Anxiety and depression in non-cardiac surgical patients and associations with adverse outcomes: A systematic review and meta-analysis

Abstract

Background

Mental health disorders such as anxiety and depression are associated with advanced disease at presentation, reduced treatment uptake, and worse postoperative outcomes. Despite this, preoperative screening remains underutilized. This systematic review and meta-analysis aimed to characterize the prevalence of preoperative anxiety and depression in non-cardiac surgery patients and assess their associations with adverse postoperative outcomes.

Methods

The protocol was registered with PROSPERO (CRD42024578357) and followed PRISMA guidelines. MEDLINE, Embase, PsycINFO, Cochrane Database of Systematic Reviews, and CENTRAL were searched. All studies included were prospective cohorts that enrolled a minimum of 100 adults undergoing non-cardiac surgery, used validated preoperative mental health screening tools, and reported postoperative outcomes. Studies involving cardiac, neurosurgical, or bariatric procedures were excluded. Random-effects meta-analysis generated pooled prevalence, mean difference (MD), and odds ratios (OR) with 95 % confidence intervals (CI) and 95 % prediction intervals (PI).

Results

Forty-six studies (n = 23,628; mean age 58.5 ± 19.2 years; 61 % female) met inclusion criteria. Pooled prevalence of anxiety was 26 % (95 % CI, 21 %–31 %; 95 % PI, 20 %–33 %; I² = 95 %) while depression was 23 % (95 % CI, 14 %–31 %; 95 % PI, 14 %–36 %; I² = 100 %). The prevalence of anxiety ranged from 19 % in orthopedic to 29 % in cancer surgery. It was associated with a greater hospital length of stay (MD 0.55 days; 95 % CI, 0.29–0.82; 95 % PI, −0.04–1.14; I² = 0 %). The prevalence of depression ranged from 17 % in orthopedic to 30 % in cancer surgery and 24 % in mixed surgery. Preoperative depression was associated with increased odds of postoperative delirium (OR 2.33; 95 % CI, 1.74–3.11; 95 % PI, 1.5–3.5; I² = 0 %).

Conclusions

Anxiety and depression affect one-fifth of non-cardiac surgical patients, with highest prevalence in cancer surgery. Preoperative anxiety was associated with a greater hospital length of stay, while preoperative depression was associated with twofold risk of postoperative delirium. Therefore, routine preoperative mental health screening could identify high-risk patients and enable targeted perioperative interventions to reduce adverse outcomes.

Introduction

Preoperative anxiety and depression are prevalent, yet under-recognized conditions in surgical patients. Estimates show that up to 30 % of all surgical patients and 50 % of patients over 60 years old experience clinically significant anxiety or depression before surgery.1, 2 Surgical patients with mental health disorders often present with more advanced disease and are less likely to receive guideline-recommended treatment.3, 4, 5 Importantly, patients with mental health disorders experience worse postoperative outcomes, including higher rates of surgical complications, delirium, length of stay (LOS), readmission and mortality.6, 7 Additionally, patients with anxiety and depression have heightened pain perception and are less adherent to recovery protocols, further increasing the risk of adverse outcomes.8, 9 This contributes to significant morbidity for patients with preexisting anxiety and depression and greater costs to healthcare systems.1, 10 It is thought that anxiety and depression contribute to worse postoperative outcomes through increased systemic and neuroinflammation as well as gut microbiota dysbiosis that favours the production of inflammatory cytokines.11, 12

Despite these risks, screening for anxiety and depression remains underutilized in the preoperative setting, allowing these disorders to go undetected.¹³ This results in missed opportunities for early intervention such as timely referral to psychiatric services or identifying patients who may benefit from additional perioperative support.

Previous systematic reviews have explored the relationship between mental health burden and postoperative outcomes. However, there are no systematic reviews or meta-analyses that focus specifically on prospective studies on anxiety and depression in non-cardiac surgical patients. The objective of this systematic review and meta-analysis is to provide a comprehensive update on the prevalence of preoperative anxiety and depression in patients undergoing non-cardiac surgery. We also aim to assess the association between preoperative anxiety and depression and adverse postoperative outcomes.

Methods

Search strategy

The protocol of this systematic review and meta-analysis was registered in the International Prospective Register of Systematic Reviews (PROSPERO) (CRD42024578357). This review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines (Table S1).¹⁴

A comprehensive literature search strategy was developed and performed with the help of a library information specialist (ME) experienced in systematic reviews. The original comprehensive search strategy was conducted in the following databases: MEDLINE, MEDLINE ePubs & In-Process Citations, Embase Classic + Embase, APA PsycINFO, the Cochrane Database of Systematic Reviews, and Cochrane CENTRAL (all via the Ovid platform), from inception to April 18, 2024. Search concept blocks consisted of controlled vocabularies and text words on: (mental illness) AND (preoperative) AND (surgery) AND (study types), and were limited to English language, humans, and adults. Our full search employing comprehensive controlled vocabulary (MeSH, Emtree, PsycINFO Thesaurus) and extensive text word synonyms can be accessed in the Supplementary.

To capture studies published after the initial search, we performed an updated search in July 2025 using PubMed (translated strategy), Google Scholar, and forward citation of included studies. Newly identified records were screened with the same eligibility criteria.

