- Question: What is the relationship between preoperative antidepressant and antianxiety medication and postoperative hospital length of stay?
- Findings: Preoperative antidepressant and antianxiety medications are associated with an increased postoperative hospital length of stay.
- Meaning: Either due to underlying psychiatric disease or medication effects, patients on preoperative antidepressant and antianxiety medication stay in the hospital longer after surgery and may require greater attention to hasten recovery.
As the incidence of antidepressant and anxiolytic medication use continues to rise among the general population, the effects of these medications, or the syndromes they represent, on perioperative outcomes and health care resource utilization have become increasingly significant. It is estimated that between 1999 and 2012, the use of antidepressant medications among adults in the United States has nearly doubled from 6.8% to 13%, while the use of anxiolytic medications has also increased from 4.1% to 6.1% in that timeframe.1 While studies and reports that do exist suggest the continuation of antidepressants in the perioperative period,2 , 3 because more patients maintained on these medications present for surgical procedures, the benefits and risks of their continuation through the perioperative period must be further identified.
The management of patients on psychoactive medications in the perioperative period has largely relied on clinician experience, with few specific guidelines outlining the perioperative management of these medications. Not only may these drugs have significant anesthetic implications through altered patient physiology and interaction with other medications, but the nature of the psychiatric condition itself may pose challenges to the anesthesiologist and individuals involved in the patient’s care. While several studies have attempted to investigate the perioperative effects of antidepressant and anxiolytic medications, there remains a paucity of literature examining their effect on hospital length of stay outside of cardiovascular surgery.4 , 5 A number of recent studies have also demonstrated that clinically significant depression and anxiety are associated with worse surgical outcomes and increased length of stay; however, these studies are often limited in their scope to a specific surgical field or procedure.6–8 Whereas the use of the medications may not be modifiable, the underlying psychiatric diseases have the potential to be optimized similarly to other medical conditions.
The primary objective of this study was to examine the relationship between antidepressant and anxiolytic medication use during the perioperative period and hospital length of stay following noncardiac surgical procedures. We hypothesize that the use of these medications is associated with a longer postoperative hospital length of stay.
Patient data obtained for this study did not contain identifying patient information and involved no direct interaction with human subjects. This study was thus exempt from our institutional review board process, documentation for which is kept on file.
This study is a retrospective analysis of an administrative database of surgical patients from January 2011 to December 2014, which included age, sex, surgical service, American Society of Anesthesiologists (ASA) score, body mass index, preoperative hemoglobin, and emergency status. Data were obtained from our institutional electronic medical record (Epic, Verona, WI) for the purpose of performing this analysis. Presence of medical comorbidities (including congestive heart failure, atrial fibrillation, chronic obstructive pulmonary disease, dementia, diabetes mellitus, human immunodeficiency virus, hypertension, liver disease, coronary artery disease, chronic kidney disease, cancer, alcohol abuse, peripheral vascular disease, and hypothyroidism) was determined by presence of diagnosis in the electronic health record by International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code at the time of surgery. Comorbidities for inclusion were selected based on presence within the Charlson and Elixhauser comorbidity indices, which have been associated with postoperative length of stay,9 , 10 and were defined using previously validated coding algorithms for ICD-9-CM comorbidity codes.
Also included was whether the patient was taking medications prescribed specifically for treatment of depression (including α-2 receptor antagonists [mirtazapine only], monoamine oxidase inhibitors, norepinephrine and dopamine reuptake inhibitors, selective serotonin reuptake inhibitors, serotonin–norepinephrine reuptake inhibitors, selective serotonin reuptake inhibitors and 5-hydroxytryptamine-1A partial agonists, and tricyclic antidepressants) or medications prescribed specifically for treatment of anxiety (including benzodiazepines, meprobamate, prazosin, and buspirone) on the day of the surgery. At our institution, all antidepressant and antianxiety medications are currently recommended to be continued up to and including the day of surgery.
Only adult patients and noncardiac procedures were included. Procedures canceled on the day of surgery, and those with obviously erroneous data were not included. Patients who died in the hospital were excluded from analysis. Cases with missing hemoglobin data were imputed using multiple imputation with 5 iterations, which is considered adequate for the proportion of missing data. Patient demographics are outlined in Table 1.
