Risk factors for symptomatic hematoma following cervical spine surgery: a systematic review and meta-analysis

Abstract

Objective

This study aimed to investigate the risk factors for symptomatic hematoma (SH) after cervical spine surgery, thereby providing evidence-based guidance for the early prevention, timely intervention, and appropriate management.

Methods

Relevant observational studies were retrieved from PubMed, Embase, the Cochrane Library, and Web of Science from inception to September 2025. Meta-analyses were performed to assess potential risk factors across several domains, including patient demographics, comorbidities, antithrombotic therapy, preoperative evaluation, and surgical factors. The odds ratio (OR), weighted mean difference (WMD), and 95% confidence interval (CI) were adopted to evaluate associated factors. Subgroup analyses, meta-regression, and sensitivity analyses were conducted.

Results

Seventeen studies involving 564,700 patients were included. The overall incidence of SH was 0.11% (608/564,700), with individual study estimates ranging from 0.03 to 1.51%. The meta-analysis identified the male sex (OR = 1.68, 95% CI 1.39 to 2.03), advanced age (WMD = 2.53 years, 95% CI 1.57 to 3.48), presence of ossification of the posterior longitudinal ligament (OPLL) (OR = 3.38, 95% CI 1.54,7.41), and undergoing anterior cervical corpectomy and fusion (ACCF) (OR = 1.71, 95% CI 1.26 to 2.31) as being significantly associated with an increased risk of SH after cervical spine surgery Meta-regression revealed that male proportion significantly modified the OPLL-SH association. The subgroup analysis results showed that in study populations with a lower proportion of male participants, OPLL was significantly associated with an increased risk of postoperative SH (when the male proportion < 60%, OR = 7.89, 95% CI 4.02 to 15.49); whereas no significant association was observed in study populations with a higher male proportion (when the male proportion ≥ 60%, OR = 1.42, 95% CI 0.87 to 2.33). Prolonged operative duration was associated with SH (WMD = 13.66 min, 95% CI 3.97 to 23.35), but this relationship was substantially influenced by factors related to surgical complexity, as differences in the number of surgical segments explained a substantial portion (76.96%) of the heterogeneity observed across studies. No significant associations were observed for body mass index (BMI), smoking history, common comorbidities, antithrombotic therapies, and most laboratory parameters. A statistically significant but clinically small difference was noted for preoperative PT (WMD = 0.20 s, 95% CI 0.07 to 0.33).

Conclusions

Male sex, advanced age, OPLL, and ACCF were identified as being significantly associated with an increased risk of SH after cervical spine surgery. Importantly, male sex also acts as an effect modifier, substantially influencing the association between OPLL and hematoma risk. Prolonged operative duration was linked with an increased risk of SH, though this relationship was substantially influenced by factors related to surgical complexity. These findings underscore the importance of comprehensive preoperative risk assessment that considers both individual factors and their potential interactions, alongside meticulous surgical technique, for effective hematoma prevention.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13018-025-06586-2.

Introduction

As one of the frequently performed surgeries by spinal surgeons, cervical spine surgery is associated with a low incidence of complications and a high safety profile [1, 2]. Postoperative imaging after cervical spine surgery often reveals varying degrees of soft tissue swelling, fluid collection, hematoma formation, and even compression of the dural sac, while most of these findings are asymptomatic and require no intervention [3, 4]. Nevertheless, postoperative hematoma may lead to symptoms such as acute airway obstruction (AAO) and progressive neurological deficits, and can even be life-threatening [5]. Collectively termed symptomatic hematoma (SH), it can be classified into retropharyngeal hematoma (RH) and spinal epidural hematoma (SEH) based on the hematoma site [6], being a rare yet critical complication of cervical spine surgery. Due to its potentially severe consequences, early postoperative identification and timely evacuation of the hematoma are paramount. Although multiple studies have evaluated the risk factors for SH after cervical spine surgery, their findings regarding the same factors are often inconsistent and at times contradictory. Furthermore, a comprehensive systematic review and meta-analysis that quantitatively synthesizes this evidence is lacking. To this end, this study conducted a systematic review and meta-analysis of previous relevant studies to comprehensively and quantitatively investigate the risk factors for SH after cervical spine surgery, aiming to provide evidence-based guidance for early identification and timely intervention.

Methodology

The design and implementation of this systematic review and meta-analysis adhered to the PRISMA 2020 guidelines [7], and the study protocol was registered in PROSPERO (ID: CRD420251131061).

Inclusion and exclusion criteria

Inclusion criteria

1. Observational studies: The study included cohort studies, case–control studies, or cross-sectional studies.

2. The subjects were patients undergoing cervical spine surgery, with primary surgical types including anterior cervical surgery and posterior cervical surgery.

3. The primary outcome was the occurrence of postoperative SH. Diagnostic criteria for postoperative SH were as follows: ① presence of hematoma-related symptoms such as respiratory distress, dysphagia, neurological deficits, or intractable pain after cervical spine surgery, with confirmation of postoperative hematoma via imaging or surgical exploration; or ② postoperative hematoma requiring evacuation surgery.

4. Studies reported one or more associated factors, including demographics, comorbidities, preoperative blood indices, or surgical-related factors.