Inclusion and exclusion criteria

The inclusion criteria were: (1) patients ≥ 18 years old undergoing non-cardiac surgery, (2) preoperative assessment of anxiety and/or depression using validated screening tools with binary cutoff thresholds, (3) sample size of ≥ 100, (4) prospective studies, (5) at least one postoperative outcome. The exclusion criteria were: (1) cardiac, neurosurgical, bariatric, or minor surgeries, (2) non-English articles, and (3) studies with overlapping data.

Study selection and data extraction

All included articles were uploaded on Covidence (Covidence.org), an online systematic review manager tool. After duplicate articles were removed, four reviewers independently performed title, abstract, and full-text screening in pairs (JC, JA, SA, TS). Conflicts were resolved either through discussion or by another reviewer (EY). Data from the included articles were independently extracted by three reviewers in duplicate (JC, JA, SA) and recorded on a standardized Excel spreadsheet, where conflicts were resolved through consultation with a third reviewer (EY). Extracted data included study characteristics (surgery type, study design, site characteristics), demographics (sample size, age, sex, comorbidities), preoperative anxiety and depression assessment (screening tools, specific cutoffs, and prevalence), and postoperative outcomes (delirium, pain, LOS, etc). Due to the raw nature of the data extracted, co-variates were not adjusted for, however they are described in Table S4.

Quality assessment

Two reviewers (JA, SA) independently performed quality assessment of all included studies using both the Newcastle-Ottawa Scale (NOS) and the Meta-Analyses of Observational Studies in Epidemiology (MOOSE) Checklist.15, 16 The results were compiled, with discrepancies resolved through discussion or by a third independent reviewer (EY). Using the NOS, studies were assessed for representativeness of the exposed cohort (one point), selection of the non-exposed cohort (one point), ascertainment of exposure (one point), demonstration that outcome of interest was not present at the start of the study (one point), comparability of cohorts on the basis of the design or analysis (two points), assessment of outcome (one point), duration of follow-up for outcomes to occur (one point), adequacy of follow-up of cohorts (one point) for a maximum of nine points. The MOOSE checklist evaluated the following domains: clearly identified population, clear outcomes and outcome assessment methods, selective loss of patients in the follow-up, and identification of confounding variables.

Statistical analysis

Both quantitative and qualitative analyses were performed. Data were synthesized and presented in a tabular format including study characteristics, patient demographics, preoperative anxiety and depression, measurement tools, and their associated postoperative outcomes. Studies were organized by surgery type into orthopedic, cancer, or mixed surgeries. If a study included mixed indications for surgery, they were grouped into cancer if > 50 % of surgeries were for cancer.

Meta-analyses were performed to obtain the pooled prevalence of anxiety and depression preoperatively and the odds ratios (OR) for postoperative outcomes. Meta-analysis was performed if two or more studies used similar anxiety or depression assessment tools, cutoff scores, outcomes, and/or time points. Specifically, they were only performed using the Hospital Anxiety and Depression Scale-Anxiety (HADS-A) for anxiety and Hospital Anxiety and Depression Scale–Depression (HADS-D), Geriatric Depression Scale (GDS), the Patient Health Questionnaire (PHQ-9), and the Center for Epidemiologic Studies Depression Scale (CES-D) for depression, as other assessment tools were utilized in a limited number of studies (<3).17, 18, 19, 20, 21 We performed meta-analyses using the Cochrane Review Manager Version 5.4 and R 4.4.2. A p-value of 0.05 or less was considered significant. Effect sizes were reported as OR with 95 % confidence interval (95 % CI) for binary variables and mean difference (MD) with 95 % CI for continuous variables.

Heterogeneity was quantified by I², and random-effects modelling was used. We calculated 95 % prediction intervals (PI) using τ² to reflect between-study variability. Subgroup analysis by surgery type and screening tool was also performed. We conducted leave-one-out sensitivity analyses to check the effect of each study on heterogeneity by systematically excluding studies and recalculating heterogeneity. Publication bias was assessed by Egger’s test and visual inspection of funnel plots with 10 or more studies.²²

We performed a meta-regression analysis using R 4.4.2²³ to determine the effect of age, female sex, anxiety/depression screening tools, surgery type, and sample size on the pooled prevalence of anxiety and depression. Age, female sex, and sample size were assessed as continuous variables. Screening tools and surgery type were examined as categorical variables. Meta-regression was performed when there were 10 or more included studies.²⁴

Results

Study characteristics

The results of our search are summarized in the PRISMA flowchart (Fig. 1). Of 23,226 records obtained from our original search, 15,762 titles and abstracts were screened following the removal of duplicates. After 241 articles were reviewed for eligibility, 37 met the inclusion criteria. Five articles were identified through citation searching and another four articles through the updated search, yielding a total of 46 included articles in this review.

Fig. 1.

PRISMA 2020 Study Selection Flow Diagram. Identification of studies via databases/registers and other methods.

Across forty-six studies, there were 23,628 patients with a mean age of 58.5 ± 19.2 years and 61.1 % being female. Most studies were conducted in the United States (n = 12), followed by China (n = 12), Canada (n = 4), the Netherlands (n = 3), Brazil (n = 2), France (n = 2), Serbia (n = 2), and nine other countries with one study each. Eighteen studies were orthopedic surgery,25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41 eighteen were mixed surgery,4, 8, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57 and ten were cancer surgery.32, 58, 59, 60, 61, 62, 63, 64, 65, 66 Study demographics, characteristics, and prevalence of preoperative anxiety and depression are displayed in Table 1.

