Skip Navigation

European Journal of Heart Failure 2000 2(3):315-324; doi:10.1016/S1388-9842(00)00101-X
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (3)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Fagerberg, B.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Fagerberg, B.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© 2000 European Society of Cardiology

Screening, endpoint classification, and safety monitoring in the Metoprolol CR/XL Randomised Intervention Trial in Congestive Heart Failure (MERIT-HF)

Björn Fagerberg*,1

Wallenberg Laboratory for Cardiovascular Research, Sahlgrenska University Hospital SE-413 45 Gothenburg, Sweden


   Abstract
Top
 Notes
 Abstract
1. Introduction
 2. Methods
 3. Results
 4. Discussion
 Appendix A
 References
 
The MERIT-HF study was designed to investigate the effect of once-daily dosing of metoprolol succinate CR/XL added to standard therapy in patients with chronic heart failure. A screening programme, aiming at identifying eligible patients for the study was used by 231 out of 313 sites, resulting in 8912 screened patients. These patients were older, more often women, had marginally higher ejection fraction and were more often in NYHA class IV compared to the finally randomised patients. There was a positive correlation between the number of screened and randomised patients and due to the high inclusion rate the number of randomised patients was increased from 3200 to 3991 patients. The clinical events were defined by a manual and were based on identical classifications made independently by two members of an Independent Endpoint Committee. Data from the case report form and the classifications made by the Endpoint Committee were in accordance for hospitalisation due to worsening heart failure on 722 occasions. On 207 occasions the data from the case report forms and the classification made by the committee differed regarding this clinical event. The Independent Safety Committee monitored safety aspects of the study by using asymmetric group sequential procedures. The study was stopped early on recommendation from the Independent Safety Committee, when the second pre-specified interim analysis (50% of total number of expected deaths) showed that the criterion for stopping due to benefit had been met and exceeded. In conclusion, a screening programme facilitated patient recruitment and showed the characteristics of the background population, the event classification was improved by using an independent endpoint committee, and the safety monitoring allowed an early closure of the study due to mortality benefit, without jeopardising the final analyses of the results.

Key Words: Screening • Endpoint classification • Safety monitoring

Received June 1, 2000; Revised June 13, 2000; Accepted June 14, 2000


   1. Introduction
Top
 Notes
 Abstract
 1. Introduction
2. Methods
 3. Results
 4. Discussion
 Appendix A
 References
 
Clinical studies are the cornerstones of evidence-based medicine when new treatments for serious diseases are established. The realisation of such studies is a complicated issue dealing with important decisions of medical, ethical, statistical, and logistic nature [1]. Patients, healthcare personnel, and fiscal resources are precious and must be used properly. Hence, it is desirable that experiences from new trials are used in a continuous discussion of how to make clinical studies as effective as possible. Some trial design features are of particular importance for the outcome of large survival studies in patients with heart failure.

First, there is often a lack of data on the background population in clinical studies and the recruitment of patients may last over extended periods of time with few included patients per investigational centre [2]. The use of screening procedures is one method to solve these problems. Second, there is a considerable heterogeneity in the definition of modes and causes of deaths in heart failure studies, not only for sudden death but also for circulatory failure [3]. These discrepancies make it difficult to compare studies and reach conclusions on cause-specific mortality and morbidity [3]. Hence, more strict definitions of clinical events should be used. Composite endpoints are often used, necessitating strict definitions and independent endpoint classifications. Previous reports have shown that there may be important differences in the classification of events made by investigators and centralised event committees in cardiovascular studies [4,5]. Finally, safety monitoring of survival studies is necessary from both the ethical and scientific perspective [1,6]. However, when interim monitoring of trials is performed meticulously defined procedures have to be used in order to avoid failures to document the true efficacy or safety of an intervention [4–6].

The Metoprolol CR/XL Randomised Intervention Trial in Heart Failure (MERIT-HF) was designed to investigate the effect of once-daily dosing of metoprolol succinate controlled release/extended release (CR/XL) added to standard therapy in patients with chronic heart failure [7–9]. A total of 3991 patients were recruited to this international, double-blind, randomised, placebo controlled survival study. The study was stopped early on recommendation from the Independent Safety Committee due to a favourable effect on total mortality in the metoprolol CR/XL group. It took 28 months from the start of this study to the publication of mortality data [8].

