No CrossRef data available.
Response to an optimistic viewpoint
Published online by Cambridge University Press: 04 August 2005
Extract
We appreciate the interests in our work and observe that we agree that earlier thrombolysis for AMI can reduce the AMI case fatality. The question is how much and at what price. There are no data available to directly address this issue, and we developed a simulation model to quantify costs and health consequences of less thrombolytic delay by using public awareness campaigns, telemedicine, or a combination of the two. Inevitably, such a model needs to be based on several uncertain parameter values. We performed a range of sensitivity analyses so readers of the analysis could see the effect of replacing our base case values with others that the reader might consider more appropriate. Due to space restrictions, we had to omit a table with sensitivity analyses that would have addressed several of the concerns Terkelsen and coworkers have.
- Type
- LETTERS TO THE EDITOR
- Information
- International Journal of Technology Assessment in Health Care , Volume 21 , Issue 3 , July 2005 , pp. 416 - 419
- Copyright
- © 2005 Cambridge University Press
To the Editor:
We appreciate the interests in our work and observe that we agree that earlier thrombolysis for AMI can reduce the AMI case fatality. The question is how much and at what price. There are no data available to directly address this issue, and we developed a simulation model to quantify costs and health consequences of less thrombolytic delay by using public awareness campaigns, telemedicine, or a combination of the two. Inevitably, such a model needs to be based on several uncertain parameter values. We performed a range of sensitivity analyses so readers of the analysis could see the effect of replacing our base case values with others that the reader might consider more appropriate. Due to space restrictions, we had to omit a table with sensitivity analyses that would have addressed several of the concerns Terkelsen and coworkers have.
Ad§1. With reference to a Dutch study (10), Terkelsen and coworkers claim that 25% of the AMI population will have prehospital thrombolytic within the first hour after onset of symptoms. We argue that prehospital thrombolytic therapy does not seem very realistic or relevant in Denmark for at least two reasons:
- The delay between calling an ambulance and the patient arriving at the nearest hospital rarely exceed 30 minutes in Denmark (total prehospital delay) and, in an urban setting, less.
- The time elapsing from the ambulance arriving to the patient until the prehospital diagnostic procedure and preparation of the thrombolytic infusion (streptokinase/anistreplase) is accomplished would exceed the period spent in the ambulance (20 minutes). Data from the Dutch study (10) and a Swedish study (2) support this assumption. In the Dutch study, the time elapsed from arrival of the ambulance until start of thrombolytic infusion was 27 minutes (10). In the Swedish study, it is claimed that prehospital thrombolytic therapy is recommended when the prehospital delay exceeds 30 minutes (2).
Based on two earlier studies of prehospital telemedicine AMI diagnostics where door-to-needle delay (hospital delay) was reduced by approximately 30 minutes (6;7), we assumed that hospital delay as base case would be reduced from 60 minutes to 30 minutes. In a Danish feasibility study of prehospital telemedicine diagnostics (16), hospital delay was reduced from 81 minutes to 38 minutes, and it was estimated that the prehospital diagnostic procedure itself will last 13 minutes. Thus the study does demonstrate the feasibility of prehospital telemedicine diagnostics but not the feasibility of prehospital thrombolysis in Denmark.
We explicitly based our analysis of the telemedicine strategy on reduction of hospital delay. Even with a minimum of patient delay and a total prehospital delay of 20– 30 minutes, no patients would be treated within the first hour if the hospital delay is reduced 60 minutes to 30 minutes.
As described in the study (9), the magnitude of the reduction in hospital delay was tested in the sensitivity analysis. If hospital delay is totally eliminated (corresponding to a delay reduction of 1 hour), the reduction in case fatality more than doubles. This estimate is mainly explained by the fact that some patients in this situation will be treated during the first hour after onset of symptoms. However, based on a Danish study (13), we estimated that only approximately 7% of Danish AMI population (i.e., of all AMI patients arriving alive at the hospital) arrive at the hospital within the first hour after symptom debut and that only a fraction of those arrive by ambulance. Thirty-four percent arrive at the hospital within the first 2 hours (12). Thus if the hospital delay is eliminated as a consequence of the telemedicine strategy, approximately 7 percent of the Danish AMI population could be treated within the first hour—if all arrived by ambulance and if thrombolytic therapy were beneficial for all. If a greater proportion of patients were transported by ambulance shortly after onset of symptoms, the health benefit of the telemedicine strategy would increase!
In the Dutch study mentioned by Terkelsen and colleagues, 21 percent of the included AMI patients who were treated with prehospital thrombolytic infusion (as opposed to those treated with bolus injection) were treated within the first hour. The delay from arrival of the ambulance to treatment was 27 minutes. In Danish routine practice, however, it is unlikely that 21 percent of the AMI population have a patient delay of less than 33 minutes—and that all of them would be transported by ambulance and also would benefit from thrombolytic therapy.
Our arguments about prehospital thrombolytic therapy is supported by Stern and Arntz (14), who state that “As is evidenced by several studies, most benefit in terms of myocardial salvage and short- and long-term mortality is achieved with initiation of therapy within the first 60–90 minutes after onset of symptoms. Nearly exclusively, prehospital initiation of thrombolysis makes it possible to take advantage of this early time window. Since it has been shown that prehospital diagnosis of an acute myocardial infarction is reliable and out-of-hospital initiation of therapy has no additional specific risk, patients seen within the first 60–90 minutes after onset of symptoms or for whom a relevant time gain of more than 90 minutes can be expected are ideal candidates for, and therefore should receive, prehospital thrombolysis.”
