The Rate and Risk Factors of Deep Brain Stimulation–Associated Complications: A Single-Center Experience

Abstract

BACKGROUND AND OBJECTIVES:

Deep brain stimulation (DBS) is an established neurosurgical treatment of a variety of neurological disorders. DBS is considered a safe and effective neurosurgical procedure; however, surgical complications are inevitable, and clinical outcomes may vary. The aim of this study was to describe DBS complications at a large clinical center in the United States and to investigate the relationship between patients' baseline characteristics, surgical technique, and operative complications.

METHODS:

We identified all patients who underwent DBS lead implantation at our center between 1st January 2012 and 1st January 2020. We extracted relevant information regarding patient demographics, surgical details, clinical complications, and clinical outcomes from the electronic medical records.

RESULTS:

A total of 859 leads were implanted in 481 patients (153 men, 328 women). The mean patient age at the time of the surgery was 65 years, with the mean disease duration of 13.3 years. There were no mortalities and 57 readmissions within 30 days (mean = 14.2 days). The most common complications included pneumocephalus (n = 661), edema (n = 78), altered mental state (n = 35), implantable pulse generator discomfort (n = 34), hemorrhage (n = 26), and infection (n = 23). Most notably, the use of general anesthesia, hypertension, heart disease, and depression were associated with significantly longer postoperative stay. High preoperative body mass index was associated with higher rates of surgery-related infections and lead revision/explantation. The intraoperative mean arterial pressure, anesthesia type, and frame apparatus were all important predictors of postoperative pneumocephalus.

CONCLUSION:

In this report, we described the rates and types of complications associated with DBS surgery at a large neurosurgical center in the United States. The novel insights highlighted in this study present an opportunity to further improve the clinical outcomes and patient selection in DBS surgery.

ABBREVIATIONS:

GPi

globus pallidum

IPG

implantable pulse generator

MAC

monitored anesthesia care

MAP

mean arterial pressure

STN

subthalamic nucleus.

Deep brain stimulation (DBS) is a well-established minimally invasive neurosurgical technique used to treat a variety of neurological disorders. The technique involves precise delivery of electrical current to subcortical brain structures, thus strengthening the synaptic connections and modifying the local tissue excitability.^1-3 The range of anatomical targets and clinical indications for DBS is constantly expanding—the technique is currently approved for the treatment of Parkinson disease, essential tremor, severe refractory epilepsy, and obsessive-compulsive disorder.⁴ The most common electrode targets include the subthalamic nucleus (STN), ventral intermediate nucleus of the thalamus, and globus pallidum internus (GPi),⁵ with many other undergoing preclinical and clinical evaluations.^6-10

DBS is generally considered a safe neurosurgical procedure¹¹; however, surgical complications are inevitable.¹² According to estimates, roughly 2.7% to 3.4% of DBS surgeries will result in intracranial complications.^13,14 Because of morbidity and mortality associated with these complications, it is of great interest to the neurosurgical community to better understand the patient and surgery-related factors that may increase or decrease the likelihood of developing such outcomes. Ultimately, this would help broaden the reach of DBS therapy and provide valuable treatment where it is needed the most.

AIM

The aim of this study was to evaluate the rate of operative complications associated with DBS surgery at a large neurosurgical center from January 1st, 2012, to March 1st, 2021. In addition, we aimed to gain a better understanding of the relationship between patients' baseline characteristics and surgical technique and their influence on operative complications. Finally, we critically appraised the results within the context of the published literature as well as discussed patient selection and surgical technique strategies that may mitigate the overall complication risk.

METHODS

This study was conducted in line with the “Strengthening the Reporting and Observational Studies in Epidemiology” statement.¹⁵

We performed a retrospective analysis of all perioperative complications (ie, within 30 days of surgery) in patients who underwent DBS lead implantation between 1st January, 2012, and 1st March, 2021, at a large neurosurgical center. During this period, 3 surgeons were actively involved in the placement of DBS leads. The exclusion criteria comprised (1) patients with incomplete or inaccessible data points required for the analysis, (2) individuals younger than 18 years, and (3) patients undergoing surgical lead DBS revisions. The latter 2 groups were excluded because of their unique complication risk profile, which could have biased the study.^16,17

The full list of data points extracted from the electronic medical records and used in this analysis is present in the Supplemental Digital Content (Part 1), http://links.lww.com/ONS/B146.