Table 1.

Demographics and preoperative prevalence of anxiety and depression in patients undergoing non-cardiac surgery.

Author year	Country	Surgery (n)	Age	% Female	Anxiety Assessment criteria	Anxiety Prevalence: n (%)	Depression Assessment Criteria	Depression Prevalence: n (%)
Orthopedic
Aalders 202441	Netherlands	TKA (349)	69 ± 8.2	68.7	HADS-A ≥ 8	67 (19.1)	HADS-D ≥ 8	71 (20.3)
Zhang 202325	China	Lumbar (268)	64.7 ± 6.5	66.0	-	-	BDI−21 ≥ 10	102 (38.1)
Chan 202126	USA	Hip fracture (199)	81.9 ± 7.7	72.9	-	-	GDS−15 ≥ 6	30 (15.1)
Beaupre 202127	Canada	TKA/THA (622)	67.5 ± 10.1	59.8	-	-	CES-D ≥ 16	126 (20.3)
Ma 202128	China	THA (325)	52.2 ± 13.7	43.4	HADS-A ≥ 7	95 (29.2)	-	-
Ren 202129	China	TKA/THA (263)	74.2 ± 7.3	71.9	HADS-A ≥ 8	40 (15.2)	-	-
Wood 202130	Canada	TKA/THA (463)	67.8 ± 9.7	56.8	HADS-A ≥ 8	70 (15.1)	HADS-D ≥ 8	52 (11.2)
Mahdi 202031	Sweden	TKA (458)	69.7	52.2	HADS-A ≥ 8	70 (15.3)	HADS-D ≥ 8	44 (9.6)
Xu 202067	Australia	TKA (136)	71.5 ± 7.6	53.8	HADS-A ≥ 8	NR	HADS-D ≥ 8	NR
Visser 201933	USA	TKA (260)	61.7 ± 8.8	31.2	-	-	PHQ−9 (NR)	34 (13.1)
Datema 201834	Netherlands	Carpal tunnel release (227)	60.0 ± 17.9	74.0	-	-	CES-D ≥ 16	23 (10.1)
Culley 2017 35	USA	TKA/THA (211)	72.0 ± 6.0	60.0	-	-	GDS−15 ≥ 5	16 (7.6)
Koorevaar 201636	Netherlands	Shoulder surgery (315)	52.0 ± 16.0	47.0	4DSQ ≥ 8	43 (13.0)	4DSQ ≥ 3	24 (7.6)
Kosar 201437	USA	TKA/THA (459)	76.3 ± 5.2	59.7			GDS−15 ≥ 6	113 (24.6)
Radinovic 201438	Serbia	Hip fracture (234)	71.2 ± 8.3	72.2	-	-	GDS−15 ≥ 6	75 (32.1)
Dubljanin-Raspopovic 201468	Serbia	Hip fracture (112)	75.8 ± 6.2	81.3	-	-	GDS−30 ≥ 21	15 (13.4)
Rakel 201239	USA	TKA (215)	61.7 ± 9.8	58.1	-	-	GDS−5 ≥ 2	32 (14.9)
Givens 200840	USA	Hip fracture (126)	79.0 ± 8.0	78.6	-	-	GDS−15 ≥ 6	31 (24.6)
Mixed
Alhamdah 202556	Canada	Mixed (307)	72.3 ± 6.7	56.0	-	-	GDS−15 ≥ 5	62 (20.2)
Periañez 20258	Brazil	Abdominal (118)	48.6 ± 13.2	78.8	HADS-A ≥ 8	66 (50.8)	HADS-D ≥ 8	19 (16.1)
Chen (2) 202457	China	Thoracic (540)	55.1 ± 12.3	61.5	HADS-A ≥ 8	121 (22.4)	-	-
Gu 202342	China	Gynecologic (330)	NR	100.0	APAIS > 10	100 (30.3)	-	-
Jin 202343	China	Mixed (1065)	70.0 ± 8.0	53.8	HADS-A ≥ 8	343 (32.2)	HADS-D ≥ 8	394 (37.0)
Moss 202344	USA	Pelvic (129)	59.0 ± 15.7	100.0	BAI ≥ 10	40 (26.8)	-	-
Wang 20234	USA	Hemorrhoidectomy (340)	43.3 ± 14.4	62.1	SAS−20 > 50	51 (15.0)	SDS−20 > 53	160 (47.1)
Cheong 202145	Malaysia	Mixed (447)	73.2 ± 5.9	53.7	-	-	GDS−15 ≥ 5	130 (29.1)
Tang 202046	USA	Mixed (558)	77.0 ± 5.0	58.2	-	-	GDS−15 ≥ 6	69 (12.4)
Ospina 201947	Canada	ENT (142)	40.0 ± 20.8	43.7	-	-	PHQ−9 ≥ 10	22 (15.5)
Smith 201848	USA	Transplant (273)	59.0 ± 13.4	33.0	-	-	BDI−21 ≥ 14	56 (20.5)
Bradshaw 201649	Trinidad and Tobago	Mixed (304)	44.9 ± 13.2	67.4	HADS-A ≥ 11	131 (43.1)	HADS-D ≥ 11	82 (27.0)
Scavée 201550	Belgium	Vascular (340)	48.3 ± 14.2	71.5	HADS-A ≥ 8	61 (17.9)	-	-
Kinjo 201251	USA	Mixed (350)	73.4 ± 6.2	49.1	-	-	GDS−15 ≥ 6	52 (14.9)
Corruble 201152	France	Transplant (339)	47.3 ± 11.8	41.3	-	-	BDI−13 ≥ 4	175 (51.6)
Caumo 200253	Brazil	Abdominal (346)	44.3 ± 9.6	73.9	STAI > 46th percentile	State anxiety: 157 (45.4) Trait-anxiety: 86 (24.9)	MADRS > 13	80 (23.1)
Greene 200954	USA	Mixed (100)	64.6 ± 7.7	9.0	-	-	GDS−15 (NR)	NR
Leung 200555	USA	Mixed (219)	73.6 ± 6.2	51.6	-	-	GDS−15 ≥ 6	26 (11.9)
Cancer
Chen 202458	China	Mixede (8350)	52.3 ± 8.1	59.2	GAD−7 ≥ 5	604 (7.2)	PHQ−9 ≥ 10	910 (10.9)
Liu 202359	China	GI (120)	69.3 ± 5.3	27.5	BAI ≥ 16	23 (19.2)	-	-
Tuveri 202260	Italy	Pancreas (161)	66.0 ± 9.0	49.1	STAI ≥ 40	34 (21.1)	PHQ−9 ≥ 10	26 (16.1)
Araya 202261	Japan	Mixed (173)	63.4 ± 12.9	49.1	-	-	HADS ≥ 11	67 (38.7)
Liu 202162	China	GI (282)	65.8 ± 12	39.0	HADS-A ≥ 8	56 (19.8)	HADS-D ≥ 8	39 (13.8)
Han 202063	China	Gastric (400)	59.5 ± 11.3	43.8	HADS-A ≥ 8	112 (28.0)	HADS-D ≥ 8	87 (21.8)
Xia 202032	China	Colorectal (381)	59.4 ± 11.7	35.5	HADS-A ≥ 8	132 (34.6)	HADS-D ≥ 8	150 (39.4)
Kim 201664	South Korea	Breast (1499)	50.0 ± 10.3	100.0	-	-	CES-D ≥ 16	1212 (80.9)
Orri 201565	France	Liver (591)	58.3 ± 13.2	46.1	HADS-A ≥ 8	238 (40.3)	HADS-D ≥ 8	142 (24.0)
Kristjansson 201066	Norway	Colorectal (182)	80.0	57.1	-	-	GDS−30 > 13	18 (9.9)

