Statistical analysis plan for aggressive hydraTion in patients with ST-elevation myocardial infarction undergoing primary percutaneous coronary intervention to prevenT contrast-induced nephropathy (ATTEMPT) study.

BACKGROUND: The ATTEMPT study is a multicenter, randomized controlled trial which is investigator-based and open label in nature. For the study, 560 patients with ST-segment elevation myocardial infarction (STEMI) underwent primary percutaneous coronary intervention (pPCI) have been randomized (1:1) for treatment with periprocedural aggressive hydration (treatment group) or general hydration (control group). To improve the quality of the study's analysis and to minimize analysis bias based on the study's findings.
METHODS: The design of the statistical analysis plan (SAP) was created by chief investigators and statisticians and received permission from the aggressive hydraTion in patients with ST-Elevation Myocardial infarction undergoing Primary percutaneous coronary intervention to prevenT contrast-induced nephropathy (ATTEMPT) management committee. Treatment allocation and research data were reviewed by the Data Safety and Monitoring Committee and researchers were kept blind. We produced data shells based on a pre-existing published protocol and produced detailed descriptions of statistical analyses. This study includes primary, secondary and safety endpoints. Relevant statistical comparisons were planned and discussed in a transparent manner. They are publicly available, verifiable and were determined prior to the data collection process being completed.
RESULTS: We developed a SAP for the ATTEMPT study and an outline and list of mock tables were also created. We produced descriptions of analyses of baseline characteristics, patient care approaches, efficacy measures, and outcomes. This study defined five previously specified subgroups and compared the statistics of groups within these subgroups.
CONCLUSIONS: This SAP has been developed for the ATTEMPT study and has high-quality standards of internal validity to minimize analysis bias. TRIAL REGISTRATION: ClinicalTrials.gov number, NCT02067195. 2020 Annals of Translational Medicine. All rights reserved.

Introduction

Patients who have ST-segment elevation myocardial infarction (STEMI) carry a high-risk of contrast-induced acute kidney injury (CI-AKI) following primary percutaneous coronary intervention (pPCI) (1). However, for this high-risk group, the extent of the effects of optimal hydration strategy is yet to be fully established (2,3). The aggressive hydraTion in patients with ST-Elevation Myocardial infarction undergoing Primary percutaneous coronary intervention to prevenT contrast-induced nephropathy (ATTEMPT) study is the first study to evaluate the peri-procedural aggressive hydration in STEMI patients undergoing pPCI (4). Here, we describe the statistical analysis plan (SAP) for the ATTEMPT study, prior to patient enrollment being completed (this was completed in June 2018) and the locking of the database for analysis.

This SAP was written by the study statistician and the principal investigator, both of whom were kept blind to the course of treatment allocated to the patients. We prospectively defined each analysis described in this SAP.

Methods

Overview of design

The ATTEMPT study is a multicenter, randomized controlled trial which is investigator-based and open label in nature. A total of 560 patients with STEMI undergoing pPCI will be randomized (1:1) to undergo treatment either by peri-procedural aggressive hydration (treatment group) or general hydration (control group). The ATTEMPT study could prove valuable for the possible identification of optimal hydration regimens for STEMI patients undergoing pPCI (4).

Inclusion criteria

All consecutive patients with STEMI who were at least 18 years of age and were prepared for pPCI were taken into consideration for enrollment in the study.

❖ Contrast medium administration within the 2 weeks prior to the procedure or the following 3 days;

❖ End-stage renal failure or renal transplantation, and refusal of pPCI or death while the procedure is taking place;

❖ Heart failure-induced cardiac shock or New York Heart Association class IV (these patients were excluded because intravenous hydration administration could potentially prove harmful);

❖ A recent acute kidney injury, which was considered to be an absolute increase of 0.5 mg/dL in serum creatinine (SCr) over baseline in the 24 hours prior;

❖ The existence of lactation, pregnancy;

❖ A tumor of a malignant nature or a predicted life expectancy of 1 year or less;

❖ An allergic reaction induced by contrast medium, peri-procedural receipt of metformin, or non-steroidal anti-inflammatory drugs in the 2 days prior and while the study is taking place;

❖ Planned renal catheterization or heart valvular surgery.

