Abstract

Background. Neuropsychiatric symptoms (NPS) are a common and potentially serious manifestation of Parkinson’s disease (PD) but are frequently overlooked in favor of a focus on motor symptomatology. Here, we conducted a literature review of the prevalence and type of NPS experienced by PD patients with a clinically defined course of their illness. Methods. We identified reports of NPS in patients with PD and mean disease duration over 3 years. Three databases—PubMed, Scopus, and Dialnet—were searched for relevant literature published between 2010 and 2020. Predefined exclusion criteria were applied prior to a descriptive analysis of the literature base. Results. In all, 87 unique reports were identified and 30 met inclusion and exclusion criteria. These included 7142 patients with PD (male: 67.3%; mean age: 66.2 years; mean disease duration: 6.7 years). The most frequent NPS were mood disorders (apathy, depression, and anxiety), psychosis, and impulse control disorders (ICD). Treatment with dopamine agonists was identified as an important risk factor for ICD. Co-occurrence of NPS and cognitive dysfunction was also evidenced in a number of studies. Patients with more significant cognitive deficits and higher levels of NPS appeared to be of older age with a longer disease duration and to have more severe motor symptoms. Conclusions. NPS, most commonly mood disorders (apathy, depression, and anxiety), psychosis, and ICDs are frequent manifestations of PD. The results of this review reflect the need to develop unified validated assessment protocols for NPS in PD, as well as to improve their management in clinical practice.

1. Introduction

Parkinson’s disease (PD) is a neurological process of chronic course, characterized by a complex clinical pattern of motor and non-motor symptoms. The precise etiology of PD remains unknown but is thought to involve a combination of both environmental and genetic factors. PD is the second most frequent neurodegenerative disorder after Alzheimer’s disease [1] with an estimated prevalence of 0.3–1.0% for the general population and an incidence of around 3.0% among individuals aged >80 years [2, 3]. PD is more prevalent among males compared with females with an incidence ratio of around 2 : 1 [3].

The main symptoms associated with PD include rigidity, bradykinesia, tremor, and instability, as a result of an impairment in the striatal dopaminergic pathway (Table 1). Dopamine-replacement therapy is the current mainstay of treatment for such symptoms. However, patients with PD also experience non-motor symptoms including cognitive and psychiatric disorders, pain, and autonomic nervous system dysfunction. The prodromal phase of the disease may extend up to 20 years before the manifestation of motor symptoms and is additionally characterized by the presence of anosmia, depression, constipation, and rapid eye movement (REM) sleep behavior disorders [4]. Although the burden of non-motor symptoms is usually larger and more prolonged than that conferred by motor symptoms, they are typically underrecognized despite their significant contribution to the functional impairment patient’s experience [5].

Due to the marked functional impairment associated with PD, management strategies have focused principally on the palliation of the motor symptoms of the disease. As the disease progresses, up to 90% of patients experience some form of NPS including mood disorders, fatigue, psychosis, cognitive impairment, sleep problems, and addictions [6]. However, despite the high prevalence of the NPS and the insidious impairment provoked on patients’ and their caregivers’ quality of life (QoL), there is no standardized evaluation criteria for NPS in clinically defined PD. Importantly, the manifestation of NPS during the different stages of the disease and the impact of current treatments for motor symptoms on NPS in these patients is not well defined. For this reason, we undertook a review of the recent literature on the prevalence and nature of NPS during the initial years following a diagnosis of PD in order to inform rational consideration of appropriate treatments and management strategies to address all the manifestations of this progressive and debilitating disease.

The objective of the present literature review and descriptive analysis was to summarize the prevalence, nature, and the current stage of NPS among patients with PD, specifically focusing on recent studies involving patients experience during the initial years following their clinical diagnosis. Additional objectives were to identify the most common NPS among PD patients in relation to their clinical characteristics and the prevalence of cognitive impairment and to explore the relationship between NPS and current approaches to the treatment of PD.

