Materials and Methods
Study Design, Location, Participants and Data Acquisition
This prospective observational study was performed at a tertiary academic teaching hospital, where eligible patients were recruited from a specialized AKI follow-up clinic. Adult (>17 years) patients were eligible if their AKI event occurred recently (<1 year), and they had KDIGO Stages 2 and 3 AKI. Patients who were diagnosed as KDIGO Stage 1 AKI were eligible for the AKI follow-up clinic if they had pre-existing CKD and/or incomplete kidney recovery at hospital discharge [serum creatinine (Cr) ≥25% of pre-AKI baseline]. Patients were excluded if they had any documented history of stroke, neurodegenerative disease, neurocognitive impairment and psychiatric diagnosis (e.g. schizophrenia), were kidney transplant recipients or had a baseline estimated glomerular filtration rate (eGFR) <15 mL/min/1.73 m2. Informed consent was provided before any neuropsychological assessments were performed or clinical data were abstracted. The cause of AKI (nephrologist-adjudicated), comorbidities, baseline Cr (defined as the closest value that was 7–365 days prior to the episode of AKI), highest hospital Cr, serum Cr and eGFR values at both neurocognitive test times, along with the AKI KDIGO stages, were abstracted from the clinic reports. Demographic variables such as ethnicity, handedness and highest education were collected at the first neuropsychological assessment. The study was approved by the Queen's University and Affiliated Hospitals Health Sciences Research Ethics Board and was performed in accordance with the declaration of Helsinki.
Healthy Control and Active Control Groups
For this study, patients recovering from AKI were compared with both a healthy control population, as well as an active control group that was matched for cardiovascular risk factors (hypertension and diabetes mellitus), but absent history of kidney disease. The active control group was selected as a comparator based on their shared cardiovascular risk factors (hypertension and diabetes mellitus) with vascular neurocognitive impairment, which allowed for greater certainty that detected neurocognitive impairment was due to the episode of AKI, and not these cardiovascular comorbidities or vascular neurocognitive impairment. This active control group was selected from a convenience sample of patients awaiting cardiac surgery.[15,16] Furthermore, only pre-operative scores were used as post-operative neurocognitive impairment may be caused by cardiac surgery.
The neurocognitive assessment was performed twice during the study. The first assessment was performed either on the day of consent or at the next clinic appointment. The second assessment was then completed on the patient's next clinic appointment following their previous assessment. The mean time between the AKI event and assessments is shown in Table 1. The neurocognitive assessment consisted of a standardized clinical assessment for measuring cognitive decline or improvement: the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS), and a robotic assessment that quantifies sensory, motor and cognitive performance: Kinarm Endpoint Lab (Kinarm). The assessment took on average 1 h and 20 min to complete, and was administered by trained research staff. We recently demonstrated the feasibility of this assessment battery on patients with CKD.
The RBANS domains and components, Kinarm task descriptions and affiliated neurocognitive domains are shown in Table 2. Specifically, the RBANS measures five domains: immediate memory, delayed memory, attention, visuospatial and language. These five domains are summarized by a composite score, the Total Scale Score. A score of <75.25 [1.65 standard deviations (SDs) below the mean of 100] was considered impaired. The mean and SD for the RBANS normative population are derived from healthy, age-matched control individuals.
In comparison, Kinarm quantifies performance in the sensory, motor and cognitive domains to produce a summarized performance score known as the task score. This task score is based on task parameters, along with spatial and temporal aspects of movement.[21,22] Additionally, the task score is standardized by age, sex and handedness, and is a normalized measure of performance that is generated from a large cohort of healthy individuals (range = 94–494, depending on the task). This healthy cohort provides the values for each task parameter that are used to create Box–Cox transformations, which converts this data into a standard normal distribution that includes regression models to factor out the influence of age, sex and handedness on performance. These normalized parameters are then aggregated together and transformed into z-task score before a final transformation to a positive value, such that task scores near 0 reflect best performance. These equations are implemented in the software used to run the Kinarm tasks (Dexterit-E, version 3.9) generating a normalized measure of performance for each participant in this study. Impairments were defined as a task score >1.96, which corresponds to performance below the 95th percentile of the healthy cohort.
To minimize learning effects on the subsequent assessment, different RBANS versions were used for each assessment. For Kinarm, previous research has found that there was good retest reliability along with minimal practice effects for most tasks in both adults and children, although there is some learning effect for the reverse visual-guided reaching task. The Kinarm also uses different variations of the tasks for trail making test forms A and B, along with different targets for the task of object hit and avoid.
Descriptive statistics were used to characterize the patients' neurocognitive performance. To determine whether there were any significant clinical or demographic differences between the AKI cohort and the active controls, a t-test or Fisher's exact test was performed as appropriate, with a P < 0.05 being deemed significant. Z-task scores were used for statistical comparisons between groups, regarding Kinarm data. An analysis of variance followed by a Tukey's honestly significant difference post hoc test was used to determine if there were any significant differences in performance between the two cohorts at either testing time, while accounting for multiple comparisons when investigating if AKI severity impacted performance. All graphs and analyses were performed using R software.
Nephrol Dial Transplant. 2022;37(2):285-297. © 2022 Oxford University Press