What is the Association Between Clinically Diagnosed Psychiatric Illness and Total Joint Arthroplasty? A Systematic Review Evaluating Outcomes, Healthcare Use, and Patient-reported Outcome Measures

Abstract

Background

Studies evaluating the effects of a psychiatric illness on orthopaedic surgical outcomes have yielded mixed results. Because awareness of patient comorbid mental health disorders has become increasingly important to tailor treatment plans, the aim of our systematic review was to present the findings of all studies reporting on the association between clinically diagnosed psychiatric illnesses and total joint arthroplasty (TJA) outcomes and evaluate the quality of evidence to provide a comprehensive summary.

Question/purpose

Is there a consistently reported association between comorbid psychiatric illness and (1) complication risk, (2) readmission rates, (3) healthcare use and discharge disposition, and (4) patient-reported outcome measures (PROMs) after TJA?

Methods

The PubMed, EBSCO host, Medline, and Google Scholar electronic databases were searched on April 9, 2022, to identify all studies that evaluated outcomes after TJA in patients with a comorbid clinically diagnosed mental health disorder between January 1, 2000, and April 1, 2022. Studies were included if the full-text article was available in English, reported on primary TJA outcomes in patients with clinically diagnosed mental health disorders, included patients undergoing TJA without a psychiatric illness for comparison, and had a minimum follow-up time of 30 days for evaluating readmission rates, 90 days for other perioperative outcomes such as length of stay and complications, and 1-year minimum follow-up if assessing PROMs. Studies that used a mental health screening examination instead of clinical diagnoses were excluded to isolate for verified psychiatric illnesses. Additionally, systematic reviews, case reports, duplicate studies between the databases, and gray literature were excluded. Twenty-one studies were included in our final analysis comprising 31,023,713 patients with a mean age range of 57 to 69 years. Mental health diagnoses included depression, anxiety, bipolar disorder, schizophrenia, major personality disorder, and psychosis as well as concomitant mental disorders. Two reviewers independently evaluated the quality of included studies using the Methodological Index for Nonrandomized Studies (MINORS) tool. The mean MINORS score was 19.5 ± 0.91 of 24, with higher scores representing better study quality. All the articles included were retrospective, comparative studies. Given the heterogeneity of the included studies, a meta-analysis was not performed, and results are instead presented descriptively.

Results

Patients with schizophrenia were consistently reported to have higher odds of medical and surgical complications than patients without psychiatric illness, particularly anemia and respiratory complications. Among studies with the largest sample sizes, patients with depression alone or depression and anxiety had slightly higher odds of complications. Most studies identified higher odds of readmission among patients with depression, schizophrenia, and severe mental illness after TJA. However, for anxiety, there was no difference in readmission rates compared with patients without psychiatric illness. Slightly higher odds of emergency department visits were reported for patients with depression, anxiety, concomitant depression and anxiety, and severe mental illness across studies. When evaluating healthcare use, articles with the largest sample sizes reporting on depression and length of stay or discharge disposition found modestly longer length of stay and greater odds of nonhome discharge among patients with depression. Although several studies reported anxiety was associated with slightly increased total costs of hospitalization, the most robust studies reported no difference or slightly shorter average length of stay. However, the included studies only reported partial economic analyses of cost, leading to relatively superficial evidence. Patients with schizophrenia had a slightly longer length of stay and modestly lower odds of home discharge and cost. Likewise, patients with concomitant depression and anxiety had a slightly longer average length of stay, according to the two articles reporting on more than 1000 patients. Lastly, PROM scores were worse in patients with depression at a minimum follow-up of 1 year after TJA. For anxiety, there was no difference in improvement compared with patients without mental illness.

Conclusion

Our systematic review found that individuals with psychiatric illness had an increased risk of postoperative complications, increased length of stay, higher costs, less frequent home discharge, and worse PROM scores after TJA. These findings encourage inclusion of comorbid psychiatric illness when risk-stratifying patients. Attention should focus on perioperative interventions to minimize the risk of thromboembolic events, anemia, bleeding, and respiratory complications as well as adequate pain management with drugs that do not exacerbate the likelihood of these adverse events to minimize emergency department visits and readmissions. Future studies are needed to compare patients with concomitant psychiatric illnesses such as depression and anxiety with patients with either diagnosis in isolation, instead of only comparing patients with concomitant diagnoses with patients without any psychiatric illnesses. Similarly, the results of targeted interventions such as cognitive behavioral therapy are needed to understand how orthopaedic surgeons might improve the quality of care for patients with a comorbid psychiatric illness.

Introduction

As of 2020, approximately 21% of adults in the United States had a mental health disorder that met the criteria for clinical diagnosis, excluding substance use and developmental disorders [57]. This proportion is expected to grow, with the WHO predicting that by 2030, major depressive disorder will be the largest worldwide cause of disease burden [27]. Because patients with a comorbid mental health disorder have been demonstrated to have higher costs, increased length of stay, and worse patient-reported outcome measure (PROM) scores, surgeons must navigate the additional challenges these patients may present in order to deliver high-quality, tailored care [18, 51, 53, 81, 94]. Additionally, depression and anxiety have been demonstrated to be more frequent among people with musculoskeletal afflictions such as osteoarthritis than in the general population [8, 63].

Studies evaluating the association between psychiatric illness and orthopaedic surgical outcomes have yielded mixed results. This may be partly explained by variations in how psychiatric illness is identified, with some studies using preoperative surveys and others ensuring that appropriate clinical diagnostic criteria have been met among studied patient populations [8, 21, 37, 52, 70, 76]. Although some studies have reported worse perioperative outcomes among patients with diagnosed psychiatric illnesses [45, 48, 73, 83, 84], other research suggests these patients have a lower risk of complications, shorter lengths of stay (LOS), and higher costs of care than patients without a psychiatric illness [11, 59, 78]. Given the variability of findings presented in current studies, a systematic review is needed to evaluate the quality of evidence and provide a comprehensive summary. We aimed to further characterize and clarify the potential association between total joint arthroplasty (TJA) outcomes to address any disparities appropriately.

We asked, is there a consistently reported association between comorbid psychiatric illness and (1) complication risk, (2) readmission rates, (3) healthcare use and discharge disposition, and (4) PROMs after TJA?