Exclusion criteria

1. Reviews, letters, comments, case reports, non-English studies, and non-human studies.

2. Studies that did not provide clear diagnostic criteria for postoperative SH.

3. Different studies containing duplicated subjects published by the same authors or institutions.

4. Studies with ambiguous or unextractable data.

Data sources and search strategy

Relevant clinical observational studies on SH after cervical spine surgery were identified through systematic searching of PubMed, Embase, Cochrane Library, and Web of Science from their inception to September 2025. The search strategy combined Medical Subject Headings (MeSH) such as “cervical spine” and “hematoma” with text word terms. To avoid omissions, manual screening was performed on the references of relevant literature and previous systematic reviews.

Study selection and data extraction

Two researchers independently screened the retrieved records according to the inclusion and exclusion criteria. Titles and abstracts were reviewed for initial exclusion, followed by a full-text assessment of potentially eligible articles to determine final inclusion. The results from the two researchers were cross-checked. Any disagreement was resolved through discussion with a third researcher to reach the consensus on inclusion. Data extraction was performed for the included studies, encompassing: (1) basic characteristics of the included studies: first author’s name, publishing year, country, study type, sample size, surgical type, and diagnostic criteria for postoperative hematoma; (2) baseline data of the subjects, such as age and sex; (3) incidence of postoperative SH and associated risk factors.

Study quality assessment

The quality of included studies was assessed via the Newcastle–Ottawa Scale (NOS), with a maximum score of 9. Studies scoring ≥ 7 were considered high-quality. Two researchers independently performed the quality assessment. Any disagreement was resolved through discussion with a third researcher.

Statistical analysis

All statistical analyses were performed in Stata version 14.0 (StataCorp, College Station, TX, USA) with its official meta module. For continuous variables, the pooled effect was calculated as the weighted mean difference (WMD) with a 95% confidence interval (CI). For categorical variables, the pooled effect was expressed as the odds ratio (OR) with a 95% CI. In studies with zero events, we applied a 0.5 continuity correction to all cells to stabilize the variance. Heterogeneity was assessed by the Cochran Q chi-square test and I² statistic. I² values < 25%, 25–50%, and > 50% were considered indicative of low, moderate, and high heterogeneity, respectively. An I² value < 50% and a Q-test P-value > 0.05 indicate low heterogeneity among studies, and a fixed-effect model was adopted for the meta-analysis. Conversely, if significant heterogeneity was detected, subgroup, meta-regression, and sensitivity analyses were conducted to explore its sources. If heterogeneity could not be reduced, a random-effects model was used. Sensitivity analysis was performed to evaluate the robustness of the results. If the results were unstable, a quantitative random-effects model analysis was abandoned in favor of a qualitative systematic review. Publication bias was assessed by visual inspection of funnel plots and Begg’s and Egger’s tests. A P-value < 0.05 was considered statistically significant.

Results

Study selection

The initial database search identified 18,125 articles. After removal of 6359 duplicates and exclusion of 4355 articles such as reviews, systematic reviews, letters, and animal experiments, the remaining 7411 articles were independently screened by two researchers based on titles and abstracts (inter-rater reliability of kappa = 0.833). Following the exclusion of 7378 irrelevant articles, the full texts of the remaining articles were assessed for eligibility, of which 16 articles that did not meet the eligibility criteria were excluded. Ultimately, a total of 17 articles were included. The study selection process, following the PRISMA guidelines, is summarized in Fig. 1.

Fig. 1.

PRISMA 2020 flow diagram [7] of literature search and selection

Characteristics of included studies and quality assessment (Table 1)

Table 1.

Characteristics of included studies and quality assessment

First author	Year	Nation	Research type	Type of surgery	Diagnosis of PSH	Sample Size	Patients with PSH	PSH type	PSH formation time	NOS
O’Neill et al. [8]	2014	US	Retrospective	ACCF/ACDF	②	2392	17	RH	2.76 ± 3.13d	7
Qu et al. [15]	2024	China	Retrospective	ACDF	①②	10,615	18	RH	8.5 (4–24)h	8
Yin et al. [9]	2014	China	Retrospective	ACCF/ACDF/ Posterior surgery	②	2338	12	SEH	5.0 (0.6–15.8)h	8
Goldstein et al. [10]	2015	Canada	Retrospective	Posterior surgery	①②	529	8	SEH		8
Abola et al. [12]	2021	US	Retrospective	Anterior surgery/Posterior surgery	①②	53,233	198	SEH	3 (0–30)d	8
Xia et al. [14]	2022	China	Retrospective	Anterior surgery/Posterior surgery /Combined approach	①②	18,220	43	SEH	2.5 (1.43, 4.83)h	7
Boudissa et al. [11]	2016	France	Retrospective	Anterior surgery	①②	2319	12	RH/SEH	16 (3–72)h	7
Miao et al. [6]	2018	China	Retrospective	ACCF/ACDF	①	1258	15	RH/SEH	5 (1–90)h	7
Wang et al. [5]	2023	China	Retrospective	ACDF	①	1150	11	RH/SEH		8
Tian et al. [17]	2025	China	Retrospective	Anterior surgery	①	13,523	55	RH/SEH		8
Avetisian et al. [16]	2025	US	Retrospective	ACDF	①	430,542	140	RH	–	8
Takenaka et al. [13]	2021	Japan	Retrospective	Anterior surgery/Posterior surgery	②	5015	25	–	–	6
Masuda et al. [18]	2019	Japan	Retrospective	Anterior surgery/Posterior surgery	①②	3034	16	SEH	–	6
Schroeder et al. [19]	2017	US	Retrospective	Anterior surgery/Posterior surgery	①	16,582	15	SEH	4.67 ± 7.90d	6
Hao et al. [20]	2022	China	Retrospective	ACCF/ACDF	①	551	7	SEH	–	6
Aono et al. [21]	2011	Japan	Retrospective	ACCF/ACDF/ Posterior surgery	①	1376	5	SEH	–	6
Pivazyan et al. [22]	2025	US	Retrospective	ACDF/ Posterior surgery	①②	2023	11	SEH	–	6