The outcome of interest was postoperative hospital length of stay, in which the day of surgery was considered day zero, and each day thereafter as 1 postoperative day. Expected length of stay was determined by proprietary procedure-specific University Health Consortium (UHC) models utilized at our institution. The UHC risk-adjusted calculation for expected length of stay is created from the data contributed by participating academic institutions. The relative weight assigned to each variable in the UHC length of stay calculation reflects the aggregate data from all hospitals. The UHC risk-adjusted model for length of stay is proprietary; each of the 300+ models is specific to a certain Medicare Severity Diagnosis Related Groups and changes from year to year.
All descriptive values are presented as mean with standard deviation, median with interquartile range, or frequency and percentage as deemed appropriate. Normality was assessed using qq plots and histograms. Comparisons between normal continuous data were performed using t test or 1-way analysis of variance, between non-normal continuous data using Mann–Whitney or Kruskal–Wallis test, and those between categorical data using χ2 or Fisher exact test.
Missing data that were considered missing at random were treated with multiple imputation, with number of iterations dictated by proportion of missing data, using the aregImpute function in the Hmisc package in the R statistical software (version 3.1.1; R Foundation for Statistical Computing, Vienna, Austria).
We assessed the association between the primary outcome of length of stay and the exposures of presence of antidepressant and antianxiety medication (and their interaction) in a multivariable negative binomial model adjusting for age, sex, ASA score, body mass index, hemoglobin, surgery type, emergency status, expected length of stay, and the previously mentioned medical comorbidities as independent covariates. A negative binomial model, which does not permit zero counts, was appropriate because only inpatient procedures with a positive integer value for length of stay were included (ie, no ambulatory surgeries). In an additional analysis, we assessed the interaction between expected length of stay and presence of antidepressant or antianxiety medication in 2 separate multivariable negative binomial models with actual length of stay as the outcome and adjusting for the same covariates as above. A third analysis was performed with each category of antidepressant and antianxiety medication included as a binomial independent variable, but otherwise the same as the primary analysis (a multivariable negative binomial model with length of stay as the outcome). Incidence rate ratios (IRRs) were calculated by coefficient exponentiation for the exposure variable of interest, along with 99% confidence intervals (CIs). IRR values >1 indicate an increase in risk of incurring additional time in the hospital. Independent factors were assessed univariately and included in the multivariable analyses if the univariable P value was <0.2. Ultimately, all of the above variables were included within the analysis. We assessed model covariates for collinearity and assessed for removal variables with variance inflation factors >4. Continuous data that were examined and that could not be assumed to be linear in relation to the outcome variable were treated with restricted cubic splines with 3 knots in the regression analyses. Knots were chosen by plotting the relationship between the given variable and the outcome variable and determination of the apparent inflection points by visual inspection. Sample size was based on the sample made available by the data from the electronic medical record over the given time period and was large enough to adequately satisfy requirements for stable regression analyses. Our sample size of over 48,000 cases would have >90% power to detect a clinically relevant IRR of >1.02 between the exposure and the outcome with a significance of 0.01. A P value of 0.01 was set for significance due to multiple comparisons. All statistical operations were performed using the R statistical software (v. 3.1.1; The R Foundation for Statistical Computing).
Propensity Score Matching Sensitivity Analysis
Propensity score matching was performed as a sensitivity analysis on the primary outcome of hospital length of stay. The Matching package for the R statistical software was utilized for propensity score matching and assessment of postmatch balance. Propensity scores for preoperative antidepressant or antianxiety medications were determined using 2 separate logistic regression models based on preoperative medical comorbidities, including coronary artery disease, history of stroke, diabetes mellitus, hypertension, atrial fibrillation, obesity, dementia, cancer, human immunodeficiency virus, and chronic obstructive pulmonary disease; BMI; age; ASA score; surgical service; admission type; and gender. One to one nearest neighbor propensity score matching without replacement using Mahalanobis distance metric weighting and a caliper width of 0.2 SDs of the logit of the propensity score was performed based on the derived propensity scores, using antidepressant or antianxiety medication as the treatment variable, and estimating the average treatment effect on the treated. Match balance was considered adequate for standardized mean difference <0.1, and by visualizing qq plots when applicable for continuous variables. Two univariate negative binomial regression analyses were performed on the matched sets using antidepressant or antianxiety medication each as the independent variables and length of stay as the dependent variable.