Materials and Methods

Search Strategy

The PubMed, EBSCO host, Medline, and Google Scholar electronic databases were searched on April 9, 2022, to identify all studies published between January 1, 2000, and April 1, 2022 that evaluated TJA procedures among patients with a diagnosed mental health disorder. The following keywords and Medical Subject Headings terms were used in combination with the “AND” or “OR” Boolean operators: “Mental Disorders”[MeSH] OR “Mental Health”[MeSH] OR “Psychotropic Drugs”[MeSH] OR “Depression”[MeSH] OR “Anxiety”[MeSH] OR “mental illness” OR “psychiatric disorder”) AND (“Arthroplasty, Replacement, Hip”[MeSH] OR “Arthroplasty, Replacement, Knee”[MeSH] OR “Arthroplasty, Replacement”[MeSH] OR “total joint arthroplasty” OR “total knee arthroplasty” OR “total hip arthroplasty” OR “TJA” OR “THA” OR “TKA.” For this study, we defined TJA as comprising THA and TKA only.

Eligibility Criteria

Articles were included if full-text articles in English were available, the study described primary TJA procedures, the study reported on the relationship between a clinically diagnosed mental health disorder and postoperative outcomes, and it included patients without any psychiatric illness for comparison. Only studies with a minimum follow-up time of 30 days for evaluating readmission rates or 90 days for other perioperative outcomes such as LOS and complications were included. Because recent studies [12, 68] have suggested that a minimum of 1 year of follow-up for PROMs captures the greatest magnitude in change and reaches minimal clinically important thresholds, only studies with a minimum of 1 year of follow-up when assessing PROMs were included. PROMs were defined as any health outcome directly reported by a patient. We restricted our study to only include studies reporting on patients with clinically diagnosed disorders, in contrast to screening tests that only detect potential indicators of a disorder, to allow our study to more effectively isolate verified psychiatric illnesses (Supplemental Table 1; http://links.lww.com/CORR/A976). Commonly used screening examinations such as the Hospital Anxiety and Depression Scale have lower predictive power, and various cutoffs are used, leading to variance in their accuracy to screen for those who would meet diagnostic criteria [17, 69, 91]. Additionally, the following were excluded from our analysis: systematic reviews, case reports, duplicate studies between the databases, gray literature such as abstracts and articles on preprint servers, and articles reporting on revision TJA.

Study Selection

This review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Two independent reviewers (CJH, AFK) assessed the eligibility of each included article. Disagreements were discussed, and a third independent reviewer (RJB) was consulted to achieve consensus. After removal of duplicates, the initial query yielded 1286 articles, which were then screened for appropriate studies that aligned with the purpose of our review. After title and abstract screening, 68 studies were selected for further consideration. The full text of each article was reviewed, 21 of which fulfilled our inclusion criteria. Most full-text screened articles that were excluded did not use clinical diagnoses to verify the presence of a psychiatric disorder [1-4, 6, 8, 10, 16, 19, 21, 23-25, 29-43, 46, 52, 56, 58, 61, 62, 64, 66, 67, 72, 76, 80, 85-89, 92, 93]. A review of each study’s reference list did not yield any additional articles (Fig. 1).

Fig. 1.

This PRISMA diagram depicts the article selection process for inclusion.

Study Characteristics

All 21 included studies were retrospective in design; five matched patients with a psychiatric illness to one or more patient without to minimize confounding variables. The other 16 studies had patients without a psychiatric illness as a comparison group, but individual patients between the two groups were not matched to limit demographic differences. Overall, 31,023,713 patients were included with a mean age range of 57 to 69 years. The percentage of women in each study cohort ranged from 51% to 78%. Eleven percent of included patients had at least one mental health diagnosis. The mental health diagnoses among the studies were depression (76% [2,593,071 of 3,432,939]), anxiety (6% [218,994 of 3,432,939]), and schizophrenia (0.9% [29,479 of 3,432,939]). Studies with one or more psychiatric illness reported anxiety and depression, bipolar disorder, personality disorder, and psychosis (16% [565,797 of 3,432,939]). The sample size of individual studies ranged from 280 to 8,738,690. With the large range in sample sizes, articles were weighed based on the quality of evidence provided when synthesizing their results in our narrative review. Our criteria for determining which studies presented more robust results were studies with larger sample sizes that matched psychiatric patients to nonpsychiatric patients to limit confounding variables, studies that analyzed each included illness in isolation compared with grouping multiple under “severe mental illness,” studies with a higher Methodological Index for Nonrandomized Studies (MINORS) score, and studies in which a power analysis was conducted. The included studies did not assess the typical comorbidities of TJA in addition to the variable of psychiatric illness (Table 1).

Table 1.

