In 1912, Herrick (7) attributed myocardial infarction to coronary artery thrombus. Fierce debate followed for the next 68 years until DeWood et al. (4) convincingly demonstrated the primary role of thrombus in 1980. Even before this debate was settled, Fletcher et al. (5) reported the first use of thrombolysis in acute myocardial infraction (AMI) in 1958. Between 1959 and 1988, thirty-three trials comparing intravenous streptokinase with placebo or no therapy were reported. A 1992 retrospective cumulative meta-analysis of these trials significantly favored treatment (15). Indeed, the case for thrombolytic drugs could have been considered proven by 1973, at which time ten studies had been conducted involving the randomization of 2,544 patients (1). To date, in excess of 200,000 patients have been randomized in clinical trials (19). Thrombolytic therapy arguably revolutionized the management of AMI in the 1980s, pushing clinical care more to active myocardial salvage. However, take-up in the United Kingdom has been inhibited by difficulty with timely delivery as well as concerns about contraindications and complications.
Influences upon the adoption and diffusion of medical technologies, such as thrombolysis, are wide ranging and include technical, medical, social, and economic factors (9). In general, three main influences on diffusion have been put forward: actors in the process—involvement of clinicians, patients, and health-care purchasers; structure and environment—health services and commercial market; and the characteristics of the innovations—technology type and cost (2). Ultimately, the actors involved, notably the medical professionals, are the final influence on adoption within health services. Elucidation of the key influences on medical professionals and other players is not complete, but one means put forward is the publication of key research results, whereas other influences are thought to include evidence-based guidance and guidelines. In the case of thrombolysis described for the Trent region of England from 1987 to 1992, the former were suggested as critical, based principally on the slope of the adoption curve (12).
This study uses a novel means of using three separately briefed groups of senior cardiologists to verify a best estimate of the diffusion curve and investigate the links between both diffusion and the time line events using an analytical method. Time lines that illustrate the events that may have influenced thrombolysis adoption and diffusion in England from 1980 are set out in Figure 1 with supportive text in Box 1. Figure 2 using data supplied by IMS Health, a commercial agency that collects data nationally on drug use and sales, shows a diffusion curve for thrombolytic agents that clearly demonstrates a sharp rise in 1987 to a plateau in the early 1990s with a second rise in the late 1990s (Packer et al., this issue).
Time lines of the pivotal events in the use of thrombolysis in the treatment of acute myocardial infarction in England. Time lines were developed in collaboration with clinical and policy experts. A: Primary research: The trial name/abbreviation or the first author is given in bold. For additional information and references see Box 1. SK,streptokinase; tPA, tissue plasminogen activator; accel. tPA, accelerated tPA. B: Secondary research: guidelines, guidance, and reviews. C: Process and licensing. Boxes with the dashed border illustrate when the particular thrombolytic first appeared in the British National Formulary (BNF) with a specific indication of myocardial infarction or acutemyocardial infarction.
Estimated doses of thrombolytic agents in England. Source:IMS Health.
METHODS
A sample of cardiologists was taken from the specialists registered on an internet directory (www.specialistinfo.com) with supplementary information from the General Medical Council's database (www.gmc-uk.org). Cardiologists were selected if they had a work-based address in England and a date of first qualification before 1977. One hundred and thirty-eight study eligible cardiologists were identified, from which 69 (50 percent) were randomly selected and randomized to one of three study groups.
GROUP A: Received the time lines (Figure 1—with supportive text, shown in a shortened form in box 1) and a blank diffusion grid. Respondents were asked to grade each of the thirty-nine events outlined on the time lines according to the effect it had, in their opinion, on the overall diffusion of thrombolytics in the treatment of AMI in England. The grades were as follows: “1”, marked increase; “2”, increase; “3”, little/no effect; “4”, decrease; “5”, marked decrease; and “X”, not familiar or difficult to say.
In addition, this group was asked to sketch a diffusion curve for thrombolysis use from 1980 using the grid while considering their grading responses. On the y-axis, thrombolytic use was marked from “no usage” at the origin to “maximum usage.” Maximum usage was defined as “the numbers of patients presenting to medical services with an AMI.”
GROUP B: Received the diffusion curve (Figure 2) and was asked to describe, with no prompts, the events that may explain its shape.
GROUP C: Received the time lines and the diffusion curve. As in Group A, respondents were asked to assign a code to each event considering the diffusion curve. Respondents in Groups B and C were instructed to make amendments to the shape of the diffusion curve, if they wished, so as to reflect their opinion of thrombolysis use in England.
All documentation was comprehensively piloted before the main study. The questionnaires were sent out to the main sample in 2002. A £50 book token incentive was offered for the return of a completed questionnaire. A week after the deadline, follow-up calls were made to those cardiologists who had not responded. Twenty completed questionnaires were received. On exclusion of the one invalid response, nineteen questionnaires were analyzed—seven in Group A, five in Group B, and seven in Group C.
RESULTS
Group A—Time Line Only
An increase in thrombolytic use during the late 1980s to early 1990s was highlighted in all the sketches from the respondents. Six respondents included a plateau in diffusion after this initial adoption. Three respondents also sketched a second rise in the mid- to late 1990s following this plateau.