To assess the relationship between preoperative antidepressant and antianxiety medications on other postoperative outcomes, 2 additional outcomes were assessed on the matched sets: venous thromboembolism (VTE), and major adverse cardiac events (new congestive heart failure, new atrial fibrillation, or myocardial infarction), as these have been reported to be associated with antianxiety medication.11 These outcomes were designated by presence of the corresponding ICD-9 codes during that admission, as well as lack of a “present on admission” flag, denoting that the diagnosis was made during the admission and was not a preexisting comorbidity.
Each outcome was assessed by univariate logistic regression analysis on the matched sets using antidepressant or antianxiety medication each as the independent variable in the respective set.
The deidentified database contained 56,393 unique surgical encounters. After exclusion criteria were applied, 48,435 patients remained (Figure 1). There were 5111 (10.5%) patients on antidepressant medications and 4912 (10.1%) patients on antianxiety medications. Patient characteristics for full cohort and separated by antidepressant or antianxiety medications use are detailed in Tables 1 and 2, respectively. There were 1273 (2.6%) cases with missing hemoglobin data that were imputed. No other data were missing. Examination of the data revealed no important differences between complete cases and those with missing data. The median length of stay for all patients was 3 days (interquartile range [IQR] = 2–6). The median length of stay was 4 days (IQR = 2–7) for patients on antidepressant medication and 4 days (IQR = 2–6) for patients on antianxiety medication.
The multivariable risk-adjusted analysis demonstrated a significant relationship between each of preoperative antidepressant (IRR = 1.04; 99% CI, 1.0–1.08; P = .006) and antianxiety medication (IRR = 1.1; 99% CI, 1.06–1.14; P < .001) and length of stay (Supplemental Digital Content, Table 1, http://links.lww.com/AA/C628). An interaction term between antidepressant and antianxiety medications was not significant (P = .99) and ultimately not retained in the analysis. We further found a significant interaction between antidepressant medication (yes/no) and expected length of stay on the outcome of length of stay (P < .001), which suggests that the association between antidepressant medication and length of stay was stronger for increasing expected length of stay. We did not find an interaction between antianxiety medication (yes/no) and expected length of stay (P = .12) on the same outcome. The results of the analysis in which each category of antidepressant and antianxiety medication was included separately are presented in Table 3. No variance inflation factors were >4, suggesting negligible collinearity.
Propensity Score Matching Sensitivity Analysis
After propensity score matching for antidepressant use, there were 3962 patients within each cohort. The balance of the covariates between cohorts improved to acceptable levels after propensity score matching (Figure 2; Supplemental Digital Content, Tables 2–3, http://links.lww.com/AA/C628). Presence of preoperative antidepressants was associated with an IRR of 1.45 (99% CI, 1.31–1.61; P < .001) for hospital length of stay. After propensity score matching for antianxiety use, there were 3929 patients within each cohort. The balance of the covariates between cohorts improved to acceptable levels after propensity score matching (Figure 3; Supplemental Digital Content, Tables 4–5, http://links.lww.com/AA/C628). Presence of preoperative antianxiety medications was associated with an IRR of 1.15 (99% CI, 1.06–1.24; P < .001) for hospital length of stay.
In the full dataset, 274 (0.57%) patients developed VTE, and 1083 (2.2%) patients had an adverse cardiac event. Within the antidepressant medication matched set, 100 (1.3%) patients developed VTE, and 374 (4.7%) patients developed an adverse cardiac event. Of patients on antidepressants, 54 (1.4%) developed VTE and 191 (4.8%) had an adverse cardiac event. Antidepressant medication was not associated with VTE (OR = 1.2; 99% CI, 0.70–2.0; P = .42) or adverse cardiac events (OR = 1.0; 99% CI, 0.85–1.3; P = .67). Within the antianxiety medication matched set, 100 (1.3%) patients developed VTE, and 293 (3.7%) patients had an adverse cardiac event. Of patients on antianxiety medication, 47 (1.2%) developed VTE and 169 (4.3%) had an adverse cardiac event. Antianxiety medication was not associated VTE (OR = 0.89; 99% CI, 0.59–1.3; P = .55), but was associated with adverse cardiac events (OR = 1.4; 99% CI, 1.1–1.7; P = .008).