Characteristics of articles included in the final analysis

Study (year)	Type of study	Data source	Procedure	Diagnostic criteria	Psychiatric diagnoses	Sample size, n	Minimum time until follow-up	MINORS scorea	Primary outcomes
Browne et al. (2014) [11]	Retrospective	NIS	Both	ICD-9-CM	Depression	497,222	1 year	19	Complications, LOS, cost, mortality
Buller et al. (2015) [14]	Retrospective	NHDS	Both	ICD-9-CM	Depression, anxiety, schizophrenia	8,379,490	1 year	19	Complications, discharge disposition
Chuang et al. (2021) [15]	Retrospective	NHDS	TKA	ICD-9-CM	Depression	32,761	1 year	19	Complications, LOS, discharge disposition, mortality
Gholson et al. (2018) [26]	Retrospective	NIS	Both	ICD-9-CM	Schizophrenia	505,840	1 year	20	Complications, LOS, discharge disposition, mortality
Gold et al. (2016) [28]	Retrospective	California Healthcare Cost and Utilization Project database	Both	ICD-9-CM	Depression	196,654	90 days	19	Readmission
Kamalapathy et al. (2021) [42]	Retrospective	Mariner Claims database	Both	ICD-9-CM and ICD-10-CM	Depression and/or anxiety, schizophrenia, bipolar disorder, other	31,982	1 year	19	Complications, readmissions survivorship analysis
Klement et al. (2016) [45]	Retrospective	Medicare database	TKA	ICD-9-CM	Depression, anxiety, schizophrenia, bipolar disorder	1,487,531	2 years	19	Complications
Klement et al. (2016) [46]	Retrospective	Medicare database	THA	ICD-9-CM	Depression, anxiety, schizophrenia, bipolar disorder	677,665	2 years	19	Complications
Kohring et al. (2018) [47]	Retrospective	Hospital-specific	Both	Clinical diagnosis in EHR	Depression	280	1 year	18	PROMs
Kooner et al. (2021) [48]	Retrospective	ABJHI database	Both	Clinical Risk Group data	Depression, anxiety, bipolar disorder, major personality disorder, psychosis, other	1828	1 year	21	Complications, LOS, discharge disposition, PROMs
Melnic et al. (2020) [54]	Retrospective	Multi-hospital database	TKA	ICD-10	Anxiety, schizophrenia, mood disorder, substance use disorder, multiple	1392	1 year	19	PROMs
Pan et al. (2019) [59]	Retrospective	NIS	TKA	ICD-9-CM	Depression, anxiety, both	7,153,752	1 year	21	Complications, LOS, cost
Paredes et al. (2020) [60]	Retrospective	CMS claims database	Both	ICD-9 and ICD-10-CM	Depression, anxiety, schizophrenia, bipolar disorder, other	1,889,032	1 year	19	Complications, readmission, LOS
Seagrave et al. (2021) [70]	Retrospective	Evidence-based Processes and Outcomes of Care	Both	Self-reported	Depression, anxiety	1669	1 year	21	Complications, readmission, LOS, PROMs
Singh and Lewallen (2013) [74]	Retrospective	Mayo Clinic Total Joint Registry	TKA	Mayo Clinic’s Hospital Adaptation of International Code for Diseases	Depression, anxiety	11,373	2 years	20	PROMs
Singh and Lewallen (2014) [75]	Retrospective	Mayo Clinic Total Joint Registry	TKA	ICD-9-CM	Depression, anxiety	13,787	2 years	21	PROMs
Stone et al. (2019) [77]	Retrospective	Hospital-specific	Both	Clinical diagnosis in EHR	Depression, anxiety, both, other	3020	90 days	19	Readmission, LOS, discharge disposition
Stundner et al. (2013) [78]	Retrospective	NIS	Both	ICD-9-CM	Depression, anxiety, both	1,212,493	1 year	20	Complications, LOS, cost
Vakharia et al. (2019) [84]	Retrospective	CMS Claims Database	TKA	ICD-9	Schizophrenia	49,176	2 year	20	Complications, readmission, LOS, cost
Vakharia et al. (2020) [83]	Retrospective	Humana claims database	TKA	ICD-9-CM	Depression	138,076	2 year	20	Complications, readmission, LOS, cost
Zalikha et al. (2021) [95]	Retrospective	NIS	Both	ICD-9-CM	Depression, anxiety, both	8,738,690	1 year	18	Complications, LOS, discharge disposition, cost

^aThe maximum MINORS score is 24 for comparative studies, with higher scores indicating better study quality. MINORS = Methodological Index for Nonrandomized Studies; NIS = National Inpatient Sample; ICD-9-CM = International Classification of Disease, Ninth Revision, Clinical Modification; LOS = length of stay; NHDS = National Hospital Discharge Survey; ICD-10-CM = International Classification of Disease, Tenth Revision, Clinical Modification; EHR = electronic health record; PROMs = patient-reported outcome measures; ABJHI = Alberta Bone and Joint Health Institute; CMS = Centers for Medicare and Medicaid Services

Risk of Bias in Individual Studies

The risk of bias was assessed by two independent reviewers (CJH, AHK) using the MINORS tool. This is a validated assessment tool that grades comparative studies from 0 to 24 based on 12 criteria related to study design, outcomes assessed, and follow-up, with higher scores indicating better study quality. Across these domains, each item is scored 0 if not reported, 1 when reported but inadequate, and 2 when reported and adequate. Any discrepancies in grading were resolved with a third reviewer (RJB). The mean MINORS score was 19.5 ± 0.91.

Primary and Secondary Study Outcomes

Our primary study goal was to characterize potential associations between various comorbid psychiatric illnesses and outcomes after TJA: complication risk, 30-day and 90-day readmission rates, emergency department (ED) use, LOS, cost, discharge disposition, and PROMs. To achieve this, using the quality and size of the study as a basis, we considered each study’s findings and reported magnitudes of effect when assessing their mixed results to allow for a discussion of the overall trends in light of each study’s robustness. For the outcome measure of cost, the included studies only reported partial economic evaluations such as the total cost of hospitalization or reimbursement claims; thus, these relatively superficial analyses were deemed comparable in the quality of evidence they provided. Fifteen articles evaluated the association between perioperative complications and a diagnosed comorbid illness [11, 14, 15, 26, 42, 45, 46, 48, 59, 60, 70, 78, 83, 84, 95]. Eight articles reported on readmission rates after TJA [28, 42, 48, 60, 70, 77, 83, 84]. Thirteen articles reported on the relationship between a psychiatric illness and TJA healthcare use and disposition [11, 14, 15, 26, 48, 59, 60, 70, 77, 78, 83, 84, 95]. Six studies examined the relationship between psychiatric illness and PROMs [47, 48, 54, 70, 74, 75].

Our secondary study goal was to evaluate whether concomitant diagnoses had different risk profiles than either of the concomitant psychiatric diagnoses in isolation. To achieve this, using the quality and size of the study as a basis, we considered each study’s findings and reported magnitudes of effect for concomitant psychiatric diagnoses compared with each illness alone when assessing their mixed results to allow for a discussion of the overall trends in light of each study’s robustness. Four studies reported on the perioperative outcomes of patients undergoing TJA who had concomitant diagnoses of anxiety and depression [59, 77, 78, 95] and one study on any concomitant mental disorders [54].