①presence of hematoma-related symptoms such as respiratory distress, dysphagia, neurological deficits, or intractable pain after cervical spine surgery, with confirmation of postoperative hematoma via imaging or surgical exploration; or ②postoperative hematoma requiring evacuation surgery

ACCF, anterior cervical corpectomy and fusion; ACDF, Anterior cervical discectomy and fusion; SHE, spinal epidural hematoma; RH, retropharyngeal hematomas; PSH, postoperative symptomatic hematoma

The 17 included studies [5, 6, 8–22] encompassed 564,700 patients who underwent cervical spine surgery, among whom 608 developed postoperative SH. The basic characteristics and quality assessment results of the included studies are presented in Table 1. All studies were assessed as having moderate to high quality (NOS scores ≥ 6). However, the NOS evaluation revealed recurrent methodological limitations, primarily in the inadequate control for key confounders such as surgical method and the type of hematoma in the ‘Comparability’ domain, and the lack of reporting on non-response rates in the ‘Exposure’ domain. This review only included studies reporting risk factors for SH after cervical spine surgery and did not incorporate all available studies related to incidence of SH. Therefore, a systematic evaluation was performed for only the incidence of SH after cervical spine surgery. The incidence of SH reported across the individual studies ranged widely from 0.03 to 1.51%, yielding an overall proportion of 0.11% (608 / 564,700) across all included studies.

Results of meta-analysis on risk factors for SH following cervical spine surgery (Table 2)

Table 2.

Results of meta-analysis on risk factors for SH following cervical spine surgery

Risk factors	Subgroups	No of studies	No of patients		Heterogeneity test		Effect model	Meta-analysis results
			SH	NoSH	P value	I2(%)		Effect size	95%CI	P value
Demographic	Sex/ male	11 [5, 6, 8–12, 14–17]	529	491,425	0.174	27.6	FEM	OR = 1.68	1.39 to 2.03	0.000
	Age (years)	10 [5, 6, 8, 9, 11, 12, 14–17]	521	488,602	0.175	28.3	FEM	WMD = 2.53	1.57 to 3.48	0.000
	BMI (kg/m2)	7 [5, 6, 11, 12, 14, 15, 17]	352	55,801	0.000	92.4	REM	WMD = -0.77	−1.00 to 2.55	0.394
	Smoking history*	8 [5, 8, 9, 11, 12, 15–17]	463	487,273	0.106	39.2	FEM	OR = 1.23	1.00 to 1.52	NA*
Comorbidities	Hypertension*	8 [5, 9, 11, 12, 14–17]	489	484,984	0.098	40.5	FEM	OR = 1.18	0.95 to 1.45	NA*
	Diabetes	8 [5, 8, 9, 11, 12, 14–16]	444	487,125	0.708	0.0	FEM	OR = 1.16	0.90 to 1.48	0.247
	Cardiopathy*	3 [6, 11, 12]	214	54,184	0.961	0.0	FEM	OR = 2.91	1.12 to 7.53	NA*
	Respiratory disease	3 [5, 11, 12]	221	54,198	0.645	0.0	FEM	OR = 1.47	0.89 to 2.42	0.129
	Coagulopathy*	3 [11, 12, 16]	350	483,461	0.289	20.1	FEM	OR = 1.61	1.02 to 2.53	NA*
OPLL	Overall#	7 [5, 6, 8, 13, 15–17]	281	440,423	0.002	70.7	REM	OR = 3.38	1.54 to 7.41	0.000#
	Low male proportion (< 60%)	3 [6, 8, 16]	172	434,020	0.970	0.0	FEM	OR = 7.89	4.02 to 15.49	0.000
	High male proportion (> = 60%)	4 [5, 13, 15, 17]	109	6403	0.239	28.9	FEM	OR = 1.42	0.87 to 2.33	0.162
Antithrombotic therapy	Anticoagulant therapy	3 [5, 10, 14]	62	1746	0.843	0.0	FEM	OR = 2.48	0.81 to 7.52	0.110
	Antiplatelet therapy*	3 [5, 11, 14]	66	1249	0.287	20.6	FEM	OR = 4.42	1.54 to 12.68	NA*
	Anticoagulant or antiplatelet therapy	2 [16, 17]	195	430,622	0.291	10.2	FEM	OR = 0.73	0.22 to 2.40	0.599
Preoperative evaluation	INR*	5 [5, 9, 14, 15, 17]	139	1523	0.594	0.0	FEM	WMD = 0.01	−0.00 to 0.02	NA*
	APTT (s)*	5 [5, 9, 14, 15, 17]	139	1523	0.011	69.6	REM	WMD = 0.53	−0.71 to 1.78	NA*
	PT(s)	5 [5, 9, 14, 15, 17]	139	1523	0.386	3.6	FEM	WMD = 0.20	0.07 to 0.32	0.002
	Platelet count (109 /L)	4 [9, 14, 15, 17]	127	384	0.167	40.7	FEM	WMD = -2.33	−10.95 to 15.62	0.731
	ASA physical status classification ≥ Ⅲ*	4 [5, 11, 12, 17]	271	54,408	0.682	0.0	FEM	OR = 1.44	1.06 to 1.96	NA*
Surgical factors	ACCF	7 [5, 8, 11, 12, 14, 17, 20]	343	57,423	0.234	24.5	FEM	OR = 1.71	1.26 to 2.31	0.001
	Anterior approach*	7 [9, 12, 14, 18, 19, 21, 22]	300	96,576	0.067	49.0	FEM	OR = 0.74	0.57 to 0.97	NA*
	Operative duration (min)^	9 [5, 6, 8, 9, 11, 12, 14, 15, 17]	381	58,200	0.000	75.2	REM	WMD = 13.66	3.97 to 23.35	0.006^
	Surgical segments (Levels)*	7 [5, 6, 8, 11, 14, 15, 17]	171	5141	0.000	79.8	REM	WMD = 0.42	0.04 to 0.81	NA*
	Blood loss (ml)	6 [5, 6, 9, 14, 15, 17]	154	2763	0.139	40.0	FEM	WMD = 3.46	−2.98 to 9.90	0.292
	Drain volume (ml)	3 [5, 6, 17]	81	2599	0.699	0.0	FEM	WMD = 10.00	−10.21 to 30.21	0.332

