Introduction

Parkinson's Disease (PD) is the fastest growing progressive neurodegenerative syndrome globally [1]. The global incidence was noted to be 2 million in 1990 and approximately 6 million in 2015 [1, 2. Projections suggest that by 2040, there will be more than 17 million people living with PD [2]. Although both genetic and environmental factors are thought to influence a multifactorial origin of PD, a lack of consensus remains regarding the true pathophysiology and early symptoms of the disease [1, 3, 4]. Certain studies have shown that aggregation of abnormally folded proteins such as alpha synuclein, the development of reactive oxygen species (ROS), and the result of oxidative stress are each potential factors at play [1, 4, 5]. Regardless of origin, all pathological mechanisms result in the loss of dopaminergic neurons in the substantia nigra [4]. This results in the well-established clinical presentation of PD that includes bradykinesia, tremors, postural instability, rigidity, and several non-motor symptoms [4]. The motor symptoms that are associated with PD appear to increase the risk of falls with approximately 33% of these falls resulting in fractures. This incidence of fractures is more than twice as high as that observed in the general population 6. Due to poor bone mineral density (BMD) associated with PD, and subsequent risk of instability or malunion, many fractures require surgical fixation [7, 8].

While appendicular fractures are the most common fracture following a fall, osteoporotic vertebral compression fractures (OVCF) represent a prevalent fracture pattern seen in patients with PD, in addition to traumatic spinal fractures [8, 9]. Due to complex clinical presentations, PD patients commonly require surgical intervention for spinal injuries. However, these patients represent challenging surgical candidates due to associated frailty, sarcopenia, spinal deformity, and osteopenia [10,11,12]. A paucity of evidence exists concerning the collective evidence pertaining to surgical outcomes in patients with PD undergoing spinal instrumented fusions [2, 8, 13]. The purpose of this study was to consolidate the literature concerning outcomes of spinal instrumented fusions in PD patients. Our study aimed to compare both intra- and post-operative complications for people with and without PD, so that future treatment strategies can be optimised accordingly.

Materials and methods

Search strategy and study selection

Two independent reviewers (AIA and JMD) performed a literature search following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [14]. Any disagreements regarding study inclusion were resolved by consulting the opinion of a senior author (SD or JSB). A comprehensive search was performed for eligible articles using the PubMed/Medline, Embase, and Cochrane databases to include studies up to, and including December 28th, 2023. Search terms included “Parkinson’s Disease/PD” AND (“spine” OR “spine surgery” OR “spine fusion”) AND (“outcomes” OR “complications” OR “revision surgery” OR “morbidity” OR “mortality”) AND (“follow-up”). Reference lists of full-text articles were reviewed and screened for further studies meeting the inclusion criteria.

Eligibility criteria

The inclusion criteria were (i) all studies pertaining to (ii) patients with PD who (iii) underwent spinal instrumented fusions (iii) that reported comparative outcomes to a control cohort and (iv) were written in English /had translated version available. The exclusion criteria included (i) non-comparative outcomes (case series or case reports).

Data extraction

All relevant information was collected by two independent reviewers (AIA and NW). The Methodological Quality of Evidence (MQOE) was assessed using the Risk of Bias in Non-Randomized Studies (ROBINS) tool developed by Cochrane for evaluating bias in non-randomized studies [15].

Outcomes analysed and statistics

Outcomes analysed included cost of surgery, length of stay, intra- and post-operative complications, revision surgeries, and mortality. Statistical analysis was performed using The R Project for Statistical Computing (version 4.1.2). Heterogeneity between studies was quantified using the I2 statistic. Meta-analysis was performed on parameters reported in ≥ 2 studies. A random effects model and binary outcomes model were employed. Results were expressed as mean for continuous outcomes and risk ratio (RR) for dichotomous outcomes, with a 95% confidence interval (CI). Heterogeneity was quantified per Cochrane values; (i) 0%–40% = low degree of heterogeneity (ii) 30%–60% = moderate degree of heterogeneity (iii) 50%–90% = substantial degree of heterogeneity (iv) 75%–100% = considerable degree of heterogeneity. A p-value of < 0.05 was considered statistically significant. This study was registered on PROSPERO, with ID# CRD42024497778.

Results

The search strategy yielded 247 results. After duplicate removal and review of titles and abstracts, 28 studies remained for full-text review, as outlined in Fig. 1. Following full-text review, 16 studies were identified and included for meta-analysis.

Fig. 1
figure 1

PRISMA flowchart

Full size image

In total, 2,323,650 patients were included across the 16 studies. Of those, 2,308,949 (99.37%) were patients without PD (non-PD), while 14,701 (0.63%) patients had been diagnosed with PD prior to surgery. Overall, the mean age of cohorts was reported in 13 studies. The collective mean age was 68.23 years (PD: 70.14 years vs non-PD: 64.86 years). Aside from three studies that did not report sex [16,17,18], there were 857,237 males (46%) and 1,006,753 females (54%). Of the males, 4,574 (0.53%) had PD and 852,663 (99.47%) did not. Of the females, 3,213 (0.32%) had PD and 1,003,540 (99.68%) did not. The individual characteristics of studies included are outlined in Table 1.