Ethical Consideration

The primary authors received unrestricted access to the institutional DBS database. The data were anonymized at the extraction stage and safely stored in a protected network. The institutional review board exemption was granted because of the retrospective nature of this study and the lack of identifiable patient data after extraction. Patient consent was not required.

Statistical Analysis

Descriptive statistics were used to summarize the data: means and SDs for normally distributed continuous variables and medians and interquartile ranges for skewed variables. Categorical variables were presented as frequencies and percentages.

General linear mixed-effects models were used to investigate associations between predictor variables and time to discharge, with a random intercept for each patient. Separate generalized linear mixed-effects models were conducted for predictors (except 3: asthma, surgical target, and bilateral leads) and pneumocephalus occurrence. Because of convergence issues, simple logistic regression was used for the remaining predictors. For binary outcomes (except pneumocephalus), logistic regression was chosen over mixed-effects models because of an imbalance between categories. Repeated measures were handled by including only the first surgery for patients with multiple surgeries and by treating single surgeries with both sides done on the same day as 1 surgery. Because of the exploratory nature of the study, correction for multiple testing was not applied.^18,19 All the statistical analyses were designed and executed by a biostatistician at our institution, with the full report enclosed in the Supplemental Digital Content (Part 2), http://links.lww.com/ONS/B147. The analysis was conducted in R statistical software (The R Project for Statistical Computing), and the P-value of <.05 was considered statistically significant.

RESULTS

The total sample included 481 patients with 859 leads. There were additional 84 patients who were excluded from the analysis because of missing vital information data points such as body mass index (BMI), disease duration, intraoperative blood pressure, and procedure duration. The patients' characteristics and their respective surgical outcomes are further summarized in Tables 1 and 2.

TABLE 1.

Patients' Demographics and Surgical Details

Demographics and surgical details	Per patient	Per lead
General
Age (median, IQR)	65 (59-70)	65 (59-70)
Sex	Female—153 (%)	Female—280 (%)
	Male—328 (%)	Male—579 (%)
Body mass index (mean, SD)	27.9 (5.84)	28.4 (5.90)
Disease duration (mean, SD)	13.3 (12.3)	13.4 (11.9)
Clinical comorbidities
Hypertension	210 (43.5%)	397 (46%)
Depression	140 (29%)	234 (27%)
Heart disease	95 (20%)	166 (19%)
Anxiety	87 (18%)	156 (18%)
History of cancer	57 (12%)	107 (12%)
Diabetes	52 (11%)	99 (11.5%)
Chronic pulmonary disease	33 (7%)	53 (6%)
Asthma	27 (5.5%)	55 (6.5%)
Clinical indications
Parkinson	333 (69%)	598 (70%)
Essential tremor	113 (24%)	204 (24%)
Dystonia	13 (2.5%)	23 (2.5%)
Central pain syndrome	12 (2.5%)	16 (2%)
Stroke	4 (1%)	8 (1%)
Obsessive compulsive disorder	3 (0.5%)	4 (0.5%)
Tourette	3 (0.5%)	4 (0.5%)
Neuroacanthocytosis	2 (0.5%)	2 (0.5%)
Surgical details
Anesthesia type	Local/MAC—427 (89%)	Local/MAC—755 (88%)
	General—54 (11%)	General—103 (12%)
Laterality	Bilateral—386 (80%)	Bilateral—711 (84%)
	Unilateral—95 (20%)	Unilateral—148 (16%)
Stereotactic technique	Leksell—403 (84%)	Leksell—735 (85.5%)
	Starfix—71 (14.5%)	Starfix—114 (13%)
	Frameless—7 (1.5%)	Frameless—10 (1.5%)
Intraoperative MAP high/low (mean, SD)	High—107.02 (17.60)	High—106.84 (17.63)
	Low—64.19 (7.66)	Low—64.08 (7.52)
Surgical targets
STN	279 (58%)	490 (57%)
VIM	116 (24%)	210 (24.5%)
GPi	69 (14%)	131 (15%)
VC/VS	11 (2.5%)	18 (2%)
DN	4 (1%)	8 (1%)
VPM-VPL	2 (0.5%)	2 (0.5%)

DN, dentate nucleus; GPi, globus pallidum; MAC, monitored anesthesia care; MAP, mean arterial pressure; STN, subthalamic nucleus; VC, ventral capsule; VIM, ventral intermediate nucleus of the thalamus; VPL, ventral posterolateral (nucleus of the thalamus); VPM, ventral posteromedial (nucleus of the thalamus); VS, ventral striatum.

