Underreporting of Delirium in Statewide Claims Data: Implications for Clinical Care and Predictive Modeling

Author's Accepted Manuscript

Under-Reporting of Delirium in State-Wide Claims Data: Implications for Clinical Care and Predictive Modeling Thomas H. McCoy Jr MD, Leslie Snapper BS, Theodore A. Stern MD, Roy H. Perlis MD, MS

PII: DOI: Reference:

S0033-3182(16)30056-1 http://dx.doi.org/10.1016/j.psym.2016.06.001 PSYM651

To appear in:

Psychosomatics

Cite this article as: Thomas H. McCoy Jr MD, Leslie Snapper BS, Theodore A. Stern MD, Roy H. Perlis MD, MS, Under-Reporting of Delirium in State-Wide Claims Data: Implications for Clinical Care and Predictive Modeling, Psychosomatics, http://dx.doi.org/10.1016/j.psym.2016.06.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

McCoy et al 1 Under-reporting of Delirium in State-Wide Claims Data: Implications for Clinical Care and Predictive Modeling Thomas H McCoy Jr, MD1,2 Leslie Snapper, BS1 Theodore A. Stern, MD2 Roy H Perlis, MD, MS1* (1) Center for Experimental Drugs and Diagnostics, Department of Psychiatry and Center for Human Genetic Research, Massachusetts General Hospital, Simches Research Building 6th Floor, 185 Cambridge St, Boston, MA 20114 (2) Avery D. Weisman Psychiatry Consultation Service, Massachusetts General Hospital Bulfinch Building 4th Floor, 55 Fruit St, Boston, MA 02114.

* Correspondence: Roy Perlis, MD MSc Simches Research Building/MGH 185 Cambridge St, 6th Floor Boston, MA 02114 617 726-7426 [email protected]

McCoy et al 2 Abstract Background – Delirium is an acute neuropsychiatric syndrome that portends poor prognosis and represents a significant burden to the health care system. Although detection allows for efficacious treatment, the diagnosis is frequently overlooked. This under-diagnosis makes delirium an appealing target for translational predictive algorithmic modeling; however, such approaches require accurate identification in clinical training data sets. Methods – Using the Massachusetts All-Payers Claims Database (APCD), encompassing health claims for Massachusetts residents for 2012, we calculated the rate of delirium diagnosis in index hospitalizations by reported ICD9 diagnosis code. We preformed a review of published studies formally assessing delirium to establish an expected rate of delirium when formally assessed. Secondarily, we reported a sociodemographic comparison of cases and non-cases. Results – Rates of delirium reported in the literature vary widely, from 3.6% to 73% with a mean of 23.6%. The state-wide claims data (APCD) identified the rate of delirium among index hospitalizations to be only 2.1%. For APCD hospitalizations, delirium was coded in 2.8% of patients > 65 years old and for 1.2% of patients < 65. Conclusion – The lower incidence of delirium in claims data may reflect a failure to diagnose, a failure to code, or a lower rate in community hospitals. The relative absence of the phenotype from large databases may limit the utility of data-driven predictive modeling to the problem of delirium recognition. Keywords: Delirium, claims data, prevalence, predictive modeling

McCoy et al 3 Introduction Delirium is an acute neuropsychiatric syndrome characterized by disturbances in attention and orientation as well as alterations in cognition and perception. (1) Delirium portends a poor prognosis: Those diagnosed with delirium are more likely to remain intubated longer, to remain hospitalized longer, to require more costly care, to have worse cognitive and functional outcomes, to die in the hospital, and to die at a higher rate following discharge. (3-10) In addition to these individual consequences, delirium represents a significant burden to the healthcare system, with an estimated attributable cost in excess of $150 billion dollars per year. (11) Strategies that diminish risk for delirium represent an opportunity to reduce morbidity and to diminish healthcare costs. Delirium can be prevented and/or treated via both pharmacological and non-pharmacological means; mounting evidence suggests that these interventions improve outcomes. (12-17) Likewise, early detection during a hospital stay may reduce many of the burdens associated with delirium. (6, 8) Despite its clinical significance and tractability, the diagnosis of delirium is frequently overlooked. (18-22) This failure to diagnose is a missed opportunity for helping patients, and may contribute to worse outcomes. (23) Many risk factors have been identified with the goal of improving recognition, and establishing the etiology, as well as improving risk stratification of delirium. (24-30) The predictors of delirium - age, sex, co-morbid diagnosis, specific interventions [e.g., intubation, surgery] and/or exposure to classes of drugs - are familiar and easily operationalized concepts. Increasingly, data-driven research in health care has identified useful, but less conceptually familiar and more difficult-to-operationalize, predictors of many important health outcomes. In this paradigm predictors are identified by mining electronic health record (EHR) data collected in the process of routine care. (31-37) The literature on delirium has started to move in this direction, with early efforts returning promising results. (30, 38) The rapidly growing adoption of EHRs has both enabled this research (by facilitating the collection and maintenance of vast databases) and provided means of translation, as the electronic medical record allows complex models to be applied to patients at the point-of-care. (39-42)

