novel new drug. Total Prescribing Volume within the drug tive Applications in the Social Sciences, No. 106. Thousand Oaks, class of a follow-on drug was also a significant predictor of new follow-on drug adoption. Office-based and board-certified 8 P&T® • November 2004 • Vol. 29 No. 11

Source: http://routinerecovery.com/Portals/0/images/PDFLibrary/Demographic,%20Practice%20and%20Prescribing%20Characteristics%20of%20Physicians%20Who%20Are%20Early%20Adopters%20of%20New%20Drugs_Glass_Rosenthal.pdf