2. Methods

2.1. Identification of Relevant Literature

The literature was systematically searched on January 13–15, 2020, and again on December 18, 2020, taking into account only articles published in peer-reviewed journals. Three electronic databases—PubMed, Scopus, and Dialnet (Table 2)—were chosen according to the following factors: accessibility, availability, and relevance for the research question to be addressed. The Boolean operators used to search the databases are detailed in Table 3. Duplicate reports were removed and predefined inclusion criteria applied to identify relevant reports as follows: (i)Clinical studies of patients with a diagnosis of PD and with a mean disease duration over 3 years. This criterion was applied in order to reduce a potential confounding effect due to the subsequent development of other neurological signs and to ensure the exclusion of patients with atypical PD(ii)Published between 2010 to 2020(iii)Language: English and Spanish(iv)Participant sample of least of 30

Predefined exclusion criteria were then applied: (1)Reviews or nonexperimental articles(2)Animal studies(3)Reports not directly related to the research objectives(4)Reports not available through the specified databases(5)Participant sample<30(6)Reports of patients with PD illness duration after clinical diagnosis <3 years(7)Reports of patients diagnosed with any other disorder that could interfere in the final results

2.2. Analyses

A descriptive, narrative analysis of the literature base was undertaken. No meta-analyses or formal hypothesis testing was undertaken. Descriptive statistics are presented for the frequency of NPS across the reports included. Average age and disease duration were calculated using the arithmetic mean where such data were available or where it was possible to calculate them based on the data presented.

3. Results

A total of 130 articles were identified (PubMed, ; Scopus, ; Dialnet, ), of which 83 were unique reports and 30 met the predefined inclusion and exclusion criteria (Figure 1). Supplementary Table 1 details the reports that were excluded from the analysis and the reasons for exclusion. The incidence of each exclusion criteria was as follows 28.1%, 5.3%, 11.5%, 1.8%, 14.0%, 38.6%, and 7.0% for exclusion criteria from 1 to 7, respectively.

The 30 reports accepted for further review included a total of 7142 patients with PD located in countries in Africa (2), Asia (5), Europe (15), North America (5), South America (1), and Oceania (2). The top-line demographics of the patient cohort for each included report are summarized in Table 4. The mean PD patient age was years, 67.3% were male, and the average duration since diagnosis of PD was years [736].

A summary of the neuropsychiatric assessments undertaken and the results of the assessments in each report are included in Table 5.

3.1. Prevalence of NPS

The most commonly reported NPS encountered in the studies included in the current review were mood disorders, particularly apathy, depression and anxiety, psychosis, and impulse control disorders (ICDs; Figure 2).

3.2. Mood Disorders: Apathy, Depression, and Anxiety

The most commonly encountered mood disorders were depression (47.2%) [11, 16, 20, 24, 25, 29, 30, 33, 36], apathy (45.5%) [8, 11, 20, 24, 25, 36], and anxiety (42.9%) [11, 16, 20, 24, 25, 29, 30, 33, 36].

In one study, a cohort of 492 patients with PD found that both the presence and severity of apathy had a significant negative impact on patient QoL (as measured using the PDQ-8) [8]. The presence of apathy and mood alterations were associated with the highest correlation coefficient (0.63; ) and effect size (0.62; ) for all the NPS identified in this study (including psychotic symptoms and ICD) [8].

The relationship between depression (as measured using the UPDR-S) and the severity of PD (as measured using the Hoehn-Yahr scale) was examined in one study [30]. The presence of depression was associated with female sex (; ) and Hoehn-Yahr Scale score (Hoehn-Yahr stage III, OR =2.01, ; Hoehn-Yahr stages IV + V , ) [30]. A separate investigation identified a positive relation between UPDR-S motor score and BDI score (; ), as well as between an increasing severity of depression and Hoehn-Yahr scale score (; ) [29]. However, a further study found no statistically significant correlation between depression as measured using the BDI and Hoehn-Yahr scale score (; ) [22]. In an examination of depression among patients with PD with and without motor symptoms, patients with motor symptomatology were of older age compared to those without motor symptomatology ( vs , respectively; ), had a longer disease duration ( vs , respectively; ), and had higher PD severity (Hoehn-Yahr stage score: vs , respectively; UPDRS motor score: vs , respectively; ) [33]. Moreover, compared to patients without motor symptoms, those with motor symptoms were more likely to have depression (35.1% vs18.4%; ), anxiety (29.4% vs 12.6%; ), psychosis (hallucinations: 17.2% vs 5.7%; delusions: 13.7% vs 3.4%; for both), and dementia (19.5% vs 4.6%; ) [33].