Results

Complications

Patients with schizophrenia or severe mental illness were consistently reported to have higher odds of complications than patients without a psychiatric illness. However, patients with depression alone or both depression and anxiety had mixed results; some showed higher or no difference in odds (Table 2). All three studies reporting the overall odds of complications—all three of which assessed more than 100,000 patients with depression—found slightly higher odds of adverse events [14, 84, 95]. Patients with depression were consistently reported to have lower odds of cardiac and gastrointestinal complications and slightly higher odds associated with acute kidney injury, pneumonia, periprosthetic fracture, periprosthetic joint infection, altered mental status, anemia, and respiratory complications. Studies with greater than 100,00 patients with depression found the odds of pulmonary embolism were slightly higher [59, 83, 95], while smaller studies reported mixed results [11, 15, 46, 78]. The odds of deep vein thrombosis were also mixed across sample sizes. Vakharia et al. [83] reported modestly higher odds after matching patients with depression to patients without depression to minimize confounding variables.

Table 2.

Studies evaluating the association between psychiatric illness and TJA complications

Study	Key findings	Factors controlled for
Browne et al. 2014 [11]	Although patients with depression had slightly higher odds of anemia, infection, and PE as well as moderately higher odds of postoperative psychosis, they had slightly lower odds of cardiac and gastrointestinal complications from TJA.	Age, procedure year, race, gender, CCI
Buller et al. 2015 [14]	Patients with depression had slightly and schizophrenia had modestly higher odds of adverse events. Patients with anxiety had slightly lower odds of adverse events after TJA.	Age, gender, procedure, comorbidities, discharge disposition, days of care
Chuang et al. 2021[15]	No difference in odds of having any complications after TKA between patients with depression and those without. Also, there was no difference in odds of wound infection, blood transfusion, DVT, or PE.	Age, race, insurance status
Gholson et al. 2018 [26]	Patients with schizophrenia had largely increased odds of developing medical complications and modestly more likely to develop surgical complications after TJA. Specifically, the odds of pulmonary complications and postoperative altered mental status were markedly higher while GU complications, transfusion, and acute postoperative hemorrhagic anemia were modestly increased. No difference reported in odds of wound infection, PJI, dislocation, thrombocytopenia, PE, DVT, or cardiac or GI complications.	Age, sex, smoking status, race, CCI
Kamalapathy et al. 2021 [42]	30-day and 90-day medical as well as surgical complication rates for patients with anxiety or depression were not elevated. However, patients with severe mental illness were more likely to experience a medical complication by 30 and 90 days. No difference reported in 30-day or 90-day severe mental illness surgical complication odds.	Age, sex, comorbidities, substance use
Klement et al. 2016 [46]	Patients undergoing TKA with any psychiatric illness had higher odds of medical and surgical complication rates at 30 and 90 days, particularly the odds of CHF, DVT, PE, AKI, acute postoperative hemorrhagic anemia, bleeding complications, and blood transfusion were modestly higher. PJI, wound complications, and CHF had largely higher odds. However, there was slightly less odds of MI or developing arrythmias reported.	Age, sex, comorbidities, tobacco use
Klement et al. 2016 [45]	At 30 and 90 days, patients undergoing TKA with depression had slightly elevated odds of bleeding, transfusion, DVT, PE, and CHF as well as largely increased odds of respiratory failure, wound complications, PJI, and periprosthetic fracture. However, there was slightly less odds of MI or developing arrythmias reported. Patients with schizophrenia had modestly decreased odds of developing arrythmias with or without atrial fibrillation and no difference in odds for MI. However, these patients also had modestly higher odds of DVT, PE, bleeding, transfusion, and CHF as well as largely increased odds of respiratory failure, wound complications, PJI, and periprosthetic fracture. Patients with bipolar disorder had slightly increased odds of transfusion, MI, arrythmias with atrial fibrillation, and bleeding complications. They also had largely higher odds of CHF, DVT, PE, PJI, and periprosthetic fracture as well as markedly increased odds of wound complications and respiratory failure.	Gender, age
Kooner et al. 2021 [48]	TJA patients with a psychiatric diagnosis were not at higher odds of experiencing a mechanical or medical adverse event overall. Also, as the only specific complications mentioned, no difference in odds for transfusion and postoperative infection.	Surgery type, year of surgery, age, sex, and medical comorbidities
Pan et al. 2019 [59]	After TKA, patients with depression had both slightly higher odds of AKI, anemia, MI, PE, pneumonia, and stroke but had slightly lower odds for postoperative infection, cardiac, GI, and CNS complications. No difference in odds for hemorrhage. Patients with anxiety had slightly higher odds of anemia, MI, pneumonia, and stroke. No difference in odds for AKI, postoperative infection, PE, cardiac, GI, GU, or CNS complications. Patients with both anxiety and depression had slightly higher odds of anemia, PE, pneumonia, and stroke compared to patients with no psychiatric illnesses. No difference in odds of AKI, hemorrhage, MI, cardiac, GI, GU, and CNS complications. Also, patients with both diagnoses had largely lower odds of postoperative infection.	Sex, age, race, hospital location, bed size, LOS
Paredes et al. 2020 [60]	Having anxiety and/or depression was associated with slightly higher odds of TJA complications overall, with a severe mental illness diagnosis having modestly higher odds.	Age, sex, CCI, race, discharge disposition
Seagrave et al. 2021 [70]	No difference in the odds of major TJA complications, VTE, wound dehiscence, fracture at surgical site, dislocation, or joint bleed between patients with anxiety and/or depression and patients without a psychiatric illness in covariate-adjusted analysis. However, TKA patients with anxiety and/or depression had slightly higher odds of developing a surgical site infection.	Age, sex, BMI, comorbidities, education level, LOS
Stundner et al. 2013 [78]	There was no difference in the overall odds for developing complications in TJA patients with depression compared to those without any psychiatric illness, but they had slightly increased odds for sepsis, cardiac complications excluding MI, and pneumonia. No difference reported in odds for VTE, MI, shock, pulmonary complications, or stroke. Patients with anxiety had slightly lower overall odds of developing complications, as well as stroke and cardiac complications including MI. There was no difference in odds of pulmonary complications, sepsis, and shock. Patients with anxiety had slightly higher odds of VTE and pneumonia. Also, there was no difference in overall odds of developing any complication for patients with concomitant anxiety and depression compared to patients without either diagnosis. No difference was reported in odds of VTE, pneumonia, cardiac complications, sepsis, or pulmonary complications. There were not enough data to assess for odds of MI and stroke.	Age, sex, race, hospital size, hospital location, and teaching status
Vakharia et al. 2019 [84]	TKA patients with schizophrenia had markedly higher odds of total medical complications and modestly higher odds of total implant complications compared to patients without schizophrenia. Transfusions, acute post-hemorrhagic anemia, and pneumonia all had markedly higher odds in schizophrenics. Also, there were modestly higher odds of stroke, UTI, AKI, thrombocytopenia, PE, DVT, dislocation, periprosthetic fractures, and PJI reported. No difference in odds for mechanical loosening or broken prosthetic implants reported.	Age, sex, region, comorbidities, CCI, discharge status, state, year, quarter, months, and date of index operation
Vakharia et al. 2020 [83]	90-day medical complications were largely and 2-year implant complications were modestly increased after TKA in patients with a depressive disorder. Transfusions, thrombocytopenia, UTI, stroke, and pneumonia had largely higher odds in patients with depression while acute respiratory insufficiencies, AKI, acute post-hemorrhagic anemia, paralytic ileus, PE, DVT, broken prosthetic joint, periprosthetic fracture, mechanical loosening, dislocation, and PJI all had modestly elevated odds.	Age, sex, and medical comorbidities associated with depressive disorders
Zalikha et al. 2021 [95]	Patients undergoing TJA who had depression, anxiety, or both had slightly higher odds of experiencing any in-hospital complication. Patients with depression had slightly decreased odds of cardiac, respiratory, GI, and GU complications as well as slightly increased odds of wound dehiscence, infection, PE, and postoperative anemia. No difference in odds of DVT reported. Patients with anxiety had slightly lower odds of GI complications as well as no difference in the odds of cardiac, respiratory, GU complications, wound dehiscence, infection, DVT, or PE. Patients with anxiety as well as concomitant depression and anxiety had a modestly increased odds of developing anemia postoperatively. Patients with concomitant depression and anxiety had no difference in cardiac or GU complications, infection, or PE. There were not enough data to assess for odds of respiratory or GI complications or wound dehiscence.	Age, BMI, race, hospital size, location, region of hospital

