2. Doctors who maintained a variety of contacts with a large number of colleagues, the socially integrated doctors, typically introduced the new drug into their practices months before their relatively isolated colleagues. The degree of a doctor's integration was measured by the number of his colleagues who named him, in response to certain interview questions, as an advisor, frequent discussion partner, or frequently visited friend.

3. Among the integrated doctors, the use of the new drug spread at an accelerating rate, indicating an interpersonal process of diffusion, while among the isolated doctors use of the new drug spread at a constant rate, indicating largely individual responses to constant stimuli outside the community of doctors. 4. The hypothesis that a doctor and his friend, a doctor and his advisor, and a doctor and his discussion partner would tend to introduce the new drug at about the same time was not borne out for the period as a whole.

5. During the early months following the drug's release, however, doctors who introduced the drug tended to follow closely upon any associates who had adopted it earlier.

6. This phenomenon was strongest during the very earliest months in the case of pairs of discussion partners and advisor-advisee pairs. In the case of pairs of friends, it reached its peak about 2 months later. In all three cases, the phenomenon occurred among the relatively isolated doctors as well as among the integrated doctors, but it reached its peak much later in the case of the isolated doctors.

7. The apparent greater effectiveness of contacts with colleagues during the early months was attributed to the greater uncertainty about the new drug that prevailed at that time. This interpretation is supported by comparisons of uncertain and clear-cut situations of another sort; pairs of related doctors were found to be more alike in the drugs they use for essential hypertension than in the drugs they use for respiratory infections.

A second article will inquire into the possible relationship of early gammanym introduction by a doctor to a more general tendency to be receptive to innovations.

REFERENCES

1. Coleman, James, Katz, Elihu, and Menzel, Herbert: Doctors and New Drugs (tentative title), Glencoe, The Free Press (in preparation).

2. Menzel, Herbert, and Katz, Elihu: Social Relations and Innovation in the Medical Profession: The Epidemiology of a New Drug, Public Opinion Quarterly 19:337, 1955. 3. Coleman, James, Katz, Elihu, and Menzel, Herbert: The Diffusion of an Innovation Among Physicians, Sociometry 20:253, 1957.

4. Menzel, Herbert: Public and Private Conformity Under Different Conditions of Acceptance in the Group, Journal of Abnormal and Social Psychology 55:398, 1957. Ridenour, Louis N., Shaw, Ralph R., and Hill, Albert G.: Bibliography in an Age of Science, Urbana, Ill., 1951, University of Illinois Press, pp. 30-35.

5.

6.

Sherif, Muzafer: Group Influences Upon the Formation of Norms and Attitudes, in Maccoby, Newcomb, and Hartley, editors, Readings in Social Psychology, New York, 1958, Henry Holt, pp. 219-233.

2.

DIMENSIONS OF BEING "MODERN" IN MEDICAL PRACTICE

HERBERT MENZEL, NEW YORK, N. Y.,
JAMES COLEMAN, and Elihu Katz,
CHICAGO, ILL.*

From the Bureau of Applied Social Research, Columbia University

(Received for publication Sept. 26, 1958)

IN ORDER to learn something of the way in which medical innovations make

their way into the doctor's practice, intensive interviews were conducted with 216 physicians in four midwestern cities, and their replies were subjected to statistical analysis. A previous article' described how the use of a new medication spread gradually among most of the 125 general practitioners, pediatricians, and internists among them. (This was ascertained by examining pharmacists' prescription records covering a 16-month period.f) This account constitutes a case history of the diffusion of a single innovation, a relatively minor variant of a well-established family of drugs. One cannot presume that the findings presented so far would necessarily hold true for innovations of other kinds or under other conditions. One wonders, nevertheless, whether the doctors who were first to use this particular drug would again be first the next time a new drug is introduced and whether they are progressive in other respects as well. This is an intriguing question; for, on the one hand most doctors know that they evaluate each proposed innovation on its own merits and, on the other hand, one keeps hearing some doctors described as "quick to jump on the bandwagon," or as "individuals that try everything that comes along," while others are characterized as "stodgy doctors" who "are not interested in hearing about new techniques and want things to go on as they are." Are such descriptions mere caricatures, or is there really some generalized tendency or trait of "innovatingness" which is strong in some doctors and weak or absent in others? This article is designed to answer this question, to the extent that an answer is possible on the basis of the data gathered. We will first look into the timing of the introduction of new drugs into the individual physician's practice and then into other aspects of the use of modern drugs. Later we will turn to some broader questions of modern practice.

This is the second of two articles and may be identified as Publication No. A-274 of the Bureau of Applied Social Research, Columbia University. Preparation was facilitated by funds provided by a grant to the Bureau of Applied Social Research from the Eda K. Loeb Fund.

*Coleman and Katz have transferred to the Department of Sociology, University of Chicago, since this study was conducted.