Randomization

All patients meeting the eligibility criteria and provided informed consent will be randomized into two study groups. A computer will carry out randomization of eligible patients by producing numbers at random at a 1:1 ratio. Randomization will be accomplished with using a block random method with 8 units in each group. Some offset or variability will be inserted to prevent anticipation of the next treatment. Randomization will be stratified based on age (<60, 60–75, >75 years), sex (male or female), and STEMI location (anterior wall or no-anterior wall). Study coordinators at each site will be responsible for obtaining a randomized treatment assignment for each eligible patient. Study sites will be provided with a web-based randomization program for this purpose. This web-based http://crdms.echobelt.org program will be tested at each site prior to the start of the trial and will be reviewed. Since this is an open-label study, the procedure of blinding will not take place.

Sample size

The total sample size of 560 patients was calculated based on our previous findings. The incidence of primary end point was estimated to decrease to 11.5% (50% relative reduction) in the aggressive hydration group from 23% in the control group with general hydration (5). We established a sample size according to nQuery + nTerim 3.0 (Statistical Solutions Ltd, Ireland) by employing a 2-sided χ2 test, a power of 90%, a significance level of 0.05, and a dropout rate <20%.

The formal statement of the null hypothesis

We fully describe the CI-AKI outcome based on the aggressive hydration versus general hydration intervention. The null hypothesis is that the two treatment groups (aggressive hydration and control) do not differ in terms of the proportion of subjects who experience CI-AKI. The alternative hypothesis is that the absolute difference in the incidence of CI-AKI between the aggressive hydration group and the control group.

Intervention

A pre-procedural loading dose 250 mL of normal saline for 30 minutes (125 mL for patients with congestive heart failure, Killip II/III or NYHA III) will be administered to the patients in the treatment group in an emergency department or cardiac catheterization lab over 30-minutes prior to the pPCI. After this, the patients will receive intravenous hydration at a general rate (1 or 0.5 mL/kg/h for patients with congestive heart failure, Killip II/III or NYHA III) until LVEDP measurement.

Patients will then undergo 4 hours of post-procedural aggressive hydration guided by LVEDP [5 mL/kg/h (LVEDP <13 mmHg), 3 mL/kg/h (LVEDP 13–18 mmHg), 1.5mL/kg/h (LVEDP >18 mmHg), and 0.5 mL/kg/h (LVEDP >20 mmHg)] and continuous intravascular hydration at the normal rate for the 24 hours following PCI. Control group patients will receive peri-procedural general hydration with ≤500 mL normal saline (within 6 hours) at a normal rate (0.5 or 1 mL/kg/h).

Interim analysis

According to the study design, interim analysis to determine if either intervention shows a substantial beneficial effect will not be carried out.

Timing of analysis

At the end of the study, the final audit should be performed usually within four to six weeks of the end of the last follow-up of the last subject in the study center. The preparations and procedures for the study of closed visits are generally the same as for a regular supervisory visit. The last patient was enrolled on 10th June 2018.

❖ Analysis is to be conducted based on adjusted intention-to-treat (ITT) (randomization must be finished before emergency surgery, and we will not include the patients who need to be excluded from the final analysis).

❖ We will not impute missing values, unless specified. The number of subjects included in an analysis will be reported if there is a substantial amount of missing data. The last observations will not be carried forward. Multiple imputation will be used if >5% of patients have missing data on the primary outcome.

❖ Each of the tests will be two-tailed, and a P value of less than 0.05 will be considered to be statistically significant.

❖ Only analyses conducted up to 1 year after randomization will feature in this analysis plan and in the primary manuscript.

❖ Pre-specified subgroup analyses will be conducted whether or not a statistically significant treatment effect on the primary outcome is seen across the total sample.

❖ We use the t-test for normally distributed continuous variables and expressed as mean ± SD, and the Wilcoxon rank-sum test was used in non-normal distribution variables and presented as median and interquartile range.

❖ For categorical variables, we used Pearson χ2 or Fisher’s exact tests to compare baseline characteristics and study’s endpoints between the aggressive and general hydration groups; these will be expressed as percentages.

❖ Logistic regression testing to assess whether the recorded treatment effect was consistent across random stratified variable (age, gender, STEMI location). Odds ratios will be reported alongside their related 95% CIs.

