Head Injury and Parkinson Disease: Updated Evidence from Meta-Analysis Studies


 Background Published studies on head injury and Parkinson's risk(PD) were inconsistent. We performed a meta-analysis study to explore the association.Methods We retrieved articles published in English from PubMed, Web of Science, Scopus and ScienceDirect between January 1, 1990 and December 31, 2019. The pooled effect of head injury and PD risk was calculated by a random effect model.Results In the meta-analysis, there were 21 studies, including 214763 individuals and 39209 PD patients. The pooled OR estimates(ORs) showed an increased risk of PD was correlated with head injury(OR = 1.46, 95% CI 1.29–1.66). Considering the unconscious state, head injury with LOC showed significant association with PD(OR = 1.49, 95%CI 1.28–1.74). However, head injury without LOC had no significant association with PD (OR = 0.57, 95% CI 0.29–1.12). Sensitivity analysis showed that, when any one study was excluded, the results did not change significantly.Conclusions Our research shows that head injury was associated with PD risk.This study provides a basis and reference for further study on head injury and PD.


Background
As a serious global public health problem, head injury have received great attention from various countries. Head injury had become the main cause of deaths and mutilations and more than 10 million people were affected each year, leading to death or hospitalization [1]. Parkinson's disease (PD) is a common neurodegenerative disease that can cause progressive dyskinesia, such as resting tremor, stiffness and bradykinesia, which were often reported in patients with head injury [2,3]. A person with head injury will increase the risk of PD, this possibility has great social and medical signi cance.
Studies have shown that neuroin ammation is one of the pathogenesis of PD [4]. Head injury could cause neuroin ammation, which may be the most reasonable explanation for the correation between them. Moreover, some researches show that head injury would damage the blood-brain barrier, leading to leukocyte in ltration and microglia activation [5]and could damage mitochondrial function, causing glutamate excitotoxicity, which are related to Neurodegenerative diseases, including PD [6][7][8]. To date, many studies related to head injury and PD have been published. Although some studies have reported head injury can increase the PD risk [9][10][11], there are also studies that cannot con rm the association [12][13]. In view of the inconsistency between the existing research literature, we conducted a metaanalysis to quantitatively evaluatd the correlation.

Methods
We completed the meta-analysis following the guidelines published by the MOOSE group [14] (S1Table) and PRISMA group (S1Appendix) in this meta-analysis studies [36].

Literature Search
We retrieved relevant articles published in English from PubMed, Scopus, Web of Science and ScienceDirect between January 1, 1990 and December 31, 2019. The search terms include"head injury", "head injuries", "Parkinson disease","brain injury", "Parkinson's disease","brain injuries", "parkinsonism","head trauma", "traumatic brain injury", two reviewers(N.L.and J.X.)independently checked the full text articles to extract data on study characteristics. Only cohort studies and case-control studies were retrieved. We also retrieved a reference list of related studies. The owchart of literature search was showed in Fig. 1.

Eligibility Criteria
If the following conditions were met, the study was eligible for inclusion: (1) head injury was taken as the target variable; (2)PD diagnosed by doctors was de ned as the target result; and (3)provided ORs and corresponding 95% CIs or provided enough information to calculate.

Data Extraction and Quality Evaluation
We extracted the data included author, publication year, sample size, study design, PD de nition, exposure variables or CI and adjustment factors. When multiple estimates were reported, adjusted estimates are included. If no other factors are adjusted, rough risk estimates were included.
The Newcastle-Ottawa Quality Assessment Scale [15] was used to evaluate the quality of studies. It included 9 questions and the full score was 9 points, each satis ed answer received 1 point. Only most questions are considered satisfactory (ie not less than 7 points), we considered the study as high methodological quality. There had two independent researchers (Y.L. and C.S.)performed the data extraction and quality assessment. Discrepancies between these results were resolved with a joint reassessment and a consensus was reached.