We summarized the grades given to each event into two categories—“increase” and “no change.” “Increase” represents the number of respondents in Group A that assigned either a “1” (marked increase) or a “2” (increase) to that particular event. “No change” represents the total number of respondents in this group that assigned a “3” (little/no effect) to the event. A very small number assigned either a “4” (decrease) or an “X” (not familiar/difficult to say). No respondent assigned a “5” (marked decrease) to an event. The grading of the events on the time lines is summarized in Table 1 for the respondents in Group A.
The events thought to have increased the use of thrombolysis were as follows:
- from all seven respondents: ISIS-2 (Second International Study of Infarct Survival—clinical trial of intravenous streptokinase, oral aspirin, both, or neither), first appearance of streptokinase in the British National Formulary (BNF) for this indication, “fast-track” thrombolysis (i.e., rapid access to cardiac-care team), and thrombolysis in Accident & Emergency (A&E) departments;
- from six respondents: GISSI (Gruppo italiano per lo studio della streptochinasi nell'infarcto miocardio—clinical trial of intravenous streptokinase versus placebo), rise in evidence-based medicine, British Cardiac Society/Royal College of Physicians & British Heart Foundation (BCS/RCP & BHF) guidelines for AMI management;
- from five respondents: the National Health Service (NHS) Plan & National Service Framework for Coronary Heart Disease (NSF for CHD), direct admission to the cardiac-care unit at the request of the patient's general practitioner, Birkhead audits (measurement of delay between onset of symptoms and thrombolysis provision and proportion of patients receiving thrombolysis within specified time targets).
Group B—Diffusion Curve Only
One respondent in this group amended the diffusion curve provided by sketching a rise in use from 1998 onward. Table 2 sets out the main themes to emerge from the five respondents in Group B.
Group C—Time Line and Diffusion Curve
No respondent amended the curve. The grading of the events on the time lines is summarized in Table 3 for the respondents in Group C. The events thought to have increased the use of thrombolysis were as follows:
- from seven respondents: GISSI, ISIS-2 and the first appearance of streptokinase in the BNF for this indication
- from six respondents: thrombolysis in A&E departments
- from five respondents: BCS/RCP & BHF guidelines, NHS Plan & NSF for CHD, the first appearance of alteplase in the BNF for this indication, “fast-track” thrombolysis, Birkhead audits, nurse-assessed thrombolysis
DISCUSSION
The construction of retrospective diffusion curves is always problematic because of a variety of data difficulties (Packer et al., this issue; 18). This study, however, gives confidence that the daily dose data for a class of drugs has been acceptably accurate. Only one respondent from the twelve that received the diffusion curve amended it by sketching an increase from 1998 onward (in contrast to the decline illustrated). The respondents in Group A who had no data-driven curve to go on produced a remarkably consistent composite picture, including a plateau or a slowing of diffusion after the initial adoption in most cases. We can say that we have no evidence to suggest that the diffusion curve in Figure 2 does not represent the adoption and diffusion pattern of thrombolytic agents in England.
The clinical trials GISSI (6) and ISIS-2 (11) were deemed to be key influences upon thrombolysis diffusion in this study. It is worth noting that, while grading ISIS-2, respondents may have been considering this trial's interim results that were published in 1987 (10). It is possible that this publication heightened awareness of the trial before the main results were published in 1988. Previously, evidence from trials has been demonstrated to have a variable impact on cardiology practice, although rigorously conducted, highly relevant randomized control trials published in high impact journals have been shown to result in a measurable influence on clinical behavior (3;13;14).
To our respondents, the influence of service developments, for example, A&E thrombolysis and fast-track thrombolysis and the drive for improved performance initiated by the clinical audit were also important (8;16). National guidelines were influential but were lower order compared with clinical trials. Indeed, concern has been expressed around implementation costs of cardiology guidelines in the United Kingdom (17).
The novel method outlined here represents a viable approach to the investigation of influences upon technology diffusion. Clear themes emerged from the study groups, and although all three had differing data sets, they all came to the same broad conclusions. However, it would require several further case studies with varying technologies before any generalizable statements could be inferred about diffusion influences from the perspective of clinicians. In the case study presented here, we believe that our sample was large enough to ensure that we have probably heard most of the perceptions that might be important.
Although it is apparent that the overall influences on adoption and diffusion of thrombolysis were multiple, clinical trials, service developments and national guidelines all were judged to have played a part. The GISSI and ISIS-2 clinical trials were confirmed as the major influence on initial adoption.
Policy Implications
This study challenges the assumption that guidelines/guidance are an overriding influence in the directing of clinician behavior. The power of the landmark clinical trial is clearly critical in this example. It will be interesting to observe whether this continues to be the case in the United Kingdom, where compulsory guidance is issued by the National Institute for Clinical Excellence (NICE). Further such research could inform policy-makers, who wish to influence the adoption of clinically effective health technology.
The authors thank the cardiologists involved in the pilot and the main survey. Mr. Peter Stephens of IMS Health kindly provided national data on the supply and use of thrombolytic agents. Claire Packer and Alison Cook are funded by the Department of Health and Andrew Stevens by the National Health Service for England.