①*Quantitative result is presented but was found to be unstable on sensitivity analysis; conclusions should be based on the qualitative discussion in the main text. ②#The overall result should be interpreted with caution due to significant heterogeneity. Refer to the subgroup analyses in the main text for further insight. ③^Due to substantial heterogeneity whose source remains unclear, the overall pooled result should be interpreted with extreme caution. A detailed interpretation is provided in the Results section of the main text. ④Bold values indicate risk factors with a statistically significant association (P < 0.05) with postoperative SH. ⑤SH, symptomatic hematoma; OR,odds ratio; WMD, weighted mean difference; CI, confidence interval; FEM, fixed effects model; REM, random effects model; NA, Not Applicable; BMI, body mass index; OPLL, ossification of the posterior longitudinal ligament; INR:international normalized ratio; APTT, activated partial thromboplastin time; PT, prothrombin time; ASA, American Society of Anesthesiologists; ACCF, anterior cervical corpectomy and fusion

Meta-analyses were performed to assess potential risk factors across several domains, including demographics, comorbidities, antithrombotic therapy, preoperative evaluation, and surgical factors.

Demographic factors

Among demographic factors, the meta-analyses for sex, age, and smoking demonstrated low to moderate heterogeneity, and thus fixed effect models were employed (Figs. 2 and 3, Additional file 1: Fig. S2). The meta-analysis for body mass index (BMI) revealed substantial heterogeneity (Additional file 1: Fig. S1). Sensitivity analysis revealed that excluding the study by Miao et al. [6] reduced the heterogeneity from 92.4 to 25.4%, without altering the statistical significance of the results. As the underlying cause of heterogeneity attributable to this study could not be definitively identified, the study was retained, and a random-effects model was adopted. The meta-analysis demonstrated that male sex (OR = 1.68, 95% CI 1.39 to 2.03) and advanced age (WMD = 2.53 years, 95% CI 1.57 to 3.48) were significant risk factors for SH after cervical spine surgery. However, no significant difference in BMI was found between the SH group and the non-SH group.

Fig. 2.

Forest plot of the association between Sex/male and postoperative SH. SH, symptomatic hematoma; SEH, spinal epidural hematoma; RH, retropharyngeal hematoma; OR, odds ratio; CI, confidence interval

Fig. 3.

Forest plot of the association between Age and postoperative SH. SH, symptomatic hematoma; SEH; spinal epidural hematoma; RH, retropharyngeal hematoma; WMD, weighted mean difference; CI, confidence interval

Sensitivity analysis indicated that the meta-analysis results for smoking were unstable. Therefore, a qualitative systematic review was provided for smoking. Six studies [8, 9, 12, 15–17] suggested that smoking did not increase the risk of postoperative SH; Wang et al. [5] reported a higher risk of SH among smokers; Boudissa et al. [11] found that smoking increased the risk of SEH but not that of RH.

Publication bias was assessed for the meta-analyses of sex and age. Both Begg’s test and Egger’s test yielded P-values > 0.05, suggesting symmetric funnel plots and a low risk of publication bias (Figs. 4 and 5).

Fig. 4.

Funnel plots of the association between Sex/male and postoperative SH. SEH, spinal epidural hematoma; RH, retropharyngeal hematoma; CI, confidence interval

Fig. 5.