The percentage values are rounded to the nearest 0.5%.

TABLE 2.

Complication Rates in Deep Brain Stimulation Surgery (Within the First 30 Days of Surgery)

Complications	Incidence (per patient)	Incidence (per lead)
General
Pneumocephalus	352 (73%)	661 (77%)
Edema	39 (8%)	78 (9%)
Hemorrhage	17 (3.5%)	23 (2.5%)
Infection	13 (3%)	23 (2.5%)
Stroke	2 (0.5%)	3 (0.5%)
Pulmonary embolus	1 (0.5%)	3 (0.5%)
Myocardial infarction	1 (0.5%)	1 (0.5%)
Systemic
Vomiting	10 (21%)	20 (2.5%)
Urinary retention	7 (1.5%)	13 (1.5%)
Hypotension	5 (1%)	12 (1.5%)
Respiratory distress	3 (1%)	5 (0.5%)
Hematuria	2 (0.5%)	8 (1%)
Arrhythmia	2 (0.5%)	4 (0.5%)
Neurological
Altered mental state	18 (4%)	35 (4%)
Headache	12 (2.5%)	25 (3%)
Altered consciousness	9 (2%)	15 (2%)
Slurred speech	7 (2%)	8 (1%)
Seizure	4 (1%)	8 (1%)
Vasovagal attack	3 (1%)	7 (1%)
Paresthesia	2 (0.5%)	5 (0.5%)
Paresis/paraplegia	2 (0.5%)	2 (0.5%)
Hardware-related
IPG discomfort	18 (4%)	34 (4%)
IPG erosion	5 (1%)	9 (1%)
Lead erosion	2 (0.5%)	4 (0.5%)
Lead migration	1 (0.5%)	2 (0.5%)
Lead failure	1 (0.5%)	1 (0.5%)
IPG failure	1 (0.5%)	1 (0.5%)
Extension fracture	0 (%)	0 (0%)
Extension malfunction	0 (%)	0 (0%)
Lead fracture	0 (%)	0 (0%)

IPG, implantable pulse generator.

The percentage values are rounded to the nearest 0.5%.

The impact of demographic characteristics, medical history, surgical factors, and procedural details on surgical outcomes was evaluated.

Time to Discharge

The mean and median time to discharge were 1.68 and 1 days, respectively (SD 1.35; range: 0–17). On average, patients with essential tremor were discharged 0.26 days earlier compared with the patients with Parkinson disease (P = .01). Hypertension, heart disease, and depression were found to significantly affect length of stay, resulting in 0.23 day (P = .01), 0.23 day (P = .04), and 0.22 day (P = .03) longer stays, respectively. In addition, the use of the Leksell (Elekta) system was associated with, on average, a 0.75 days’ shorter time to discharge compared with those who underwent the surgery with the Starfix device (FSH Inc) (P < .0001). Patients with local + monitored anesthesia care (MAC) anesthesia had 0.83 days’ shorter time to discharge compared with those who underwent general anesthesia (P < .0001). Interestingly, higher BMI was associated with shorter length of stay, with 0.02 shorter stay per each BMI unit (P < .0001). Post hoc analysis revealed that anesthesia type confounded the relationship between BMI and length of stay because patients with higher BMI more frequently received local + MAC (associated with shorter time to discharge) instead of general anesthesia. When controlling for anesthesia type, the association between BMI and time to discharge was eliminated. Finally, the choice of GPi as the surgical target was associated with a significantly longer time to discharge compared with both STN and ventral intermediate nucleus of the thalamus (0.71 days, P < .0001; 0.87 days, P < .0001) (Figure 1).

