Delirium is a form of acute brain failure that commonly afflicts older hospitalized patients.1 The association between delirium and benzodiazepines is well-documented,2 but the association with other psychotropic medication classes is less clear. Using a novel liquid chromatography-mass spectrometry (LC-MS) assay, we observed supratherapeutic psychotropic drug levels (SPDLs) occurred in 17% of delirium cases and were associated with prolonged delirium episodes. Over half of SPDLs were secondary to selective and serotonin-norepinephrine reuptake inhibitors (SSRI/SNRIs),3 which are not considered deliriogenic.2 Most studies examining SSRI/SNRIs and delirium relied on medication lists, which do not account for SPDLs.4, 5 They also did not account for preexisting dementia, which may increase vulnerability to developing delirium, even at nontoxic, therapeutic drug concentrations.4, 5 We sought to determine if serum SSRI/SNRI concentrations were associated with prolonged delirium duration in older hospitalized adults and examined if this association was modified by preexisting dementia. This was an exploratory analysis of a prospective cohort study.3, 6 Patients ≥65 years old who were admitted to the hospital, had serum available for LC-MS measurements, and enrolled within 4 h of emergency department presentation were included. All delirious and a random selection (~17%) of non-delirious patients were enrolled between March 2012 and November 2014. Delirium was assessed daily using the modified Brief Confusion Assessment Method (bCAM), which is 82% sensitive and 96% specific for delirium.7 Serum SSRI/SNRI drug levels were measured using a LC-MS assay that measured 30+ psychotropic medications (Precera Bioscience, Inc., Franklin, TN). Standardized serum SSRI/SNRI concentrations were calculated by dividing each measurement by the upper limit of normal published in the literature. To establish the validity of our standardization approach, we analyzed standardized serum benzodiazepine concentrations. A patient was considered to have pre-illness dementia if they had: (i) a pre-illness dementia assessment (IQCODE) greater than 3.38,8 (ii) documented dementia diagnosis in the medical record, or (iii) prescribed cholinesterase inhibitors prior to admission. To determine if standardized serum concentrations of SSRI/SNRIs or benzodiazepines were associated with delirium duration, proportional odds logistic regression was performed adjusting for age, pre-illness dementia, functional status, and depression, comorbidity burden, pre-illness severity, kidney/liver dysfunction, and central nervous system diagnosis.6 A standardized serum concentration*pre-illness dementia interaction was incorporated; effect modification was considered present if the p-value was <0.20. Adjusted odds ratios (aORs) with 95% confidence intervals (95% CI) are reported. Statistical analyses were performed on SAS version 9.3 (SAS Institute, Cary, NC). Patient characteristics for the 158 patients included in this analysis can be seen in Supplemental Table 1. The median (interquartile range) delirium duration was 3 (2, 5) days. SSRI/SNRIs and benzodiazepines were detected in the sera in 54 (34.2%) and 26 (16.5%) patients, respectively. Of note, five out of 56 (8.9%) had SSRI/SNRIs and eight out of 26 (30.8%) had benzodiazepines detected by LC-MS, but were not listed in the electronic health record. As standardized serum concentrations of SSRI/SNRI and benzodiazepines increased, delirium duration also increased (Figure 1). The aOR (95% CI) for the proportional odds logistic regression models evaluating the association between serum SSRI/SNRI and benzodiazepine concentrations and delirium duration are seen in Figure 2. In the adjusted analysis, standardized serum SSRI/SNRI concentrations were significantly associated with longer delirium duration in patients with preexisting dementia only (aOR = 2.4, 95% CI: 1.2, 4.7, p-value for interaction = 0.126). Standardized serum benzodiazepine concentrations were significantly associated with delirium duration, regardless of preexisting dementia status (aOR = 3.1, 95% CI: 1.1, 9.0, p-value for interaction = 0.265). We observed that standardized serum SSRI/SNRI concentrations were associated with prolonged delirium duration in older hospitalized adults. The association was more prominent in those with preexisting dementia, who are the most vulnerable to developing delirium. This has multiple possible explanations: SSRI/SNRIs worsen delirium, may represent polypharmacy, serve as markers of advanced dementia with behavioral disturbances, or indicate aging physiology and reduced cognitive reserve. Further work is necessary to elucidate if SSRI/SNRIs have a causative effect on delirium duration. Most SSRI/SNRIs are not considered potentially inappropriate medications by established criteria.9 Our study suggests that older patients taking prescribed SSRI/SNRIs with delirium may benefit from serum concentration measurement. If elevated, dose reduction may be considered, especially if the patient has preexisting dementia. Because this was an exploratory analysis, our findings require confirmation. We used a novel LC-MS assay to measure serum drug concentrations of psychotropic medications, which is a more accurate method to characterize drug exposure. Studies using medication lists can miss an average of two medications and do not take into account nonadherence.10 In our cohort, 9% and 31% of SSRI/SNRIs and benzodiazepines, respectively, were detected in the serum but were not documented in the electronic health record. LC-MS serum drug measurements can also account for toxic serum drug levels, which may lead to delirium and other adverse outcomes. This approach seems valid as standardized serum benzodiazepine concentrations were also associated with prolonged delirium. Our study has several limitations. Residual confounding likely exists. We also did not evaluate the effect of SSRI/SNRIs on long-term cognition. A larger prospective cohort study is needed to examine this in older patients. JHH designed the original DELINEATE study. JHH designed the exploratory analysis, supervised data collection, and performed the statistical analysis. JOJ, JHH had access to the entire data set. JHH performed the statistical analysis. JOJ, JHH wrote the primary manuscript draft. All authors provided substantial revisions and approved the final manuscript. We would like to thank Ryan Morrison, Timothy Ryan, and Scott Daniels of Precera Bioscence, Inc. for performing the serum drug measurements and providing expertise in liquid-chromatography mass-spectrometry measurement. Dr. Jin H. Han and this study were funded by the National Institutes of Health under award number K23AG032355. This study was also supported by the National Center for Research Resources, Grant UL1 RR024975-01, and is now at the National Center for Advancing Translational Sciences, Grant 2 UL1 TR000445-06. Precera Bioscience, Inc, provided the serum drug concentration measurements for this study. Dr. Eduard E. Vasilevskis and Dr. Sandra F. Simmons are supported by the National Institutes of Health under award number R01AG53264. Dr. Jin H. Han, Dr. Eduard E. Vasilevskis, and Dr. Sandra F. Simmons are also supported by the Veteran Affairs Geriatric Research, Education, and Clinical Center (GRECC). All authors certify that they have NO affiliations with or involvement in any organization or entity with any financial interest or nonfinancial interest in the subject matter or materials discussed in this manuscript. The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. Supplemental Table 1. Patient characteristics. The Older American Resources and Services Activities of Daily Living (OARS ADL) characterizes seven basic and seven instrumental activities of daily living with higher scores representing greater dependence. Charlson morbidity score was used to characterize comorbidity burden. The Acute Physiology Score was used to characterize severity of illness. IQR, interquartile range. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.