Background
Subsyndromal delirium (SSD) is a frequent condition characterized by a less severe cognitive impairment in comparison to delirium, in which some, but not all, diagnostic criteria for delirium are met [
1]. However, to date there is no published consensus on the definitions for a subclinical form of delirium and SSD has been commonly reported as an intermediate stage between delirium and normal mental states [
1]. The most frequently employed delirium screening tools consider the diagnosis of SSD when the Intensive Care Delirium Screening Checklist (ICDSC) score is 1–3 out of 8 [
2] or when the Confusion Assessment Method (CAM) score is positive in two items out of four items [
3].
The Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-V), Neurocognitive Disorders Workgroup used the term “attenuated delirium syndrome” to describe a condition very similar to SSD but without specific diagnostic criteria and has been discussing whether SSD should be added as a subcategory of delirium in parallel with a new category, mild neurocognitive disorder. Of note, neither the DSM-IV-TR nor the DSM-V Workgroup determine nor distinguish whether subsyndromal presentations do or do not progress to delirium [
4,
5].
Moreover, recent studies did not show a consistent association between SSD and increased mortality rates or clinical outcomes in intensive care unit (ICU) patients [
6‐
8]. Nevertheless, despite limited available knowledge, intervention studies were recently performed aimed at the reduction of conversion from SSD to clinical delirium as a means to improve outcomes [
9,
10]. We therefore conducted a systematic review of studies that evaluated SSD in the ICU. Our main objective was to produce quantitative estimates of the prevalence of SSD and to explore the association between SSD and clinically relevant outcomes of ICU patients such as the mortality, ICU and hospital length of stay (LOS), duration of mechanical ventilation (MV), and conversion to delirium.
Discussion
In the present systematic review we synthesized the data on the prevalence of SSD in patients admitted to the ICU, as well as the association between SSD and delirium and clinical outcomes in critically ill patients. We identified six studies enrolling a total of 2630 patients. SSD was present in 950 (36%) patients. Despite marked variations between studies (from 33% to 57%) at least one in every three patients fulfilled the criteria for SSD, confirming the notion that it is a highly prevalent condition.
The SSD was not consistently associated with increased mortality or worse outcomes, as opposed to current data on delirium [
18]; however, our meta-analysis found an increase in hospital stay (Fig.
2). Only one study evaluated the association of SSD with duration of MV and was not able to report a clinically relevant outcome [
7], although the current literature supports the notion that delirium is independently associated with an increase in MV duration [
18‐
20].
Studies that evaluate non-ICU patients have described the outcomes of SSD (i.e., cognitive decline, functional decline, increased hospital LOS, and increased rates of admission to long-term institutions and death) as being poor in older people [
1,
21‐
23]. Cole et al. [
1] in a recent systematic review of non-ICU older patients described that SSD had an elevated prevalence (23% (9–42%)), and is a clinically important condition that falls on a continuum between no symptoms and full delirium considering hospital LOS, post-discharge mortality, and functional decline [
1]. Despite the apparent importance of SSD in non-ICU settings the increased mortality described above is not observed in the ICU population. Moreover, post-ICU discharge information about the role of SSD on long-term outcomes such as cognitive impairment or functional decline is not currently available.
There are important differences when we compare ICU and non-ICU patients regarding SSD that may explain the observed discrepancies in the outcomes. SSD may be a marker for underlying medical conditions not severe enough to cause full delirium in the non-ICU population where the cognitive trajectory and baseline severity of illness leads to a slowly increasing number of risk factors. In contrast, studies in ICU patients reported that at least 11 variables were considered to have moderate or strong evidence for contributing to delirium and they often occur simultaneously and are usually present in the first days following ICU admission [
24,
25]. This high burden of non-modifiable risk factors present early at the onset of critical illness can contribute to the occurrence of delirium without a prodromal phase or SSD in the ICU. Moreover, as described by Patel et al. [
26], the rapidly reversible sedation-related delirium showed fewer MV days (2.5 (1.6–2.8) vs 6.2 (3.7–12.0),
p < 0.001), ICU days (4.5 (2.2–7.2) vs 13.1 (8.8–19.1),
p = 0.001), and hospital days (6.7 (3.8–16.4) vs 25.4 (13.6–29.6),
p < 0.001) than persistent delirium. Those patients with rapidly reversible sedation-related delirium had lower hospital mortality in comparison with persistent delirium (0% vs 36%,
p = 0.001), which was similar to subjects with no delirium [
26]. This may also indicate that the occurrence of SSD (a condition of lower severity as compared with rapidly reversible delirium) may not be sufficient to generate worse outcomes or may be a transient condition from a worse neurologic state (such as coma or delirium) before re-establishing normal cognition.
