Sepsis and Cognitive Assessment

Sepsis disproportionally affects people over the age of 65, and with an exponentially increasing older population, sepsis poses additional risks for cognitive decline. This review summarizes published literature for (1) authorship qualification; (2) the type of cognitive domains most often assessed; (3) timelines for cognitive assessment; (4) the control group and analysis approach, and (5) sociodemographic reporting. Using key terms, a PubMed database review from January 2000 to January 2021 identified 3050 articles, and 234 qualified as full text reviews with 18 ultimately retained as summaries. More than half (61%) included an author with an expert in cognitive assessment. Seven (39%) relied on cognitive screening tools for assessment with the remaining using a combination of standard neuropsychological measures. Cognitive domains typically assessed were declarative memory, attention and working memory, processing speed, and executive function. Analytically, 35% reported on education, and 17% included baseline (pre-sepsis) data. Eight (44%) included a non-sepsis peer group. No study considered sex or race/diversity in the statistical model, and only five studies reported on race/ethnicity, with Caucasians making up the majority (74%). Of the articles with neuropsychological measures, researchers report acute with cognitive improvement over time for sepsis survivors. The findings suggest avenues for future study designs.


Introduction
Sepsis is one of the most common, expensive, and inadequately managed syndromes. A 2016 task force introduced an updated definition (Sepsis-3), explaining that sepsis is "a life-threatening organ dysfunction caused by a dysregulated host response to infection" [1][2][3][4]. Approximately 2 million adults are affected by sepsis within the United States annually with sepsis being responsible for one out of three hospital deaths [5,6]. Although the sepsis-associated hospitalization frequency has minimally changed for young individuals aged 18-49 years old, the hospitalization frequency has risen dramatically for those aged 65 and older, such that sepsis is currently considered a "disease of the aged" [2]. With increasing age, individuals with sepsis are more likely to endure critical illness with major organ damage resulting in chronic care conditions.
For the brain, sepsis might act as a major inflammatory stimulus and potentially increase the brain's susceptibility to neurodegenerative disease [7,8]. Fitting with postulates of a brain reserve theory and threshold, sepsis may stimulate the deterioration of cognitive ability or enhance the risk for future progressive cognitive deterioration [9]. Brain health and cognition is also relevant to the risk of developing acute organ dysfunction [10]. By 2050, people aged 65 and older will reach 1.6 billion worldwide [11] and our healthcare system will face greater numbers of individuals with early to late-stage neurodegenerative disorders such as Alzheimer's disease (AD) [12].
To assist sepsis cognitive research going forward, we conducted the current review to provide insight into the strengths and weaknesses of published research addressing cognition following sepsis. We summarize study designs, the inclusion of a cognitive expert on the team, if authors considered the impacts of education, sex, race/ethnicity, the timeline of testing and inclusion of a baseline, the time of testing, the type of cognitive measures, the statistical approach, and findings. We provide conclusions regarding current limitations and strengths of this literature.

Materials and Methods
The literature review process included a thorough PubMed search for publications up to 13 January 2021 (See Figure 1). Key terms included: "sepsis and cognition", which yielded 850, "sepsis and cognitive" (850), "sepsis and memory" (699), "sepsis and thinking" (642), and "sepsis and attention" (1962). Across these search terms a total of 3050 unique articles were identified, with 2816 articles removed based on keywords missing from the article titles, and with 234 qualifying as full text reviews. Of these, 18 met all inclusion criteria. To be eligible for inclusion, articles needed to have a population with a mean age of ≥18 years, a quantitative assessment of a cognition post-septic episode, they needed to be peer-reviewed (i.e., no dissertations), and they needed to be written in English. No limits were placed on the date of publication or timeframe of the cognitive assessment.