❖ Risk ratio (RR) and absolute risk difference (ARD) with their corresponding 95% confidence interval (CIs) are used to describe the interaction effect (primary, second, and safe events). The number needed to treat (NNT) for preventing one CI-AKI-related event was calculated by inverse of the ARD.

Analysis of primary outcome

The primary analysis will be based on adjusted intention to treat principles. Since the intervention is given and the primary outcome is observed for a very short duration (72 hours), we expect only a few dropouts or crossovers.

As ATTEMPT study’s protocol pointed out primary outcome would use multivariable logistic regression with (age, sex, creatinine clearance, and left ventricular ejection fraction). However, basic on principal component analysis, we have changed to use multivariable logistic regression to evaluate the intervention effect with random stratified variable [e.g., age (<60, 60–75, >75 years), gender, STEMI location]. The analysis for each variable will be performed by OR with their corresponding 95% CIs to describe the intervention effect. All tests will be two-tailed, and a P value less than 0.05 will be considered statistically significant.

Subgroup analyses

We will undertake analysis of five pre-specified subgroups defined by the following baseline criteria: age (<60, 60–75, >75 years), gender (Male or female), STEMI location (anterior wall or no-anterior wall), and LVEF (≥40% or <40%), eGFR (≥90 or <90 mL/minute/1.73 m2). eGFR formula (Modified relative dose response) = [186 × serum Cr (mg/dL)]^− 1.154 × age (yr) ^− 0.203 × (multiply by 0.742 for women).

Within each subgroup, summary measures will include raw counts and percentages within each treatment arm. The analysis for each subgroup will be performed by RR with their corresponding 95% CIs to describe the interaction effect. The results will be shown on a forest plot including the P-value for heterogeneity corresponding to the interaction term between the intervention and the subgroup variable.

Analysis of secondary outcomes

Our secondary objectives will be tested using χ2 tests, and the 95% CI of the rate difference of RR and ARD will be calculated using the method described by Altman et al (reported in Newcombe and recommended by the Food and Drug Administration and Clinical and Laboratory Standards Institute. The secondary end-points will be analyzed based on the adjusted ITT principle.

Sensitive analyses

To evaluate the stability of primary outcome, we will use logistic regression to analyze the influence of compliance and use baseline value comparison to assess the differences between the two groups of excluded cases. It will be performed by OR with their corresponding 95% CIs. All tests will be two-tailed, and a P value less than 0.05 will be considered statistically significant.

For patients: the volumes of urine and oral hydration (water in milliliters) in the 24 hours following the procedure will be recorded.

For medical staff: intravenous hydration information will be gathered in the 24 hours following the procedure. A preoperative renal function test and details on postoperative SCr, failure to hydrate according to protocol, and the absence of SCr will be obtained in the follow-up. All of this information, laboratory tests and vital signs will be carefully collected by the research staff.

Central effect analysis

To evaluate the central effect, we will use One-way analysis of primary outcome with different centers. We describe the central effect by OR with their corresponding 95% CIs. All tests will be two-tailed, and a P value less than 0.05 will be considered statistically significant.

Analysis of safety outcomes

One of the important adverse reactions to aggressive hydration is the increased risk of heart failure. We will record all information relating to acute heart failure: acute pulmonary edema, cardiogenic shock, and further auxiliary examination such as ECG, chest X-ray, laboratory assessment (with specific biomarkers), and echocardiography.

The Clinical Event Committee (CEC) will be responsible for determining the endpoints in clinical studies and avoiding deviations in event determination between centers in order to achieve a more accurate assessment of the test results. We will use RR and ARD with their corresponding 95% CIs to describe the safety endpoint.

Treatment of missing data

We conservatively estimate that up to 20% of subjects may be lost to follow-up SCr within 72 hours and exclusive patients. The site coordinators will make every effort to identify such subjects including at least two laboratory tests of SCr after pPCI. Multiple imputation will be used if >5% of patients have missing data on the primary outcome.

Statistical software

All data analyses will be performed using SAS version 9.4 (SAS Institute, Cary, NC, USA) and R soft-ware (version 3.6.1; R Core Team, Vienna, Austria).

Protocol deviation

Any program deviations need to be recorded, and the Guangdong Provincial Key Laboratory of Coronary Heart Disease will create a program deviation table to ensure that events are tracked correctly. The efficacy of the deviation basis will be determined. If sub-centers still experience serious program deviation after remedial training, the Data Monitoring Committee (DMC) may recommend the termination of the center in question. After consulting with the Executive Committee, subjects have the right to withdraw from this study, and the program will be reported to the main research unit Institutional Review Board as required.