Funnel plots of the association between Age and postoperative SH. SEH, spinal epidural hematoma; RH, retropharyngeal hematoma; WMD, weighted mean difference; CI, confidence interval

Comorbidities and SH after cervical spine surgery

The meta-analyses for diabetes, cardiopathy, and respiratory diseases among comorbidities showed no significant heterogeneity. The meta-analyses for hypertension and coagulopathy exhibited low to moderate heterogeneity, and thus fixed-effect models were used. The meta-analysis demonstrated that comorbidities such as hypertension, diabetes, and respiratory diseases are not significantly associated with the risk of postoperative SH (Additional file 1: Fig. S3).

Sensitivity analysis revealed unstable meta-analysis results for cardiopathy and coagulopathy, prompting a qualitative systematic review. Abola et al. [12] suggested that comorbid coagulopathy increased the risk of SH after cervical spine surgery. Other relevant studies indicated no association between cardiopathy or coagulopathy and SH after cervical spine surgery.

The meta-analysis for ossification of the posterior longitudinal ligament (OPLL) showed significant heterogeneity. To investigate the source of this heterogeneity, we performed a meta-regression analysis, which indicated that the proportion of male participants significantly influenced the association between OPLL and postoperative SH risk (β = –0.085, 95% CI –0.15 to –0.02; P = 0.016), as shown in Fig. 6. This variable accounted for 97.0% of the between-study heterogeneity, with low residual heterogeneity (I²_res = 8.55%). We subsequently divided the studies into two subgroups based on the proportion of male participants (< 60% vs. ≥ 60%). The two subgroups showed low within-group heterogeneity, but their pooled results differed significantly (Fig. 7). In the subgroup of low male proportion, OPLL was significantly associated with an increased risk of SH (OR = 7.89, 95% CI 4.02 to 15.49). In contrast, this association was absent in the subgroup of high male proportion (OR= 1.42, 95% CI 0.87 to 2.33). Given that our findings also indicate that male sex is associated with an elevated risk of postoperative SH, a high proportion of male participants in a study population appears to raise the baseline risk, thereby diminishing the discernible added risk from OPLL. These results suggest that the proportion of male participants is an important effect modifier of the association between OPLL and postoperative hematoma risk.

Fig. 6.

Bubble plot of meta-regression on male proportion modulating the association between OPLL and postoperative SH. OR, odds ratio

Fig. 7.

Forest plot of the association between OPLL and postoperative SH. OPLL, ossification of the posterior longitudinal ligament; SH, symptomatic hematoma; OR, odds ratio; CI, confidence interval

Other preoperative factors

Meta-analyses for all antithrombotic therapy subgroups showed low heterogeneity, and fixed-effect models were used (Additional file 1: Fig. S4). The meta-analysis revealed no significant association between SH after cervical spine surgery and either anticoagulant therapy or anticoagulant/antiplatelet therapy. The result for the antiplatelet therapy subgroup was statistically significant, whereas the sensitivity analysis indicated unstable results. We therefore conducted a qualitative systematic review. Only the study by Wang et al. [5] suggested that antiplatelet therapy increased the risk of SH after cervical spine surgery.

In preoperative evaluation, meta-analyses for international normalized ratio (INR), prothrombin time (PT), platelet count, and American Society of Anesthesiologists (ASA) physical status classification ≥ III showed low to moderate heterogeneity, and fixed-effect models were thus used for analysis (Fig. 8, Additional file 1: Fig. S5–S7). The meta-analysis indicated that the SH group had higher preoperative PT (WMD = 0.20 s, 95% CI 0.07 to 0.33) than the non-SH group, while preoperative platelet count showed no significant difference.

Fig. 8.

Forest plot of the association between PT and postoperative SH. PT, prothrombin time; SH, symptomatic hematoma; WMD, weighted mean difference; CI, confidence interval

The meta-analysis for activated partial thromboplastin time (APTT) showed significant heterogeneity (Additional file 1: Fig. S8). Excluding the study conducted by Yin et al. [9] reduced the heterogeneity from 69.6 to 6.6%, while the statistical significance of the results changed. Due to the non-normal distribution of continuous variables in this study, we extracted median (range) from the original text and converted them to mean ± standard deviation (SD) by using the Box-Cox method [23] for inclusion in the meta-analysis. This might be a potential source of heterogeneity. That study used the non-parametric Mann–Whitney U test and found no significant difference in APTT between groups. Among other studies, only the study by Tian et al. [17] reported higher preoperative APTT in the SH group. Furthermore, sensitivity analysis indicated unstable results for INR and ASA physical status classification ≥ III. Only Abola et al. [12] suggested that ASA physical status classification ≥ III increased the risk of SH after cervical spine surgery. Other studies found no association between preoperative INR or ASA physical status classification ≥ III and SH after cervical spine surgery.

Surgical factors

Among surgical factors, the meta-analysis for drain volume showed no heterogeneity, and the meta-analyses for having undergone ACCF (Fig. 9) and for blood loss exhibited low to moderate heterogeneity. Therefore, fixed-effect models were used. The meta-analysis indicated that patients who underwent ACCF had a higher risk of postoperative SH (OR = 1.71, 95% CI 1.26 to 2.31). However, no significant differences were observed in blood loss or drain volume between the groups (Additional file 1: Fig. S9–S10).

Fig. 9.