FIGURE 1.

Forest plot illustrating the relationship between predictor variables and the number of days to discharge. BMI, body mass index; ET, essential tremor; GPi, globus pallidum; MAC, monitored anesthesia care; MAP, mean arterial pressure; PD, Parkinson's Disease; STN, subthalamic nucleus; VIM, ventral intermediate nucleus of the thalamus.

Rate of Readmission Within 30 Days

A total of 57 unplanned readmissions within 30 days after surgery were observed within this cohort. The mean and median time to readmission were 14.20 and 10 (SD: 9.88; range: 1–30). Importantly, none of the evaluated predictor variables were correlated with the rate of readmission within 30 days of surgery.

Lead Revision or Explantation

A total of 27 patients had their lead surgically revised while 15 had their lead permanently explanted. The reasons included hardware infection (n = 16), loss of efficacy (n = 13), erosion (n = 3), end of a research study (n = 5), hemorrhage (n = 1), discomfort (n = 1), and elevated impedance (n = 1). In addition, in 2 cases, the clinical notes were insufficient to discern the exact reason for revision/explanation. The analysis suggests that the risk of requiring lead revision and explanation is diminished by increasing age (odds ratio [OR] 0.95, CI: 0.93–0.98; P < .0001) and increased by increasing BMI (OR 1.07, CI: 1.01–1.14; P = .02). In addition, the patients who received local + MAC anesthesia had lower rates of lead revisions and explanations compared with those who underwent the procedure under general anesthesia (OR 0.35, CI: 0.14-0.86; P = .02) (Figure 2).

FIGURE 2.

Forest plot illustrating the relationship between predictor variables and the risk of lead revision and/or explantation. BMI, body mass index; ET, essential tremor; GPi, globus pallidum; MAC, monitored anesthesia care; MAP, mean arterial pressure; PD, Parkinson's Disease; STN, subthalamic nucleus; VIM, ventral intermediate nucleus of the thalamus.

Wound Infection

A total of 23 events of wound infections were reported. The risk of infection was correlated with the patient's BMI (OR 1.1, CI: 1.01-1.19; P = .02), with the odds of developing an infection increasing by 10% for every unit increase of BMI. In addition, the infection risk was lower in older individuals (OR 0.96, P = .02) (Figure 3).

FIGURE 3.

Forest plot illustrating the relationship between predictor variables and the risk of wound infection. BMI, body mass index; ET, essential tremor; GPi, globus pallidum; MAC, monitored anesthesia care; MAP, mean arterial pressure; PD, Parkinson's Disease; STN, subthalamic nucleus; VIM, ventral intermediate nucleus of the thalamus.

Hemorrhage

A total of 23 events of postoperative hemorrhage were reported on computed tomography (CT) imaging acquired within the initial 24 hours after the surgical procedure. This included 16 intraparenchymal, 4 subdural, 2 intraventricular, and 1 subarachnoid hemorrhage. The choice of the target seems to have influenced the risk of developing a hemorrhagic event, with the odds of experiencing hemorrhage in STN patients being significantly lower than GPi (OR 0.31, CI: 0.10-0.91; P = .03) (Figure 4). There were no significant differences in rates of hemorrhage when comparing STN with thalamus or GPi with thalamus.

FIGURE 4.

Forest plot illustrating the relationship between predictor variables and the risk of intracranial hemorrhage. BMI, body mass index; ET, essential tremor; GPi, globus pallidum; MAC, monitored anesthesia care; MAP, mean arterial pressure; PD, Parkinson's Disease; STN, subthalamic nucleus; VIM, ventral intermediate nucleus of the thalamus.

Edema

A total of 78 events of postoperative edema, including perilead edema, were reported on CT imaging acquired within the initial 24 hours after the surgical procedure. Importantly, none of the evaluated predictor variables were correlated with the likelihood of developing this complication.