Values are reported as n (%) or mean ± SD where appropriate. Mixed surgery includes a combination of general, cancer, thyroid, orthopedic, breast, thoracic, abdominal, gynecological, urological, vascular, and other surgery types.

Abbreviations: 4DSQ, Four-Dimensional Symptom Questionnaire; APAIS, Amsterdam Preoperative Anxiety and Information Scale; BAI, Beck Anxiety Inventory; BDI, Beck Depression Inventory; BPI, Brief Pain Inventory; CAM, Confusion Assessment Method; CES-D, Center for Epidemiologic Studies Depression Scale; GAD-7, Generalized Anxiety Disorder 7-item; GDS, Geriatric Depression Scale; HADS-A, Hospital Anxiety and Depression Scale-Anxiety Subscale; HADS-D, Hospital Anxiety and Depression Scale-Depression Subscale; MADRS, Montgomery-Åsberg Depression Rating Scale; PHQ, Patient Health Questionnaire; SAS, Zung Self-Rating Anxiety Scale; SDS, Zung Self-Rating Depression Scale; STAI, State Trait Anxiety Inventory; THA, total hip arthroplasty; TKA, total knee arthroplasty; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index.

Measures of anxiety, depression, and postoperative outcomes

Sixteen articles examined both anxiety and depression,4, 8, 30, 31, 32, 36, 41, 43, 49, 53, 58, 60, 62, 63, 65, 67 seven examined anxiety alone, 28, 29, 42, 44, 50, 57, 59 and twenty-three examined depression alone.25, 26, 27, 33, 34, 35, 37, 38, 39, 40, 45, 46, 47, 48, 51, 52, 54, 55, 56, 61, 64, 66, 68 The most common screening tool for anxiety was the HADS-A tool (n = 15), while other anxiety screening tools were utilized in one or two studies with different cutoffs. The most utilized screening tools for depression were the GDS (n = 14),¹ HADS-D (n = 12),¹⁷ PHQ-9 (n = 4),¹⁹ and CES-D (n = 3)²¹ (Table 1). Table 2 summarizes adverse postoperative outcomes stratified by follow-up time points, highlights the direction of association, and describes the number of contributing studies and commonly used outcome-reporting measures.

Table 2.

Overview of postoperative adverse outcomes associated with preoperative anxiety and depression.