The aim of this report is to describe the screening programme, the endpoint classification procedures and the safety monitoring in the MERIT-HF study and examine how these trial design features influenced the realisation of the study.


   2. Methods
Top
 Notes
 Abstract
 1. Introduction
 2. Methods
3. Results
 4. Discussion
 Appendix A
 References
 
2.1. Design and organisation
The study outline has been described in detail elsewhere [7–9]. In summary, the first primary endpoint was total mortality and the second primary endpoint was total mortality or all cause hospitalisation (time to first event). The design was a randomised double-blind placebo-controlled parallel-group study preceded by a 2-week single-blind placebo period. There was a common closing date for all patients in the study. Eligible patients were 40–80 years old, with a reduced ejection fraction (≥0.40) and symptoms of chronic heart failure [New York Heart Association functional (NYHA) classes II–IV]. The study ran at 313 sites in the United States, and 13 European countries (Belgium, Czech Republic, Denmark, Finland, Germany, Hungary, Iceland, the Netherlands, Norway, Poland, Sweden, Switzerland, UK). The power calculation showed that 3200 patients had to be recruited and followed for a mean follow-up time of 2.4 years [7–9]. The study was approved by the ethics committees at each site and the patients provided written informed consent.

The International Steering Committee consisted of the members of the International Executive Committee, national coordinators from the participating countries, and non-voting members from the sponsor. The International Steering Committee had the scientific responsibility for the study. An Independent Endpoint Committee whose five members were unaware of the treatment status classified all events from copies of medical charts and other documents according to prespecified definitions. An Independent Safety Committee including two cardiologists, one biostatistician, and one data analyst monitored safety issues during the study.

2.2. Screening
A screening programme was constructed and tested at three hospitals prior to the start of the study. A concept of retrospective screening of known patients and prospective screening of new patients was applied. The first approach used registers at echocardiography laboratories and corresponding units or registers of diagnoses for previously treated patients in order to identify individual patients. The second approach was to identify patients under on-going treatment in different units ranging from emergency rooms to outpatient units. The medical records (charts) were thereafter used to complete a screening form, which provided key data necessary for a preliminary judgement of eligibility.

The screening forms were faxed to the Wallenberg Laboratory for Cardiovascular Research at Sahlgrenska University Hospital in Gothenburg, Sweden where the responsibility for the screening programme rested. The data were fed into a computer, incorrectly completed forms were identified and questions on these were faxed to the investigator, who also received monthly updates of the screening outcome. A goal was that at least 60% of the included patients should be in NYHA class III or IV and screening data were used to give feed-back and to encourage the investigators. The screening procedure resulted in a number of identified patients, who might be eligible for participation in the MERIT-HF study.

Eleven out of the 14 countries in the MERIT-HF study participated in this screening programme and approval from ethics committees were obtained. In the three remaining countries there were other and varying screening programmes to identify eligible patients.

2.3. Endpoint classification
During the course of the study the investigators completed case record forms regarding clinical events and adverse events which were immediately faxed to the Co-ordinating Centre and received by a datafax system. This continuous reporting identified clinical events, which were classified by the Independent Endpoint Committee. Further clinical events were also detected by the Endpoint Committee members when they read the copies of the clinical documentation.

An Event Classification Manual defined the clinical events which were classified by the Independent Endpoint Committee, using all available information such as photocopies of hospital records, death certificates, or postmortem reports blinded as regards patient identification and treatment assignment. The definitions of the clinical events are shown in Table 1. Cardiovascular death was defined as all deaths where a non-cardiovascular cause was not identified, and in case of competing causes, judged to be of equal importance, the cardiovascular cause should take preference. A hospitalisation was classified as being due to a cardiovascular cause if a non-cardiovascular cause was not identified. Each clinical event was classified independently by two members of the Endpoint Committee, and each of them completed a classification data form. If the two members disagreed, the case was discussed, and if necessary, a third member made his judgement in order to obtain consensus between two members of the Endpoint Committee, which constituted the final classification.