Table 1 Characteristics of studies
Full size table

Risk of bias assessment

The ROBINS tool was used to evaluate bias across several domains for the studies included. Overall, a high degree of bias was noted in seven studies [16, 18, 21, 23, 25, 26, 30]. To score an overall high degree of bias, an inferred high degree of bias had to be present in at least one domain. Of those seven studies, three studies scored a high risk of bias for selection of participants into the study as patients with traumatic spinal injury were excluded [18, 21, 30]. Six studies scored a high risk of bias due to missing data as follow-up for patients was not included [16, 18, 21, 25, 26, 30]. Two studies scored a high risk of bias in selection of reported result as means and medians were provided without supporting results of standard deviation and/or error [21, 23]. The comprehensive ROBINS assessment is outlined below in Fig. 2

Fig. 2
figure 2

Risk of bias in non-randomized studies (ROBINS) assessment of studies included

Full size image

Surgical outcomes were compared to assess if patients with PD had poorer outcomes than non-PD patients. Results from individual studies were collated and compared for total complications, mechanical and surgical complications, surgical site infections (SSI), post-operative delirium, pneumonia, venous thromboembolic events (VTE), revision surgeries and mortality. The authors required at least two studies having a statistic for a parameter in order for it to be analysed. Authors further analysed each of these outcomes based on the type of surgery, including the following: All thoracolumbar/lumbar fusion surgeries, anterior cervical discectomy and fusion (ACDF), thoracolumbar/ lumbar fusion surgeries excluding OVCF patients, and all spinal instrumental fusions (both ACDF and thoracolumbar spine fusions) combined. The results for individual parameters are described below.

Peri/post operative outcomes in all spinal instrumental fusions

Total post/peri operative complications in all spinal instrumental fusions

Fourteen studies have reported at least one complication in patients that have undergone a spinal instrumental fusion. The two studies that did not report were Avila et al. [16] and Spindler et al. [27] The collective rate of complications among the PD cohort was 23.6% (3333/14133), compared to 4.6% (101,774/2,236,309) in the non-PD cohort. The difference in complication rates proved insignificant on quantitative meta-analysis (RR: 2.46; 95% CI: 1,6.02; p = 0.05), as outlined in Fig. 3.

Fig. 3
figure 3

Comparative rates of total post/peri operative complications for PD and non-PD cohorts undergoing spinal instrumental fusions. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Mechanical/surgical complications in all spinal instrumental fusions

In total, seven studies reported on mechanical and/or surgical complications in all spinal instrumental fusions [20, 22,23,24, 27,28,29]. In total, there were 196,706 patients included (PD: 0.6%, 1211 patients vs. non-PD 99.4%; 195,495 patients) across the seven studies. Six studies reported increased mechanical and/or surgical complications in their PD cohorts [20, 22, 23, 27,28,29]. However, one study resulted in increased mechanical complications for the non-PD cohort [24]. Overall, PD patients were associated with increased mechanical and/or surgical complications on collective analysis (RR: 1.44; 95% CI: 0.73;2.82; p = 0.07), as depicted in Fig. 4.

Fig. 4
figure 4

Comparative rates of mechanical/surgical complications for PD and non-PD cohorts undergoing spinal instrumental fusions. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

SSI in all spinal instrumental fusions

Eight studies reported rates of post-operative SSI [20, 23, 24, 26, 28,29,30,31]. Total sample size was 1,350,717 (PD: 0.4%, 4935 patients vs. non-PD 99.6%; 1,345,782 patients). One study had higher rates of SSI in the non-PD cohort [20]. However, the comparative rates of SSIs among groups were 0.9% (44/4935) in the PD cohort, versus 0.02% (238/1,345,782) in the non-PD group which indicate higher association with the PD cohort. When compared, this proved statistically significant (RR: 4.11; 95% CI: 1.46,11.55; p = 0.01), highlighted in Fig. 5.

Fig. 5
figure 5

Comparative rates of SSIs for PD and non-PD cohorts undergoing spinal instrumental fusions. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Postoperative mortality in all spinal instrumental fusions

Seven studies reported on comparative post operative mortality rates in spinal instrumental fusions [16, 18, 21, 23, 26, 28, 30]. Total sample size for included studies was 1,495,771 (PD: 0.5%, 7148 patients vs. non-PD 99.5%; 1,488,623 patients). Of these seven studies, three studies [21, 28, 30] reported reduced rates of mortality in PD patients, while 4 studies [16, 18, 23, 26] reported higher rates of mortality. Overall, there were higher rates of mortality noted in the PD cohort (2.1%, 149/7148) when compared to the non-PD cohort (0.2%, 2400/1,488,623), which did not prove statistically significant as seen in Fig. 6 (RR: 2.42; 95% CI: 0.89,6.55; p = 0.07).

Fig. 6
figure 6

Comparative rates of postoperative mortality for PD and non-PD cohorts undergoing spinal instrumental fusions. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

VTE in all spinal instrumental fusions

Seven studies reported on post-operative VTE [20, 23, 25, 26, 28, 30, 31]. Total sample size was 1,376,700 patients (PD: 0.4%, 5622 patients vs. non-PD 99.6%; 1,371,078 patients). Rates of post-operative VTE were higher in PD cohorts in all the studies reported except an equal incidence between PD and non-PD cohorts in the Westermann et al. [31] and Hollern et al. [28] studies. On collective analysis, the PD cohort had higher rates of postoperative VTE events (PD: 1%, 56/5622 vs. 0.2%, 2948/1,371,078). This comparison was statistically significant on meta-analysis, as evident in Fig. 7 (RR: 1.84; 95% CI: 1.14,2.97; p = 0.02).