Outcome	Follow-up time point	Anxiety	Depression	Measurement tools
Pain	Acute (POD1–7)	? (6)	↑ (7)	VAS, NRS, BPI, QoR−40 Pain
	Subacute (1–3 month)	↑ (3)	↑ (4)	VAS, NRS, BPI, WOMAC-Pain
	Long-term (6–12 month)	↑ /? (5)	↑ (9)	VAS, NRS, WOMAC-Pain, KOOS-Pain
	Extended (≥ 2 year)	? (1)	↑ /? (2)	VAS, NRS
Function	1 month	-	↓ (1)	WOMAC-Function
	3–6 month	↓ (1)	↓ (2)	OKS, KOOS-PS, WOMAC-function
	1 year	↓ /? (3)	↓ (6)	OKS, KOOS-PS, WOMAC-function, HHS, KSS, IADL, DASH
	≥ 2 year	↓ (1)	↓ (1)	OKS, KOOS-PS
Delirium	POD1–5 / by discharge	↑ /? (2)	↑ (6)	CAM, DSM-IV/medical records
Complications (all-cause)	3 month	↑ /? (4)	↑ /? (3)	Clavien-Dindo, CCI
Severe complications	3 month	↑ /? (3)	? (2)	Clavien-Dindo III+
Length of stay	Until discharge	↑ /? (6)	↑ /(6)	LOS in days (mean±SD or median [IQR])
Readmission	30–90 day	↑ /? (3)	↑ /? (2)	Hospital records (all-cause)
Mortality	30–90 d, long-term (> 6 month)	↑ /? (2)	↓ /? (4)	Medical records, registry

Prevalence of preoperative anxiety

Of the 13 studies (n = 4924) reporting preoperative anxiety (defined as HADS-A ≥7/8), the pooled prevalence was 26 % (95 % CI, 21 %–31 %; 95 % PI, 20 %–33 %; I² = 95 %)8, 28, 29, 30, 31, 32, 41, 43, 50, 57, 62, 63, 65 (Fig. 2). In the subgroup analysis by surgery type, the pooled prevalence of anxiety was 19 % (95 % CI, 14 %–23 %; 95 % PI, 13 %–26 %; I² = 85 %) in orthopedic surgery,28, 29, 30, 31, 41 35 % (95 % CI, 20 %–50 %; 95 % PI, 0.6 %–98 %; I² = 97 %) in mixed surgery,8, 43, 50 and 29 % (95 % CI, 21 %–37 %; 95 % PI, 17 %–44 %; I² = 94 %) in cancer surgery32, 57, 62, 63, 65 (Fig. 2).

Fig. 2.

Forest plot of pooled prevalence of preoperative anxiety (HADS-A ≥ 7/8) in surgical patients overall (A) and by surgery type (B). Note: Between-study variability may reflect differences in cut-off points and types of surgeries. Variation across surgical subgroups may reflect procedure-specific patient characteristics and recovery profiles. HADS-A; Hospital Anxiety and Depression Scale-Anxiety. Mixed surgery: it includes a combination of general, cancer, thyroid, orthopedic, breast, thoracic, abdominal, gynecological, urological, vascular, and other surgery types.

Prevalence of preoperative depression

Of 26 studies (n = 18,218) reporting preoperative depression, the overall prevalence was 23 % (95 % CI, 14 %–31 %; 95 % PI, 14 %–36 %; I² = 100 %), based on a pooled analysis of four commonly used screening tools (GDS, HADS-D, PHQ-9, and CES-D)17, 18, 19, 21. In the subgroup analysis by surgery type, the pooled prevalence of depression was 17 % (95 % CI, 13 %–22 %; 95 % PI, 12 %–23 %; I² = 93 %) in orthopedic surgery,26, 27, 30, 31, 34, 35, 37, 38, 40, 41 24 % (95 % CI, 15 %–32 %; 95 % PI, 16 %–35 %; I² = 97 %) in mixed surgery,8, 43, 45, 46, 47, 51, 55, 56 and 30 % (95 % CI, 4 %–56 %; 95 % PI, 7 %–70 %; I² = 100 %) in cancer surgery32, 60, 62, 63, 64, 65, 69 (Fig. 3). The pooled prevalence of depression was similar in studies using the HADS-D (21 % [95 % CI, 14 %–29 %; 95 % PI, 13 %–32 %; I² = 97 %])8, 30, 31, 32, 41, 43, 62, 63, 65 or GDS (22 % [95 % CI, 16 %–29 %; 95 % PI, 15 %–32 %; I² = 95 %])26, 35, 37, 38, 40, 45, 46, 51, 55, 56. While pooled depression prevalence was highest when using the CES-D (37 % [95 % CI, −11 %–85 %; 95 % PI, 0 %–100 %; I² = 100 %]),27, 34, 64 and lowest in studies using the PHQ-9 (13 % [95 % CI, 9 %–17 %; 95 % PI, 1.9 %–54 %; I² = 64 %])47, 58, 60 (Fig. 4, Fig. S1). The prevalence of preoperative anxiety and depression by type of surgery is summarized in a bar chart in Fig. 5.

Fig. 3.

Forest plot of pooled prevalence of preoperative depression in surgical patients overall (A) and by surgery type (B). Variation in prevalence estimates may arise from heterogeneity in different depression screening tools, scoring thresholds, and types of surgery. Variation across surgical subgroups may reflect procedure-specific patient characteristics and recovery profiles as well as differences in depression screening tools. Mixed surgery: it includes a combination of general, cancer, thyroid, orthopedic, breast, thoracic, abdominal, gynecological, urological, vascular, and other surgery types.