Results

Flowchart of recruitment

The patient sorting processes throughout the different stages of the trial will be shown in the form of a Consolidated Standards of Reporting Trials diagram () (6-8).

Figure 1

CONSORT flowchart. CONSORT, Consolidated Standards of Reporting Trials.

The diagram will show the total numbers for eligible patients from those screened, patients included in the study, and explanations for those patients who were not included. Consent status will be outlined in a second diagram ().

Figure 2

Consent details.

Patient characteristics and baseline comparisons

Comparisons of patients’ baseline characteristics will be set out according to treatment group, as shown in the tables (see ).

Table S1

Proposed format of data tables and figures for main results publication

Characteristic	Aggressive group (N=xxx)	Control group (N=xxx)
Age, yr	xxx (xx)	xxx (xx)
Age >75 yr, No. (%)	xxx (xx)	xxx (xx)
Sex (male), No. (%)	xxx (xx)	xxx (xx)
Weight (kg)	xxx (xx)	xxx (xx)
Anterior myocardial infarction, No. (%)	xxx (xx)	xxx (xx)
Killip class >1, No. (%)	xxx (xx)	xxx (xx)
Creatine kinase MB, U/L	xxx (xx)	xxx (xx)
Serum creatinine, μmol/L	xxx (xx)	xxx (xx)
Estimate glomerular filtration rate, mL/min/1.73 m2	xxx (xx)	xxx (xx)
Estimate glomerular filtration rate <90 mL/min/1.73 m2, No. (%)	xxx (xx)	xxx (xx)
Cystatin C, mg/L	xxx (xx)	xxx (xx)
LVEF, %	xxx (xx)	xxx (xx)
LVEF <40%, No. (%)	xxx (xx)	xxx (xx)
Hypertension, No. (%)	xxx (xx)	xxx (xx)
Diabetes mellitus, No. (%)	xxx (xx)	xxx (xx)
ACEI/ARB, No. (%)	xxx (xx)	xxx (xx)
Beta-Blockers, No. (%)	xxx (xx)	xxx (xx)
Statin, No. (%)	xxx (xx)	xxx (xx)
Abciximab, No. (%)	xxx (xx)	xxx (xx)
Diuretic, No. (%)	xxx (xx)	xxx (xx)
Volume of contrast medium, mL	xxx (xx)	xxx (xx)
Time from diagnosis to reperfusion, min	xxx (xx)	xxx (xx)
Intra-aortic balloon pump, No. (%)	xxx (xx)	xxx (xx)
Mehran scores	xxx (xx)	xxx (xx)
Peri-procedures intravenous hydration volume (mL)	xxx (xx)	xxx (xx)
Pre-angiography	xxx (xx)	xxx (xx)
Procedure	xxx (xx)	xxx (xx)
0–4 hours post-angiography	xxx (xx)	xxx (xx)
4–24 hours post-angiography	xxx (xx)	xxx (xx)
Post-procedures oral hydration volume (mL)	xxx (xx)	xxx (xx)
Post-procedures urine volume (mL)	xxx (xx)	xxx (xx)

LVEF, left ventricular ejection fraction; ACEI/ARB, angiotension-converting enzyme inhibitor/angiotension receptor blocker.