TJA = total joint arthroplasty; PE = pulmonary embolism; CCI = Charlson comorbidity index; DVT = deep vein thrombosis; GU = genitourinary system; PJI = periprosthetic joint infection; GI = gastrointestinal system; CHF = congestive heart failure; AKI = acute kidney injury; MI = myocardial infarction; CNS = central nervous system; LOS = length of stay; VTE = venous thromboembolism; UTI = urinary tract infection.

For patients with anxiety, there was no difference in the overall odds of complications. However, slightly higher rates of anemia [59, 95] and pneumonia [59, 78] were consistently reported. For patients with schizophrenia, three studies reported higher overall rates of complications [14, 26, 84], with deep vein thrombosis, pulmonary embolism, anemia, infection, respiratory complications, periprosthetic joint infection, and periprosthetic fracture each having modestly to largely higher odds than that for controls. Lastly, for patients with concomitant depression and anxiety, two studies with more than 1000 patients found slightly higher rates of overall complications than for patients with no psychiatric illnesses. However, Stundner et al. [78] found no difference in odds with a much smaller sample size. Only anemia was consistently reported to have higher odds in patients with both depression and anxiety. The remaining complications (pulmonary embolism, periprosthetic joint infection, pneumonia, and cardiac) showed no difference.

Readmission Rates

Depression, schizophrenia, and severe mental illness were most commonly reported to have higher odds of readmission after TJA, while anxiety showed no difference (Table 3). Vakharia et al. [83] and Gold et al. [28] found slightly higher odds of readmission for patients with depression than for those without any psychiatric illness, while Stone et al. [77] reported no difference in odds among their small sample. Only Stone et al. [77] reported the odds of readmission for anxiety alone, in which no difference was noted compared with patients without psychiatric illness. Studies that combined “depression and/or anxiety” as a group were split, with Paredes et al. [60] reporting higher odds of readmission with their relatively large sample size, compared with no difference reported in the markedly smaller samples of Kamalapathy et al. [42] and Seagrave et al. [70]. For concomitant depression and anxiety, Stone et al. [77] reported no difference. Vakharia et al. [84] reported modestly higher odds of readmission for patients with schizophrenia. Paredes et al. [60] found moderately higher readmission rates for patients with severe mental illness with a much larger sample. Lastly, Kamalapathy et al. [42] and Stone et al. [77] reported slightly higher odds of ED visits for patients with depression, anxiety, concomitant depression and anxiety, and severe mental illness.

Table 3.

Studies evaluating the association between psychiatric illness and TJA readmissions

Study	Key findings	Factors controlled for
Gold et al. 2016 [28]	Patients with depression undergoing TJA have slightly higher odds of 90-day readmission.	Age, gender, race, year of surgery, CCI, and insurance status.
Kamalapathy et al. 2021 [42]	No difference in odds of 30-day and 90-day readmission for patients with depression and/or anxiety undergoing TKA or THA, but patients with severe mental illness had slightly higher odds 90 days after TKA. In contrast, odds of ED visit at 30 and 90 days were slightly elevated for TKA and THA patients with depression and/or anxiety or with severe mental illness.	Age, gender, comorbidities, substance use
Kooner et al. 2021 [48]	No difference in 30-day readmission rates between patients with or without a psychiatric diagnosis.	Surgery type, year of surgery, age, sex, and medical comorbidities
Paredes et al. 2020 [60]	Slightly higher odds of 30-day readmission after TJA are associated with anxiety or depression as well as modestly higher odds for severe mental illness.	Age, sex, CCI, race, discharge disposition
Seagrave et al. 2021 [70]	No difference in the odds of 30-day readmission after TJA for patients with anxiety and/or depression.	Age, sex, BMI, comorbidities, education level, LOS
Stone et al. 2019 [77]	No difference in the odds of 30-day readmission after TJA for patients with depression, anxiety, both, or other severe mental illness. However, patients with anxiety, depression, both, or severe mental illness had variable but generally slightly higher odds of visiting the ED.	Age, sex, BMI, LOS, discharge disposition
Vakharia et al. 2019 [84]	Schizophrenia was associated with modestly higher odds of 90-day readmission after TKA.	Age, sex, region, comorbidities, CCI, discharge status, state, year, quarter, months, and date of index operation
Vakharia et al. 2020 [83]	Depression was associated with slightly higher odds of 90-day readmission after TKA.	Age, sex, and medical comorbidities associated with depressive disorders

TJA = total joint arthroplasty; CCI = Charlson comorbidity index; ED = emergency department; LOS = length of stay.

