Statistical analyses for neurological recovery will include comparisons of means of motor and sensory function, functional independence, and serum measurements (e.g., glucose, insulin, ALT, and AST) between the two groups, as well as comparisons within each group of the changes from baseline to rehabilitation to discharge (which should take approximately 5 weeks). Descriptive statistics, including measures of central tendency (sample mean, sample median) and dispersion (variance) will be calculated for each group. For safety outcomes, instead of comparing the two groups to the established standard safety values, we will compare circulating levels of proposed outcomes at the various time points among the two groups. If there are no significant group differences in these outcomes, we will conclude that the KD intervention is safe. We will also calculate 95% confidence intervals for the mean group differences for these outcomes as an additional check on safety and feasibility. The primary method of analysis will be analysis of covariance (ANCOVA), regressing the 5-week outcomes (safety, motor and sensory recovery, functional independence, and glycemic state) on the randomly assigned diet, baseline values of the measures, and other covariates such as age and sex. Repeated measures ANCOVA will also be used. Statistical assumptions will be assessed using box plots, residual plots, and normal probability plots. Secondary analyses, accounting for all time points, will be conducted using mixed linear models, including repeated measures models. An appropriate structure for the covariance matrix (e.g. the unstructured covariance matrix) will be selected for these models using the final data. When a model term is statistically significant, the Tukey-Kramer multiple comparisons test will be used to determine which specific pair of means are significantly different. This method will allow us to compare changes over time (within-group changes) and differences between groups simultaneously. Covariates (e.g., length of enteral and solid feeding and length of stay in acute and rehabilitation care) will be accounted for in these analyses. Overall cross-sectional between-group comparisons, such as baseline comparisons used to examine pre-test parameters, may be performed using the two-group t-test, and overall within-group comparisons may be performed using the paired t-test. If assumptions of normality of distribution for the above tests are not tenable, variables may be transformed prior to analysis or appropriate nonparametric tests, such as the Wilcoxon rank-sum and signed-rank tests, may be used. Statistical tests will be two-sided and will be performed using a 5% significance level. Stratified analyses accounting for different injury levels will also be performed. These analyses will be specific to the distribution of data in each individual subgroup, which will minimize heterogeneity and allow us to analyze subgroups that may be more statistically homogenous and clinically relevant. SAS software, version 9.4 or later, will be used to conduct the statistical analyses. Multiple imputation methods will be used to address missing data for variables with moderate amounts of missing data (≥ 10%) and after examining whether data are missing completely at random (MCAR), at random (MAR), or not at random (MNAR). Missing data analysis will not be performed on variables with < 10% missing data to avoid statistical bias larger than that obtained through the use of complete case analysis.
For microbiome analysis, alpha diversity indices will be analyzed using linear mixed-model with time (pre and post) and group (KD and SD) as fixed factors and participant as a random effect. Permutational multivariate analysis of variance (PERMANOVA) testing will be used to assess the effects of time, treatment, and time X treatment on beta diversity indices. The principal coordinate analysis will be performed to visualize the beta diversity distance matrices (18).
For targeted proteomics, three specific analytic approaches will be utilized. First, two-sample t-tests, or Wilcoxon rank-sum tests if the normality assumption is questionable, will be used to examine differences in the number of selected peptides and proteins between diet groups at discharge (approximately 5 weeks). Because of the number of proteins to be examined, a false discovery rate of 0.2 will be used to account for multiple hypothesis tests. The second approach will examine the manner in which levels of selected peptides vary across the longitudinal time points of baseline, rehabilitation, and discharge. To examine these patterns, mixed linear models such as repeated measures models using time as a within-person factor and diet as a between-person factor will be constructed. Finally, Pearson’s correlation will be used to measure the strength and direction of linear associations between selected peptide levels and 5-week outcomes of safety, neurological, functional, and glycemic levels. Stratified analyses accounting for different injury levels will also be performed. These analyses will be specific to the distribution of data in each individual subgroup, which will minimize heterogeneity and allow us to analyze subgroups that may be more statistically homogenous and clinically relevant. SAS software, version 9.4 or later, will be used to conduct the statistical analyses.
Trial monitoring
No regular external trial auditing is scheduled. However, the Data and Safety Monitoring Committee (DSMC) includes 2 external safety monitors and 3 study personnel including the principal investigator, study physician, and a post-doctoral fellow. Following key investigators and study staff provide DSMC with information regarding safety endpoints, data quality and validity quarterly over the course of the study. All human participant data, ranging from recruitment/screening to diet intervention and testing to laboratory tests, are reviewed quarterly (every 3 months) attended by key study staff, the PI, and the DSMC committee. This includes updates on any new hazards, risks or adverse events, and plans of action. Additional investigators and staff will be asked to participate as the need for their input or expertise arises. If during the course of these meetings, particular unforeseen hazards or risks are identified that may predispose to an unusually high number of serious adverse events, the PI consults the appropriate members of the investigative team, as well as the UAB IRB and NIH/NINR to determine if the study should be terminated or altered in some way. Any procedure that is deemed hazardous is eliminated from the study and replaced with an alternative if one with reasonable risk can be identified.