McCoy et al 4

Data-driven studies typically draw on a collection of insurance claims and/or electronic medical record (EMR) data. These data sources each provide an opportunity to mine large clinical data repositories; nevertheless, they differ in important ways. Medical records exhaustively record the details of the episodes of care within the entity, keeping a single medical record; however, many patients will receive care from multiple providers from disparate systems. As such, no single medical record is likely to be complete, an example of the open-system problem. Insurance claims data, by comparison, tend to be broad but shallow; they exhaustively report all benefit utilization by a subscriber (and thus a closed system), but are limited to the rudimentary structured data required to process a claim. Large databases – be they claims or EMR - are dependent on the quality of the source data captured: quantity alone is not enough. Furthermore, even in EHR datasets claims data are often used to index the relevant records; that is, when researchers define case and noncase it is frequently by reference to the code providers assigned at the time of care. Prior experience with data-driven research suggests that claims data are particularly susceptible to under-reporting of non-primary complaints and delirium is often under-diagnosed and under-treated. (18, 43) If delirium is systematically underreported, data-driven work on this presentation will be limited; therefore, we sought to measure the rate of delirium in claims data and to compare this with the rate of delirium diagnosed in study settings. To do so we utilized a large state-wide claims database in concert with a literature review. Secondarily we utilized sociodemographic and comorbidity information to assess the extent to which identified cases demonstrated the expected risk factors. Materials and methods Data were drawn from the Massachusetts All-Payer Claims Database (APCD), encompassing all claims data paid for every state resident independent of insurance type; it included all public and private payers with the exception of Medicare. The American Reinvestment and Recovery Act (ARRA) created the APCD program as a data infrastructure

McCoy et al 5 to enable comparative research. Within the 2012 Massachusetts APCD, we analyzed the first hospital admission that occurred during the observation period for each hospitalized resident of the state. For patients admitted multiple times, subsequent encounters were excluded. A single, the first, hospitalizations was selected for ease of interpretation given the non-independence of subsequent hospitalizations and the desire to be most comparable with the literature values which are typically single encounter studies. The primary outcome – diagnosis of delirium – was defined using ICD-9 diagnoses as reported with claims. The sociodemographic and clinical features of this cohort were also abstracted from the same entries in the APCD. These features were at the intersection of those associated with delirium in prior research and those available in APCD. These variables included age in years, sex, the length of admission in days, and whether the admission occurred via an Emergency Department presentation. Additionally we identified admissions that were covered by public payers and calculated the age-adjusted Charlson Comorbidity Index, a composite metric of disease severity. (44) These sociodemographic and clinical features were collected to allow identified cases to be compared with non-cases as a face validity assessment of true case status. To identify the expected rate of delirium, we reviewed the biomedical literature using Pubmed for published rates of delirium in hospitalized patients, with particular attention paid to the setting of care, the sample, and the outcome. The Partners Institutional Review Board and the Institutional Review Board of the Massachusetts State Department of Health approved the health claims study and allowed a waiver of informed consent under 45 CFR 46.116, as only de-identified data was used. All analysis was performed using Python v2.7. (45, 46) Given the unbalanced class sizes, we selected kernel density estimates over histograms to display the relevant distributions. Statistical inferences on equality of means and of proportions across the delirium and no-delirium group were done as appropriate. Given the preeminent importance of age as a risk factor and the role of age > 65 years to health insurance