Fig. 7
figure 7

Comparative rates of VTE for PD and non-PD cohorts undergoing spinal instrumental fusions. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Pneumonia in all spinal instrumental fusions

Five studies reported on post operative incidence of pneumonia [20, 25, 26, 28, 30]. Total sample size was 1,376,016 patients (PD: 0.4%, 5298 patients vs. non-PD 99.6%; 1,370,718 patients). Rates of post-operative pneumonia were noted to be higher in non-PD cohorts in two of the studies [20, 26] whereas three studies [25, 28, 30] had higher rates in the PD cohort. Overall, there was a trend towards higher incidences of pneumonia in PD patients (PD: 1.7%, 90/5298 vs. 0.51%, 7004/1,370,718). However, this has not proven statistical significance on meta-analysis, as evident in Fig. 8 (RR: 1.23; 95% CI: 0.66,2.28; p = 0.41).

Fig. 8
figure 8

Comparative rates of pneumonia for PD and non-PD cohorts undergoing spinal instrumental fusions. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Delirium in all spinal instrumental fusions

Five studies reported on post operative delirium [20, 24, 28, 29, 31]. Total sample size was 813 patients (PD: 20.2%, 164 patients vs. non-PD 79.8%; 649 patients). Rates of post-operative delirium were noted to be higher in PD cohorts in all studies but one [24]. Overall, there appeared to be a trend towards higher incidences of delirium in the PD cohort (PD: 11.6%, 19/164 vs. 4.5%, 29/649). However, this was not significant on meta-analysis, as shown in Fig. 9 (RR: 3.15; 95% CI: 0.79,12.54; p = 0.08).

Fig. 9
figure 9

Comparative rates of delirium for PD and non-PD cohorts undergoing spinal instrumental fusions. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Revision surgeries in all spinal instrumental fusions

Ten studies reported on revision surgeries [17,18,19, 23,24,25, 27,28,29, 31]. The total sample size was 665,174 (PD: 1.1%, 7289 patients vs. non-PD 98.9%; 657,885 patients). All studies revealed increased revision surgery in the PD cohort compared to the non-PD cohort. When revision rates were collated for spinal instrumental fusions, the overall rate of revision surgery was higher in PD patients (8.9%, 647/7289) compared to non-PD patients (2.7%, 18,079/657885). As shown in Fig. 10, this was statistically significant on meta-analysis (RR: 3.61; 95% CI: 1.11;11.77; p = 0.04).

Fig. 10
figure 10

Comparative rates of revision surgeries for PD and non-PD cohorts undergoing spinal instrumental fusions. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Surgical outcomes in the ACDF cohort (excluding delirium. 27)

Total complications in ACDF Cohort

Six studies reported on at least one complication which occurred secondary to ACDF [17, 18, 22, 26, 28, 30]. Common complications among reported studies were post-surgical mortality, SSI, pneumonia, deep vein thrombosis, pulmonary embolism and mechanical complications. The collective rate of complications among the PD cohort was 24.4% (1595/6529), compared to 22% (16,610/75335) in the non-PD cohort. The difference in complication rates proved statistically insignificant on quantitative meta-analysis (RR: 4.2; 95% CI: 0.46,38.21; p = 0.16), as outlined in Fig. 11.

Fig. 11
figure 11

Comparative rates of total peri- and post-operative complications for PD and non-PD cohorts undergoing ACDF. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Mechanical and surgical complications in ACDF cohort

In total, two studies reported on mechanical and/or surgical complications [22, 28] Mechanical and surgical complications were grouped together for the purpose of this analysis and were inclusive of device- or implant-related issues, in addition to progression of injury or deformity despite surgical fixation. For example, mechanical complications were described by Miller et al. [20] as “mechanical complications of internal orthopaedic device”, while Hollern et al. [28] described it as “any surgical complication.” In total, there were 195,433 patients included (PD: 0.43%, 843 patients vs. non-PD 99.6%; 194,590 patients) across the two studies. Both studies reported increased mechanical and/or surgical complications in their PD cohorts. PD patients were associated with increased mechanical and/or surgical complications on collective analysis (RR: 1.19; 95% CI: 0.97;1.46; p = 0.06), as depicted in Fig. 12.

Fig. 12
figure 12

Comparative rates of mechanical and/or surgical complications for PD and non-PD cohorts undergoing ACDF. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

SSI in ACDF cohort

Three studies reported rates of post-operative SSI [26, 28, 30]. Total sample size was 1,349,435 (PD: 0.3%, 4547 patients vs. non-PD 99.7%; 1,344,888 patients) [17, 18, 21, 22]. The comparative rates of SSIs among groups were 0.44% (20/4547) in the PD cohort, versus 0.02% (210/1,344,888) in the non-PD group. When compared, this proved statistically insignificant (RR: 9.42; 95% CI: 0.58,153.33; p = 0.07), highlighted in Fig. 13.

Fig. 13
figure 13

Comparative rates of SSI for PD and non-PD cohorts undergoing ACDF. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Post-operative VTE in ACDF cohort

Three studies reported on post-operative VTE [26, 28, 30]. VTE events included deep vein thrombosis and pulmonary embolism. Total sample size was 1,349,435 patients (PD: 0.3%, 4547 patients vs. non-PD 99.7%; 1,344,888 patients). Rates of post-operative VTE were noted to be higher in PD cohorts in two of the studies [26, 30]. On collective analysis, PD cohort experienced increased rates of postoperative VTE (PD: 0.55%, 25/4547 vs. 0.2%, 2506/1,344,888). One study had equal rates of VTE [28]. This did not prove significant on meta-analysis, as evident in Fig. 14 (RR: 2.32; 95% CI: 0.58,9.26; p = 0.12).