The conversion of mental status or percentage of transition from SSD to delirium was only evaluated in two small-sample clinical trials to describe the effect of antipsychotics in preventing the conversion from SSD to delirium [
9,
10]. In the study of Al-Qadheeb et al., 1358 patients were evaluated but only 68 patients were classified as SSD and received intervention [
9].
Therefore, as the course of delirium can be heterogeneous and unpredictable it is unclear if the presence of SSD during the trajectories of delirium or even subsequent development of residual SSD after recovery can be implicated in negative outcomes.
From the two studies that evaluated the use of antipsychotic drugs in SSD patients, only one study could show a reduced conversion from SSD to delirium [
10]. Moreover, none of these studies demonstrated any other positive impact in relevant outcomes such as ICU and hospital LOS or mortality. These findings appear to be in concordance with the current literature on delirium which provides no conclusive evidence that pharmacologic treatment of ICU delirium modifies clinically relevant outcomes other than agitation [
27].
Finally, the absence of a formal definition of SSD contributes to the heterogeneity between studies and may partially explain the conflicting results described. Some researchers have defined SSD as having at least two or more a priori selected core symptoms [
21,
22], whereas others have specifically focused on attention and cognitive impairments [
28] or have used specific cut-off points on diagnostic scales for delirium, such as in the ICDSC [
6]. The ICDSC scale was developed to diagnose and graduate the delirium symptoms [
2] and it is more in line with the proposed diagnosis of SSD. The CAM was used by authors to diagnosis SSD in ICU [
8,
17] and non-ICU settings [
1], but considers only the presence of some symptoms not fulfilling the criteria for delirium. At this point, the CAM forces a dichotomization in diagnosis of mental status and it is not clear if the items of the CAM represent the same relevance in cognitive dysfunction or the same impact in the outcomes. Recently, two new delirium rating scales were described, called CAM-S [
29] and the CAM-ICU-7 [
30]. They were derived from CAM and CAM-ICU, respectively, and were able to provide a graded scale for delirium severity assessment. This is in contrast to ICDSC, which plateaus at the threshold of clinical delirium and does not provide further predictive discrimination. However, the data available were limited to delirious patients only. For future studies it will be necessary to clarify assessment of SSD using the new scales.
The clinical entity and the implications of SSD are not fully understood or precisely defined. It remains unclear whether SSD represents an early stage of manifest full delirium, an independent diagnosis, or simply a description for an array of symptoms with no major clinical consequence.
The present study has some limitations and several of them are actually related to the aforementioned absence of a clear definition of SSD. First, we consider that, despite being a major drawback, the diagnoses of SSD for patients in the present study are those being used by clinicians in their practice as well as investigators in clinical studies. We believe that differences in the design, definitions, and tools to diagnose delirium and available data on outcomes and disease severity as well as substantially diverse patient populations could account for the substantial differences in prevalence and mortality. Second, none of the studies evaluated SSD using the CAM-ICU. ICU providers do not typically use the CAM, considering that many patients are not able to communicate or it takes too long and requires special training, which could contribute to an underdiagnosis of delirium. Third, the assessment of publication bias through the funnel plot analysis was impaired due to the small sample size; the power of the test is too low to distinguish chance from real asymmetry. Fourth, we did not conduct a grey literature search which might contribute to an overestimation of the size effect in small trials. Finally, since data on long-term outcomes are not available in the current literature, this relevant aspect of SSD could not be evaluated.