Healthcare Use and Disposition

Although patients with schizophrenia consistently had longer LOS, higher odds of nonhome discharge, and increased costs, patients with depression, anxiety, or concomitant depression and anxiety had mixed results (Table 4). All articles reporting on depression and LOS or discharge disposition found modestly longer LOS and greater odds of nonhome discharge in patients with depression except for one [15], which had a much smaller sample size than the other studies did [11, 14, 59, 77, 78, 83, 95]. There were mixed results among the included articles on the total cost of care for patients with depression compared with controls, both among metrics used to quantify cost and among studies with more rigorous designs [11, 59, 78, 83, 95].

Table 4.

Studies evaluating the association between psychiatric illness and TJA resource use and disposition

Study	Key findings	Factors controlled for	Metric to quantify cost utilized
Browne et al. 2014 [11]	Despite slightly longer average LOS, patients with depression undergoing TJA had slightly lower average costs than those without depression.	Age, procedure year, race, gender, CCI	Total inflation-adjusted hospital charges
Buller et al. 2015 [14]	Depression and anxiety are associated with slightly higher and schizophrenia with modestly higher odds of discharge to an inpatient facility after TJA.	Age, gender, procedure, comorbidities, discharge disposition, days of care
Chuang et al. 2021 [15]	LOS for patients with depression undergoing TKA was on average slightly less than that of patients without depression. No difference in discharge disposition odds to rehabilitation facilities reported.	Age, race, insurance status
Gholson et al. 2018 [26]	Patients with schizophrenia undergoing TJA have a slightly higher average LOS and have largely increased odds of being discharged to an inpatient facility.	Age, sex, smoking status, race, CCI
Kooner et al. 2021 [48]	Any psychiatric diagnosis is associated with a slightly lower percentage of patients with TJA discharged to home as well as a slightly increased average LOS.	Surgery type, year of surgery, age, sex, and medical comorbidities
Pan et al. 2019 [59]	Although patients with anxiety undergoing TKA had a slightly shorter average LOS, their average cost of hospitalization was slightly higher. Patients with depression had moderately longer average LOS and slightly higher costs. Patients with concomitant anxiety and depression had no difference in LOS but was associated with slightly higher costs.	Sex, age, race, hospital location, bed size, length of stay	Total cost of hospitalization
Paredes et al. 2020 [60]	The odds of being above the 75th percentile for LOS in TJA patients with anxiety and/or depression were modestly higher than patients without a psychiatric illness, and the odds for those with a severe mental illness are largely higher. Anxiety and/or depression was associated with modestly higher and severe mental illness with largely higher odds of non-home discharge.	Age, sex, CCI, race, discharge disposition
Seagrave et al. 2021 [70]	No difference in average LOS between patients with anxiety and/or depression and those without any diagnosis.	Age, sex, BMI, comorbidities, education level, LOS
Stone et al. 2019 [77]	Average LOS and percentage of patients discharged to a skilled nursing facility after TJA were slightly increased for depression, anxiety, and other psychiatric illness.	Age, sex, BMI, LOS, discharge disposition
Stundner et al. 2013 [78]	Patients with depression, anxiety, or concomitant depression and anxiety who underwent TJA had slightly longer average LOS than patients with no psychiatric illness. Patients with depression, anxiety, or both were slightly more likely to be in the top quartile of TJA patients with respect to total hospital charges.	Age, sex, race, hospital size, hospital location, and hospital teaching status	Total cost of hospitalization
Vakharia et al. 2019 [84]	Day of surgery and 90-day costs of care for TKA patients with schizophrenia were modestly elevated. They also had a slightly longer average LOS.	Age, sex, region, comorbidities, CCI, discharge status, state, year, quarter, months, and date of index operation	Reimbursement claims
Vakharia et al. 2020 [83]	Day of surgery and 90-day costs of care for TKA patients with depression were modestly elevated. They also had a modestly longer average LOS.	Age, sex, and medical comorbidities associated with depressive disorders	Reimbursement claims
Zalikha et al. 2021 [95]	Patients with depression had a modestly increased average LOS and odds for non-home discharge. However, no difference in cost reported. Patients with anxiety had slightly increased average cost, but no difference in discharge disposition or LOS. Patients with concomitant anxiety and depression had slightly increased odds of non-home discharge, but no difference in LOS or cost compared to patients with no psychiatric illnesses.	Age, BMI, race, hospital size, location, region of hospital	Total hospital charges

TJA = total joint arthroplasty; LOS = length of stay; CCI = Charlson comorbidity index.

Among patients with anxiety, the average LOS was no different from or slightly shorter than that of patients without a psychiatric illness in the two largest studies reporting such information [59, 95]. In contrast, two smaller studies reported slightly longer LOS [7, 78]. Likewise, Buller et al. [14] found slightly higher odds of nonhome discharge in their relatively large sample compared with Zalikha et al. [95], who reported no difference in odds. Studies agreed that patients with anxiety had a slightly increased healthcare cost after TJA, using total cost of hospitalization in their analyses [59, 78, 95].

Patients with concomitant depression and anxiety had slightly longer average LOS per the two articles reporting on more than 1000 patients [59, 95], compared with the smaller studies of Stone et al. [77] and Stundner et al. [78], which reported no difference. Although Stundner et al. [78] found patients with concomitant depression and anxiety were slightly more likely to have total hospitalization costs within the top quartile of all patients undergoing TJA, Zalikha et al. [95] found no difference in average total hospitalization costs.

Studies agreed that patients with schizophrenia had slightly longer LOS and modestly lower odds of home discharge and cost [14, 26, 84]. Lastly, Stone et al. [77] noted that although fewer than 10% of patients with a psychiatric illness taking no medications were discharged to a skilled nursing facility. This percentage increased to 20% when a patient took two or more psychiatric medications, suggesting that more severe mental health disorders may be associated with longer postoperative recovery.