Fig. 14
figure 14

Comparative rates of post-operative VTE for PD and non-PD cohorts undergoing ACDF. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Revision surgeries in ACDF cohort

Three studies reported on revision surgeries [17, 18, 28]. The total sample size of studies included was 329,418 patients (PD: 0.4%, 1215 patients vs. non-PD 99.6%; 328,203 patients). Two of the three studies revealed increased revision surgery in the PD cohort compared to the non-PD cohort [17, 28]. Reasons for revision surgery include implant failure, deformity progression, excisional debridement of wound, infection or burn, and exploration of instrumentation, among others. The Steinle et al. [18] study however was in favour of the PD cohort where none of their patients opted for post operative revision surgery, whereas 189 non-PD patients have. When revision rates were collated for studies, the overall rate of revision surgery was higher in PD patients (3.7%, 45/1215) compared to non-PD patients (0.07%, 226/328,203). As shown in the forest plot in Fig. 15, this did not prove statistically significant on analysis (RR: 8.82; 95% CI: 0,30,282.33; p = 0.37).

Fig. 15
figure 15

Comparative rates of revision surgeries for PD and non-PD cohorts undergoing ACDF. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Mortality in ACDF cohort

Five studies reported on comparative post-operative mortality rates in ACDF [16, 18, 26, 28, 30]. Galivanche et al. [26], Hollern et al. [28] and Martini et al. [30] defined mortality rates as “in-hospital mortality”, while Avila et al. [16] merely described “inpatient mortality”. Steinle et al. [18] referred to mortality as “30 day post operative mortality”. Total sample size for included studies was 1,442,786 (PD: 0.4%, 5185 patients vs. non-PD 99.6%; 1,437,601 patients). Of these five studies, two studies [28, 30] reported reduced rates of mortality in PD patients, while three studies [16, 18, 26] reported higher rates of mortality. Overall, there were higher rates of mortality noted in the PD cohort (1.14%, 59/5185) when compared to the non-PD cohort (0.14%, 2063/1437601), which did not prove statistically significant as seen in Fig. 16 (RR: 3.68; 95% CI: 0.73,18.68; p = 0.09).

Fig. 16
figure 16

Comparative rates of post-operative mortality for PD and non-PD cohorts undergoing ACDF. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Post-operative pneumonia in ACDF cohort

Three studies have reported on post operative incidence of pneumonia [26, 28, 30]. Total sample size was 1,349,435 patients (PD: 0.3%, 4547 patients vs. non-PD 99.7%; 1,344,888 patients). Rates of post-operative pneumonia were noted to be higher in non-PD cohorts in two of the studies [26, 30] whereas one study [28] had higher rates in the PD cohort. Overall, there seems to be similar incidences of pneumonia between PD and non-PD cohorts across all three studies (PD: 0.44%, 20/4547 vs. 0.44%, 5970/1,344,888). This did not prove significant on meta-analysis, as evident in Fig. 17 (RR: 0.72; 95% CI: 0.26,1.95; p = 0.29).

Fig. 17
figure 17

Comparative rates of post-operative pneumonia for PD and non-PD cohorts undergoing ACDF. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Post/peri operative outcomes in thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed (excluding pneumonia [25])

Total complications in thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed

Seven studies have reported at least one complication in patients that have undergone a thoracolumbar/lumbar spinal fusion with OVCF patients removed [18, 19, 21, 23, 25, 29, 31].  Common complications among reported studies were post-surgical mortality, SSI, deep vein thrombosis, post operative delirium, pulmonary embolism and mechanical complications. The collective rate of complications in the PD cohort was 22.8% (1723/7567), compared to 12.5% (44,848/357536) in the non-PD cohort. The difference in complication rates proved statistically slightly insignificant on quantitative meta-analysis (RR: 1.36; 95% CI: 0.98,1.88; p = 0.06), as outlined in Fig. 18.

Fig. 18
figure 18

Comparative rates of total peri- and post-operative complications for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Mechanical/ surgical complications in thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed

In total, three studies reported on mechanical and/or surgical complications in thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed [23, 27, 29]. Mechanical/surgical complications was regarded as “progression of vertebral collapse” by Shah et al. [23] and Kawabata et al. [29] mentioned it as “mechanical complications/failure.” Spindler [27] referred to it as “hardware failure.” In total, there were 929 patients included (PD: 35.6%, 331 patients vs. non-PD 64.3%; 598 patients) across the three studies. All three studies reported increased mechanical and/or surgical complications in their PD cohorts. PD patients were associated with increased mechanical and/or surgical complications on collective analysis (RR: 2.44; 95% CI: 0.12;48.24; p = 0.33), as depicted in Fig. 19.

Fig. 19
figure 19

Comparative rates of total peri- and post-operative mechanical/surgical complications for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

SSI in thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed

Three studies reported rates of post-operative SSI [23, 29, 31]. Total sample size was 917 (PD: 38.3%, 351 patients vs. non-PD 61.7%; 566 patients). The comparative rates of SSIs among groups were 6.6% (23/351) in the PD cohort, versus 3.5% (20/566) in the non-PD group which indicate higher association with the PD cohort. When compared, this proved statistically insignificant (RR: 1.56; 95% CI: 0.69,3.52; p = 0.15), highlighted in Fig. 20.

Fig. 20
figure 20

Comparative rates of total peri- and post-operative SSI for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Mortality in thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed

Three studies reported on comparative post operative mortality rates in thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed [18, 21, 23]. Kleiner et al. [21] and Shah et al. [23] defined mortality rates as “in-hospital mortality”. Steinle et al. [18] referred to mortality as “30 day post operative mortality”. Total sample size for included studies was 52,985 (PD: 3.7%, 1963 patients vs. non-PD 96.3%; 51,022 patients). Of these three studies, one study [21] reported reduced rates of mortality in PD patients, while two studies [18, 23] reported higher rates of mortality. Overall, there were higher rates of mortality noted in the PD cohort (4.6%, 90/1963) when compared to the non-PD cohort (0.66%, 337/51022), which did not prove statistically significant as seen in Fig. 21 (RR: 1.15; 95% CI: 0.27,4.79; p = 0.72).