Fig. 4.

Forest plot of pooled prevalence of depression using validated screening tools (HADS-D ≥ 8, GDS ≥ 5/6, PHQ ≥ 9, CES-D ≥ 16). Differences between subgroups highlight how instrument choice and cut points may influence pooled estimates. Variation may also reflect heterogeneity in surgery types utilizing each tool. HADS-D: Hospital Anxiety and Depression Scale-Depression (cutoff = 8); GDS: Geriatric Depression Scale (= 5/6); CES-D: Center for Epidemiological Studies-Depression (= 16); PHQ: Patient Health Questionnaire (= 9). Surgeries include orthopedic, mixed and cancer. Mixed surgery: it includes a combination of general, cancer, thyroid, orthopedic, breast, thoracic, abdominal, gynecological, urological, vascular, and other surgery types.

Fig. 5.

Overall pooled prevalence of preoperative anxiety (A) and depression (B) by type of surgery. Mixed surgery: it includes a combination of general, cancer, thyroid, orthopedic, breast, thoracic, abdominal, gynecological, urological, vascular, and other surgery types.

Preoperative anxiety and adverse outcomes

The association between preoperative anxiety and adverse outcomes are shown in Table 2, Fig. 6A, Table S2, and Fig. S2. Postoperative pain in anxious cohorts was the most explored outcome in twelve studies.4, 28, 30, 31, 41, 42, 43, 49, 50, 57, 58, 59, 67 A meta-analysis of four studies reporting pain on postoperative day 1 showed no significant difference between anxious and non-anxious cohorts4, 28, 42, 49 (Fig. S2). However, with longer follow-up periods, anxious patients were more likely to report a significantly higher level of pain within the first five days,28, 42, 49, 50, 57 at 3 months,43, 58 6 months,⁴³ and 12 months (Table 2).30, 31 A meta-analysis of four studies showed that anxious patients had a longer LOS than non-anxious patients (MD: 0.55 days; 95 % CI, 0.29–0.82; 95 % PI, –0.04–1.14; p < 0.01, I² = 0 %) (Fig. 6A).28, 41, 57, 60

Fig. 6.

Associations between preoperative anxiety (A) and depression (B) and postoperative outcomes.

There are contradictory findings in studies examining anxiety and other postoperative outcomes. Of two studies, only one found a significant association between anxiety and postoperative delirium.28, 29 Of four studies, two studies showed a higher incidence of all-cause complications, specifically pulmonary complications, in anxious patients.50, 57, 60, 65 One study found that those with anxiety had significantly higher unplanned hospitalization rates,⁵⁰ but two studies found no significant difference.28, 60

Of three studies, one found differences in functional outcome following orthopedic surgery in anxious individuals at 1 year.30, 36, 41 Although three studies found no significant impact of anxiety on overall survival or mortality rates at 3 months,52, 60, 65 one found a significantly reduced overall survival in those with preoperative anxiety at 3 years.⁵⁹

Preoperative depression and adverse outcomes

The association between preoperative depression and adverse outcomes are shown in Table 2, Fig. 6B, Table S3, and Fig. S2. A meta-analysis of six studies (n = 632) found that depressed patients had a significantly higher incidence of postoperative delirium compared to non-depressed patients (OR 2.33; 95 % CI, 1.74–3.11; 95 % PI, 1.5–3.5; p < 0.01, I² = 0 %) (Fig. 6B).26, 37, 45, 54, 55 Pain following surgery in depressed cohorts was the most explored outcome in seventeen studies. Nine studies showed significantly higher pain scores in depressed patients until 5-year time points, while the other studies did not report significant differences.4, 25, 27, 30, 31, 33, 34, 37, 39, 43, 47, 49, 53, 58, 61, 64 Other studies exploring depression and postoperative outcomes showed mixed findings (Table 2, Table S3). Of six studies, three showed a significantly greater functional impairment in depressed patients.27, 30, 33, 36, 41, 46 One study found higher postoperative complications and another showed a greater 90-day readmission rate amongst depressed patients.60, 65 However, the meta-analyses found no association between depression and prolonged LOS, worse functional outcomes, or mortality (Fig. S2).

The number of contributing studies were shown together with the measurement instruments. Symbols indicate effect direction: ↑, higher in anxious/depressed patients; ↓, lower;?, mixed/inconclusive. Abbreviations: BPI, Brief Pain Inventory; CAM, Confusion Assessment Method; CCI, Comprehensive Complication Index; DASH, Disabilities of the Arm, Shoulder, and Hand questionnaire; DSM-IV, Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; HHS, Harris Hip Score; IADL, Instrumental Activities of Daily Living; IQR, interquartile range; KOOS, Knee Injury and Osteoarthritis Outcome Score; KOOS-PS, KOOS Physical Function Shortform; KSS, Knee Society Score; LOS, length of stay; NRS, numeric rating scale; OKS, Oxford Knee Score; POD, postoperative day; QOR-40, Quality of Recovery-40; VAS, visual analogue scale; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index.