This study represents a large patient cohort examining the effects of antidepressant and antianxiety medications on hospital length of stay, with our results demonstrating a statistically significant relationship between both. After controlling for confounding variables, we found that the use of preoperative antidepressant medications was associated with an IRR of 1.04 (99% CI, 1.00–1.08) and antianxiety medication with an IRR of 1.1 (99% CI, 1.06–1.14) for postoperative hospital length of stay, suggesting a longer length of stay in patients with preoperative use of these medications.
While there are currently no published guidelines by the ASA with regards to the continuation of antidepressant and antianxiety medications through the perioperative period, it is generally assumed that the benefits of continuing these medications outweigh the risks. A number of studies have been conducted to determine the safety of continuing these medications with conflicting evidence. Some studies have reported an increased incidence of adverse events, including bleeding and mortality, in patients taking antidepressant medications and undergoing surgical procedures,12–14 while others report a low incidence of intraoperative hemodynamic events.15 Furthermore, the use of antidepressant medications has also suggested benefit with regards to postoperative pain scores,16 postoperative nausea and vomiting,17 and recovery time,18 although research continues to be performed to further evaluate these relationships and the contexts in which they are applicable as data seem to be conflicting at this time.19–21 Intraoperative benzodiazepine administration, specifically midazolam, has been associated with a reduction in postoperative nausea and vomiting.22 , 23 It is unclear if the effects of benzodiazepines extend to patients chronically taking long-acting benzodiazepines, although 1 study suggests that preoperative lorazepam, a long-acting oral benzodiazepine, resulted in reduced pain after hysterectomy.24
There have also been a number of prior studies examining the effects of depression on hospital length of stay in cardiac surgery,25 breast surgery,26 total hip and knee arthroplasty,27 and even as an independent risk factor across all procedures.4 Similarly, antianxiety medications have also shown to be associated with increased morbidity or mortality in patients undergoing surgical procedures, which is corroborated by the results of our secondary analysis, in which we report an association between antianxiety medications and major adverse cardiac events.11 While these studies bring to light an important consideration in identifying patients at risk for prolonged hospitalization, further statistical analysis of their conclusions is limited by their small sample sizes. One other study examining the association between preoperative antidepressant use and hospital length of stay determined no significant relationship after evaluating approximately 3600 patients; however, the authors failed to perform a multivariable analysis on their cohort and examined the effects between 2 groups that were significantly different.5 Results of our study may also differ from those of other studies due to differences in which antidepressant and antianxiety medications were included for analysis. The results of our study support those of a study examining the relationship between psychiatric medication and prolonged length of stay in patients undergoing total hip and knee arthroplasty, although that study was exclusive of antianxiety medication and inclusive of antipsychotic medication, utilized a different measure for prolonged length of stay, and included only a specific perioperative population as opposed to the broad population of the current study.27 They are also supported in a study of cardiac surgery patients in which a relationship was seen between selective serotonin reuptake inhibitors and serotonin–norepinephrine reuptake inhibitors and increased length of stay, though that study did not show an association with bleeding.28
Although this study demonstrates a significant correlation between the use of antidepressant and antianxiety medications and increased hospital length of stay, there exist several important limitations that limit the strength of conclusions which can be reached. A major limitation of this type of study is missing confounders that may affect the results, in this case the lack of evaluation of analgesic consumption or nausea/vomiting in the postoperative period. As with many retrospective cohort studies, the ability to assess a temporal relationship between variables and outcomes is difficult, and thus, it is not possible to conclude a causal relationship from the observed correlation. Furthermore, there is substantial logistical difficulty designing and implementing large prospective trials to further assess this relationship. With many guidelines recommending continuation of antidepressant and antianxiety medications through the perioperative period due to the risk of both physiologic and psychiatric withdrawal, randomized clinical trials addressing this question are unlikely to be seen in the near future. For this reason, also, the ability to further identify modifiable risk factors for increased length of stay in the patient population is rather limited.