McCoy et al 6 coverage in the United States, we conducted a subgroup analysis (groups > 66 and < 66) parallel to the overall analysis. Results The rate of delirium cited in the literature varies widely depending on setting and methodology: from 3.6% in Thai surgical ICUs and 8.3% in North American Emergency Departments (EDs) to 73% in North American surgical ICUs (Table 1). Table 1 shows the marked variability in rate both across and within settings (ICU, ED, ward) and population (surgical non-surgical). When only studies using a CAM outcome are considered the collected literature median was 17% and the rate was mean 22%. In the state-wide claims data (APCD) we identified delirium in only 2.1% of index hospitalizations (Table 2). For APCD hospitalizations of patients > 65 years old delirium was coded in 2.8% of admissions, whereas for those < 65 years delirium was coded in 1.2% of hospital admissions (Table 3). When all admissions, as opposed to only index admissions, were considered in a secondary sensitivity analysis the overall rate of delirium was 2.2% (full results not shown). Characteristics of those with a diagnosis of delirium differed from those without this diagnosis. In both the pooled (Table 2) and age-grouped (Table 3) samples, those with a diagnosis of delirium tended to have longer hospitalizations, greater Charlson index scores, and to be older. Additionally, those with a delirium diagnosis were more likely to be privately insured, male, and to have presented to the index hospitalization via the ED (Table 2 and 3). Discussion Delirium carries significant morbidity and potential mortality risk; however, it is treatable and potentially preventable. Treatment is predicated on diagnosis and prevention is more effective in high-risk groups than is treatment. As such, recognition of and risk stratification for delirium is important for practitioners of psychosomatic medicine and for research. Large repositories of clinical data – for example, insurance claims – enable datadriven approaches to risk-factor identification and patient-level risk stratification.

McCoy et al 7 Application of these methods will be most effective if they occur in clinically-representative datasets. Here, we report evidence that delirium is likely to be under-reported in claims data, and may be under-reported by 10-fold or more (1.2-2.8% in APCD versus Table 1 values). This finding is consistent with prior comparison of claims and clinical data. (43) This under-reporting reduces the expected value of this otherwise rich source of physician experience. The extent to which delirium is under-reported in claims data as compared to studies focused on delirium raises a number of questions. The literature reviewed for this study made clear that delirium, as defined within explicit studies, is a clinically-significant outcome with a much higher rate than that seen in Massachusetts claims data. Inasmuch as claims data allow, we see similar trends towards clinical meaning – longer stays – and risk factors as is observed in the existing literature, with delirium occurring in older, sicker (by Charlson) patients that were admitted through the ED. These differences between cases and non-cases are those we would expect in true cases of delirium. That this appeal to face valid comorbidity, is the best possible control on false positive rate speaks to the limitations of claims data. The lower incidence of delirium in claims data may reflect a failure to diagnose, a failure to code, or a lower rate in community hospitals that are well represented in claims data and rarely the site of prospective research. It is possible that delirium is routinely diagnosed at or near the rates provided in the literature, but that this code is not submitted with the final resulting insurance claim. This diagnosis without coding could be an expected side effect of charge bundling or minimal financial incentive to make secondary or associated diagnosis. Alternatively it is possible that delirium is not routinely diagnosed in clinical practice and significant educational efforts are needed to detect and treat this condition. A growing literature suggests that under-diagnosis may be the norm in clinical practice .(19-21) Finally, it is possible that the rate observed in claims data may reflect dramatic differences between the average hospital within a state and the average hospital within a study of delirium. This possibility may bespeak a need to increase the representation of smaller community hospitals in the psychosomatic literature. More encouragingly, it is possible that significant headway has been made on the diagnosis and prevention of delirium such that the rate observed in Massachusetts