View this table:
[in this window]
[in a new window]

  		Table 1 Definition of endpointsa

 
2.4. Principles for safety monitoring
The Independent Safety Committee formally monitored the mortality in the trial at approximately 25, 50, and 75% of the total number of expected deaths. An adjustment to the overall significance level was made for testing both the primary endpoint, all-cause mortality, and the second interrelated primary endpoint, the combined endpoint of all-cause mortality and all-cause hospitalisations (time-to-first-event). The combined endpoint was tested at the 1% level (two-sided P-value of 0.010), and the primary endpoint of all-cause mortality was tested at the cumulative 4% level (two-sided P-value of 0.0415, including interim analyses). This means a total value (significance level) of less than 0.0500 (two-sided) for the two primary endpoints together, assuming that hospitalisations and mortality were three times more common than mortality alone.

This study used asymmetric group sequential procedures to monitor the primary outcomes. A Peto-type boundary [10,11] was used for monitoring a positive trend (where the beta-blocker was better than placebo). This approach required a large critical value for the test statistics to be declared significant at all interim analyses before the end of the trial (see Table 2). Thus, any adjustment for repeated testing at the scheduled end of the trial was minor, and the conventional critical value for the test statistics (in this case: Z=2.05, {alpha}=0.02, one-sided test) could be used at the end of the trial.


View this table:
[in this window]
[in a new window]

  		Table 2 Monitoring bounds for a positive trend (benefit)

 
The stopping rule for efficacy was based on total number of expected deaths, analysed on an intention to treat principle which requires all randomised patients and all primary outcomes to be included. In the case of a positive trend, the probability of premature stopping for benefit at one the three predefined points in time before the end of the trial would be 0.0036 (cumulative) (see Table 2). These monitoring bounds refer only to the mortality outcome. The secondary primary endpoint, i.e. the combination of total mortality and all hospitalisations (time to first event), was not formally monitored with interim analyses during the course of the trial.

In the event that a negative mortality trend emerged during MERIT-HF, the Independent Safety Committee charter suggested that the trial should proceed until a definitive result had been obtained. Although a negative trend may be sufficient to rule out any possible positive benefit, the Independent Safety Committee should continue the trial until a harmful effect could be distinguished from neutrality. The rationale for this was that not being able to distinguish between a possible harmful mortality effect and a non-harmful effect was important in this patient population since metoprolol CR/XL may be used for other beneficial effects than just mortality.

If the negative trend would be substantial, but still not statistically significant at the interim analyses for either 50 or 75% of the expected number of deaths or information time, the Independent Safety Committee might need to increase the frequency of their interim analyses. However, the trial should not continue beyond the point where harm had been established. In order to implement this, the lower monitoring boundary to evaluate the statistical test comparing mortality for statistical significance would be implemented using the alpha spending function method of Lan and DeMets [12]. In particular, the uniform spending function, at t (t=information time), would be selected where t (0≤t≤1) is the proportion of total expected deaths observed at the interim analysis. The lower alpha level was set at 0.02 (i.e. 0.04/2). This alpha spending function approach does not require the number of interim analyses or the exact time of analyses to be pre-specified. However, the exact spending function must be specified in advance and cannot be changed during the trial. The prespecified overall alpha level would be protected, regardless of the number and timing of the interim analyses for harm (see Table 3). This increase in frequency is in contrast to positive trends for mortality benefit where primary interim analyses for benefit would be carried out only at the prespecified times (25%, 50%, 75%).


View this table:
[in this window]
[in a new window]

  		Table 3 Monitoring bounds for a negative trend (harm)

 

   3. Results
Top
 Notes
 Abstract
 1. Introduction
 2. Methods
 3. Results
4. Discussion
 Appendix A
 References
 
3.1. Screening
Two hundred and thirty-one sites in 11 countries screened 8912 patients and randomised 2852 patients, whereas 1139 patients were randomised from 82 sites not participating in the specially designed screening programme. A comparison between the screened population and the patients who were randomised into the study showed that the former were more often women, were older, had marginally higher ejection fraction and differed in NYHA functional class (Table 4).


View this table:
[in this window]
[in a new window]

  		Table 4 Characteristics of the patients who were screened and those who were randomised to the study in the 11 countries participating in the screening programme

 
The reasons for not being eligible for participation in the MERIT-HF study are shown in Table 5. The preliminary estimation was that 75% (n=6662) of the screened patients might be possible candidates for the study.