Age, Education, Sex, Race and Ethnicity
The mean age ranged from 49 to 80.81 years across all studies, for a grand mean of 58.30 years. Analytically, 35% of studies considered premorbid intellectual abilities such as education, with a wide educational attainment range. Three studies (16.7%) reported a 7th grade education on average, four studies (22.2%) were comprised mainly of high school graduates (12 years), and two studies (11.1%) included mainly college graduates (16 years).
Fifteen studies (83.3%) reported on sex, which were generally evenly split across sexes with slightly more females (n = 1067) than males (n = 995).
Only five studies (27.8%) reported on race and/or ethnicity, with the majority of patients identifying as Caucasians (79.5%). Only three studies (16.7%) reported on the inclusion of patients identifying as Black/African-American, one study (5.6%) reported on individuals identifying as Hispanic, and one study (5.6%) reported on individuals identifying as Native American or other Pacific Islanders. See Table 1 for additional demographic information.

Baseline and Time of Testing
Few studies (17%) included baseline cognitive testing (pre-sepsis data), making it difficult to accurately identify cognitive change from premorbid abilities. Initial cognitive testing occurred within two months of hospital discharge for nine studies (50%), with seven of those assessments being within 48 h of discharge. Twelve articles (66.7%) had at least one follow-up assessment, and eleven of them conducted at least two follow-up testing sessions, providing data for the course of cognitive changes and recovery post sepsis. See Table 2 for more information on cognitive testing.

Type of Measures, Congitive Domains, and Reported Scores
Seven studies (39%) only administered cognitive screening tools, with the remaining using a combination of standard neuropsychological measures. Cognitive domains typically included declarative memory, attention and working memory, processing speed, and executive function. Eleven studies (61%) used raw scores or mean raw scores, two studies (11.1%) reported percentages of scores falling 1.5 or 2 standard deviations below the mean, one study (5.6%) reported the percentage of cognitive impairment, two studies (11.1%) included t-scores (age and/or education adjusted), and two studies (11.1%) included z-scores.