In relation to Guo et al. (2015) and Solla et al. (2011), they observed that depression (53.2% and 30.9%, respectively) and anxiety (44.8% and 25.2%, respectively) were the most prevalent neuropsychiatric condition in their sample and more frequent among females [11, 33].

A number of studies examined the relationship between depression, anxiety, and apathy and a number of other factors, manifestations, and treatments including genetics [9, 12, 23, 34, 36], parkinsonian impairments [31], pharmacological treatments [35], non-pharmacological treatments [7, 13], hypertension [20] correlation with ICD [27], olfactory dysfunction [18, 32], and the comparison and comorbidity in contrast to healthy controls or baseline data [10, 17, 19, 26, 29]. Moreover, some investigations focused on the caregivers’ distress as a consequence of PD [16, 25].

3.3. Psychosis

Psychosis is characterized by the presence of positive symptomatology (hallucinations, illusions, and delusions) and negative symptoms (mood impairment). The mean prevalence of psychosis was 19.4% [8, 11, 20, 24, 25, 29, 30, 33, 36].

Two studies reported on the rates of hallucinations and delusions among patients with PD [29, 30]. Both studies reported higher rates for hallucinations (23.8% and 11.5%, respectively) than delusions (13.5% and 2.2%, respectively). The most common types of hallucinations were visual (20.6%), somatic (13.5%), auditory (7.2%), and olfactory (1.6%) [29]. Only two studies reported the manifestation of minor hallucinations and concretely visual misperceptions [15, 30]. No information related was found in any other article included in this review.

One study reported on the rates of negative symptomatology among patients with PD and found that negative symptoms (as measures using the SANS) were more severe among those with PD than among a control group of patients without PD (SANS score for patients with PD, ; SANS score for controls: ; ) [22]. However, there were no significant differences between patients with PD and the control group with regard to the severity of positive symptoms (SAPS score for patients with PD, ; SAPS score for controls, ; ) [22]. Moreover, no statistically significant association was identified between the presence of positive and negative symptomatology in relation to PD severity measured according to Hoehn-Yahr scale (SAPS: Hoehn-Yahr scale ; ; SANS: Hoehn-Yahr scale ; ) [22].

In a separate study, compared with PD patients without psychosis, those PD patients with psychosis were significantly older (PD + psychosis, years; PD without psychosis: years; ) and had a longer disease duration (PD + psychosis, years, PD without psychosis, ; ) [29]. In this cohort, no association between the manifestation of psychotic symptoms and type, dose, or combination regimen of anti-Parkinsonian drugs was identified [29]. However, there was a statistical trend toward higher daily dose of levodopa ( mg vs  mg; ) and higher levodopa equivalent daily dose ( mg vs  mg; ) among those with psychotic symptoms [29].

Several studies examined the influence of psychosis in their sample in relation to other factors, manifestations, and treatments including genetics [9, 12, 23, 36], non-pharmacological treatments [7], hypertension [20], correlation with motor symptoms [33], ICD [27], olfactory dysfunctions [18, 32], neurophysiology and neuroanatomy [15], and the comparison and comorbidity in contrast to healthy control subjects [19, 26, 29]. Moreover, some investigations focused on patients’ QoL and caregivers’ distress due to PD [8, 16, 25].

3.4. Impulse Control Disorders (ICDs)

The prevalence of ICDs among patients with PD was reported in two studies [8, 33]. A prevalence of 18.5% for ICDs among patients with PD was reported [8, 33]. The presence of ICDs were associated with a detrimental impact on QoL (ICD severity: PDQ8 ; ) [8].