PROMs

Depression, particularly when untreated, was associated with poorer PROM scores after 1 year of follow-up after TJA (Table 5). Anxiety and grouped psychiatric illnesses showed no difference in improvements compared with controls. Kohring et al. [47] found patients with treated depression had comparable physical function computer adaptive test scores to those who were not depressed, but patients with untreated depression had less improvement. Depression was consistently reported to be associated with worse knee functional status 2 and 5 years postoperatively, and there was a higher odds of reporting moderate to severe pain 2 years after TKA [74, 75].

Table 5.

Studies evaluating the association between psychiatric illness and TJA PROMs

Study	Key findings	Factors controlled for
Kohring et al. 2018 [47]	After 1 year follow-up at minimum, patients undergoing TJA with untreated depression do not show clinically meaningful improvement in PF-CAT scores, but those who were treated showed similar improvements to those who were not depressed.	Age, gender, BMI, ASA score
Kooner et al. 2021 [48]	After 1 year follow-up at minimum, no difference in improvement for WOMAC scores between patients with a psychiatric diagnosis and those without a diagnosis who underwent TJA.	Surgery type, year of surgery, age, sex, and medical comorbidities
Melnic et al. 2020 [54]	No difference in improvements in PROMIS PF10a or PROMIS Mental scores in patients with any psychiatric illness compared to patients with no diagnoses after 1 year follow-up.	ASA score, reoperation rate, BMI, race, sex, age
Seagrave et al. 2021 [70]	After 1 year follow-up, no difference in Oxford Hip or Knee scores between people with anxiety and/or depression and those without. After adjusted analysis, no difference reported in patient-reported global improvement or satisfaction.	Surgery, age, sex, education level, and medical comorbidities
Singh and Lewallen 2013 [74]	Anxiety was associated with slightly higher odds of moderate to severe pain 2 years after TKA. Depression and anxiety were each associated with slightly higher odds for 5 years after TKA.	Age, sex, BMI, ASA score, CCI
Singh and Lewallen 2014 [75]	Depression was associated with modestly higher odds of reporting poorer knee functional status 2 years after TKA, with no difference reported after 5 years. No difference in reported functional status after 2 or 5 years from patients with anxiety.	Age, sex, BMI, ASA score, CCI, distance from medical center, surgery

TJA = total joint arthroplasty; PROM = patient-reported outcome measure; PF-CAT = physical function computer adaptive test; ASA = American Society of Anesthesiologists; PROMIS PF10a = Patient-Reported Outcome Measurement Information System Physical Function short form 10a; CCI = Charlson comorbidity index.

Anxiety was also associated with slightly higher odds of reporting moderate to severe pain 2 and 5 years after TKA. However, there was no difference in reported functional status during these time periods [74, 75]. In studies grouping patients with any psychiatric illness for analysis, there were no differences reported in WOMAC or Patient-Reported Outcome Measure Information System scores compared with patients without mental disorders after 1 year of follow-up [48, 54].

Discussion

Because a growing proportion of patients undergoing TJA are expected to have a concomitant psychiatric diagnosis, orthopaedic surgeons must consider the impact of their patients’ mental health perioperatively to ensure equitable care is provided. The aim of our systematic review was to present the findings of all studies reporting on the association between clinically diagnosed psychiatric illnesses and TJA outcomes and evaluate the quality of evidence to provide a comprehensive summary. Our analysis found patients with a psychiatric illness broadly had increased odds of complications including anemia, thromboembolic events, and hemorrhage compared with patients without a psychiatric illness. The association between a psychiatric illness on healthcare use and disposition was largely consistent among the studies, in which longer average LOS, higher costs, and less frequent discharge to home were increased in patients undergoing TJA who had a comorbid psychiatric illness. Similarly, among the studies reporting PROM scores, most reported worse pain and function among patients with mental health disorders than among those without, with inadequate pain management potentially contributing to increased healthcare use and drug-drug interactions between pain-relieving and psychiatric medicines that may increase the likelihood of adverse events. Orthopaedic surgeons should include mental health as a component of patient risk stratification and education before TJA to establish expectations and foster shared decision-making to improve patient outcomes while also ensuring that identifying higher-risk patients does not perpetuate current disparities.

Limitations

Our analysis has several limitations. We only included studies that evaluated the effects of a clinically diagnosed psychiatric illness on TJA outcomes; therefore, the conclusions of this study are not applicable to patients who may have an undiagnosed disorder. Second, multiple included studies noted that patients who have a psychiatric illness but do not have a formal diagnosis may have been included in the cohorts without psychiatric illnesses, potentially underestimating the effect of comorbid psychiatric illness on TJA outcomes. Third, some studies grouped psychiatric illnesses under umbrella terms such as “other” or severe mental illness, typically comprising schizophrenia, bipolar disorder, and other more uncommon psychiatric illnesses. Although beneficial in elucidating overall trends across psychiatric disorders, these groupings did not allow us to assess each included condition individually, which may have skewed our results for schizophrenia in particular. Fourth, the included studies were heterogenous in reporting the odds of developing complications, with some reporting only categories such as “cardiac complications,” while others listed specific cardiac complications. Similarly, the metric with which studies determined healthcare cost varied across studies, and all studies were only partial economic analyses. Fifth, the included studies were retrospective in design, with only two conducting their study prospectively [47, 48]. As a result, some included studies did not have sufficient data to report on specific complications patients experienced for each psychiatric illness. Sixth, we were unable to comment on the indications for TJA among included studies because most patients were identified by International Classification of Diseases, Ninth Revision, procedural codes. Seventh, the included studies had a wide range of sample sizes, making generalizable inferences difficult. Therefore, we considered how studies limited confounding variables (if power analyses were performed) and MINORS score when comparing results. Lastly, given the heterogeneity of the included studies, a meta-analysis was not appropriate, and results are instead presented descriptively.