Table S2

Baseline characteristics

Demographic	Aggressive group (N=xx)	Control group (N=xx)
Smoke, No. (%)	xxx (xx)	xxx (xx)
Clinical	xxx (xx)	xxx (xx)
Killip class, No. (%)	xxx (xx)	xxx (xx)
I	xxx (xx)	xxx (xx)
II	xxx (xx)	xxx (xx)
III	xxx (xx)	xxx (xx)
Pre-angiography renal function	xxx (xx)	xxx (xx)
Serum creatinine, μmol/L	xxx (xx)	xxx (xx)
Estimate glomerular filtration rate, mL/min/1.73 m2	xxx (xx)	xxx (xx)
Estimate glomerular filtration rate <90 mL/min/1.73 m2, No. (%)	xxx (xx)	xxx (xx)
Cystatin C, mg/L	xxx (xx)	xxx (xx)
Creatine kinase MB, U/L	xxx (xx)	xxx (xx)
Hematocrit, %	xxx (xx)	xxx (xx)
Hbalc, %	xxx (xx)	xxx (xx)
LVEF, %*	xxx (xx)	xxx (xx)
LVEF <40, No. (%)	xxx (xx)	xxx (xx)
Hypertension, No. (%)	xxx (xx)	xxx (xx)
Diabetes mellitus, No. (%)	xxx (xx)	xxx (xx)
Previous myocardial infarction, No. (%)	xxx (xx)	xxx (xx)
Contrast type, No. (%)	xxx (xx)	xxx (xx)
Iodixanol	xxx (xx)	xxx (xx)
Iopromide	xxx (xx)	xxx (xx)
Iopamidol	xxx (xx)	xxx (xx)
Other low osmolal agents	xxx (xx)	xxx (xx)
Contrast media volume, total (mmHg)	xxx (xx)	xxx (xx)
Left ventricular end-diastolic pressure, no./total (mmHg)	xxx (xx)	xxx (xx)
Intra-aortic balloon pump, No. (%)	xxx (xx)	xxx (xx)
Mehran scores	xxx (xx)	xxx (xx)

LVEF, left ventricular ejection fraction.

Table S3

Hydration information

Hydration information	Aggressive group (N=xx)	Control group (N=xx)
Peri-procedure hydration adjustment, no.	xxx (xx)	xxx (xx)
Intense hydration for hypotension	xxx (xx)	xxx (xx)
Did not undergo aggressive hydration	xxx (xx)	xxx (xx)
In-procedure hydration adjustment, no.	xxx (xx)	xxx (xx)
Intense hydration for hypotension	xxx (xx)	xxx (xx)
Post-procedure hydration adjustment, no.	xxx (xx)	xxx (xx)
Intense hydration for hypotension	xxx (xx)	xxx (xx)

Table S4

Inclusion and exclusion criteria

Variable name	Aggressive group (N=xx)	Control group (N=xx)
Inclusion criteria
Candidates for primary PCI	xxx (xx)	xxx (xx)
Years after age >18	xxx (xx)	xxx (xx)
Written informed consent	xxx (xx)	xxx (xx)
Exclusion criteria
Pregnancy, lactation	xxx (xx)	xxx (xx)
Allergy to contrast	xxx (xx)	xxx (xx)
Renal catheterization or heart valvular surgery	xxx (xx)	xxx (xx)
End-stage renal failure or renal transplantation	xxx (xx)	xxx (xx)
Recent acute kidney injury	xxx (xx)	xxx (xx)
Malignant tumor or life expectancy <1 year	xxx (xx)	xxx (xx)
Cardiac shock or NYHA IV	xxx (xx)	xxx (xx)
Contrast medium administration within the previous 7 days and post-procedure 72 h	xxx (xx)	xxx (xx)
Peri-procedural receipt of NSAIDs, aminoglycosides, cyclosporine or cisplatin in the past 48 h and during the study period	xxx (xx)	xxx (xx)
Inferior and/or right ventricle myocardial infraction combined with hypotension on admission (systolic pressure ≤90 mmHg)	xxx (xx)	xxx (xx)
Pre-procedural renal insufficiency eGFR ≤60 mL/min/1.73 m2	xxx (xx)	xxx (xx)

PCI, percutaneous coronary intervention; eGFR, estimate glomerular filtration rate.

Study outcomes

The primary endpoint is CI-AKI, which is considered to be a >25% or 0.5 mg/dL increase in SCr from baseline in the 48–72 hours immediately following the procedure (, ) (9).

Table 1

Analysis timing of measurement for endpoints

Endpoint	Analysis timing
Primary endpoint
CI-AKI	48 to 72 hours after the procedure
Secondary endpoint
CI-AKI48 h	48 hours after the procedure
CI-AKICysC	24 hours after the procedure
Major adverse cardiovascular events	Within the first year after enrollment
Major adverse clinical events	Within the first year after enrollment
CI-PKI requiring dialysis	Within the 3-month after enrollment
Total hospitalization costs	Discharge from hospital
Length of stay	Discharge from hospital
Safety endpoint
Acute heart failure	Within the hospitalization after enrollment

CI-AKI, contrast-induced acute kidney injury; CI-PKI, contrast-induced persistent renal damage.