In a study of pathological gambling and other variants of ICD (ICD-not otherwise specified [NOS]) among patients with PD, it was observed that both conditions were associated with a longer duration of PD in comparison to PD patients without ICD (PD + pathological gambling vs PD without ICD: ; PD + ICD-NOS vs PD without ICD: ; PD with pathological gambling vs PD with ICD-NOS: ) [27].

The presence of ICDs was positively associated with the consumption of dopaminergic agonists () and more severe psychotic symptomatology (PPRS: PD+ pathologic gambling vs PD without ICD: ; PD + ICD-NOS vs PD without ICD: ) [27]. The most notable positive symptoms were visual hallucinations (PD + ICD-NOS vs PD without ICD; ), paranoid ideations (PD + ICD-NOS vs PD without ICD; ), and illusions (PD + ICD-NOS vs PD without ICD; ) [27]. ICDs in PD are suggested to arise as a result of dopaminergic involvement in the reward circuitry. However, one study found no correlation between ICDs and dopaminergic agonists [10]. This may be explained by the short duration of this longitudinal study of 2 years.

Finally, in a study led by Solla et al. (2011), it was found that PD patients with dyskinesias (22.2%; ) manifested a higher frequency of ICDs than PD patients without motor complications (3.4%), PD patients with motor complications (12.2%), and PD patients with motor fluctuations (11.8%; ) and that 92.9% of male patients with PD showed significant manifestations of hypersexuality in contrast to females () [33].

3.5. Pharmacological and Non-pharmacological Treatment: Drugs and Surgeries

A number of studies examined the association between anti-Parkinson drugs and NPS. No correlation was found neither between UPDRS subscale I psychotic symptoms and dopaminergic agonists () nor between NPS and amantadine (), monoamine oxidase B inhibitors (MAOI-B) () or catechol-O-methyl-transferase inhibitors (COMTI) () [17]. Similarly, Pérez-Pérez et al. (2015) undertook a head-to-head comparison of the neuropsychiatric effect of dopamine agonists—pramipexole, ropinirole and levodopa—in PD [24]. Only pramipexole was shown to exert a positive effect on NPS with a significantly lower frequency of clinically meaningful apathy (NPI apathy score ≥4) among those treated with this agent compared to those patients treated with either ropinirole or L-dopa () [24].

Subthalamic nucleus deep brain stimulation (STN-DBS) has been shown to be an effective alternative for motor symptoms in PD, but data are lacking with regard to the impact on neuropsychiatric and behavioral complications [7]. In a study with 6 years follow-up, a significant reduction of both neuropsychiatric non-motor fluctuations in OFF (dysphoria state) (39.1% vs 10.1%, for baseline and endpoint; ) and ON (euphoria state) (37.7% vs 1.4, for baseline and endpoint; ) was noted for patients with PD following STN-DBS [7]. The frequency of apathy (2.9% vs 24.6% for baseline and endpoint; ), depression (5.8% vs 13%, for baseline and endpoint; ), and psychosis (0% vs 5.8%, for baseline and endpoint; ) increased [7]. On the other hand, hyperdopaminergic behaviors were markedly reduced in the follow-up: nocturnal hyperactivity, creativity, hobbyism, risk-taking behaviors, compulsive shopping, pathological gambling, dopaminergic addiction, and excess in motivation were significantly less common after STN-DBS surgery () [7].

Weintraub et al. (2010), Lamberti et al. (2016), and Radziunas et al. (2020) also reported on the effect of drugs and surgery on the neuropsychiatric profile of PD patients [13, 28, 35].