Forest plot of the association between ACCF and postoperative SH. ACCF, anterior cervical corpectomy and fusion; SH, symptomatic hematoma; SEH, spinal epidural hematoma; RH, retropharyngeal hematoma; OR, odds ratio; CI, confidence interval

The meta-analyses for the surgical approach (anterior approach vs. others) and the number of surgical segments (levels) showed significant heterogeneity (Additional file 1: Fig. S11–S12), and sensitivity analysis indicated unstable results. Thus, a qualitative systematic review was conducted. Xia et al. suggested that the anterior approach was associated with a lower risk of postoperative SH than other approaches (posterior or combined). Additionally, Boudissa et al. (in the RH subgroup) and Xia et al. consistently reported that more surgical segments were involved in the SH group. Other relevant studies found no significant difference in surgical approach or the number of surgical segments between the SH group and the non-SH group.

While the meta-analysis of operative duration indicated significantly longer operative duration in the SH group compared to the non-SH group (WMD = 13.66 min, 95% CI 3.97 to 23.35), significant heterogeneity was observed among the studies (Fig. 10). To explore potential sources of this heterogeneity, we performed a meta-regression using the difference in the mean number of surgical segments between groups as a continuous variable (Fig. 11). This revealed that a greater difference in surgical segments was strongly associated with a greater difference in operative time (β = 22.19, 95% CI 5.23 to 39.16, P = 0.019), explaining 76.96% of the between-study heterogeneity. It should be noted that while the meta-analysis included 10 studies, the meta-regression was based on 8 studies that provided sufficient data on intergroup differences in surgical segments. Among the two remaining studies that described compositional differences in surgical segments without reporting means, Abola et al. reported that the SH group underwent significantly more multi-level surgeries with correspondingly longer operative times, while Yin et al., despite observing a higher proportion of multi-level surgeries in the SH group, reported no statistically significant difference in operative duration between groups. When we stratified studies based on whether they reported a significant difference in either the mean number of segments or the proportion of multi-level surgeries, heterogeneity remained high (Additional file 1: Fig. S13). These inconsistent findings and the persistent unexplained heterogeneity suggest that other unmeasured factors related to surgical complexity may also influence the relationship between operative duration and SH risk. Therefore, although longer operative duration is associated with SH, the heterogeneity across studies warrants cautious interpretation of this result.

Fig. 10.

Forest plot of the association between Operative duration and postoperative SH. SH, symptomatic hematoma; SEH, spinal epidural hematoma; RH, retropharyngeal hematoma; WMD, weighted mean difference; CI, confidence interval

Fig. 11.

Bubble plot of meta-regression on the difference in surgical segments modulating operative duration. SEH, spinal epidural hematoma; RH, retropharyngeal hematoma; WMD, weighted mean difference

Discussion

SH after cervical spine surgery typically manifests acutely, with most reported cases developing symptoms within the first 24 h postoperatively [6, 9, 11, 14, 15]. Abola et al. emphasized that the clinical progression of SH depended on its location and expansion rate, not merely its presence or size, and highlighted the critical importance of vigilant monitoring during the acute postoperative phase [12]. The potential sources of bleeding for SH after cervical spine surgery include vascular structures, muscles, and exposed cancellous bones [14]. Direct triggers may involve direct or indirect vascular injury during surgical manipulation, intramuscular hemorrhage due to excessive intraoperative retraction, inadequate hemostasis or delayed bleeding secondary to poorly controlled postoperative blood pressure [24]. Additionally, patients’ underlying conditions and comorbidities may increase the bleeding risk by exacerbating vascular injury or impairing hemostasis. Unexpectedly, although factors such as hypertension, coagulopathy, and antithrombotic therapy are widely considered potential bleeding risks, this meta-analysis found no significant association between these factors and SH after cervical spine surgery. The results of this study indicated that the male sex, advanced age, presence of OPLL, prolonged preoperative PT, and undergoing ACCF are significantly associated with the risk of SH after cervical spine surgery.

This study found that the mean age of the SH group was higher than that of the non-SH group. This difference might be closely linked to age-related vascular pathological changes. Aging promotes endothelial dysfunction and degeneration of elastic fibers, leading to impaired vasomotor function and increased vascular fragility [25]. This compromises the vascular compensatory capacity against surgical mechanical trauma and increases susceptibility to intraoperative microvascular rupture and delayed postoperative bleeding [26]. Meanwhile, older patients are more likely to have comorbidities such as hypertension, diabetes, and cardiovascular diseases, and therefore often require long-term preoperative use of anticoagulants/antiplatelet, or antihypertensive medications. Even with appropriate preoperative discontinuation, residual anticoagulant effects and the timing of postoperative resumption may complicate perioperative bleeding risk management. Notably, although this study found no significant association between individual comorbidities (e.g., hypertension, diabetes, and respiratory diseases) or antithrombotic therapies and SH after cervical spine surgery, the complex interplay of multiple comorbidities may increase the risk of delayed bleeding through occult mechanisms. Furthermore, the incidence of OPLL is higher in older patients [27], and this study confirms that presence of OPLL increases the risk of SH after cervical spine surgery.