Table S5

Primary and secondary endpoints

Endpoint	Aggressive group (N=xx)	Control group (N=xx)	Relative ratio (95% CI)	Absolutely risk difference (95% CI)	P value
Primary endpoint, (%)
CI-AKI	xxx (xx)	xxx (xx)	xxx (xx-xx)	xxx (xx-xx)	xx
Secondary endpoint, (%)
CI-AKI 48h	xxx (xx)	xxx (xx)	xxx (xx-xx)	xxx (xx-xx)	xx
CI-AKICysC	xxx (xx)	xxx (xx)	xxx (xx-xx)	xxx (xx-xx)	xx
Major adverse cardiovascular event, (%)	xxx (xx)	xxx (xx)	xxx (xx-xx)	xxx (xx-xx)	xx
Major adverse clinical event, (%)	xxx (xx)	xxx (xx)	xxx (xx-xx)	xxx (xx-xx)	xx
CI-PKI (%)	xxx (xx)	xxx (xx)	xxx (xx-xx)	xxx (xx-xx)	xx
Total hospitalization costs, $	xxx	xxx	—	—	xx
Length of stay, d	xxx	xxx	—	—	xx
Safety endpoint, (%)
Acute heart failure	xxx (xx)	xxx (xx)	xxx (xx-xx)	xxx (xx-xx)	xx

CI-AKI, contrast-induced acute kidney injury; CI-PKI, contrast-induced persistent renal damage. $ Exchange rate, 1 $= 7.03 ¥ (2019.08.13).

Table S6

Clinical events

Variable name	Aggressive group (N=xx)	Control group (N=xx)
Total events, No. (%)	xxx (xx)	xxx (xx)
CI-AKI	xxx (xx)	xxx (xx)
CI-AKI48 h	xxx (xx)	xxx (xx)
CI-AKICysC	xxx (xx)	xxx (xx)
Death	xxx (xx)	xxx (xx)
Target vascular revascularization	xxx (xx)	xxx (xx)
Nonfatal myocardial infarction	xxx (xx)	xxx (xx)
Dialysis	xxx (xx)	xxx (xx)
Acute heart failure	xxx (xx)	xxx (xx)
Cardiac shock	xxx (xx)	xxx (xx)
Stroke	xxx (xx)	xxx (xx)
Bleeding	xxx (xx)	xxx (xx)
Arrhythmias	xxx (xx)	xxx (xx)
Infection	xxx (xx)	xxx (xx)

CI-AKI, contrast-induced acute kidney injury.

The secondary outcomes consist of: (I) CI-AKI, defined as a >50% or 0.3-mg/dL absolute increase in SCr from baseline in the 48 hours immediately after the procedure; (II) CI-AKI, defined as a >10% or 0.3-mg/dL absolute increase in serum cystatin-C during the 24 hours immediately following the procedure (9); (III) persistent renal damage, which is considered to be residual impairment of renal function demonstrated by a >25% reduction in creatinine clearance at 3 months compared with baseline (10); (IV) major adverse cardiovascular events, which include all-cause mortality, target vascular revascularization, and non-fatal myocardial infarction; (V) major adverse clinical events which take place in hospital following the procedure, including acute pulmonary edema, cardiogenic shock, stroke, clinically significant arrhythmias, and bleeding; (VI) total hospitalization costs; (VII) length of stay.

The safe outcome is Acute heart failure (AHF) during hospitalization, defined as signs/symptoms of heart congestion and/or hypoperfusion by physical examination and further auxiliary examination such as ECG, chest X-ray, laboratory assessment (with specific biomarkers), and echocardiography (11).

Primary outcome analysis

All consented and randomized subjects will be accounted for and reported in the CONSORT diagram for the study; however, only those randomized subjects who have started either intravenous hydration intervention or have finished test for SCr at least once on pre-procedure and 72 hours post-procedure (i.e., did not drop out or withdraw prior to the start of the allocated intervention) will be considered as an adjusted ITT subject to be included in the DMC reports and primary efficacy analysis.

Conclusions

The ATTEMPT study will investigate the efficacy and safety of adequate hydration during the perioperative period among patients who require pPCI treatment. This SAP is intended to minimize the analysis bias of the study.

The article’s supplementary files as