View this table:
[in this window]
[in a new window]

  		Table 5 Causes of non-eligibility among the screened patients (n=8894). One patient may have several causes

 
As demonstrated in Fig. 1 there was a positive association between the number of screened and randomised patients when all sites (n=255) in the 11 countries were examined (r=0.73, P<0.0001). Due to the fact that the inclusion rate was higher than planned, the pre-specified recruitment period, lasting for 14 months could be used to increase the number of included patients from 3200 to 3991 leading to an increase in the power of the study.


Figure 1
View larger version (12K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 1 Scattergram of the number of screened (n=8894) and randomised patients (n=3018) in 255 sites in the eleven countries participating in the screening programme.

 
A comparison of entry characteristics of the patients divided by utilisation of the screening programme showed that the 61.9% of patients from the screening sites were in NYHA class III or IV to be compared to 51.8% in the patients from non-screening sites [mean difference 10.1% (95% CI 6.7–13.5%)]. Correspondingly, patients from the screening sites were more often women [mean difference 4.2% (95% CI 1.4–7.0%)] whereas the ejection fraction was marginally higher [mean difference 0.010 (95% CI 0.005–0.015)] compared with non-screening sites.

3.2. Endpoint classification
The Independent Endpoint Committee classified 768 hospitalisations as due to worsening heart failure. Forty-six of these events were not ticked as hospitalisations due to worsening heart failure by the investigators in the case report forms. Conversely, the investigators but not the Independent Endpoint Committee, indicated hospitalisation due worsening heart failure on 161 occasions.

3.3. Safety monitoring
The Independent Safety Committee had monthly teleconferences, and performed the first interim analysis on 1 April 1998, and the second interim analysis on 21 September 1998 when 25 and 50%, respectively, of all expected deaths had occurred. At the first interim analysis (25%) the comparisons between the mortality rates in the metoprolol CR/XL and placebo groups yielded a Z-value of 2.550 for the log rank test comparing the two survival curves (Fig. 2). The MERIT-HF trial was stopped after the second interim analysis (50% point) of total mortality. The analysis was based on 296 deaths reported until 15 September 1998: 115 deaths in the metoprolol CR/XL group and 181 deaths in the placebo group giving a risk ratio of 0.637 and a log rank Z-value of 3.807 (well over the prespecified Z-value of 2.97), and a P-value=0.00014 (unadjusted). The Independent Safety Committee met with the International Executive Committee on 21 September 1998 and recommended early closure of the study. The Executive Committee met with the International Steering Committee on 2 October when it was decided to stop the study by 31 October 1998. For safety and ethical reasons it was not possible to close the study earlier.


Figure 2
View larger version (10K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 2 Monitoring bounds for total mortality in the MERIT-HF study. The triangle shows the outcome of the first, and the square the second interim analyses which were made by the Independent Safety Committee.

 

   4. Discussion
Top
 Notes
 Abstract
 1. Introduction
 2. Methods
 3. Results
 4. Discussion
Appendix A
 References
 
The recruitment of patients to the MERIT-HF study was very successful in terms of number of included patients and high event rates among randomised patients in the placebo group. Thus, 25% more patients than was originally planned could be randomised during the pre-specified 14-month inclusion period, and in the placebo group the observed annual mortality rate exceeded the rate that was expected (11.0 vs. 9.4%, respectively). In general terms, a large number of included patients at high-risk increases the likelihood of obtaining conclusive results in a survival study within a limited period of time. One explanation for the successful outcome in the present study seems to be the screening programme. Investigators who actively used this programme not only randomised more patients but also included 10% more patients in NYHA class III–IV in comparison with those not using this programme (61.9 vs. 51.8%).

Patients enrolled in a trial should be representative of patients treated in clinical practice. In the community setting the typical heart failure patient is older than 70 years, the sex distribution is even, diastolic left ventricular dysfunction is common and co-morbidity is frequent [13]. The large intervention trials in chronic heart failure have so far been focused on patients with left ventricular systolic dysfunction, often in groups with a mean age below 70 years, and with a predominance of men [14–17]. The results from the screening in the present study seem to reflect the selection of patients from the general heart failure population. Thus, the screened population tended to be somewhat older, to contain more women, and more patients with severe heart failure, and also to have a higher mean ejection fraction than 0.40 [13]. A state of health that did not allow adherence to the protocol, indicating co-morbidity, and an ejection fraction above 0.40 were common causes of non-eligibility for the screened patients.