Fig. 21
figure 21

Comparative rates of total peri- and post-operative mortality for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

VTE events in thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed

Four studies reported on post-operative VTE [23, 25, 29, 31]. Total sample size was 27,176 patients (PD: 4%, 1076 patients vs. non-PD 96%; 26,100 patients). Rates of post-operative VTE were noted to be higher in PD cohorts in all the studies reported except an equal incidence between PD and non-PD cohorts in the Westermann et al. [31] study. On collective analysis, (PD: 3.3%, 35/1076 vs. 1.7%, 455/26100). This did not prove significant on meta-analysis, as evident in Fig. 22 (RR: 1.51; 95% CI: 0.94,2.42; p = 0.07).

Fig. 22
figure 22

Comparative rates of total peri- and post-operative VTE for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Delirium in thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed

Two studies have reported on post operative incidence of delirium [29, 31]. Total sample size was 341 patients (PD: 18.5%, 63 patients vs. non-PD 81.5%; 278 patients). Overall, there seems to be higher incidences of delirium if you had PD (PD: 12.7%, 8/63 vs. 5.4%, 15/278). This did not prove significant on meta-analysis, as evident in Fig. 23 (RR: 3.67; 95% CI: 0.78,17.37; p = 0.06).

Fig. 23
figure 23

Comparative rates of total peri- and post-operative Delirium for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Revision surgeries in thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed

Seven studies reported on revision surgeries [18, 19, 23, 25, 27, 29, 31]. The total sample size of studies included was 335,713 patients (PD: 1.8%, 6063 patients vs. non-PD 98.2%; 329,650 patients). All studies revealed increased revision surgery in the PD cohort compared to the non-PD cohort. When revision rates were collated for this type of surgery, the overall rate of revision surgery was higher in PD patients (9.9%, 599/6063) compared to non-PD patients (5.4%, 17,872/329,650). As shown in the forest plot in Fig. 24, this was statistically significant on analysis (RR: 1.66; 95% CI: 1.21;2.28; p < 0.01).

Fig. 24
figure 24

Comparative rates of total Revision Surgeries for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Post/peri operative outcomes in all thoracolumbar spinal fusions including OVCF patients

Total peri/post operative complications in all thoracolumbar spinal fusions including OVCF patients

Nine studies reported at least one complication in patients that have undergone a thoracolumbar/lumbar spinal fusion [21,19,20,21, 23,24,25, 29, 31]. The collective rate of complications among the PD cohort was 27.3% (2078/7604), compared to 12.5% (44,904/357864) in the non-PD cohort. The difference in complication rates proved statistically significant on quantitative meta-analysis (RR: 1.43; 95% CI: 1.08,1.9; p = 0.02), as outlined in Fig. 25.

Fig. 25
figure 25

Comparative rates of total peri/post operative complications for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort including OVCF patients. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Total mechanical/surgical complications in all thoracolumbar spinal fusions including OVCF patients

In total, five studies reported on mechanical and/or surgical complications in thoracolumbar/lumbar spinal fusions including OVCF patients [20, 23, 24, 27, 29]. Mechanical/surgical complications was regarded as “progression of vertebral collapse” by Shah et al. [23] and Kawabata et al. [29] mentioned it as “mechanical complications/failure.” Spindler [27] referred to it as “hardware failure.” Nakajima [24] described it as “Progression of vertebral collapse.” Watanabe [20] referred to it as “mechanical failure.” In total, there were 1273 patients included (PD: 28.9%, 368 patients vs. non-PD 71.1%; 905 patients) across the five studies. Four studies reported increased mechanical and/or surgical complications in their PD cohorts [20, 23, 27, 29]. However, one study [24] resulted in increased mechanical complications for the non-PD cohort. Overall, PD patients were associated with increased mechanical and/or surgical complications on collective analysis (RR: 1.65; 95% CI: 0.49;5.63; p = 0.32), as depicted in Fig. 26.

Fig. 26
figure 26

Comparative rates of mechanical/surgical complications for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort including OVCF patients. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

SSI in all thoracolumbar spinal fusions including OVCF patients

Five studies reported rates of post-operative SSI in all thoracolumbar spinal fusions including OVCF patients [20, 23, 24, 29, 31] Total sample size was 1282 (PD: 30.3%, 388 patients vs. non-PD 69.7%; 894 patients). One study had a higher association with the non-PD cohort [20]. However, the comparative rates of SSIs among groups were 6.2% (24/388) in the PD cohort, versus 3.1% (28/894) in the non-PD group which indicate higher association with the PD cohort. When compared, this proved insignificant (RR: 1.55; 95% CI: 0.99,2.45; p = 0.05), highlighted in Fig. 27.

Fig. 27
figure 27

Comparative rates of SSI for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort including OVCF patients. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Postoperative mortality in all thoracolumbar spinal fusions including OVCF patients (Identical to 3.4.4 because there is no OVCF patients with regards to studies with mortality for this surgery).