Discussion
Relevant research articles reviewed herein span 13 years (February 2008 to January 2021). The 18 publications addressing cognition varied with regards to the study design, demographic reporting, cognitive test type, and statistical approach. Despite 18 published articles, we continue to have a limited understanding of sepsis and cognition. Strengths include the number of studies considering cognitive domains in addition to general cognitive screeners, as well as the inclusion of an expert in cognitive assessment as part of the study team panel. Eleven studies report at least two follow-up testing sessions. Initial cognitive testing occurred within two months of hospital discharge for nine studies, with seven of those assessments being within 48 h of discharge. Across the publications, the themes suggest an acute decrease in cognitive function, with follow-up assessments showing cognitive improvement from acute levels. However, few studies considered premorbid or demographic factors within the statistical model, and less than half included a non-sepsis peer group to calculate practice effects. These design limitations challenge the accuracy of study findings, for there are associations between premorbid status and cognitive change [31][32][33][34], and there is a value to considering practice effects in post-trauma cognitive change models [35][36][37]. Our review findings suggest avenues for improvement as the field of sepsis and cognitive research moves forward.
Although more than one third of studies limited their investigation to global cognitive screening tools, the remaining studies used a combination of neuropsychological measures primarily assessing attention/working memory and declarative memory. These domains involve neurological systems typically assaulted by common comorbidities often associated with sepsis and small vessel vascular disease such as hypertension and diabetes [38][39][40]. Two studies additionally considered subdomains of measures of language (semantics) [21,23], and two looked at subdomains of higher cortical abstract reasoning [21,29], but they were limited in sample size. Hypothetically, within larger samples, comparing semantics and higher order functions to the more "vulnerable" cognitive domains would guide insight into cortical versus subcortical/white matter contributions, neuronal vulnerability, and potential mechanisms [41,42]. Longitudinal studies comparing cognitive domains in larger samples are needed.
Control groups are discussed in eight of the published investigations. Control groups used within a statistical design can provide information on "non-disease" practice effects as well as provide normative reference sources for calculating a standardized individual composite reliable change score [34,36,37,42]; the change is larger than reasonably expected due to the measurement error alone [43]. Control groups provide a reference for the expected performance. This information is particularly valuable when a researcher is concerned about the psychometric properties of cognitive measures (e.g., range of possible scores, normal distribution) and the test appropriateness for the population of interest (i.e., one would not administer a 16-word verbal list learning test to individuals with moderate cognitive impairment, for this would likely result in a poor performance and a floor effect). Patients who are acutely ill or cognitively compromised may perform at floor level; the test may be too difficult or not appropriate for identifying further deterioration. By contrast, individuals with a superior premorbid cognitive reserve who acquire sepsis may not be challenged appropriately with general cognitive screening tests. Control groups provide the research team with an external comparison to examine expected patterns of performance without complication from the disease state [37,42].
There was also a notable absence in the publications of the reporting of the years of education. Half of the publications reported this variable, and three considered education years in the statistical model. Given the wealth of research showing how education is a proxy for cognitive reserve, we consider the absence of educational reporting a concerning finding. The years of education have been used repeatedly in aging studies to help explain variability in cognition relative to pathology [44,45]. Cognitive reserve proxies, such as education, are shown to protect against impairment from traumatic brain injury [31][32][33] or operative/anesthesia exposure [34]. Considering education in analytical approaches may explain important variabilities in the outcome. The years of education are also a vital consideration if research teams are unable to acquire a premorbid/pre-sepsis cognitive estimation using formal test tools [46].
Sociodemographic considerations were limited, as was a consideration of medical comorbidities. Animal models [47] and some human studies [48,49] report sex as an important predictor of sepsis recovery. However, no study reported findings relative to participants' sex. Although race or ethnicity were reported in five of the 18 studies, no study considered these variables relative to the cognitive trajectory. Today, we appreciate how race and ethnicity are surrogates for appreciating health disparities' contribution to neurobiological responses [50]. Social determinants of health, including systemic racism experienced by many communities of color, are known to have negative ramifications on health outcomes and recovery [51][52][53]. It remains unknown if or how the cognitive sequela of sepsis differentiates across sex, race, and ethnicity. Sex, race, and ethnicity are proxies for health disparities and may modify neurobehavioral responses. This is an important area for future research. Further, individuals within different sociodemographics have a unique risk of sepsis (i.e., renal disease, congestive heart failure, myocardial infarction, chronic pulmonary disease, liver disease, peripheral vascular disease, peptic ulcer disease, and connective tissue disease) [38], and these are worthy of consideration relative to the presepsis neuronal vulnerability status. We also identified a considerable patient heterogeneity in the current state of the literature with regards to the demographic characteristics and origin of sepsis. These are relevant design considerations.
Based on the current review, the authors consider sepsis cognitive research to be in an early developmental stage. There are noticeable methodological study design weaknesses that limit confidence in study findings. We encourage future researchers to consider study design suggestions from the fields of epilepsy (see [54] for a review of neuropsychology and epilepsy) and perioperative cognitive research-both of which have aggressively addressed challenges in longitudinal cognitive study design problems since the 1990s [42]. While sepsis poses a unique challenge due to the unexpected nature of the illness, researchers may wish to consider the benefits of including a control group, premorbid intellectual estimates, education, and other sociodemographic factors in future investigations. It should be noted that our review spanned a considerable timeframe (from February 2008 to January 2021), where the definition and management of sepsis rapidly evolved. We encourage a repeated literature review over the next 5 years.
Author Contributions: Conceptualization: L.C.J., C.C.P. and P.A.E.; Methodology: L.C.J., C.D. and C.C.P.; Data Curation: L.C.J. and C.D.; Writing-original draft preparation: L.C.J.; Writing-review and editing: C.D. and C.C.P.; Supervision: C.C.P. and P.A.E. All authors have read and agreed to the published version of the manuscript.
Funding: This research was supported by NIH K07 AG066813.