Sensitivity analysis

We performed sensitivity analyses for the prevalence of anxiety and depression in surgical patients. The heterogeneity for the pooled prevalence of anxiety and depression was 95 % and 100 %, respectively, and was contributed to equally by all studies. In subgroup analyses, we observed that the heterogeneity for anxiety prevalence in orthopedic surgery decreased from 85 % to 0 % after removing Ma et al., 2021. ²⁸ After performing leave-one-out study analysis, there was no change in the heterogeneity for studies including depressed patients undergoing orthopedic surgery or anxious or depressed patients undergoing mixed or cancer surgery. The sensitivity analysis for the screening tools in depressed patients showed that, after removing Chen et al., 2024, the heterogeneity decreased from 64 % to 0 % for the PHQ-9 tool. ⁵⁸ There was no change observed when removing other screening tools.

We also conducted leave-one-study-out analysis for the postoperative outcomes and found that the heterogeneity for pain on postoperative day 1 in anxious patients decreased from 73 % to 51 % after removing Bradshaw et al., 2016. ⁴⁹ The heterogeneity for LOS of depressed patients also decreased from 53 % to 0 % after removing Aalders et al.⁴¹

Egger’s test and Begg’s rank test confirmed that the publication bias was not significant for anxiety (p = 0.12 and p = 0.14), which was supported by the lack of asymmetry of funnel plots (Fig. S3). For depression, while the Egger’s test was non-significant (p = 0.90), the Begg’s rank correlation test was significant (p = 0.02). However, the funnel plot inspection did not reveal obvious symmetry. (Fig. S4). Taken together, these findings suggest that there may be publication bias despite a lack of obvious asymmetry.

Meta-regression

We conducted a meta-regression analysis to explore the effect of age, female sex, sample size, screening tools, and surgery type on the pooled prevalence of anxiety and depression (Table S6). The analysis showed a positive significant relationship between age (p = 0.04) and the pooled prevalence of anxiety. No significant association was observed for other covariates of female sex, sample size and surgery type. Analysis showed that there was no significant association between pooled prevalence of depression and any of the covariates.

Qualitative assessment

All the included prospective studies were assessed using the NOS and MOOSE.15, 16 Using the MOOSE, all studies clearly identified the study population and defined the outcomes and methods of assessment.15, 16 None of the studies had selective loss of patients during follow-up periods, and all studies were adjusted for multiple confounders (Table S4). The mean NOS score was 7.3 (range 5–9) and all studies demonstrated good representation of the exposed cohort, selection of non-exposed cohort, follow-up cohort and duration (Table S5). The exposed and non-exposed cohorts scored well in comparability, as all studies controlled for important confounders. However, studies with lower NOS scores had less certainty on the methods to ascertain exposure and/or the blinding of outcome assessors.

Discussion

In this systematic review and meta-analysis, we found a high pooled prevalence of preoperative anxiety (26 %) and depression (23 %) in surgical patients. Anxiety was highest in mixed surgery at 35 %, while depression was highest in cancer surgery at 30 %. Anxiety and depression were lowest in orthopedic surgery at 19 % and 17 %, respectively. Patients with preoperative anxiety had a longer LOS, and depressed patients had twofold greater odds of developing postoperative delirium compared to their non-depressed counterparts.

Our findings are consistent with global estimates of anxiety and depression in cancer patients at 31 % and 33 %, respectively, highlighting the mental health burden faced by those with cancer.⁷⁰ In a retrospective review, there were higher rates of anxiety and depression in metastatic cancer patients than patients undergoing joint arthroplasty. ⁷¹ This might be attributed to worse prognoses and mortality rates cancer patients face where surgery offers less curative relief.⁷¹

The pooled prevalence of anxiety and depression in orthopedic surgery in our systematic review and meta-analysis was higher than two other studies on orthopedic patients. In a retrospective chart review of 22 million hip arthroscopy patients, the prevalence of anxiety and depression was 19 % and 16 %, respectively, which aligns with our findings.⁷² In a prospective study of knee arthroplasty patients, the prevalence of anxiety and depression was 20.3 % and 22.7 %, respectively.⁷³ At one year after surgery, there was a decrease in prevalence of both anxiety and depression, highlighting the benefits of knee arthroplasty on mental health.⁷³

The prevalence of depression varied amongst different screening tools. The prevalence was similar between the HADS-D and GDS at 21 % and 22 %, respectively. It was the highest using the CES-D (37 %) and lowest with the PHQ-9 (13 %).17, 18, 19, 21 These different screening tools have applications for specific populations. For example, the PHQ-9 was designed for primary care settings, while the CES-D was designed for epidemiologic surveys of general populations. Interestingly, studies show that the PHQ-9 and CES-D perform similarly, suggesting that the observed differences in our depression estimates might be driven by factors other than the intrinsic qualities of the tools themselves.⁷⁴

Our meta-analysis demonstrated that anxious patients had a half day longer LOS than non-anxious patients.28, 60 This is supported by previous studies in orthopedic, cardiac, caesarean section surgeries demonstrating correlations between preoperative anxiety and prolonged LOS.75, 76, 77, 78 Consistent with previous literature,41, 70 we showed that depression was associated with more than twofold increase in delirium.26, 37, 45, 54, 55 This is important as postoperative delirium itself is associated with higher odds of postoperative complications, unplanned intensive care unit admissions, longer LOS, non-home discharge, and mortality.⁷⁹ Our findings on pain are consistent with previous reviews, indicating a significant association between preoperative anxiety and depression and higher pain scores at 72 h, 6 months, and 12 months postoperatively.80, 81 Using meta-regression techniques, we found a significant association between preoperative anxiety and increasing age. Preoperative depression was also significantly associated with female sex, higher sample size, types of surgery, and the different screening tools.