Three studies reported on comparative post operative mortality rates in thoracolumbar/lumbar spinal fusion cohort with OVCF patients removed [18, 21, 23]. Kleiner et al. [21] and Shah et al. [23] defined mortality rates as “in-hospital mortality”. Steinle et al. [18] referred to mortality as “30 day post operative mortality”. Total sample size for included studies was 52,985 (PD: 3.7%, 1963 patients vs. non-PD 96.3%; 51,022 patients). Of these three studies, one study [21] reported reduced rates of mortality in PD patients, while two studies [18, 23] reported higher rates of mortality. Overall, there were higher rates of mortality noted in the PD cohort (4.6%, 90/1963) when compared to the non-PD cohort (0.66%, 337/51022), which did not prove statistically significant as seen in Fig. 28 (RR: 1.15; 95% CI: 0.27,4.79; p = 0.72).

Fig. 28
figure 28

Comparative rates of postoperative mortality for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort including OVCF patients. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

VTE in all thoracolumbar spinal fusions including OVCF patients

Five studies reported on post-operative VTE [20, 23, 25, 29, 31]. Total sample size was 27,498 patients (PD: 4%, 1102 patients vs. non-PD 96%; 26,396 patients). Rates of post-operative VTE were noted to be higher in PD cohorts in all the studies reported except an equal incidence between PD and non-PD cohorts in the Westermann et al. study [31]. On collective analysis, (PD: 3.3%, 36/1102 vs. 1.7%, 457/26396). This comparison is insignificant on meta-analysis, as evident in Fig. 29 (RR: 1.56; 95% CI: 0.99,2.44; p = 0.05).

Fig. 29
figure 29

Comparative rates of VTE for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort including OVCF patients. PD = Parkinson’s Disease. RR = relative risk, 95% CI = 95% confidence interval

Full size image

Pneumonia in all thoracolumbar spinal fusions including OVCF patients

Two studies reported on post operative incidence of pneumonia [20, 25]. Total sample size was 26,581 patients (PD: 2.8%, 751 patients vs. non-PD 97.2%; 25,830 patients). Rates of post-operative pneumonia were noted to be higher in non-PD cohorts in one of the studies [25] whereas one study [20] had higher rates in the PD cohort. Overall, there seems to be higher incidences of pneumonia in PD (PD: 9.3%, 70/751 vs. 5.6%, 1439/25830). This has proven statistical significance on meta-analysis, as evident in Fig. 30 (RR: 1.72; 95% CI: 1.33,2.22; p = 0.02).

Fig. 30
figure 30

Comparative rates of pneumonia for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort including OVCF patients. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Delirium in all thoracolumbar spinal fusions including OVCF patients

Four studies have reported on post operative incidence of delirium [20, 24, 29, 31]. Total sample size was 685 patients (PD: 17%, 100 patients vs. non-PD 83%; 585 patients). Rates of post-operative delirium were noted to be higher in PD cohorts in all studies but one [24]. Overall, there seems to be higher incidences of delirium in the PD cohort (PD: 16%, 16/100 vs. Non-PD: 5%, 29/585). This did not prove significant on meta-analysis, as evident in Fig. 31 (RR: 2.86; 95% CI: 0.43,18.86; p = 0.17).

Fig. 31
figure 31

Comparative rates of delirium for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort including OVCF patients. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Revision surgeries in all thoracolumbar spinal fusions including OVCF patients

Eight studies reported on revision surgeries [18,19,20, 24, 25, 27, 29, 31]. The total sample size of studies included was 3,35,756 patients (PD: 1.8%, 6074 patients vs. non-PD: 98.2%; 329,682 patients). All studies revealed increased revision surgery in the PD cohort compared to the non-PD cohort. When revision rates were collated for this type of surgery, the overall rate of revision surgery was higher in PD patients (9.9%, 602/6074) compared to non-PD patients (5.4%, 17,875/329682). As shown in the forest plot in Fig. 32, this was statistically significant on analysis (RR: 1.67; 95% CI: 1.25;2.22; p < 0.01).

Fig. 32
figure 32

Comparative rates of revision surgeries for PD and non-PD cohorts undergoing thoracolumbar/lumbar spinal fusion cohort including OVCF patients. PD = Parkinson’s disease, RR = relative risk, 95% CI = 95% confidence interval

Full size image

Length of stay and cost of surgery from all spinal instrumentations

While not included for comparable quantitative meta-analysis (due to absence of standard deviations), PD patients were also associated with a higher mean length of stay and cost of surgery in respective individual studies. Regarding length of stay, Kleiner et al. [21] (6.4 days vs 5.2 days, p < 0.0001), Miller et al. [22] (3 days vs 1 day, p < 0.001), Shah et al. [23] (5.6 days vs 4.4 days; p = 0.001), Galivanche et al. [26] (2 days vs 1 day, p < 0.001), Spindler et al. [27] (20.2 days vs 14.1 days, p = 0.03), Hollern et al. [28] (6.4 days vs 4.1 days, p = 0.046), Westermann et al. [31] (8 days vs 6 days, p < 0.01), Martini et al. [30] (5.8 days vs 3.4 days, p < 0.001) and Avila et al. [16] (4.26 days vs 3.66 days, p < 0.001) all studies reported higher mean length of stay for PD patients. Similarly for cost of surgery, Kleiner et al. [21] ($129,212 vs $110,324; p = 0.0001), Miller et al. [22] ($65,561 vs $50,892; p = 0.34), Shah et al. [23] ($187,707 vs $126,610; p < 0.001), Hollern et al. [28] ($69,565 vs $57,388; p = 0.248), Martini et al. [30] ($112,970 vs $87,408; p = 0.005) and Berreta et al. [25] ($14,451 vs $12,098; p = 0.007) all report higher mean costs associated with operation for PD patients. It is important to note that Kleiner et al.’s study concerned lumbar procedures, Miller et al.’s study concerned cervical decompression and fusion, while Shah et al.’s study concerned thoracolumbar spinal fusion, which may have influenced overall costs.