The quality of endpoint classifications in a clinical study is based on the availability of relevant data and on well-defined criteria. This is a small problem in studies with robust response variables such as total mortality or all cause hospitalisations which were primary endpoints in the MERIT-HF study. However, for more complicated endpoints such as the clinical patient condition or results from examinations, the classification of endpoints is more problematic. In studies where the endpoints are diagnoses such as myocardial infarction or results of examinations such as blood flow after angiographic thrombolysis, the classifications made by endpoint committees have dramatic effects on the statistical significance of treatment in comparison to data given by the investigators [4,5]. In the SAVE trial the Classification Committee modified the initial myocardial infarction rate by using strict criteria [18]. It is likely than an investigator often may have to base the decision to report a clinical event mainly on information provided by the patient. A central endpoint committee may modify such information from investigators by getting access to more data than was available to the investigator at the completion of the case report form, and by using strict predefined criteria. As an example of this discrepancy we examined the classification of one important clinical endpoint, i.e. hospitalisation due to worsening heart failure. Although the case report form data and the Independent Endpoint Committee classification were identical in 722 cases, differences were found in 207 cases. Thus, a centralised event classification process guided by strict criteria can have a substantial impact on important clinical endpoints [4,5,18].

A well-defined classification of mortality in chronic heart failure may help to identify pathways leading to death and hence suggest therapies to improve survival [3]. A review of 27 heart failure studies disclosed that definitions of sudden death were heterogeneous and that the majority of studies failed to describe how circulatory failure was defined [3]. In the MERIT-HF study a proposed classification of mortality was used resulting in a careful description of modes of death and of the intervention effect of metoprolol CR/XL treatment [8].

The MERIT-HF was stopped early due to beneficial effects according to predefined rules and reported conclusive results on all the major endpoints [8,9]. Whereas statistical considerations have been discussed above, interpretation of interim data is very complex and requires both clinical and statistical experts reviewing the data in concert.

Interim data safety reports pose well recognised statistical problems related to the multiplicity of statistical tests to be conducted on the accumulating set of data. The basic problem is well known and is referred to as ‘sampling to a foregone conclusion’ (Cornfield) or the problem of repeated significance tests [19–21]. Armitage et al. have shown that with successive repeated significance tests at the usual 0.05 significance level, the likelihood of a Type I (false positive) error increases from {alpha}=0.05 for the first test to {alpha}=0.14 after five tests, 0.19 after 10 tests, 0.32 after 50 tests, and eventually to {alpha}=1.0 [20].

An obvious solution to the problem of repeated tests would be to require a smaller P-value for statistical significance at each successive test so that the realised overall Type I error level would still be within the desired limits. For example, if up to five data analyses were conducted for a single primary question, based on the work of Armitage et al. [20] and Pocock [22,23], each of these five reports should employ an alpha level of 0.016 for significance (critical Z value of 2.413 rather than 1.96) in order to ensure an overall (two-sided) cumulative Type I error level of {alpha}=0.05. Procedures such as this one are referred to as ‘group sequential’ procedures. There are many other such procedures which also control the Type I error or false positive rate, and such a procedure was used in the MERIT-HF study.

There are examples from the literature of improperly terminated trials in which the principles for interim analyses were violated or data were analysed incorrectly [6], demonstrating the absolute need of strict predefined written rules and guidelines for safety monitoring and the expertise of an experienced, multidisciplinary safety committee.

In conclusion, a screening programme facilitated patient recruitment and selected a higher risk population, the event classification was improved by using an independent endpoint committee, and the safety monitoring allowed an early closure of the study due to mortality benefit, without jeopardising the final analyses of the results.


   Appendix A
Top
 Notes
 Abstract
 1. Introduction
 2. Methods
 3. Results
 4. Discussion
 Appendix A
References
 
Writing Group: David DeMets, Björn Fagerberg, Hans Wedel, John Wikstrand.

International Steering Committee: Åke Hjalmarson MD, PhD, (co-chairman), Sidney Goldstein MD (Co-chairman), Björn Fagerberg MD, PhD (Secretary), Hans Wedel PhD (Biostatistician), Finn Waagstein MD, PhD, John Kjekshus MD, PhD, John Wikstrand MD, PhD, Dia El Allaf MD, Jirí Vítovec MD, PhD, Jan Aldershvile MD, PhD, Matti Halinen MD, PhD, Rainer Dietz MD, Karl-Ludwig Neuhaus MD, András Jánosi MD, DSc, Gudmundur Thorgeirsson MD, PhD, Peter HJM Dunselman MD, PhD, Lars Gullestad MD, Jerzy Kuch MD, Johan Herlitz MD, PhD, Peter Rickenbacher MD, Stephen Ball MD, PhD, Stephen Gottlieb, MD, Prakash Deedwania, MD.