Pneumocephalus

Pneumocephalus was the most common radiological complication, with as many as 616 unique observations on postoperative CT scans. The use of the Leksell (Elekta) frame was associated with approximately 0.493 times lower incidence compared with Starfix (FSH Inc) (P = .01). In addition, the use of the local + MAC anesthesia protocol was associated with 3.68 higher odds of developing pneumocephalus, compared with general anesthesia (P < .0001). Finally, the risk of developing pneumocephalus was higher in patients with low intraoperative mean arterial pressure (MAP) recordings (MAP maximum—OR 0.98, CI: 0.95–0.99, P = .04; MAP minimum—OR = 0.97, CI 0.94–0.99, P = .01).

Implantable Pulse Generator Discomfort

A total of 34 events of implantable pulse generator (IPG) discomfort were noted in the medical records. Importantly, none of the evaluated predictor variables were correlated with the risk of developing IPG discomfort.

Altered Mental State

A total of 35 events of altered mental state were observed, which included confusion, disorientation, and delirium. The odds of developing an AMS were 3.21 times higher in patients with coexisting cancer (OR 3.21, CI: 1.19–8.64, P = .02) and 4.26 times higher in patients with depression (OR 4.26, CI: 1.73–10.52; P = .001). The incidence of altered mental state was not affected by the surgical target (OR 0.99, 0.99, 1.00).

Nausea and Vomiting

A total of 20 incidences of postoperative nausea and vomiting were noted in the medical records. Importantly, none of the evaluated predictor variables were correlated with the risk of developing this outcome.

DISCUSSION

In recent years, several isolated attempts have been made to better understand the factors influencing the rate of DBS complications in adult population. For instance, a meta-analysis conducted by Kantzanou et al²⁰ identified the use of powder vancomycin as an important preventative factor for the development of post-DBS surgical site infections. Similarly, a team led by Atchley et al²¹ identified age at surgery as an independent risk factor for the development of DBS implantation-related seizures. In a separate investigation, a team from Shanghai identified low preoperative Mini-Mental State Examination score as an independent risk factor of postoperative delirium²² while a meta-analysis conducted by Tiefenbach et al²³ suggested that age and history of hypertension may be associated with an increased risk of perioperative intracranial hemorrhage.

The results presented within this report represent a robust attempt to further contribute to this field of research and better understand patient and surgery-related factors as they relate to DBS complications. The remaining of the Discussion section will primarily focus on the key findings identified by our analysis and its potential relevance to clinical practice.

DBS surgery is traditionally performed as an inpatient surgery, with most patients being discharged on the following day.²⁴ This is consistent with the policy at our institutions, with a mean and median time to discharge of 1.68 and 1 days, respectively. Prolonged hospital stay is traditionally associated with postoperative complications, which require active management and monitoring. Importantly, we found that certain comorbidities, such as hypertension, heart disease, and depression, were associated with a longer length of stay. In addition, we found that the use of general anesthesia led to a longer length of stay, possibly because of a higher risk of anesthesia-associated adverse outcomes.²⁵ A trend toward longer length of stay was further found among overweight patients, although this finding was confounded by anesthesia type. While a longer length of stay was observed in patients who had their lead placed using the Starfix (FSH Inc) apparatus this should be interpreted cautiously as this device was primarily utilized by 1 (out of 3) active surgeons. Finally, longer length of stay was recorded in patients who had their lead implanted into GPi and in patients with Parkinson disease, possibly because of the former group frequently requiring bilateral implantation under general anesthesia and the latter being overall a highly fragile subpopulations.

The discharged patients may infrequently be readmitted, should they experience additional complications. The reported readmission rate within 30 days of surgery in the literature is 6.6%,²⁶ which is consistent with our observations (57/859; 6.6%). Interestingly, none of the evaluated predictor variables were correlated with the risk of readmission. Another infrequent outcome involves lead revision or explantation, required because of an infection, loss of efficacy, or other reasons.²⁷ Our analysis suggests that older patients were less likely to require lead revision/explantation, possibly because of their shorter expected lifespan. Patients with high BMI were also at risk of needing lead revision/explantation, likely because of their higher propensity of developing bacterial infections. Finally, patients who underwent surgery with local + MAC anesthesia had lower rates of revision/explantation. However, this peculiar finding may be explained by a research study at our institution where 5 subjects had their leads placed under general anesthesia and later explanted, thus skewing the results.