Discussion

In the forthcoming decades, there will be an inevitable increase in the number of patients with PD undergoing spinal instrumented fusions due to concurrent increases in the prevalence of PD, and an increase in the number of spine surgeries being performed globally on an annual basis [2, 3, 32, 33]. Several longitudinal population studies published in the literature depict a consistent increase, specifically in degenerative lumbar surgeries [32, 33]. As such, there is a need to understand particular pre- and post-operative complications associated with PD patients undergoing spinal instrumented fusions.

The results of our study highlight several interesting findings. On collective analyses, the PD cohort was more likely to suffer post-operative complications. When sub-analysed, patients with PD were significantly more likely to suffer from mechanical complications and SSIs. Furthermore, the rate of revision surgery was higher in the PD cohort. This is consistent with Schroeder et al. [34], who showed that PD was an independent predictor of revision surgery. In terms of mechanical complications and increased revision surgeries in PD patients, concomitant spinal deformity is of concern and commonly associated with PD [12, 13]. While the true aetiology of PD is poorly understood and likely multifactorial, an increasing prevalence and worsening of spinal deformity is evident as PD progresses in its course. Potential contributors to the deformity emergence or progression are likely sarcopenia progression, proprioceptive disintegration, dystonia, and focal myopathy [5, 12, 26]. Scoliosis is the most common spinal deformity seen in PD, while antecollis (forward flexion of head and neck) and camptocormia (forward flexion of thoracolumbar spine) are more commonly seen in later stages of the disease [35]. Progression of spinal deformity in the post-operative period has been noted by Koller et al. [36] and can necessitate revision surgery. Other potential causes of mechanical complications include systemic inflammation, side effects of common medications, and osteosarcopenia [37].

Preliminary results from pre-clinical studies have shown that low-grade systemic inflammation can stimulate osteoclastogenesis, perpetuating failure of any instrumentation implanted to fixate the injured spinal segment [5]. Similarly, levodopa, a well-known medication used for PD, can lead to increased circulatory levels of homocysteine, a homologue of the amino acid, cysteine [37, 38]. Studies have shown a negative effect on bone caused by increased serum levels of homocysteine, as it can stimulate osteoclastogenesis and impair collagen cross-linking, in addition to generating ROS [37,38,39]. While the relationship between hardware failure and PD has been poorly elucidated in spinal instrumented fusions, certain pre-clinical studies elucidate the potential mechanisms which contribute to mechanical complications in these patients.

Additionally, it remains unclear from the studies included as to why PD patients are associated with significantly more SSIs. The process of wound healing is well-established and typically involves four distinct phases; haemostasis, inflammation, fibroblast proliferation, and remodelling. Each phase represents a complex process [40]. Certain studies in orthopaedic surgery have shown similar findings to our meta-analysis. In a study by Wang et al. [41] concerning arthroplasty patients, a diagnosis of PD was associated with increased rates of SSI (OR: 1.46; p = 0.009), periprosthetic infections (OR: 4.89; p < 0 0.001), and superficial wound infections (OR: 3.36; p = 0 0.006). Additionally, SSIs occur due to the limited mobility post operation as a result of inherent characteristics of PD and significant surgery. Studies have shown that decreased levels of dopamine can negatively impact numbers of functional micro-RNAs (miRNAs). Certain miRNAs are integral for a variety of normal physiological processes, including cellular development, cell growth and apoptosis, angiogenesis, and wound healing [42, 43]. Although more robust evidence is needed, studies have shown that PD is associated with dysregulated miRNA profiles [44]. While more concerned with the pathogenesis and improved detection of early PD, certain miRNA profiles highlighted in these studies are involved and can influence the process of wound healing [45]. Therefore, impaired miRNA function and dysregulated expression of miRNAs may present causative factors for increased SSIs in PD patients undergoing spinal instrumented fusions.

Furthermore, the difference between the type of surgery performed can have a major impact on surgical outcomes of PD patients. The authors performed further analysis by analysing differences in outcomes between the two cohorts based on three separate parameters: ACDF surgery, instrumented thoracolumbar fusions without OVCF patients, and instrumented thoracolumbar fusions including OVCF patients. For the ACDF cohort, there was equivocal incidence of postoperative pneumonia overall between the studies reporting it [26, 28, 30]. Interestingly, Hollern et al. [28] found equal rates of VTE after surgery between both PD versus non-PD cohorts. With regards to postoperative mortality, Hollern et al. [28] and Martini et al [30] had higher rates in the non-PD cohort. Among the studies that included instrumented thoracolumbar spinal fusions without OVCF patients, Westermann et al. [31] had equal rates of VTE between the two cohorts. For studies that included thoracolumbar spinal instrumentations with OVCF patients, at least one study had higher rates of incidence in the non-PD cohort for mechanical/surgical complications, SSIs, pneumonias and delirium. From these three different meta-analyses, ACDF surgeries were notably more susceptible to complications than thoracolumbar fusions as there was higher overall incidence of adverse outcomes in the PD cohort when compared to the non-PD cohort.