Adverse event monitoring and reporting
Reporting. The principal investigator will be responsible for the accurate documentation, investigation, and follow-up of all possible study-related adverse events. All adverse events are reported by the PI to the UAB IRB with a description of the event, when and how it was reported, and appropriate documentation to corroborate the event. The description of the adverse event includes all information listed on the case report forms. The IRB then determines whether additional reporting to the NIH/NINR is required.
• Serious Adverse Events. All serious adverse events are reported immediately to both the study physician and PI, and then, in turn, reported to the IRB within 48 hours. In addition, all serious adverse events that are possibly related to the study intervention are reported to the NIH/NINR within two weeks.
• Non-Serious Adverse Events. Non-serious events are reported to the IRB within five business days of the event.
Annual Reports. An annual statement summarizing the results of safety monitoring is sent to the UAB IRB and to the NIH/NINR. The annual report includes written summaries from quarterly meetings between the DSMC, key personnel and staff, descriptions of all adverse events, and a summary discussing:
(1) Whether all participants met entry criteria;
(2) Whether adverse event rates are consistent with expected rates and seriousness;
(3) Reasons for dropouts from the study;
(4) Whether continuation of the study is justified; and
(5) Whether conditions have been met for terminating the study prematurely.
Protocol Amendments
Any change to the protocol requires a written protocol amendment that must be approved by NIH/NINR and IRB before implementation. Upon acceptance from the sponsor and IRB, the PI makes updates and edits to the study record published on ClinicalTrials.gov. If the PI determines that an immediate change or deviation from the protocol is necessary for safety reasons to eliminate an immediate hazard to the participants the IRB will be notified immediately.
Confidentiality
To protect privacy and confidentiality, any original paperwork documenting a participant’s name and PHI is stored in a locked cabinet in the research coordinator's office, while any digital study records involving PHI are stored in REDCap and/or on computers requiring password authentication that are stored in locked offices and are behind secure firewalls. Records that identify study participants are kept confidential as required by law, and every effort is made to maintain the confidentiality of participants’ study records. Except when required by law or if necessary to protect their rights or welfare, study participants are not identified by name or any other identifying characteristic in records disclosed to those outside of the study staff.
Access to data
Only study staff, the UAB IRB, official overseers of clinical research at UAB, and representatives of the NIH have access to study records, data, and specimens; all access is on a need-to-know basis. All study staff are trained in HIPAA standards for protecting PHI and do not refer to PHI or confidential information in the presence of individuals outside of the study team. Moreover, each study staff member's access to participants’ data is limited to only the functions for which s/he is responsible.
Dissemination Policy
The results of our research will be disseminated to three major stakeholders: (1) the scientific community; (2) individuals with spinal cord injury (SCI); and (3) the general public.
A. Scientific Community
Timeline: Up to 18 months after the last participant completes the study intervention.
The results of this project will be presented at scientific conferences, including at the American Congress of Rehabilitation Medicine (ACRM), the American Spinal Injury Association, and the Society for Neuroscience annual meetings. In addition, results will be published in peer-reviewed journals with open access policies and/or in top-tier journals.
B. Individuals with SCI
Timeline: Up to 12 months after the last participant completes the study intervention.
• The UAB Spinal Cord Injury Model System Information Network Website. This website, which averages more than 43,000 visits per year, is newly designed and features a comprehensive collection of links (over 360 to date) to SCI-related information provided by reputable organizations, associations and educational institutions, including ones from the Model Systems Knowledge Translation Center (MSKTC) and other SCIMS centers. The site will be regularly updated with nutrition information and links to other reliable sites of interest and value to SCI consumers. All educational materials written and produced from this project will be made available free on this website.
• Pushin’ On. The UAB-SCIMS newsletter (now a fully digital newsletter) has been published for 33 years. It provides persons with SCI and their families with information of interest. Over the years, the newsletter has featured original articles along with news, information, and synopses of research of importance to individuals with SCI. The study team will work with the newsletter editor to write a synopsis of the proposed research trial, including a description of the diet-intervention and a lay summary of major findings and how to implement them.
• Paraplegia News (PN). For more than 60 years, PN has been a leader in the wheelchair community of pertinent practical news and information. Paraplegia News Online is an extension of PN easily accessible on the world-wide-web and is dedicated to bringing the very best of real-time, up to the moment news and information for wheelchair users, family members, and medical professionals on the go. The study team will work with the newsletter editor to write a synopsis of proposed research trial including the description of diet-intervention and a lay summary of major findings and how to implement them.
Timeline: Up to 18 months after the last participant completes the study intervention.
Study findings will be communicated to the general public by collaborating with journalists. The principal investigator will reach out to several of her contacts in radio, TV, and print media to plan to write a feature article on the proposed project.