McCoy et al 8 hospitals (in 2012) was notably lower than that expected based on reported rates. This would be an encouraging outcome and remarkable accomplishment for the field of psychosomatic medicine. The inability to differentiate between these multiple possible explanations for the low rate of delirium observed in our current study speaks to both the strengths and weakness of the current research. Our study is strengthened by its use of a large and inclusive state-wide database. However, it is limited by its reliance on relativity superficial claims data, lack of access to Medicare data, and reliance on claims within a single state. (47-49) One particular limitation of note is the concern with reporting of substance use disorders to the APCD, in part due to concern of the specific federal regulations – 42 CFR Part 2 – governing the management of substance user records. (50) This limitation is difficult to model, is outside of the research process, and given the important role that substance use plays as a risk factor for delirium, a significant obstacle to claims research on delirium. Future data-driven research may be able to address this through the addition of EHR data and particularly natural language-processing approaches.(51) Hearteningly, if the lower-than-expected incidence of delirium in claims data is the result of under-diagnosis or under-coding, then the consultation psychiatrist has the capacity to make a significant contribution to both patient wellbeing and the future of data-driven research within the field by diagnosis and coding thoroughly. Although cases of delirium were less common than expected, those that were diagnosed deviated from those without a diagnosis in the expected fashion. Relative to patients not receiving a diagnosis of delirium, the delirious patient in the APCD tends to be older, sicker, admitted via the ED, and have a longer hospital length of stay (LOS). Of these, the association of delirium with advancing age is particularly striking, with both the overall cohort and the age subgroup cohorts showing delirious patients’ ages skewed to the right (Figure 1). The delirium literature typically differentiates between delirium in the ICU and in the non-ICU patient; here we approach this as a question of severity of illness through the age-adjusted Charlson co-morbidity index. This index, log transformed, shows a bimodal distribution, reflecting first, the sharp spike of those who are fundamentally well and incidentally hospitalized, and then a more normally distributed second hump of those who have a Charlson-detectable burden of disease. Those with delirium are

McCoy et al 9 disproportionately represented in the second hump, as would be expected based on prior literature (Figure 2). Finally, those with delirium remain inpatients slightly more than three days longer than those without delirium, consistent with what would be expected based on prior literature (Table 2). These differences between cases of coded delirium differ from their non-delirious counterparts in expected ways, lending credibility to these as bona fide cases. Conclusion The growing availability of large clinical data-sets and availability of computational resources sufficient to work with these data-sets have enabled a new avenue for psychiatric discovery: the data-driven study. Data from health insurance claims are an appealingly complete and structured source of such data. The Massachusetts All-Payer Claims Database has made these data available at the state level, allowing large and staterepresentative studies. Here we find that the rate of delirium, an important entity for consultation psychiatrists, is much lower in large datasets than would be expected based on existing delirium research. This marked discrepancy requires further investigation; as the cases that are identified appear superficially similar to cases of delirium described elsewhere in the literature and thus may be a new and valuable data resource if carefully interpreted.

Reference 1. Association AP. Diagnostic and statistical manual of mental disorders: DSM-5. Washington, D.C.: American Psychiatric Association; 2013. 2. Maldonado JR. Nauropathogenesis of delirium: review of current etiologic theories and common pathways. The American journal of Geriatric Psychiatry. 2013;21(12):1190222. 3. Pauley E, Lishmanov A, Schumann S, Gala GJ, Van Diepen S, Katz JN. Delirium is a robust predictor of morbidity and mortality among critically ill patients treated in the cardiac intensive care unit. American heart journal. 2015;170(1):79-86. e1. 4. Salluh JI, Wang H, Schneider EB, Nagaraja N, Yenokyan G, Damluji A, et al. Outcome of delirium in critically ill patients: systematic review and meta-analysis. bmj. 2015;350:h2538.