When comparing ACDF to thoracolumbar instrumental fusions, it is important to stipulate which type of surgery is more geared towards the postoperative complications assessed in this study. As stated previously, this study found higher susceptibility to postoperative outcomes in ACDF surgeries compared to thoracolumbar surgeries. The authors theorize that this may be due to limited space manoeuvrability when operating on the cervical spine which can make operations more difficult and thus increase risk of post operative complications. Another issue is that patients with PD are more likely to experience postoperative complications such delirium and pneumonia because of underlying autonomic dysfunction, decreased mobility, and problems with medication administration. Compared to thoracolumbar fusions, ACDF surgery may increase these risks because of its proximity to the oesophagus and vocal cords, which can impair breathing and swallowing function and result in worse surgical outcomes. The thoracolumbar region is less crowded with critical structures, which reduces risks of unintentional damages during surgery. PD is associated with impaired mobility and this immobility increases risk for VTE. ACDF can exacerbate immobility because of postsurgical neck immobility with a neck brace. In contrast, thoracolumbar fusions involve the thoracic and lumbar regions of the spine which will not contribute to postoperative immobility and might even reduce risk of VTE.

While this study provides valuable insights into outcomes of spinal instrumented fusions in PD patients, several limitations exist. All studies included were inherently limited by their retrospective design, further evidenced by the ROBINS assessment, as shown in Fig. 2. Retrospective studies may omit certain characteristics such as age and gender, since they rely on reviewing charts initially not intended for research data collection. More robust prospective studies would allow for potential longitudinal follow-up and improved analysis of patient outcomes. Additionally, postoperative multidisciplinary care and discharge planning were not outlined which may influence both complication rates and revision rates. Nonetheless, this study highlights interesting findings and considerations regarding future treatment protocols. While in-hospital, adherence to an individual’s medication regime is integral, specifically on the day of operation. Appropriate input must be provided from the respective neurology and anaesthetics teams, with agreement concerning alternative medication regimes during surgery (e.g. transdermal patch). Evaluation of patient BMD should take place prior to day of surgery. Opportunistic markers of BMD on CT (Hounsfield Units) and MRI (Vertebral Bone Quality Score) are now readily available and appear to represent reliable measures of BMD, comparable to Dual Energy X-Ray Absorptiometry (DXA) scores widely used in the lumbar spine [46].

Careful attention must be given to the pre-operative investigation of BMD. Either delaying surgical intervention or avoiding surgery altogether can be crucial for bone protection. Merely 23% of women who are 67 years of age or older and have an osteoporotic related fracture get a prescription for osteoporosis therapy or have their BMD tested within six months after the fracture [47]. Also, reduced vertebral BMD, assessed quantitatively through DXA, poses a challenge for surgical interbody fusion since diminished bone strength increases the likelihood of adverse events related to implants post-operatively [48]. Dipaola et al. [49] assessed the extent of use of DXA before instrumented and non-instrumented spine surgeries by distributing a questionnaire to 114 spine surgeons. Of the 93 completed questionnaires, only 44% reported employing the use of DXA before instrumented fusions and 22% for non-instrumented fusions [49]. Reduced BMD is a direct risk factor for instrumentation failure, and thus preoperative screening is essential [48]. Overall, some of the responses from the questionnaire commented that the BMD alone was a factor in changing their surgical plan or preoperative treatment [49].

In the post-operative period, early multidisciplinary intervention from allied health professionals such as advanced wound care management, dietetics, and physiotherapy services is imperative. Ensuring patients with PD are placed in individual rooms post-operatively where possible, or close to the nurses' station where appropriate, may facilitate management of potential post-operative delirium [50]. Furthermore, the importance of community exercise programmes must be stressed, as it has been shown to potentially reduce the risk of spinal deformity progression by strengthening paraspinal musculature and mitigating sarcopenic changes seen with disease progression [50]. Nevertheless, further robust prospective evidence evaluating optimised pre- and post-operative treatment protocols are ultimately needed, such as Enhanced Recovery After Surgery (ERAS) protocols, which have resulted in positive outcomes for patients across multiple surgical specialities, including spine surgery [51, 52]. Overall, this study may help inform clinical decision making and highlight the prevalence of increased postoperative complications amongst PD patients and thereby raise awareness amongst clinicians and medical personnel to consider the additional challenges raised when caring for patients with PD.

Conclusion

The number of people diagnosed with PD and the number of spinal surgeries being performed annually, are rising simultaneously. Therefore, it is expected that the number of patients with PD undergoing spine surgery will increase drastically. PD is typically associated with low BMI, decreased mobility, and systemic low-grade inflammation, potentially making people with PD poor surgical candidates. Nevertheless, many PD patients develop spinal pathologies that require surgical intervention and may require comprehensive post-operative care. There is a paucity of collective evidence regarding their outcomes following spinal instrumented fusions. It is important to understand if PD patients are more susceptible to specific complications to optimise pre- and post-operative treatment strategies.

From the results of this meta-analysis, PD patients are more likely to suffer post-operative complications, specifically mechanical complications related to instrumentation or a SSI. For ACDF surgeries, the PD cohort suffered higher rates of postoperative complications, VTE and pneumonia. For thoracolumbar surgeries with OVCF patients, PD resulted in higher rates of total postoperative complications, revision surgeries and delirium. Lastly, in thoracolumbar surgeries without OVCF patients, PD had higher mortality rates and postoperative complication than the non-PD cohort. Although more robust prospective studies are required, the results of this study highlight the need for advanced wound care management in the post-operative period, both in-hospital and in the community, in addition to comprehensive multidisciplinary care from allied health professionals. A review of surgical techniques with regards to mechanical/hardware failure and prophylactic measures for VTE can help reduce complications.

All in all, PD presents a formidable challenge for surgeons due to its associated risks of poor postoperative outcomes. However, with significant interventions in place and an increased focus on retrospective reviews of postoperative outcomes, assumption of anticipation of a trajectory towards improved patient outcomes will occur in the future.