Update to the study protocol, including statistical analysis plan for the CopenHeartIE trial. Trine Bernholdt Rasmussen, 1,2 Ann-Dorthe Zwisler, 3 Signe Stelling Risom, 1 Kirstine Lærum Sibilitz, 1 Henning Bundgaard, 1 Christian Gluud, 4 Philip Moons, 5, 6Lau Caspar Thygesen, 7Jane Lindschou Hansen, 4 Tone M. Norekvål, 8Selina Kikkenborg Berg, 1 the CopenHeart Group.9
1
Introduction
This update relates to the CopenHeartIE trial study protocol, a randomized clinical trial comparing comprehensive cardiac rehabilitation after infective endocarditis, including physical exercise and psycho-education, with usual care. This update should be read in conjunction with the original protocol publication1.
Our initial recruitment target was 150 patients with the recruitment period estimated to two and a half years. Realizing that the inclusion rate was lower than expected we began inclusion from two additional sites (Roskilde and Herlev hospital), offered a fee for every included patient to the recruiting sites, and arranged for imbursement of transportation expenses for patients from region Zealand. Furthermore, we extended the inclusion period twice in the hope of reaching full sample size. At the end of almost five years in October 2016, we were only able to include 117 patients. Due to the following considerations, we chose to stop inclusion for the CopenHeartIE trial at that time:
- Inclusion had lasted almost 5 years, which was twice as long as first estimated. Over the years, we made different attempts/adjustments to increase inclusion with minimal effect. Reaching full sample size would have necessitated 1-1½ year’s further inclusion.
- All other CopenHeart trials were finalized. The organizational set-up including staff for exercise training, consultations and ergo-spirometry testing, was quite extensive considering the low number of patients included (on average one per month). It would have been difficult to justify the resources required for a further 1½ year.
- The treatment strategy had changed since the trial began. Many patients now have shorter hospitalizations as they receive antibiotics as partial oral treatment, via an IV pump or IV infusion at home administered by a home care nurse.
- The rehabilitation strategy has changed since the trial began. For patients having undergone heart valve surgery rehabilitation is now part of the ‘pakkeforløb’. It was difficult including participants insisting they could not participate in an exercise training programme if allocated to the control group. This meant that we would have a number of non-compliant participants in the control group, which will negatively affect data.
- By finalizing the overall CopenHeart project now, staff and resources are available to ensure all data will be compiled and assessed and results reported.
Updated sample size and power
The power calculation will be repeated, based on this revised sample size of 117 participants, once the baseline mean scores (standard deviations) have been assessed. The primary outcome is self-rated mental health measured on the continuous variable mental health component scale (MCS) on the Short Form-36 questionnaire. Based on the assumptions defined in the protocol1 of a minimally important difference of six points and a standard deviation of thirteen points2, a sample size of 150 was needed to provide 80% power to be able to reject the null hypothesis, with a type I error of 5%. The secondary outcome is the continuous variable peak oxygen uptake (VO2 peak) measured by cardiopulmonary exercise testing. Based on the assumed minimally important difference 3 mL/kg/min and standard deviation of 6.9 ml/kg/min3,4, we would have been able to reject the null hypothesis with a probability of 75.4% and a type I error probability of 5% with a sample size of 150.
Statistical analysis plan for the primary and secondary outcomes
The primary analysis essentially remains the same as defined in the protocol, but as the trial was stopped before the sample size was reached, we will also conduct sequential analysis with a more restrictive alpha to assess the results of significance testing taking sparse data and repetitive testing into consideration5. We will use the trial sequential analysis programme for this purpose6-9. We also propose slight revisions to the analysis model and handling of missing data, and also provide further specification of the handling of multiplicity of outcome testing.
Primary analysis
The level of significance is set at 5%. The intention-to-treat analysis using a mixed model with repeated measures (MMRM) for continuous outcome measures will be adjusted by the trial stratification variable ‘treating heart center, Rigshospitalet/Other’1.
Missing values and sensitivity analyses
As stated in the protocol, using MMRM ensures that missing data values will not create bias, as long as the values are missing at random10. For the primary and secondary outcomes, if a statistically significant result is obtained (P ≤0.05), we will undertake sensitivity analyses using multiple imputation11,12 and also undertake imputation that assumes not missing at random including ‘worst-case’ and ‘best-case’ scenarios.
Multiplicity
The primary and secondary outcomes (MCS and VO2 peak) will be analyzed as stated above. The various exploratory outcome measures pre-defined in the protocol1 will be analyzed with no adjustment of P-values due to multiplicity. Instead, the interpretation of each exploratory outcome measure will be assessed in the light of multiple testing; that is, statistically significant effects will be interpreted in the context of increased risk of type I error.
