A detailed picture of the annual rate of entry of NCEs into clinical investigation between 1958 and 1979 is shown in Fig. 1. After the first observation in 1958. there was a steep rise in 1959. Then beginning in 1960, there was a sharp decline for 4 consecutive years with the largest drop (67%) from 1962 to 1963. Comparing the 5-year period 1958 to 1962 to the following decade, 1963 to 1972, there was a 60% overall decline from a mean of 89 a year to 35 a year. In the last 5 years of the survey (1975 to 1979) the mean rate declined further, to 17 a year-an 81% drop from the pre-1963 average level.

Fig. I also shows the number of selforiginated NCEs studied in man each year. Since self-originated NCES account for approximately 80% of the total sample, they follow trends that are similar to those for all NCES. Entry of self-originated NCEs into clinical test

ing dropped sharply in the early 1960s and continued to decline thereafter.

From 1958 until the late 1960s. only 3% or less of self-originated NCEs were first tested abroad (Fig. 1). In the first half of the 1970s. however, a strong trend developed toward initial testing abroad. This trend peaked at 60% in 1975. The proportion has since fluctuated but in general has declined: in 1977 to 1978 only 21% of self-originated NCEs entered clinical trials abroad, although the percentage rose to 45% in 1979. The trends shown here for the nine firms are similar to those we observed for all U.S. firms over the period 1963 to 1979.23

Fig. 2 shows the number of IND/eqs filed on self-originated and acquired NCEs and the percentage of those that were self-originated. Although IND/eq filings by the nine firms have decreased over time, the self-originated percentage has remained at approximately 80%.

Time required to reach NDA approval. We compared the average time required for NCES to progress from IND/eq filing to NDA approval at the beginning and at the end of the observation period. Self-originated NCEs that entered clinical trials between 1958 and 1963 averaged 54 months from IND/eq filing to NDA approval, whereas those approved between 1972 and 1979 averaged 112 months."

Comparison of pharmacologic types of NCEs under investigation in 1958 to 1963 and in 1975 to 1979. We compared pharmacologic types of the NCEs investigated in the

With a data base of this type that has fixed start and finish doses. in order to avoid biases because of start-up artifact and right censoring of the data, one must calculam values for the early years by the dase of IND/eq filing and values for the later years by NDA approval. This removes one type of bias but makes the values not exactly comparable, so that the precise size of the increse cannot be obtained from these values.

first and last 5 years of the survey period (Table II).

The emphasis on certain pharmacologic areas has changed. In particular, psychotropic and neurotropic drugs, analgesic and anti-inflammatory drugs, and drugs acting on the motor system and on body fluids and electrolytes accounted for larger percentages of the pre-1963 NCES than the 1975 to 1979 NCES. On the other hand, cardiovascular, endocrine, and digestive system drugs accounted for smaller percentages in the pre-1963 period than they did in 1975 to 1979.

This comparison also highlights the decline in the number of NCEs entering clinical investigation. Although the 11 major pharmacologic areas in Table I accounted for approximately 90% of the NCEs under investigation both in the pre-1963 period and in the last 5 years, they

previously encompassed many more NCES (420 and 77).

New drug approvals, 1950-1980. In addition to studying the entry rate of drugs into clinical research, we also examined the approval rate of drugs for the market. The total number of NCE-NDA approvals granted to the nine firms each year from 1950 to 1980 and the number granted for self-originated NCES from 1963 to 1980 are shown in Fig. 3. This graph shows that the number of NCE-NDA approvals fell sharply in 1961 from a mean of 10.6 a year in 1950 to 1960 to a mean of 5.4 a year in 1961 to 1967 (a decline of 49%). The decline continued from the mid-1960s to the early 1970s, to a mean of three a year in 1968-1975, with an overall decline of 72% from the 1950-1960 level.

The number of self-originated NCEs approved (shown in the dashed line of Fig. 3),

declined even more (71% from the mid-1960s to the early 1970s alone) to only one a year. Subsequently the numbers recovered, so that by 1980 they had returned to the levels of the early 1960s. These trends are similar to those shown for NCE drug approvals granted in the same period to all U.S.-owned pharmaceutical irms."

Discussion

The manner in which pharmaceutical firms in the United States could test their drugs in man and obtain NDA approval for marketing changed importantly with the passage of the Kefauver-Harris Drug Amendments on Oct. 10. 1962, and issuance by the FDA of procedural and interpretative regulations that came into effect on Feb. 7, 1963.4. Before 1963 the regulations governing clinical trials on INDs did not require either an initial notice to the FDA or

subsequent reports on ongoing trials. Under the new regulations a sponsor who wished to test a new drug or antibiotic in man had to file with the FDA a notice of claimed investigational exemption" (IND) before clinical trials could commence. For the first time the FDA required substantial information before clinical work could begin. This information included data on the nature of the new drug, the preclinical toxicity tests that had been performed, the proposed plans for clinical trials, and the identity and qualifications of the investigators who had to supervise and be responsible for the trials. Informed consent of clinical subjects was also required for the first time. The subsequent clinical research was to be closely monitored, and detailed reports on its progress were to be filed regularly with the FDA. If the FDA deemed the plans inadequate or the trials unsafe, it could require corrective action or termination of the studies.