Complications

Among the highest-quality studies, patients with schizophrenia, severe mental illness, depression, or concomitant depression and anxiety had a higher odds of experiencing complications after TJA than patients with no psychiatric illness. Anxiety was associated with no difference in the overall odds of complications. However, anxiety and all other diagnoses were associated with an increased odds of anemia. Additionally, although patients with concomitant depression and anxiety had no difference in the odds of experiencing a pulmonary embolism or deep vein thrombosis, the odds for patients with depression tended to be slightly higher in more robust studies and largely higher for schizophrenia across all reporting studies. Hoirisch-Clapauch et al. [34] suggested that the increased odds of thromboembolic complications may be associated with decreased levels of tissue plasminogen activator and protein S found in patients with schizophrenia. Lahlou-Laforet et al. [49] found that patients with depression have elevated levels of plasminogen activator inhibitor-1, which also leads to an increased risk of thromboembolic complications. Consequently, surgeons could implement strategies that increase tissue plasminogen activator levels, when inadequate, in TJA candidates with psychiatric diagnoses, such as normalizing homocysteine levels; monitoring vitamins B12, B6, and B9 intake; encouraging smoking cessation; and controlling blood glucose levels [13, 34, 35]. Furthermore, as patients with depression have decreased bone mineral density and are more likely to fall [50, 90], we found that patients with depression had slightly higher odds of periprosthetic fractures than controls. Orthopaedic surgeons should mitigate periprosthetic bone resorption and osteoporosis progression through bisphosphonates to bolster bone integrity and stability [71]. Additionally, because schizophrenia was associated with largely higher odds of respiratory complications than for controls, even after adjusting for increased smoking rates among patients with schizophrenia as a confounder, these patients should be closely observed with continuous pulse oximetry, avoidance of drugs that decrease respiratory drive, and early mobilization postoperatively [26].

Readmission Rates

Depression, schizophrenia, and severe mental illness were reported to be associated with higher odds of readmission after TJA, while patients with anxiety had no difference in odds compared with controls. A recent meta-analysis by Kim and Kim [44] evaluating depression and readmission rates among TJA found slightly higher 90-day readmission rates in patients undergoing TKA and no difference for those undergoing THA compared with controls. A key distinction from the present review is that Kim and Kim [44] included studies evaluating depression through preoperative screening tools. In contrast, our study used clinically diagnosed conditions to better isolate patients with verified psychiatric illnesses. In addition to elevated readmission rates, we found slightly higher odds of ED visits for patients with depression, anxiety, concomitant depression and anxiety, and severe mental illness. Further, because postoperative pain and wound problems are the most common reason patients who underwent TJA present to the ED, studies suggest patients with a mental health disorder are more likely to experience pain catastrophizing and somatization than those without a disorder [3, 8, 65, 79]. Consequently, pain catastrophizing may mediate the increase in ED visits because most patients who present to the ED following TJA are not readmitted. Given these findings, orthopaedic surgeons should ensure that patients with mental health disorders, particularly those who catastrophize about pain preoperatively [79], have adequate follow-up immediately in the postoperative period. Because there is a need for high-quality interventions to reduce pain catastrophizing, particularly in patients with a psychiatric illness, Birch et al. [5] published a protocol for a prospective randomized controlled trial about the effectiveness of a physiotherapist in administering cognitive-behavioral education to patients undergoing TKA.

Healthcare Use and Disposition

Although schizophrenia has been consistently associated with increased healthcare use after THA, only higher-quality studies linked depression or concomitant depression and anxiety with higher use, compared with patients with no psychiatric illnesses. On the other hand, except for increased costs, anxiety alone was not associated with increased use. These findings are similar to those of Bot et al. [9], who reported higher rates of nonhome discharge for patients with clinical diagnoses of depression, anxiety, or schizophrenia identified by International Classification of Diseases, Ninth Revision, codes after undergoing shoulder arthroplasty. In contrast to our findings, Bot et al. [9] reported a slightly shorter LOS in patients with depression, which they attributed to more rapid discharge to a long-term facility. Additionally, Huang et al. [38] noted that patients with anxiety had slightly lower odds of prolonged LOS after spinal surgery than patients without any psychiatric disorders, which agrees with our findings. However, the reported magnitudes of effect in the present analysis and that of Huang et al. [38] do not exceed minimum clinically important difference thresholds [59, 95]. A thorough discussion regarding healthcare use and postoperative expectations in patients with comorbid psychiatric illness should be incorporated into preoperative evaluations to improve patient expectations and surgical outcomes.

PROMs

Patient with depression reported worse functional improvement and pain after TJA, while patients with anxiety reported no difference in functional scores compared with those without psychiatric illness. Postoperative pain in patients with anxiety is possibly slightly higher based on the limited data reported by the included studies. Although patients with treated depression tended to report comparable functional improvement in the present analysis, a review of TJA outcomes in patients with depression found comparable improvements between patients with depression who are treated and those who are not treated [82]. However, Vajapey et al. [82] did not restrict their study to include only clinically diagnosed depression, for which patients with depressive symptoms without a formal diagnosis may have less severe symptoms, leading to the discrepancy in findings. Furthermore, our finding that depression is associated with postoperative pain aligns with a growing body of researchers who consider depression a low-grade chronic inflammatory response that may trigger increased pain sensation [22, 55, 61]. Because patients with depression experience more pain postoperatively, insufficient management of pain is associated with increased healthcare use, such as longer LOS [20]. However, patients with depression or anxiety taking selective serotonin reuptake inhibitors are at a synergistically increased risk of bleeding when these inhibitors are combined with NSAID therapy [7, 10]. Consequently, orthopaedic surgeons should consider avoiding NSAID administration in patients undergoing TJA who have depression or anxiety and are taking selective serotonin reuptake inhibitors to minimize the risk of adverse bleeding events.

Conclusion

Our systematic review found individuals with psychiatric illness had an increased risk of postoperative complications, increased LOS, higher costs, less frequent home discharge, and worse PROMs after TJA than those without psychiatric illness. These findings encourage continued efforts to include the presence of a diagnosed comorbid psychiatric illness in patient risk stratification, education, and shared decision-making. Attention should be focused on perioperative interventions to minimize the risk of thromboembolic events, anemia, bleeding, and respiratory complications, as well as adequate pain management with drugs that do not exacerbate the likelihood of these adverse events to minimize ED visits and readmissions. Future studies are needed to compare patients with concomitant psychiatric illnesses such as depression and anxiety with patients who have either diagnosis in isolation instead of only comparing patients with concomitant diagnoses to patients without any psychiatric illnesses. Similarly, the results of targeted interventions such as cognitive behavioral therapy and methods to preemptively address pain catastrophizing are needed to understand how orthopaedic surgeons might improve the quality of care for patients with a comorbid psychiatric illness.