Author affiliations: Department of Cardiology, Sahlgrenska University hospital, Göteborg, Sweden (Drs Hjalmarsson, Waagstein, Herlitz). Henry Ford Hospital, Care Western Reserve University, Michigan, Detroit (Dr Goldstein). Wallenberg Laboratory for Cardiovascular Research, Sahlgrenska University hospital, Göteborg, Sweden (Drs Fagerberg, Wikstrand, the latter also non-voting, senior medical advisor, AstraZeneca). Nordic School of Public Health, Göteborg, Sweden (H. Wedel). Section of Cardiology, Rikshospitalet, Oslo, Norway (Drs Kjekshus, Gullestad). Service de Cardiologie et de réanimation cardique, Centre Hospitalier Hutois, Huy, Belgium (Dr El Allaf). 2 interní klinika, Fakultní nemocnice, Brno, Czech Republic (Dr Vítovec). Department of Medicine B, Rigshospitalet, Copenhagen, Denmark (Dr Aldershvile). Accident and Emergency Department, Kuopio University Hospital, Kuopio, Finland (Dr Halinen). Univ.-Klinikum Rudolf Virchow FU, Berlin, Germany (Dr Dietz), Städt. Kliniken Kassel/Akadem. Lehrkrankenhaus, Inntere Med. II, Kassel, Germany (Dr Neuhaus). Department of Medicine, Lanspítalinn, University Hospital, Reykavik, Iceland (Dr Thorgeirsson). Department of Cardiology, St. Ignatius Ziekenhaus, Breda, The Netherlands (Dr Dunselman). Department of Cardiology, Warzaw School of Medicine, Warsaw, Poland (Dr Kuch). Kardiologie, Kantonsspital Bruderholz, Switzerland (Dr Rickenbacher). Institute for Cardiovascular Research, Leeds General Infirmary, UK (Dr Ball). VA Medical Center, California, Fresno (Dr Deedwania). Division of Cardiology, University of Maryland, Baltimore (Dr Gottlieb).

Independent Safety Committee: Desmond G Julian (UK), David L DeMets (USA), Kanu Chatterjee (USA), Jan Feyzi (USA).

Independent Endpoint Committee: Seppo Lehto (Finland), Pal Kárpáti (Hungary), Wolfgang Motz (Germany), Ola Samuelsson (Sweden), Jan Willem Viersma (Netherlands).

The MERIT HF Study Group: The names of the investigators have been published previously (Lancet 1999;353:2001–2007.)

Funding: The study was supported by grants from AstraZeneca.


   Notes
Top
 Notes
Abstract
 1. Introduction
 2. Methods
 3. Results
 4. Discussion
 Appendix A
 References
 
* E-mail address: bjorn.fagerberg{at}medfak.gu.se (B. Fagerberg). Back

1 See Appendix A. Back


   References
Top
 Notes
 Abstract
 1. Introduction
 2. Methods
 3. Results
 4. Discussion
 Appendix A
 References
 