The criteria for approving an NDA also changed in 1962.3.6.8.9.15.16 The provision in the 1938 act that had required the FDA to approve an NDA automatically 60 days after its submission was dropped; the requirement for premarket notification was changed to a requirement for premarket approval. A requirement was added that the manufacturer should pro

*Automatic approval was granted unless the FDA deemed within that time that the information supplied was incomplets or that more time was required (up to 180 days) to review the application.

vide "substantial evidence" through "wellcontrolled investigations" to show that a drug was effective, as well as safe, for its proposed indications. The impact of the efficacy provisions on drug development in 1962 and the years immediately following is not clear-cut, however, because it took almost 8 years for the FDA to establish detailed criteria for "wellcontrolled investigations:" these regulations were not made final until May 8, 1970.00

Although the effects of the 1962 drug amendments on the number of drugs being marketed in the United States since 1962 have been analyzed extensively, 12.11.12.18.19 there have been no previous studies of the other primary intent of the amendments, namely to control clinical drug research. Our study shows how large the impact was: the amendments were associated with a steep reduction (by 60% or more) in the number of NCEs entering clinical testing. In subsequent years there was a corresponding decline in the number of NCEs reaching NDA approval, an increase in the time required to do so, and a further reduction of NCES entering clinical testing.

The temporal changes described in this paper are complex, and the reasons for them are complicated as well. The peak in the number of

The 1938 law required that the drug be safe for its intended uses. Although proof of efficacy had not formally been required, the safery judgment had presumably been made in light of the drug's intended uses and information about its efficacy.

NCES tested in man in 1959 may be a statistical fluke that represents a chance variation: for instance, in this sample the number of NCES tested in man per company in the period 1958 to 1962 ranged from as few as one to as many as 32 a year. Alternatively, a specific scientific development may have come to a peak in that year, such as the culmination of activity in many firms searching for better semisynthetic penicillins. Another possibility is that 1959 was a high-water mark of industrial enthusiasm and financial commitment after the dramatic scientific and commercial successes of the 1940 to 1950 decade. Whatever the explanation for the peak in 1959. the following years unquestionably represent a marked and permanent decline in the number of NCEs entering human investigation.

Commentaries about both investigational NCES and NDA approvals during this period tend to assume that anything occurring before or during 1962 cannot be blamed on the Kefauver-Harris amendments passed in October of that year or on the implementing regulations that followed. The 1962 legislation, however. was the culmination of 4 years of congressional hearings that attacked the pharmaceutical industry, its products, and its advertising and pricing policies. Congressman Blatnik and, later. Senator Kefauver chaired extensive hearings over those years, and the actions of the FDA and industry during the period were subject to extensive media coverage. Added to this was the growing realization of the thalidomide tragedy in Europe. Indeed, the FDA itself had published proposed new IND regulations on Aug. 10, 1962-under the existing law-2 months before the Kefauver-Harris amendments were enacted. Thus it is obvious that a change in conditions and attitude existed well before the amendinents passed and that this was one likely inhibitor of both the industry's clinical NCE studies and the FDA's NDA approvals.

The permanent decline in investigational NCEs in the early 1960s would be expected (if no compensatory factors operated) to have led to a decline in NDA approvals after a latency period corresponding to the average INDplus-NDA time. Analysis of the yearly NCENDA approvals obtained by the nine firms

showed that after the sharp decline in the early 1960s. there was indeed a further decline in approvals (Fig. 3), which was slower but of a considerable magnitude (49%), from the mid1960s until the mid-1970s. The later decline was even more marked (71%) in the case of self-originated NCES. In a separate paper dealing with the whole U.S. pharmaceutical industry, we discuss (1) wider aspects of this link between the flow of investigational NCEs and subsequent NDA approvals and (2) the possible future significance of the further NCE decline of the mid-1970s.

Other influences. perhaps more subtle but as fundamental, were also contributing to the reduction in NCE flow. Running through this whole period, but difficult to quantify. were changes in both philosophy and state of the art. Scientific attitudes are changed by many forces. including technologic progress. In the 1940s and 1950s many in industry believed that preclinical testing was not highly predictive of a drug's clinical utility and that after a modest amount of toxicity testing, a new drug should (and safely could) be tested promptly in man. However, the public's concern about adverse drug effects prompted the FDA and industry to add many preclinical tests that had not been routinely conducted previously (e.g.. tests for teratogenicity, carcinogenicity, and recently mutagenicity). Whether or not these tests vindicated the time and money spent on them is beside the point: it became almost unthinkable not to do them, and the result-for both scientific and economic reasons-would be fewer drugs left to enter clinical testing. At the same time. laboratory scientists were becoming more accurate in predicting therapeutic activity. Today, for example, it is rare for an NCE not to show the proposed therapeutic effect postulated by chemical theory and animal experimentation. Such methodologic progress justifies more nonhuman pharmacodynamic evaluation, and again the trend would be for fewer compounds to reach clinical testing.

Finally, the scientific rules for convincing scientists about efficacy were changing. The modern controlled trial became firmly established as the premiere method for demonstrating clinical activity in an unbiased and convincing