†A fuller account of our findings and procedures is being prepared for publication by The Free Press under the tentative title Doctors and New Drugs. Selected aspects are treated in detail in certain articles.3-5

THE TIMING OF THE FIRST USE OF THREE NEW DRUGS

The basic datum in our previous article was the date on which each doctor prescribed a certain medicine, termed "gammanym" in our report, for the first time. It was seen that 87 per cent of the doctors had used gammanym at least once, but that the dates on which they prescribed it for the first time ranged over the survey's entire period of 16 months from the time gammanym was released. Is a doctor's early introduction of gammanym attributable to any readiness to accept innovations generally? This question cannot be answered on the basis of experience with only one new drug. Fortunately, we have information on the doctor's introduction of two additional classes of drugs. One of these is a group of products which were first released on the American market about 6 months before "gammanym." Like gammanym, these drugs had possible application in the practice of most physicians; they will here be referred to by the name "cryptonym." Actually, all but 5 per cent of the interviewed physicians who treated the relevant conditions at all had used cryptonym drugs by the time they were interviewed. Again the time at which they had started using the new products varied widely. Were the early cryptonym users to be found among the early gammanym users, or was there no connection between these two forms of innovation? Table I shows that the early cryptonym users were indeed very likely (68 per cent) to be early gammanym users, while most (93 per cent) of the late cryptonym users were also late gammanym users. However, Table I is based on the relatively small number of doctors in the sample who used both drugs and gave precise information on the time they first used them.

*Tabulates only doctors who gave precise information about the timing of their introduction of both these drugs. Other doctors either did not use both drugs or else gave imprecise answers like "when It first came out." The doctors' own statements rather than pharmacists' records were used for this and the following table. This procedure will be discussed presently.

In order to conceal the identity of commercial products, all calendar dates are translated into "months after the release of gammanym."

A third group of drugs on which information was collected was not the same for each doctor. Each doctor had been asked, "What is the name of the drug that you have most recently tried in your practice? When did you use it for the first time?" The release date of each drug mentioned was ascertained and

the gap in time between that release date and the date on which the doctor first used the drug was measured. As before, this gap was found to vary over a wide range. Some doctors had "most recently tried" a drug which had only just been released; others, a drug that had been available for 18 months or more. Again the question arises are the doctors who were prompt in trying this selfdesignated drug to be found chiefly among doctors who were prompt to try gammanym and cryptonym? Table II provides the answer. Quite clearly, prompt introduction of the self-designated drug occurred most commonly (32 per cent) among the prompt introducers of gammanym and cryptonym. Later introduction of the self-designated drug occurred most commonly (76 per cent) among the slower introducers of gammanym and cryptonym. This finding gives us considerable confidence in the existence, within each doctor, of an over-all tendency to try out new drugs early or late.

If they gave

*Table II lists all doctors who gave precise information about the timing of their introduction of a self-designated drug and of at least one of the two drugs, gammanym or cryptonym. information on both, the average date was used.

THE DOCTOR'S OWN TESTIMONY AND HIS PRESCRIPTION RECORD

The tabulations in our earlier article were based on prescription records kept by pharmacists. Because pharmacists' prescription files were examined only for one type of drug, we are relying here on the doctors' own testimony. But is the physician's memory about the time when he first used a new drug reliable? Table III provides a partial answer.

There is evidently a fairly strong correlation between the date of first use of gammanym as stated by the doctor and that revealed by a search of the pharmacists' files. The doctors are certainly not guessing or making the dates up. Doctors who placed themselves in the class of earliest introducers tend to be found in the earliest class according to the prescription search. Those who placed themselves in the latest class tend to be found in the latest class according to the record. Those who indicated an intermediate position for themselves tend to occupy an intermediate position according to the files. In other words, the doctor's own statements produce for most, though not all, doctors the same ordering as does the search of prescription records.

*The classification used in Table III allows the doctor's statement a 2-month "headstart," for reasons which will soon become apparent. Calendar dates are replaced by "months after the release of gammanym," as before.

This, to be sure, is far from saying that the actual date given by the doctor is the same date as that revealed by the prescription search. In fact, even if we allow both to the doctor's statement and to the sampling of the pharmacists' files a modest margin of error, only about one-quarter of the doctors can qualify as accurate (defined as stating a date not more than one month before or after the prescription date).

TABLE IV. DATE OF INTRODUCTION OF GAMMANYM
(Interval between Interview and Prescription Dates)

Table IV shows that 27 per cent of the doctors on whom sufficient information is available fall within this somewhat liberalized definition of accuracy. At the same time, the number of doctors who give an earlier date than their prescription record is exactly twice that of the doctors who give a later date. This indicates that the majority of these doctors prefer to think of themselves as prompt triers of a new drug, rather than as latecomers. Hence, in instances where their memory fails to give them an accurate date, they tend, unconsciously or consciously, more often to "update" than to "downdate" themselves.*

*To some extent this discrepancy is also due to the fact that the prescription search only sampled 3 days of every month and therefore yielded dates which may frequently lag a month or so behind the actual introduction date. But this sampling error can account for only a fraction of the observed dis crepancy.