Per-protocol analysis
In addition to the primary analysis, we will also undertake a per-protocol analysis to account for the variable adherence to the prescribed rehabilitation program by intervention group participants13. We expect that the effect of the intervention will depend on the participants’ adherence to the study protocol14.
The per-protocol definition is based on the intervention consisting of two elements: physical exercise and a psychoeducational intervention. Adherence to both elements of the intervention will be described according to compliance data for self-reported training diaries, data from pulse watches, and records made in relation to the nurse consultations.
Physical exercise
The physical exercise comprises three weekly sessions conducted between one and four months after surgery, a total of 36 sessions. Participation in the intervention will be defined as participation in an individual consultation to plan exercise training, receiving instruction in the use of the training diary and pulse watch, and participation in 75% (≥27) of the exercise sessions.
Psycho-educational intervention
The psycho-educational intervention is based on the theories of RR Parse 15 and comprises one monthly consultation conducted within the first six months after surgery, a total of five consultations. Participation in the intervention will be defined as attending at least 80% (four out of five) of the psycho-educational consultations.
Author affiliations
1 The Heart Centre, Copenhagen University Hospital, Rigshospitalet, 2 Department of Cardiology, Herlev and Gentofte University Hospital, 3 Danish Centre for Rehabilitation and Palliative Care, Odense University Hospital,4 Copenhagen Trial Unit, Centre for Clinical Intervention Research, Copenhagen University Hospital, Rigshospitalet, 5 KU Leuven Department of Public Health and Primary Care, KU Leuven - University of Leuven, 6 Institute of Health and Care Science, University of Gothenburg, 7 National Institute of Public Health, University of Southern Denmark, 8 Department of Heart Disease, Haukeland University Hospital, Bergen and9 www. Copenheart.org.
References
1. Rasmussen TB, Zwisler AD, Sibilitz KL, et al. A randomised clinical trial of comprehensive cardiac rehabilitation versus usual care for patients treated for infective endocarditis--the CopenHeartIE trial protocol. BMJ Open. 2012;2(6):10.1136/bmjopen-2012-001929. Print 2012.
2. Perrotta S, Aljassim O, Jeppsson A, Bech-Hanssen O, Svensson G. Survival and quality of life after aortic root replacement with homografts in acute endocarditis. Ann Thorac Surg. 2010;90(6):1862-1867.
3. Landry F, Habel C, Desaulniers D, Dagenais GR, Moisan A, Cote L. Vigorous physical training after aortic valve replacement: Analysis of 10 patients. Am J Cardiol. 1984;53(4):562-566.
4. Lim HY, Lee CW, Park SW, et al. Effects of percutaneous balloon mitral valvuloplasty and exercise training on the kinetics of recovery oxygen consumption after exercise in patients with mitral stenosis. Eur Heart J. 1998;19(12):1865-1871.
5. Jakobsen JC, Gluud C, Winkel P, Lange T, Wetterslev J. The thresholds for statistical and clinical significance - a five-step procedure for evaluation of intervention effects in randomised clinical trials. BMC Med Res Methodol. 2014;14:34-2288-14-34.
6. Brok J, Thorlund K, Gluud C, Wetterslev J. Trial sequential analysis reveals insufficient information size and potentially false positive results in many meta-analyses. J Clin Epidemiol. 2008;61(8):763-769.
7. DeMets DL, Lan KK. Interim analysis: The alpha spending function approach. Stat Med. 1994;13(13-14):1341-52; discussion 1353-6.
8. Thorlund,K, Engstrøm J, Wetterslev J, Brok J, Imberger G, Gluud C. User manual for trial sequential analysis (TSA). . 2011.
9. Wetterslev J, Thorlund K, Brok J, Gluud C. Trial sequential analysis may establish when firm evidence is reached in cumulative meta-analysis. Journal of clinical epidemiology. 2008;61(1):64-75.
10. Verbeke G, Molenberghs G. Linear mixed models for longitudinal data. Springer; 2000.
11. Sterne JA, White IR, Carlin JB, et al. Multiple imputation for missing data in epidemiological and clinical research: Potential and pitfalls. BMJ. 2009;338:b2393.
12. Dmitrienko A, Tamhane A, Bretz F. Multiple testing problems in pharmaceutical statistics. Chapman & Hall/CRC biostatistics series; 2010.
13. Sedgwick P. What is per protocol analysis?. BMJ. 2013;346:3748.
14. Keteyian SJ, Leifer ES, Houston-Miller N, et al. Relation between volume of exercise and clinical outcomes in patients with heart failure. J Am Coll Cardiol. 2012;60(19):1899-1905.
15. Parse RR. The human becoming school of thought: A perspective for nurses and other health professionals. Thousand Oaks, Calif.: Sage; 1998:130 sider.
1