The association between sex and SH after cervical spine surgery remains controversial. This meta-analysis of 11 studies demonstrated a higher risk of postoperative SH in male patients, while only three of the included studies specifically reported data on sex differences. The underlying reasons for this sex disparity in SH incidence are not fully understood. Multiple studies suggest that the male sex is associated with an increased risk of adverse events after cervical spine surgery [28, 29]. Concurrently, the male sex is also a recognized risk factor for cervical hematoma requiring reoperation following thyroidectomy [26, 30]. Chen et al. [31] hypothesized that the greater muscle mass in the male neck might contribute to suture slippage or tearing of previously ligated vessels during emergence from anesthesia, thereby leading to hematoma formation. Intraoperatively, the more robust neck musculature in males may necessitate more extensive dissection and greater retraction force, which potentially results in increased soft tissue trauma, vascular injury, and a larger surface area susceptible to bleeding. Furthermore, the higher susceptibility of male patients to symptomatic RH may be related to sex-based anatomical differences in the upper airway. Males typically have longer airways and larger airway volumes but are at higher risk of airway collapse [32, 33]. In contrast, female airways, though shorter, exhibit more stable mechanical properties and are less prone to compression-induced collapse [34]. Qu et al. [15] also highlighted the importance of upper airway morphology in identifying the risk of symptomatic RH after cervical spine surgery. Their study identified a horizontal distance between the transverse arytenoid muscle (TAM) and the epiglottic tip ≥ 2 as an independent risk factor for symptomatic RH. They note that the airway superior to the epiglottic tip is more spacious, and the airway inferior to the TAM is protected by tracheal cartilage. However, the airway segment between the TAM and the epiglottic tip is narrower with thinner prevertebral soft tissue, which has increased the risk of compression and collapse.

Importantly, the well-established higher prevalence of OPLL in males compared to females provides a crucial link connecting male sex, OPLL, and SH risk [35–37]. Our findings reveal that males are not only associated with an increased risk of SH but also modify the effect of OPLL on SH risk. The combination of high OPLL prevalence and inherent anatomical risk factors in males creates a high baseline risk background, making the additional risk contribution from OPLL less discernible in predominantly male populations. This mechanism explains why OPLL demonstrates a strong association with SH in studies with lower male proportion, while this effect is attenuated in studies where males predominate. Compared to the patients with cervical spondylotic myelopathy, those with OPLL exhibit an increased bleeding tendency [38], potentially related to the pathological process of heterotopic ossification initiated via the endochondral pathway. Vascular invasion of the cartilaginous template plays a pivotal role in the pathophysiological endochondral ossification of bone tissue [39]. Kamogawa et al. [40] used magnetic resonance imaging (MRI) for three-dimensional visualization of the intraspinal venous plexus in cervical OPLL patients, and identified critical vascular pathological changes including congestion, atrophic collapse, and hemodynamic impairment. They proposed that increased intravascular resistance within stenotic spinal canal segments led to hemodynamic stress and elevated pressure, thereby causing persistent mechanical trauma that may induce asymmetric abnormal dilation (relative expansion and congestion) of the intraspinal venous plexus. This pathological process may result in venous plexus degeneration and reduced vascular integrity, potentially triggering venous reperfusion injury after intraoperative decompression—rupture of chronically compressed and atrophied venous plexuses due to abrupt pressure changes. The primary distribution of this venous plexus along the ventrolateral spinal cord, adjacent to neural structures, further complicates effective hemorrhage control. Indeed, abnormal structure and pathology of the intraspinal venous plexus have been recognized as one of the primary causes of spontaneous spinal epidural hematoma [41]. Moreover, the surgery for OPLL patients is typically more complex. A larger exposure area and longer operative duration will inevitably increase perioperative bleeding risk. Meanwhile, the spinal dura mater in OPLL patients often adheres densely to the ossified ligaments and may even be complicated by dural ossification. Surgical dissection carries a high risk of dural tear and consequent cerebrospinal fluid leakage [42, 43]. Yang et al. proposed that the spinal dura mater possesses inherent tension to resist hematoma pressure. As long as hematoma pressure does not exceed this tension threshold, clinical symptoms may not manifest despite the presence [44]. However, a dural tear weakens or eliminates this mechanical barrier, allowing hematoma to directly compress the spinal cord and cause neurological deficits.

Among surgical factors, our analysis identified ACCF as clearly associated with increased SH risk. In contrast, the relationships of surgical approach, surgical segments, and operative duration with SH risk remained inconclusive, though all appear to be linked to the surgical complexity. ACCF is typically indicated for multi-level conditions or OPLL. The procedure involves operating on more segments and resecting vertebral bodies with wider exposure, which not only increases the area of bleeding bony surfaces and the risk of bleeding from cancellous bone and venous plexus injury, but also prolongs operative duration and introduces multiple additional risk factors. Longer surgical duration subjects soft tissues and muscles to extended periods of mechanical retraction, thereby leading to muscular bleeding [6]. Prolonged irrigation further compounds the bleeding risk by directly diluting the local concentration of coagulation factors and impairing the efficiency of the coagulation cascade. In vitro experiments demonstrated that dilution of plasma with normal saline progressively reduces peak thrombin generation [45]. Operative duration exceeding 2 h increases the risk of hypothermia [46], which not only inhibits platelet aggregation and thrombin activity [47] but also significantly delays the initiation time of thrombin generation [48]. Both PT and APTT are significantly affected by hypothermia and hemodilution [49]. Notably, hypothermia and hemodilution exert synergistic effects on coagulopathy [48], further increasing the bleeding risk. Furthermore, surgical team fatigue during prolonged complex procedures demands consideration. Diminished attention and fine motor control in lengthy operations can directly impact the precision and effectiveness of hemostasis, creating a vicious cycle where prolonged complex operations promote fatigue, which in turn heightens technical challenges and bleeding risk. Therefore, surgical complexity should be comprehensively considered through multiple aspects, including surgical techniques, equipment requirements, operative duration, surgical segments, and surgical approach. Future research should focus on developing reliable assessment tools that integrate these diverse elements to enhance risk evaluation and facilitate personalised surgical planning, ultimately improving procedural safety in complex cervical spine surgery.