Within an enhanced recovery protocol, each individual component may not significantly affect outcomes such as length of stay in the hospital, but ideally the combination of factors has a positive impact. Greater attention to psychiatric well-being may be a powerful tool within this context, as it is potentially modifiable as opposed to fixed factors such as age, sex, or surgical procedure. Finally, other weaknesses to our study include the potential influence of unmeasured confounding variables and the reliance on accurate documentation from a number of different providers.
The current study suggests that reasons for prolonged length of stay in patients on antidepressant or antianxiety medications need to be further examined for potential avenues of perioperative optimization and modification for hastening recovery, which may be in the form of preoperative counseling, postoperative psychiatric consults, or holistic recovery approaches. These are potential targets for inclusion within enhanced recovery surgery protocols. This relationship may be due to medication effects, although considering the conflicting evidence this is questionable at this time, or as a reflection of underlying psychiatric illness or well-being, which can be particularly volatile during stressful events such as surgery. This also may be an influence in the relationship between antianxiety medications and adverse cardiac events. The presence of antidepressants or antianxiety medications is easily identifiable within the electronic medical record even before a preoperative evaluation clinic visit, during which these avenues may allow for psychiatric optimization in addition to medical optimization.
The authors would like to thank the NYU Department of Anesthesiology, Perioperative Care & Pain Medicine for supporting this work.
Name: Rishi Vashishta, MD.
Contribution: This author helped design the study, collect and analyze the data, and prepare the manuscript.
Name: Samir M. Kendale, MD.
Contribution: This author helped design the study, collect and analyze the data, and prepare the manuscript.
This manuscript was handled by: Ken B. Johnson, MD.
1. Kantor ED, Rehm CD, Haas JS, Chan AT, Giovannucci EL. Trends in prescription drug use among adults in the United States from 1999–2012. JAMA. 2015;314:1818–1831.
2. Kudoh A, Katagai H, Takazawa T. Antidepressant treatment for chronic depressed patients should not be discontinued prior to anesthesia. Can J Anaesth. 2002;49:132–136.
3. Sprung J, Distel D, Grass J, Bloomfield EL, Lavery IC. Cardiovascular collapse during anesthesia in a patient with preoperatively discontinued chronic MAO inhibitor therapy. J Clin Anesth. 1996;8:662–665.
4. Kerper LF, Spies CD, Buspavanich P. Preoperative depression and hospital length of stay in surgical patients. Minerva Anestesiol. 2014;80:984–991.
5. Sutherland AM, Katznelson R, Clarke HA, Tait G, Beattie WS. Use of preoperative antidepressants is not associated with postoperative hospital length of stay. Can J Anaesth. 2014;61:27–31.
6. McKenzie LH, Simpson J, Stewart M. A systematic review of pre-operative predictors of post-operative depression and anxiety in individuals who have undergone coronary artery bypass graft surgery. Psychol Health Med. 2010;15:74–93.
7. Balentine CJ, Hermosillo-Rodriguez J, Robinson CN, Berger DH, Naik AD. Depression is associated with prolonged and complicated recovery following colorectal surgery. J Gastrointest Surg. 2011;15:1712–1717.
8. Linnen H, Krampe H, Neumann T. Depression and essential health risk factors in surgical patients in the preoperative anaesthesiological assessment clinic. Eur J Anaesthesiol. 2011;28:733–741.
9. Kaplan GG, Hubbard J, Panaccione R. Risk of comorbidities on postoperative outcomes in patients with inflammatory bowel disease. Arch Surg. 2011;146:959–964.
10. Potretzke AM, Kim EH, Knight BA. Patient comorbidity predicts hospital length of stay after robot-assisted prostatectomy. J Robot Surg. 2016;10:151–156.
11. Ward N, Roth JS, Lester CC, Mutiso L, Lommel KM, Davenport DL. Anxiolytic medication is an independent risk factor for 30-day morbidity or mortality after surgery. Surgery. 2015;158:420–427.
12. Auerbach AD, Vittinghoff E, Maselli J, Pekow PS, Young JQ, Lindenauer PK. Perioperative use of selective serotonin reuptake inhibitors and risks for adverse outcomes of surgery. JAMA Intern Med. 2013;173:1075–1081.