Clinical implications

The high prevalence of preoperative anxiety and depression and their associations with adverse outcomes support the implementation of mental health screening and interventions in the preoperative setting, especially in cancer surgery. Practically, this might include implementing screening programs to identify patients that may benefit from targeted referrals, with one such program in the literature achieving a 66 % success rate in encouraging preoperative patients to receive mental health care or referrals.⁸² Nursing interventions involving psychological counseling and motivational interviewing were shown to lessen preoperative anxiety and depression with sustained postoperative benefits.83, 84 In ambulatory surgery, preoperative anxiety interventions have led to improved pain, surgical recovery, activity level, and patient satisfaction.⁸⁵ In cardiac surgery, patients who received brief cognitive behavioural intervention for anxiety and depression symptoms experienced shorter LOS.⁸⁶ Despite concerns that these measures may impose additional costs and further strain on our healthcare system, the efficacy of these interventions and reduction in healthcare costs from reduced complications and LOS highlight the importance of identifying and tailoring perioperative care to patients with anxiety and depression.

Incorporating mental health services into standard surgical protocols may allow for integrated models of care that benefit patients in their surgical journey, from preoperative management of symptoms and optimizing patient mental health, to the recovery phase and long-term outcomes. This holistic approach will lend to greater well-being and satisfaction amongst surgical patients with anxiety or depression.

Limitations

There are several limitations in this systematic review and meta-analysis. We observed very high heterogeneity in our pooled prevalence estimates. Although we performed subgroup analyses by surgery type and screening tool, meta-regression, and leave-one-out sensitivity analyses, substantial heterogeneity persisted. This likely reflects the clinical and methodological diversity of the included studies. Thus, our pooled estimates should be interpreted with caution. Additionally, while our meta-regression identified a significant association between age and anxiety prevalence, other variables such as sex, sample size, and surgery type were not significantly associated. These null findings should be interpreted cautiously, as they may reflect limitations in study reporting, heterogeneity, and underpowered subgroup analyses rather than true absence of effect.

In addition, we only included prospective cohort studies, meaning only correlational but not causal relationships can be inferred. The variability in screening instruments (e.g., HADS-A, HADS-D, GDS, PHQ-9, CES-D) and cut-off thresholds likely contributed to the heterogeneity in prevalence estimates. To improve comparability, we did meta-analysis on commonly used tools that shared similar cutoff scores and performed subgroup analyses by individual screening instruments. Nevertheless, heterogeneity remained high, and the lack of standardized screening tools remains a limitation.

Moreover, postoperative adverse outcomes were often measured using variable tools at different time points, limiting our ability to perform meta-analyses. Although our pooled ORs are based on raw unadjusted event counts, individual studies may have adjusted for different covariates. This variability in confounders likely contributes to heterogeneity and highlights the need for standardized reporting in future research. Finally, the countries represented in our review are biased towards higher-income countries. Given that preoperative anxiety and depression is nearly double in lower/middle income settings, further research is needed in patients from low socioeconomic backgrounds.87, 88

Conclusions

In this systematic review and meta-analysis, we found a high prevalence of preoperative anxiety (26 %) and depression (23 %) in surgical patients. Preoperative anxiety and depression were highest in cancer surgery and lowest in orthopedic surgery. Both anxiety and depression were associated with adverse outcomes, specifically longer LOS and postoperative delirium. However, the high heterogeneity and wide prediction intervals necessitate caution when interpreting these results. Nevertheless, preoperative screening and targeted interventions for patients with anxiety and depression would allow early risk stratification and improve perioperative care.

CRediT authorship contribution statement

Ellene Yan: Writing – review & editing, Writing – original draft, Visualization, Software, Methodology, Conceptualization. Sam Ali: Writing – review & editing, Writing – original draft, Visualization, Investigation. Joshua Andrusiak: Writing – review & editing, Writing – original draft, Visualization, Investigation. Jonathan Chung: Writing – review & editing, Writing – original draft, Visualization, Methodology, Formal analysis, Conceptualization. Frances Chung: Writing – review & editing, Writing – original draft, Visualization, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization. Aparna Saripella: Writing – review & editing, Writing – original draft, Visualization, Formal analysis. All authors have read and agreed to the published version of the manuscript.

Consent for publication

Not applicable.

Ethical statement

Not applicable.

Funding

FC reports research support from the Ontario Ministry of Health Innovation Grant, ResMed Foundation, University Health Network Foundation, and STOP-Bang Questionnaire proprietary to University Health Network. EY reports research support from the Canadian Institutes of Health Research (CIHR) Canada Graduate Scholarships Doctoral Award [grant number: FBD-195197].

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Frances Chung is an Editorial Board Member for this journal and was not involved in the editorial review or the decision to publish this article.

Appendix A. Supplementary material

Supplementary material

.

Supplementary material

.

Data availability

All data cited in this review are publicly available through the original publications listed in the references and supplementary.