McCoy et al 10 5. Francis J, Martin D, Kapoor WN. A prospective study of delirium in hospitalized elderly. Jama. 1990;263(8):1097-101. 6. Weinrebe W, Johannsdottir E, Karaman M, Füsgen I. What does delirium cost? Zeitschrift für Gerontologie und Geriatrie. 2016;49(1):52-8. 7. Curyto KJ, Johnson J, TenHave T, Mossey J, Knott K, Katz IR. Survival of hospitalized elderly patients with delirium: a prospective study. The American Journal of Geriatric Psychiatry. 2001;9(2):141-7. 8. Pisani MA, Kong SYJ, Kasl SV, Murphy TE, Araujo KL, Van Ness PH. Days of delirium are associated with 1-year mortality in an older intensive care unit population. American journal of respiratory and critical care medicine. 2009;180(11):1092-7. 9. O'Keeffe S, Lavan J. The prognostic significance of delirium in older hospital patients. Journal of the American Geriatrics Society. 1997;45(2):174-8. 10. McCusker J, Cole M, Dendukuri N, Belzile É, Primeau F. Delirium in older medical inpatients and subsequent cognitive and functional status: a prospective study. Canadian Medical Association Journal. 2001;165(5):575-83. 11. Leslie DL, Inouye SK. The importance of delirium: economic and societal costs. Journal of the American Geriatrics Society. 2011;59(s2):S241-S3. 12. Trogrlić Z, van der Jagt M, Bakker J, Balas MC, Ely EW, van der Voort PH, et al. A systematic review of implementation strategies for assessment, prevention, and management of ICU delirium and their effect on clinical outcomes. Critical Care. 2015;19(1):1-17. 13. Mu JL, Lee A, Joynt GM. Pharmacologic Agents for the Prevention and Treatment of Delirium in Patients Undergoing Cardiac Surgery: Systematic Review and Metaanalysis*. Critical care medicine. 2015;43(1):194-204. 14. Hshieh TT, Yue J, Oh E, Puelle M, Dowal S, Travison T, et al. Effectiveness of multicomponent nonpharmacological delirium interventions: a meta-analysis. JAMA internal medicine. 2015;175(4):512-20. 15. Fok MC, Sepehry AA, Frisch L, Sztramko R, Borger van der Burg BL, Vochteloo AJ, et al. Do antipsychotics prevent postoperative delirium? A systematic review and meta‐ analysis. International journal of geriatric psychiatry. 2015;30(4):333-44. 16. Breitbart W, Alici Y. Evidence-based treatment of delirium in patients with cancer. Journal of Clinical Oncology. 2012;30(11):1206-14. 17. Larsen KA, Kelly SE, Stern TA, Bode RH, Price LL, Hunter DJ, et al. Administration of olanzapine to prevent postoperative delirium in elderly joint-replacement patients: a randomized, controlled trial. Psychosomatics. 2010;51(5):409-18. 18. Glick REL, Sanders KM, Stern TA. Failure to record delirium as a complication of intra-aortic balloon pump treatment: a retrospective study. Journal of geriatric psychiatry and neurology. 1996;9(2):97-9. 19. Han JH, Wilber ST. Altered mental status in older patients in the emergency department. Clinics in geriatric medicine. 2013;29(1):101-36. 20. Han JH, Zimmerman EE, Cutler N, Schnelle J, Morandi A, Dittus RS, et al. Delirium in older emergency department patients: recognition, risk factors, and psychomotor subtypes. Academic Emergency Medicine. 2009;16(3):193-200.