13. Mahdanian AA, Rej S, Bacon SL, Ozdin D, Lavoie KL, Looper K. Serotonergic antidepressants and perioperative bleeding risk: a systematic review. Expert Opin Drug Saf. 2014;13:695–704.
14. Dall M, Primdahl A, Damborg F, Nymark T, Hallas J. The association between use of serotonergic antidepressants and perioperative bleeding during total hip arthroplasty–a cohort study. Basic Clin Pharmacol Toxicol. 2014;115:277–281.
15. van Haelst IM, van Klei WA, Doodeman HJ, Kalkman CJ, Egberts TC. MAOI Study Group. Antidepressive treatment with monoamine oxidase inhibitors and the occurrence of intraoperative hemodynamic events: a retrospective observational cohort study. J Clin Psychiatry. 2012;73:1103–1109.
16. Wong K, Phelan R, Kalso E. Antidepressant drugs for prevention of acute and chronic postsurgical pain: early evidence and recommended future directions. Anesthesiology. 2014;121:591–608.
17. Chen CC, Lin CS, Ko YP, Hung YC, Lao HC, Hsu YW. Premedication with mirtazapine reduces preoperative anxiety and postoperative nausea and vomiting. Anesth Analg. 2008;106:109–113.
18. Castro-Alves LJ, Oliveira de Medeiros AC, Neves SP. Perioperative duloxetine to improve postoperative recovery after abdominal hysterectomy: aprospective, randomized, double-blinded, placebo-controlled study. Anesth Analg. 2016;122:98–104.
19. Lunn TH, Frokjaer VG, Hansen TB, Kristensen PW, Lind T, Kehlet H. Analgesic effect of perioperative escitalopram in high pain catastrophizing patients after total knee arthroplasty: a randomized, double-blind, placebo-controlled trial. Anesthesiology. 2015;122:884–894.
20. Gilron I, Tu D, Holden RR, Jackson AC, DuMerton-Shore D. Combination of morphine with nortriptyline for neuropathic pain. Pain. 2015;156:1440–1448.
21. Kudoh A, Katagai H, Takazawa T. Increased postoperative pain scores in chronic depression patients who take antidepressants. J Clin Anesth. 2002;14:421–425.
22. Ahn EJ, Kang H, Choi GJ, Baek CW, Jung YH, Woo YC. The effectiveness of midazolam for preventing postoperative nausea and vomiting: asystematic review and meta-analysis. Anesth Analg. 2016;122:664–676.
23. Grant MC, Kim J, Page AJ, Hobson D, Wick E, Wu CL. The effect of intravenous midazolam on postoperative nausea and vomiting: ameta-analysis. Anesth Analg. 2016;122:656–663.
24. Ciccozzi A, Marinangeli F, Colangeli A. Anxiolysis and postoperative pain in patients undergoing spinal anesthesia for abdominal hysterectomy. Minerva Anestesiol. 2007;73:387–393.
25. Beresnevaitė M, Benetis R, Taylor GJ, Jurėnienė K, Kinduris Š, Barauskienė V. Depression predicts perioperative outcomes following coronary artery bypass graft surgery. Scand Cardiovasc J. 2010;44:289–294.
26. Törer N, Nursal TZ, Calişkan K. The effect of the psychological status of breast cancer patients on the short-term clinical outcome after mastectomy. Acta Chir Belg. 2010;110:467–470.
27. Jørgensen CC, Knop J, Nordentoft M, Kehlet H. Lundbeck Foundation Centre for Fast-track Hip and Knee Replacement Collaborative Group. Psychiatric disorders and psychopharmacologic treatment as risk factors in elective fast-track total hip and knee arthroplasty. Anesthesiology. 2015;123:1281–1291.
28. Smith MM, Smith BB, Lahr BD, et al. Selective serotonin reuptake inhibitors and serotonin-norepinephrine reuptake inhibitors are not associated with bleeding or transfusion in cardiac surgical patients. Anesth Analg. 2018;126:1859–1866.
Supplemental Digital Content
© 2019 International Anesthesia Research Society