This study observed that preoperative PT was significantly prolonged in the SH group than in the non-SH group (WMD = 0.20 s, 95% CI 0.07 to 0.33). However, the small effect size may lack substantial clinical significance because only the PT prolongation exceeding 3 s beyond normal is typically considered clinically significant in practice. Similarly, the lack of significant differences in APTT and INR between groups, consistent with previous research, supports the limited predictive value of routine coagulation screening (PT/APTT) for surgical bleeding risk [50]. Notably, the absence of significant association between comorbidities such as hypertension or coagulopathy and SH in this study might be attributed to how comorbidities were primarily defined based on medical history in the included studies. Although coagulopathy increasing perioperative bleeding risk is a well-established pathophysiological basis, such patients in the actual included studies likely received standardized perioperative coagulation management and were deemed to have controllable surgical risk. Patients with severe preoperative coagulopathy (e.g., PT or APTT significantly exceeding the reference ranges) or active hemorrhagic disorders were likely excluded. For hypertensive patients, optimal control of preoperative hypertension and effective suppression of acute postoperative fluctuations—such as blood pressure spikes triggered by pain, emergence from anesthesia, or fluid shifts—may be more critical than a history of hypertension alone. This perspective aligns with the finding of Tian et al., who identified the peak mean arterial pressure during recovery period as an independent risk factor for postoperative SH, while a history of hypertension and preoperative mean arterial pressure showed no significant effect [17]. Furthermore, although no significant difference in drain volume was observed between groups in our analysis, the strategic use of drainage might still influence hematoma prevention. Liu et al. found that drainage tube placement was associated with a significantly lower incidence of postoperative epidural hematomas in single-segment thoracolumbar surgeries [51].

Strengths and limitations

To our knowledge, this is the first meta-analysis investigating the incidence and risk factors of SH after cervical spine surgery, and it provides evidence-based insights for this field. However, several limitations should be acknowledged. Firstly, the evidence is derived predominantly from retrospective observational studies, which are inherently susceptible to unmeasured confounding and selection bias, thereby precluding causal inferences and moderating the certainty of our conclusions. Quality assessment revealed common weaknesses in the inadequate control for key confounders and the lack of reporting on non-response rates, indicating potential information bias. Second, despite our efforts to investigate heterogeneity through subgroup and meta-regression analyses, significant unexplained heterogeneity persisted for some outcomes (e.g., operative duration), necessitating cautious interpretation. Furthermore, the planned subgroup analysis by hematoma type was limited as many original studies did not provide stratified data, and consequently, subgroup results could not be presented. Finally, the restriction to English-language publications may have introduced language bias and limited the global applicability. Therefore, high-quality, large-scale prospective cohort studies are warranted to further explore the risk factors for SH after cervical spine surgery.

In summary, the overall proportion of SH after cervical spine surgery was 0.11%, with reported incidences across studies ranging from 0.03 to 1.51%. The male sex, advanced age, presence of OPLL, and undergoing ACCF were identified as significant risk factors for postoperative SH. Notably, male sex not only represents an independent risk factor but also significantly modifies the association between OPLL and SH risk, underscoring the importance of considering such interactions in risk assessment. Similarly, although preoperative PT demonstrated statistical significance, its minimal effect sizes suggest limited standalone predictive value in clinical practice. These findings collectively highlight that comprehensive preoperative evaluation should extend beyond individual risk factors to incorporate both overall surgical complexity and potential effect modifications between factors. Meticulous intraoperative hemostasis combined with systematic perioperative management represents the cornerstone of hematoma prevention and can effectively mitigate the risk associated with patients’ inherent mild bleeding tendencies. Surgeons should enhance preoperative risk assessment, prioritize intraoperative hemorrhage control and ensure adequate hemostasis, and closely monitor perioperative neurological function and blood pressure fluctuations, and intervene promptly to minimize the risk of SH.

Supplementary Information

Below is the link to the electronic supplementary material.

Authors’ contributions

ChenGuang Wang and ChengHan Xu conceived and designed the experiments, performed the experiments, analyzed the data, prepared figures and/or tables, authored drafts of the article, and approved the final draft. YinGang Zhang conceived and designed the experiments, reviewed drafts of the article, and approved the final draft.

Funding

This research was sponsored by Shaanxi Provincial Science and Technology Development Program Project (No.2025SF-YBXM-449).

Data availability

All authors and institutions can access the raw data by directly contacting the author at Gary123965@outlook.com by reasonable request.

Declarations

Ethics approval and consent to participate

This study does not require approval from the Research Ethics Committee (REC) or the Institutional Review Board (IRB) of the institution where the experiment was performed.

Consent for publication

All authors have given their informed consent for the publication.

Competing interests

The authors declare that they have no competing interests.