3.6. Cognition and NPS

As illustrated in Table 5, cognition was a factor which received a different approach between studies. For that reason, studies were classified according to the following criteria: “Patients received no cognitive nor neuropsychological assessment apart from the neuropsychiatric evaluation,” “Patients received a simple cognitive evaluation with MMSE and/or MoCA or a similar screening instrument apart from the neuropsychiatric evaluation. Neuropsychological assessment was not done,” and “Patients received both a cognitive evaluation with MMSE and/or MoCA or a similar screening instrument and a neuropsychological assessment apart from the neuropsychiatric evaluation.”

Petrovic et al. (2016) suggested that PD patients with clinically significant NPS were of older age () and had a longer disease duration (), more severe PD (), and worse cognition () than those without clinically significant NPS [25]. In a separate study, dementia was more likely to be present in patients with more advanced stages of PD than at early stages (Hoehn-Yahr scale; ) [30]. Patients with dementia were older than those without dementia (73.7 vs 69.1 years, OR =1.07) and had a higher age at PD onset (67.7 vs 63.3 years, ), although this last condition became practically insignificant () after statistical adjustment was made for age and severity [30]. Martinez-Martin et al. (2015) found that NPS were more common among patients with dementia than among those without dementia () [16]. For patients with dementia, the most prevalent NPS was apathy, while depression was more frequent in patients without dementia () [17]. However, Orfei et al. (2018) found no differences for severity of anxiety and depressive symptomatology within their experimental groups (Parkinson’s Disease Dementia [PDD], multidomain Mild Cognitive Impairment PD [mdMCI-PD], single-domain Mild Cognitive Impairment PD [sdMCIPD], no Cognitive Impairment PD [noCI-PD]). However, increased apathy was observed (PDD =15.5 ± 9.5; mdMCI-PD =7.9 ± 0.6; sdMCIPD =7.2 ± 5.5; noCI-PD =6.8 ± 5; ) [21]. The PDD subgroup performed worse on all the neuropsychological domains in comparison with the other groups (PDD =20.7 ± 4.0; mdMCI-PD =27.1 ± 2.0; sdMCI-PD =28.0 ± 1.6; noCI-PD =29.2 ± 0.9; ) [21]. In a separate study, apathy was shown to be positively correlated with phonological fluency score (, ) and number of errors in the Stroop test (, ), while depression did not appear to exert an influence in any cognitive domain [31]. On the other hand, depression (GDS: ; ), apathy (AS: rho = -0.32; ), and anxiety (STAI-S: rho = -0.37; STAI-T: ; ) were identified as predictors of learning decline [26].

Regarding to the manifestation of psychotic symptomatology in PD, it was found that patients with dementia (hallucinations: 14.9%; visual misperceptions: 12.8%; paranoid symptoms: 3.5%; delusions: 1.3%) were more likely to present positive symptomatology in comparison to PD patients without dementia (hallucinations: 5.5%; visual misperceptions: 7.3%%; paranoid symptoms: 1.3%; delusions: 0.7%) [30].

Overall, the influence of PD on cognitive performance was approached by numerous investigations, tackling this issue from different perspectives [9, 10, 12, 14, 15, 17, 18, 20, 23, 27, 33, 35, 36].

4. Discussion

Our review of the literature has shown that NPS are a common feature of the symptomatology of PD. The most frequent NPS experienced by patients with PD are mood disorders (depression, apathy, and anxiety), psychosis, and ICDs [8, 11, 16, 20, 24, 25, 29, 30, 33, 36]. Patients with PD also experience a range of other NPS including mental fatigue [16], sleep disturbances [17, 24, 25], and irritability [16, 17, 24, 25]. Depression appears to be more likely to be manifested in female patients with PD [29, 30, 33] and those with more severe disease [29]. Similarly, anxiety seems to be more predominant in female PD patients [29, 33]. A number of studies highlighted the prevalence of psychosis (specifically symptoms) among patients with PD, most commonly among those with older age [17, 25, 29] and longer disease duration [25, 29]. No relation has been found for psychosis and PD severity [22], nor psychosis and any type of anti-parkinsonian drugs (dopaminergic agonists, amantadine, MAOI-B, and COMTI) in relation with item 2 from UPDRS subscale I scores [17].