Statistical Analysis
Considering the heterogeneity between studies, we performed a random-effect model to calculate the pooled ORs. Heterogeneity of the included studies between head injury and PD was quanti ed by Isquared (I 2 ) statistic and Q-statistic. When the I 2 value is 25%, 50% and 75%, it can represent low, medium and high heterogeneity levels. Prespeci ed subgroup analyses were conducted according to unconsciousness status, publication year ( before2005 versus after 2005(included)), study region (Europe, America and Asia )and study quality score ( ≧ 7 versus < 7). Sensitivity analyses were conducted by removing researches one by one. Funnel plots was used to assess the publication bias, a symmetric inverted funne-shaped or egger test(P ≧ 0.05) indicated no publication bias. All the analyses were undertaken using Stata 12.0 software (StataCorp, College Station, TX).

Head injury and PD risk
In the results, the combined ORs for the PD risk was 1.46 (95% CI 1.29-1.66) (Fig. 2). The statistical value (I 2 = 68.7%, P < 0.001)between the studies indicated that there was substantial heterogeneity. When considering the unconsciousness status, head injury with LOC was association with the increased PD risk(OR = 1.

Publication bias
Visual evaluation of funnel chart display that the size distribution of the study was fairly symmetrical relative to the merger effect (Fig. 6), suggesting that our meta-analysis had little published bias.
Furthermore, the Egger test (P = 0.325) and Begg' test (P = 0.154) further proved that there did not have potential publication bias.

Discussion
Our meta-analysis include 21 case control studies, with more than two hundred thousand individuals and 39209 PD patients. Persons with head injury had a 46% increased risk of PD, compared to people without head injury. Tanner [21] found people who work in the agricultural eld and agricultural industry have an increased PD risk, if they ever exposed to signi cant head injury. Goldman et al [17] found that people who are unconscious for more than 5 minutes were twice as likely to have PD as normal people. A recent study [18] explored the combined effects of a-synuclein gene polymorphism length and head injury on PD. In the study, persons with head injury were not correlated with PD, but they were signi cantly more at risk of PD compared to other groups. These results provide more evidence that head injury can cause Neurodegenerative diseases.
One study [34] performed a meta-analysis using the literature before 2012 and showed that head injury with LOC was associated with PD risk. Our meta-analysis results were partially consistent with that. Our analysis included six studies with high quality published since 2012. In addition, in this study, we took the unconscious state into consideration and performed a sensitivity analysis on more factors than the quality score. Compared with other studies, the summary estimate of PD risk was more accurate (OR = 1.46,95% CI 1.29-1.66).
Subgroup analysis showed that head injury with LOC was signi cantly associated with PD, while head injury without LOC had no signi cantly associated with PD. The results of ve studies did not adjust for other factors, after removing them, head injury without LOC group did not contain any studies. Some studies failed to differentiate head injury with or without LOC, leading to few studies in the without LOC group and a low statistical power. The next phase of researches require to assess the odds ratios (ORs) of PD with and without LOC for head injury separately.
Different studies have different assessment methods for head injury. Some of the included studies about head injury assessment were mainly obtained through questionnaires and medical records, and there was a recall bias for PD patients. In addition, the current data does not explain whether a single or multiple head injury could increase the PD risk, and whether a recent or early head injury have different effects on PD. The current research data does not allow for a more accurate classi cation of head injury. The random effects method provides some hoterogeneity allowance in addition to sampling error [34]. Using a random effects model, the impact of heterogeneity can be expected to be very limited, although it may not exclude the in uence of inter-study heterogeneity. Sensitivity analysis was conducted through some research-level factors to seek heterogeneity source, However, heterogeneity remained after sensitivity analysis.
Several limitations exist in the study. First, there have recall bias in case-control study participants, which may affect the quality of information. Second, different studies may have different de nitions of head injury, leading to misclassi cation. Third, in our study, case-control studies from Europe and the Americas accounted for the largest proportion, so the meta-analysis results need to be interpreted cautiously, when they are extended to other populations and regions.

Conclusions
In conclusion, our research showed that head injury was association with an increasing PD risk. When the unconsciousness status was considered, head injury with LOC also showed signi cant correlated with PD risk. Furthermore, head injury would increase the PD risk, which was closely related to higher quality score studies, euramerica studies and papers published since 2005.

Declarations
Ethics approval and consent to participate Not applicable Consent for publication Not applicable.

Availability of data and materials
All data generated or analysed during this study are included in this published article and its supplementary information les.
Competing interests: The authors declare that they have no con ict of interest.