McCoy et al 11 21. Kishi Y, Kato M, Okuyama T, Hosaka T, Mikami K, Meller W, et al. Delirium: patient characteristics that predict a missed diagnosis at psychiatric consultation. General hospital psychiatry. 2007;29(5):442-5. 22. Sri-on J, Tirrell GP, Vanichkulbodee A, Niruntarai S, Liu SW. The prevalence, risk factors and short-term outcomes of delirium in Thai elderly emergency department patients. Emergency Medicine Journal. 2015:emermed-2014-204379. 23. Kakuma R, Fort D, Galbaud G, Arsenault L, Perrault A, Platt RW, et al. Delirium in older emergency department patients discharged home: effect on survival. Journal of the American Geriatrics Society. 2003;51(4):443-50. 24. Schuurmans MJ, Duursma SA, Shortridge‐Baggett LM. Early recognition of delirium: review of the literature. Journal of clinical nursing. 2001;10(6):721-9. 25. Elie M, Cole MG, Primeau FJ, Bellavance F. Delirium risk factors in elderly hospitalized patients. Journal of general internal medicine. 1998;13(3):204-12. 26. Meenen LC, Meenen DM, Rooij SE, Riet G. Risk Prediction Models for Postoperative Delirium: A Systematic Review and Meta‐Analysis. Journal of the American Geriatrics Society. 2014;62(12):2383-90. 27. Zaal IJ, Devlin JW, Peelen LM, Slooter AJ. A systematic review of risk factors for delirium in the ICU*. Critical care medicine. 2015;43(1):40-7. 28. Wolters AE, Zaal IJ, Veldhuijzen DS, Cremer OL, Devlin JW, van Dijk D, et al. Anticholinergic medication use and transition to delirium in critically ill patients: A prospective cohort study. Critical care medicine. 2015;43(9):1846-52. 29. Pandharipande P, Shintani A, Peterson J, Pun BT, Wilkinson GR, Dittus RS, et al. Lorazepam is an independent risk factor for transitioning to delirium in intensive care unit patients. The Journal of the American Society of Anesthesiologists. 2006;104(1):21-6. 30. Van den Boogaard M, Pickkers P, Slooter A, Kuiper M, Spronk P, Van Der Voort P, et al. Development and validation of PRE-DELIRIC (PREdiction of DELIRium in ICu patients) delirium prediction model for intensive care patients: observational multicentre study. Bmj. 2012;344:e420. 31. Castro VM, McCoy TH, Cagan A, Rosenfield HR, Murphy SN, Churchill SE, et al. Stratification of risk for hospital admissions for injury related to fall: cohort study. BMJ. 2014;349:g5863. 32. McCoy TH, Castro VM, Rosenfield HR, Cagan A, Kohane IS, Perlis RH. A clinical perspective on the relevance of research domain criteria in electronic health records. American Journal of Psychiatry. 2015. 33. McCoy T, Castro V, Cagan A, Roberson A, Kohane I, Perlis R. Sentiment Measured in Hospital Discharge Notes Is Associated with Readmission and Mortality Risk: An Electronic Health Record Study. PloS one. 2015. 34. Reis BY, Kohane IS, Mandl KD. Longitudinal histories as predictors of future diagnoses of domestic abuse: modelling study. Bmj. 2009;339:b3677. 35. Blumenthal SR, Castro VM, Clements CC, Rosenfield HR, Murphy SN, Fava M, et al. An electronic health records study of long-term weight gain following antidepressant use. JAMA psychiatry. 2014;71(8):889-96. 36. Yu S, Liao KP, Shaw SY, Gainer VS, Churchill SE, Szolovits P, et al. Toward highthroughput phenotyping: unbiased automated feature extraction and selection from

McCoy et al 12 knowledge sources. Journal of the American Medical Informatics Association. 2015;22(5):993-1000. 37. Tatonetti NP, Patrick PY, Daneshjou R, Altman RB. Data-driven prediction of drug effects and interactions. Science translational medicine. 2012;4(125):125ra31-ra31. 38. Rudolph JL, Doherty K, Kelly B, Driver JA, Archambault E. Validation of a Delirium Risk Assessment Using Electronic Medical Record Information. Journal of the American Medical Directors Association. 2015. 39. Blumenthal D, Tavenner M. The “meaningful use” regulation for electronic health records. New England Journal of Medicine. 2010;363(6):501-4. 40. Murdoch TB, Detsky AS. The inevitable application of big data to health care. Jama. 2013;309(13):1351-2. 41. Sittig DF, Teich JM, Osheroff JA, Singh H. Improving clinical quality indicators through electronic health records: it takes more than just a reminder. Pediatrics. 2009;124(1):375-7. 42. Sittig DF, Singh H. Electronic health records and national patient-safety goals. New England Journal of Medicine. 2012;367(19):1854-60. 43. Katznelson R, Djaiani G, Tait G, Wasowicz M, Sutherland AM, Styra R, et al. Hospital administrative database underestimates delirium rate after cardiac surgery. Canadian Journal of Anesthesia/Journal canadien d'anesthésie. 2010;57(10):898-902. 44. Charlson M, Szatrowski TP, Peterson J, Gold J. Validation of a combined comorbidity index. J Clin Epidemiol. 1994;47(11):1245-51. 45. Oliphant TE. Python for scientific computing. Computing in Science & Engineering. 2007;9(3):10-20. 46. Hunter JD. Matplotlib: A 2D graphics environment. Computing in science and engineering. 2007;9(3):90-5. 47. Ferver K, Burton B, Jesilow P. The use of claims data in healthcare research. Open Public Health J. 2009;2:11-24. 48. Strom BL. Data validity issues in using claims data. Pharmacoepidemiology and drug safety. 2001;10(5):389-92. 49. Fisher ES, Whaley FS, Krushat WM, Malenka DJ, Fleming C, Baron JA, et al. The accuracy of Medicare's hospital claims data: progress has been made, but problems remain. American journal of public health. 1992;82(2):243-8. 50. Wattenberg SA, MSW. Frequently Asked Questions - Applying the substance abuse confidentiality Regulations to Health Information Exchange (HIE) Legal Action Center for the Substance Abuse and Mental Health Services Administration: U.S. Department of Health and Human Services. 51. Perlis R, Iosifescu D, Castro V, Murphy S, Gainer V, Minnier J, et al. Using electronic medical records to enable large-scale studies in psychiatry: treatment resistant depression as a model. Psychological medicine. 2012;42(01):41-50. 52. McPherson JA, Wagner CE, Boehm LM, Hall JD, Johnson DC, Miller LR, et al. Delirium in the cardiovascular intensive care unit: exploring modifiable risk factors. Critical care medicine. 2013;41(2):405. 53. Bamalwa M, Mahmood S, Praharaj S. Delirium in cardiac ICU patients. Annals of clinical psychiatry: official journal of the American Academy of Clinical Psychiatrists. 2016;28(1):51-5.