Ad.§2. While an early meta-analysis concluded that the effect of thrombolysis on mortality declines linearly with increasing delay from onset of symptoms (1), Boersma and colleagues argue that the decline is exponential (i.e., much greater effect with small delays) (3). Boersma and colleagues reanalyzed the data, including also analysis of prehospital versus in-hospital thrombolytic therapy (randomization of delay). We adopted Boersma's functional form, which is in line with the opinion of Terkelsen and colleagues (as far as we understand their arguments) (15).
The crucial issue is then what proportion of unselected Danish AMI patients will have early thrombolysis (defined as the first 1–2 hours after onset of symptoms) within a telemedicine strategy. The real weakness of our analysis rather lies in the fact that we divided patients into time categories rather than treating time continuously. This strategy may have caused some underestimation of the beneficial effect of treatment during the golden hour(s). Our method, however, cannot bias the result by a factor of seven to nine. If Terkelsen and coworkers assume that prehospital thrombolysis reduces the case fatality by 15–22 per 1,000 treated—independently of the distribution of patient delay in Danish routine practice and of the time gained by pre-hospital thrombolytic therapy—they might overestimate the effect of prehospital thrombolysis in Denmark considerably.
Ad.§3. Due to space limitation, we were not allowed to present detailed cost data in the paper. These data, however, were based on the original Danish report (8) but were revised based on published comments to the report about the resources needed for the telemedicine strategy. We had discussions with the Danish ambulance operator (Falck) to get reliable data, and we observe that Terkelsen and coworkers disagree with the ambulance operator on several points.
To improve the prehospital management of patients, the Danish ambulance staff (not paramedics) needs upgrading, but this upgrading does not necessarily include prehospital telemedicine diagnostic tools. As proposed by Falck and described in the study (9), we tested the consequences of excluding upgrading costs in the sensitivity analysis. Due to space limitations, we could not present the sensitivity analysis in detail.
The defibrillators should be excluded from the base case analysis, only if new defibrillators are not required due to the introduction of the telemedicine system. But outdated equipment is not as easily sold as Terkelsen and coworkers propose. Excluding defibrillators, the total costs would be 272,225,000 DKK, corresponding to 89% of the base case estimate.
Terkelsen and coworkers claim that the costs of a 5-year telemedicine strategy would be 183,453,249 DKK, corresponding to 60% of our base case estimate. From experience with a Norwegian telemedicine project (4), we know that the real costs of telemedicine may increase by a factor of two to three compared with estimates from telemedicine enthusiasts.
Ad.§4. As explicitly stated in the study, the evidence on the effect of public campaigns is conflicting (9). We applied data from the Swedish study because of cultural similarities between Sweden and Denmark, comparable precampaign patient delay and very detailed descriptions and documentation of the campaign: importantly, the distributions of patient delay before and after the campaign. In the REACT trial (11), such detailed information was not available; moreover, both the intervention group and the comparison group experienced reduction in median delay times—this finding might be explained by diffusion of information from the intervention group to the comparison group and not merely by a secular trend.
Ad.§5. It should be obvious from our study that we consider thrombolysis to be a real benefit to AMI patients, and earlier thrombolysis an even greater benefit. The question is whether further efforts to reduce delay are cost-effective. Our study results indicate that this may not be the case in a Danish setting (see §1). Hence, we conclude that programs aimed at reducing delay of thrombolysis in patients with AMI are likely to have limited impact on AMI fatality—in Denmark.
Ad.§6. We are well aware of PCI for AMI, and one of us (I.S.K.) recently has published a study on its cost-effectiveness (5). The study that Terkelsen and coworkers criticize (9), however, was designed to evaluate thrombolysis and not primary PCI. The study was designed and implemented at a time when PCI was not routine therapy, and thrombolytic infusion alone was a relevant strategy for AMI. The introduction of new types of thrombolytic agents has cost implications. Today, PCI is routine and frequently combined with prior thrombolysis.
Terkelsen and colleagues state that “a recent study (16) has documented that a prehospital diagnostic strategy results in 81 minutes earlier initiation of primary PCI if combined with direct referral of patients to an interventional center.” However, the earlier initiation of PCI is partly explained by the fact that prehospital diagnostic bypasses the local hospital, and the finding does not invalidate our arguments and analysis. In our study, we highlighted that telemedicine in combination with primary PCI might render the telemedicine strategy more cost-effective. A cost-effectiveness analysis of primary PCI with and without the use of telemedicine—in routine practice—would be of great interest.
The health and resource consequences of reducing thrombolysis delay depends on an array of assumptions about previous medical practice and what can be achieved through various strategies to reduce thrombolytic delay. Here, Terkelsen and coworkers appear to believe in one set of assumptions and we in another. They claim that telemedicine will achieve seven to nine times greater effect than our model would indicate; however, this statement seems to be based on the assumption that prehospital thrombolysis will reduce the case fatality by 15–22 per 1,000 treated—independently of the distribution of patient delay in routine practice and the time gained by the strategy. They hope that our “pessimism” can be replaced by “optimism and inspire the readers to establish” new technologies. Our position is that science is brought forward by asking critical questions rather than excessive optimism. Finally, use of value-laden words such as “error,” “major miscalculation,” “ignore fact,” and so on is unlikely to facilitate exchange of viewpoints among researchers, especially when we are dealing with disagreement over assumptions rather than scientifically established truths.