  1. Fleming T.R., DeMets D.L. Monitoring of clinical trials: issues and recommendations. Contr Clin Trials (1992) 14:183–197.[CrossRef]
  2. Bjørn M., Brrendstrup C., Karlsen S., Carlsen J.E. Consecutive screening and enrollment in clinical studies: the way to representative patient samples? J Cardiol Fail (1998) 4:225–230.[CrossRef]
  3. Narang R., Cleland J.G.F., Erhardt L., et al. Mode of death in chronic heart failure. A request and proposition for more accurate classification. Eur Heart J (1996) 17:1390–1403.[Abstract/Free Full Text]
  4. Näslund U., Grip L., Fischer-Hansen J., Gundersen T., Lehto S., Wallentin L. The impact of an end-point committee in a large multicentre, randomized, placebo-controlled clinical trial. Eur Heart J (1999) 20:771–777.[Abstract/Free Full Text]
  5. Mahaffey K.W., Tardiff B.E., Granger K.L., Lee R.A., Harrington E.J., Topol R.M. Disagreement between site investigators and clinical event committees is common and can affect trial results. J Am Coll Cardiol (1998) 31(Suppl_A):184A.
  6. Task Force of the Working Group on Arrhythmias of the European Society of Cardiology. The early termination of clinical trials: causes, consequences, and control. Circulation (1994) 89:2892–2907.[Abstract/Free Full Text]
  7. The International Steering Committee on behalf of the MERIT-HF Study Group. Rationale, design, and organisation of the metoprolol CR/XL randomized trial in heart failure (MERIT-HF). Am J Cardiol (1997) 80(9B):54J–58J.[CrossRef][Medline]
  8. MERIT-HF Study Group. Effect of metoprolol CR/XL in chronic heart failure: Metoprolol CR/XL Randomized Intervention Trial in Congestive Heart Failure (MERIT-HF). Lancet (1999) 353:2001–2007.[CrossRef][Web of Science][Medline]
  9. Hjalmarson Å., Goldstein S., Fagerberg B., et al. Effects of controlled-release metoprolol on total mortality, hospitalizations, and well-being in patients with heart failure: the Metoprolol CR/XL Randomized Intervention Trial in congestive heart failure (MERIT-HF). MERIT-HF Study Group. J Am Med Assoc (2000) 283:1295–1302.[Abstract/Free Full Text]
  10. Haybittle J.L. Repeated assessments of results in clinical trials of cancer treatment. Br J Radiol (1971) 44:793–797.[Abstract/Free Full Text]
  11. Peto R., Pike M.C., Armitage P., et al. Design and analysis of randomized clinical trials requiring prolonged observations of each patient. I. Introduction and design. Br J Cancer (1976) 34:585–612.[Web of Science][Medline]
  12. Lan K.K.G., DeMets D.L. Discrete sequential boundaries for clinical trials. Biometrika (1983) 70:649–653.
  13. Sharpe N.M., Doughty R. Epidemiology of heart failure and ventricular dysfunction. Lancet (1998) 352(Suppl 1):3–7.[CrossRef][Web of Science][Medline]
  14. The CONSENSUS trial study group. Effects of enalapril on mortality in severe congestive heart failure. N Engl J Med (1987) 316:1429–1435.[Abstract]
  15. The SOLVD investigators.
  16. CIBIS II investigators and committees. The cardiac insufficiency bisoprolol study II (CIBIS-II): a randomised trial. Lancet (1999) 353:9–13.[CrossRef][Web of Science][Medline]
  17. Packer M., Bristow M.R., Cohn J.N., et al. The effect of carvedilol on morbidity and mortality in patients with chronic heart failure. N Engl J Med (1996) 334:1349–1355.[Abstract/Free Full Text]
  18. Pfeffer M.A., Braunwald E., Moyé L.A., et al. Effect of captopril on mortality and morbidity in patients with left ventricular dysfunction after myocardial infarction. Results of the survival and ventricular enlargement trial. N Engl J Med (1992) 327:669–677.[Abstract]
  19. Cornfield J. Sequential trials, sequential analysis, and the likelihood principal. Am Stat (1966) 20:18–23.
  20. Armitage P., McPherson C.K., Rowe B.C. Repeated significance tests on accumulating data. J R Stat Soc Ser A (1969) 132:235–244.[CrossRef]
  21. McPherson K. Statistics: the problem of examining accumulating data more than once. N Engl J Med (1975) 290:501–502.[Web of Science]
  22. Pocock S.J. Group sequential methods in the design and analysis of clinical trials. Biometrika (1977) 64:191–199.[Abstract/Free Full Text]
  23. Pocock S.J. Interim analyses for randomized clinical trials: the group sequential approach. Biometrics (1982) 38:153–162.[CrossRef][Web of Science][Medline]

Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 NEJMHome page
L. Sjostrom, K. Narbro, C. D. Sjostrom, K. Karason, B. Larsson, H. Wedel, T. Lystig, M. Sullivan, C. Bouchard, B. Carlsson, et al.
Effects of Bariatric Surgery on Mortality in Swedish Obese Subjects
N. Engl. J. Med., August 23, 2007; 357(8): 741 - 752.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (3)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Fagerberg, B.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Fagerberg, B.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?