McCoy et al 13 54. Maldonado JR, Wysong A, van der Starre PJ, Block T, Miller C, Reitz BA. Dexmedetomidine and the reduction of postoperative delirium after cardiac surgery. Psychosomatics. 2009;50(3):206-17. 55. Naughton BJ, Moran MB, Kadah H, Heman-Ackah Y, Longano J. Delirium and other cognitive impairment in older adults in an emergency department. Annals of emergency medicine. 1995;25(6):751-5. 56. Han JH, Eden S, Shintani A, Morandi A, Schnelle J, Dittus RS, et al. Delirium in older emergency department patients is an independent predictor of hospital length of stay. Academic Emergency Medicine. 2011;18(5):451-7. 57. Van de Meeberg EK, Festen S, Kwant M, Georg RR, Izaks GJ, Ter Maaten JC. Improved detection of delirium, implementation and validation of the CAM-ICU in elderly Emergency Department patients. European journal of emergency medicine: official journal of the European Society for Emergency Medicine. 2016. 58. Élie M, Rousseau F, Cole M, Primeau F, McCusker J, Bellavance F. Prevalence and detection of delirium in elderly emergency department patients. Canadian Medical Association Journal. 2000;163(8):977-81. 59. Kennedy M, Enander RA, Tadiri SP, Wolfe RE, Shapiro NI, Marcantonio ER. Delirium risk prediction, healthcare use and mortality of elderly adults in the emergency department. Journal of the American Geriatrics Society. 2014;62(3):462-9. 60. Simons KS, Laheij RJ, van den Boogaard M, Moviat MA, Paling AJ, Polderman FN, et al. Dynamic light application therapy to reduce the incidence and duration of delirium in intensive-care patients: a randomised controlled trial. The Lancet Respiratory Medicine. 2016;4(3):194-202. 61. Ouimet S, Kavanagh BP, Gottfried SB, Skrobik Y. Incidence, risk factors and consequences of ICU delirium. Intensive care medicine. 2007;33(1):66-73. 62. Pipanmekaporn T, Chittawatanarat K, Chaiwat O, Thawitsri T, Wacharasint P, Kongsayreepong S. Incidence and risk factors of delirium in multi-center Thai surgical intensive care units: a prospective cohort study. Journal of intensive care. 2015;3(1):1. 63. Carin-Levy G, Mead G, Nicol K, Rush R, van Wijck F. Delirium in acute stroke: screening tools, incidence rates and predictors: a systematic review. Journal of neurology. 2012;259(8):1590-9. 64. Bruce AJ, Ritchie CW, Blizard R, Lai R, Raven P. The incidence of delirium associated with orthopedic surgery: a meta-analytic review. International Psychogeriatrics. 2007;19(02):197-214. 65. Pandharipande P, Cotton BA, Shintani A, Thompson J, Pun BT, Morris Jr JA, et al. Prevalence and risk factors for development of delirium in surgical and trauma ICU patients. The Journal of trauma. 2008;65(1):34. 66. Ryan DJ, O'Regan NA, Caoimh RÓ, Clare J, O'Connor M, Leonard M, et al. Delirium in an adult acute hospital population: predictors, prevalence and detection. BMJ open. 2013;3(1):e001772. 67. Inouye SK, Rushing JT, Foreman MD, Palmer RM, Pompei P. Does delirium contribute to poor hospital outcomes? Journal of general internal medicine. 1998;13(4):234-42.