A total of 23 events of infection were reported, representing 2.6% (23/859) of the entire cohort. This is at the lower range of the reported range,²⁸ suggesting that the infection prevention techniques practiced at our institution, such as the use of vancomycin powder, have proven effective. Furthermore, the risk of infection was correlated with patients' BMI, which is a well-known association within the broader medical literature.²⁹ In addition, 23 events of surgery-related hemorrhages were identified, including 16 intracranial, 4 subdural, 2 intraventricular, and 1 subarachnoid hemorrhage. Interestingly, the risk of developing hemorrhage within our cohort was significantly higher in instances when the lead was implanted into the GPi, as opposed to STN—an observation suggested by Binder et al in 2005,³⁰but not definitively confirmed in a meta-analysis by Tiefenbach et al in 2023.²³

CT imaging acquired within the initial 24 hours after the surgical procedure identified 78 events of postoperative edema and 616 events of largely asymptomatic, postoperative pneumocephalus. Although none of the evaluated factors affected the risk of developing edema, several variables were associated with pneumocephalus. First, the risk of developing pneumocephalus was higher in patients with low intraoperative MAP, a finding recently identified by Wu et al.³¹ The likely explanation is that a low intraoperative MAP leads to a lower intracranial pressure, thus creating less resistance for the air to enter the intracranial cavity during the procedure. Furthermore, the use of the Leksell (Elekta) frame was associated with a lower incidence of pneumocephalus, when compared with Starfix (FSH Inc), possibly because of nuanced differences in patient positioning.³¹ Finally, local + MAC anesthesia was associated with significantly higher odds of pneumocephalus (OR 3.68), possibly because of more intraoperative motion, when contrasted with general anesthesia.

A total of 34 events of IPG discomfort and 20 events of postoperative nausea/vomiting were recorded, with none of the evaluated factors predictive of these outcomes. Furthermore, a total of 35 events of altered mental status were reported, including confusion, disorientation, and delirium. The odds of developing such an outcome were higher in patients with coexisting cancer (OR 3.21) and depression (OR 4.26), associations which are well recognized in the broader medical literature.^32,33

Strengths and Limitations

Finally, it remains essential to briefly acknowledge the main strengths and limitations of this report. First and foremost, the analysis included a high number of observations (ie, 481 patients with 859 unique leads), which allowed for a robust statistical analysis. In addition, the analysis was designed and executed by an experienced biostatistician, adding further validity to these results. As for limitations, the relative lack of “positive” outcomes for some complications precluded their analysis, thus leaving the primary aim of this work, to some degree, incomplete. Furthermore, it is important to acknowledge that all observations hail from a single center, which may thus, to an extent, limit the external generalizability of these findings. Finally, the analysis contained 5 patients who were part of an experimental study involving dentate nucleus and requiring subsequent explanation, which may have biased the revision/explanation subanalysis.

CONCLUSION

We presented the rates and types of complications associated with DBS surgery at a large tertiary center in the United States. More importantly, we evaluated the effects of many patient and surgical factors and their respective impact on surgical complications. Altogether, the reported association may prove valuable to functional neurosurgeons performing DBS surgery.

Funding

This study did not receive any funding or financial support.

Disclosures

Andre Machado received funding from the National Institutes of Health and is a consultant for Abbott. The other authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article.

SUPPLEMENTAL DIGITAL CONTENT

Part 1. Data Collection Items.

Part 2. Full Statistical Report.

COMMENTS

This is an excellent and thorough review of DBS acute complications experienced by a single institution. Complications of DBS surgery are well known to functional neurosurgeons. However, general readers may benefit from this study, learning that DBS is a rather safe procedure, even when bilateral. The authors report about the perioperative complications experienced during the 30 days after the implant of 895 electrodes in 481 patients. No mortality is reported. Hospital readmission was required by 57 patients. Lead revision was needed by 37 patients overall while permanent lead removal was required in 15.

The readers need to be aware that DBS complications are obviously not limited to the 30 days after the procedure as hardware malfunction or breakage, skin erosion, or neurological side effects became increasingly more frequent with time.

Pantaleo Romanelli

Winter Park, Florida, USA