McCoy et al 14

Table 1: Published rates of delirium Reported Rate

Sample Size

Diagnosis By

Setting

26%

200

CAM

Cardiac ICU

20.30%

590

CAM

Cardiac ICU

Bamalwa et al (53)

16%

50

CAM

Cardiac ICU

Maldonado et al (54)

34%

90

Clinical

Cardiac Surgery

Larsen et al (17)

27.50%

400

Clinical

Elective Surgery

Naughton et al (55)

39.90%

188

CAM

ER

Han et al (56)

17%

628

CAM

ER

Sri-on et al (22)

12%

232

CAM

ER

Van de Meeberg et al (57)

10%

478

CAM

ER

9.60%

447

CAM

ER

9%

700

CAM

ER

8.30%

303

CAM

ER

Simons et al (60)

35%

734

CAM

ICU

Ouimet et al (61)

31.80%

764

ICDSC

ICU

Pipanmekaporn et al (62)

3.60%

4450

ICDSC

ICU

Carin-Levy et al (63)

26%

2979 (from 20 studies)

CAM

Meta Analysis

4-53.3% in trauma; 3.6-28.3% in elective cases

26 publications

Mixed

Meta Analysis

73%

100 (46 SICU and 54 TICU)

CAM

Surgical ICU

17.60%

280

CAM

Wards

12%

727

CAM

Wards

Study McPherson et al (52) Pauley et al (3)

Elie et al (58) Kennedy et al (59) Han et al (20)

Bruce et al (64) Pandharipande et al (65) Ryan et al (66) Inouye et al (67)

McCoy et al 15 Table 2: Sociodemographics of delirium Delirium at No (n=224357) Admission Count % Male 96813.00 43 Private Insurance 68694.00 31 Emergency 107579.00 48 Department Admission Mean SD Age-adjusted 1.27 0.99 Charlson Comorbidity Index Age at admission 62.27 15.13 (years) Length of Stay 4.61 6.38

Yes (n=4733) Count 2336.00 1727.00 2883.00

% 49 36 61

X2 72.42 74.75 311.43

p <.001 <.001 <.001

Mean 1.44

SD 1.03

t 11.63

p <001

69.62

14.40

31.92

<.001

8.00

9.12

47.74

<.001

Table 3: Sociodemographics of delirium contrasted by age greater than or less than 65 years. Delirium at Admission Male Private Insurance Emergency Department Admission Age-adjusted Charlson Comorbidity Index Age at admission (years) Length of Stay

66 and Older No (n=89572) Yes (n=2491)

65 and Younger No (n=121003) Yes (n=1502)

Count 38519.00 18233.00

% 43 20

Count 1152.00 671.00

% 46 27

Count 54378.00 48038.00

% 45 40

Count 962.00 904.00

% 64 60

51328.00

57

1507.00

60

45941.00

38

856.00

57

Mean 1.76

SD 0.62

Mean 1.87

SD 0.64

Mean 1.05

SD 1.05

Mean 1.44

SD 1.14

77.16

6.86

79.31

6.77

51.25

8.85

53.55

7.79

4.82

6.07

8.08

9.36

4.44

6.70

8.35

8.45

McCoy et al 16 Figure 1: Distribution of patient ages at the time of admission

McCoy et al 17

Figure 2: Distribution of patient Charlson scores at the time of admission