ICD are also common among patients with PD and appear to be associated with longer PD duration, younger age, and sex [10, 27], as well as to the consumption of dopaminergic agonists [8]. This latter observation is consistent with the hyperdopaminergic symptomatology observed in a large proportion of patients with longer disease duration and a higher dopaminergic dose. Postsurgical patients with PD evidenced lower rates of hyperdopaminergic symptomatology and higher rates of hypodopaminergic symptomatology. This may be a consequence of the decrease in dopamine agonist treatment following surgery and the slow desensitization to its effect in these patients [7, 13].

A number of studies suggested that PD patients with cognitive impairment tended to be of older age and with longer disease duration [10, 17, 25] and were more likely to manifest a major motor disability [17, 25] and NPS [10, 25, 26] (in particular mood disorders—apathy, anxiety, and depression) [26]. Patients with PD and dementia tended to have a higher prevalence of apathy than those without dementia, while the most significant NPS among patients without dementia was depression [16].

When considering the results presented here, a number of limitations should be noted. Although we offer an integrating description of the principal and most frequent neuropsychiatric symptomatology in PD according to the last-decade scientific literature, the keywords used for the literature searches may not have captured the full spectrum of NPS. Future investigations should include a major variety of related terms in the search strategy in order to guarantee full coverage of all the NPS aspects.

Audiovisuals and other alternative data sources were not consulted. A particular challenge arose from the wide diversity of perspectives and experimental samples used when studying the neuropsychiatric aspects of PD highlighting the need for a more systematic approach to research in this area.

5. Conclusions

The results of this review of the literature support the need to evaluate and manage NPS in patients with PD during the first years following a PD diagnosis. Given the range of assessment tools currently employed to evaluate the various manifestations of NPS, there is a need to develop an unified and comprehensive approach to the assessment of NPS in patients diagnosed with PD with the development of validated tools suitable for use in routine clinical practice. Future research should seek to define the longitudinal evolution of NPS in patients with PD along with study designs to minimize the confounding effects of treatment and the presence of dementia when evaluating NPS in PD.

Abbreviations

AES:Apathy Evaluation Scale
ARS:Apathy Rating Scale
AS:Anxiety Scale
ASBPD:Ardouin Scale of Behaviour in Parkinson’s Disease
BAI:Beck Anxiety Inventory
BDI:Beck Depression Inventory
BDI-II:Beck Depression Inventory II
BSRS:Brief Psychiatric Rating Scale
CBI-R:Cambridge Behavioural Inventory-Revised
CDR:Clinical Dementia Rating Scale
CIDI:Composite International Diagnostic Interview
COMTI:Catechol-O-methyl-transferase inhibitors
ERS:Euro-QoL Scale
ESS:Epworth Sleepiness Scale
GAF:Global Assessment of Functioning Scale
GDS-15:Geriatric Depression Scale
HADS:Hospital Anxiety and Depression Scale
HARS:Hamilton Anxiety Rating Scale
HDRS:Hamilton Depression Rating Scale
ICD:Impulse control disorders
IDS:Inventory of Depressive Symptomatology
LARS:Lille Apathy Rating Scale
MADRS:Montgomery and Asberg Depression Rating Scale
MAOI-B:Monoamine oxidase B inhibitors
MBI-C:Mild Behavioural Impairment Checklist
MMSE:Mini-Mental State Examination
MoCA:Montreal Cognitive Assessment
NOS:ICD-not otherwise specified
NPI:Neuropsychiatric Inventory
NPS:Neuropsychiatric symptoms
PD:Parkinson’s disease
PDQ-8:Parkinson’s Disease Questionnaire Short Form
PPRS:Parkinson Psychosis Rating Scale
QoL:Quality of life
QUIPRS:Questionnaire for Impulsive-Compulsive Disorders in Parkinson’s Disease Rating Scale
REM:Rapid eye movement
SANS:Scale for the Assessment of Negative Symptoms
SAPS:Scale for the Assessment of Positive Symptoms
SAS:Starkstein Apathy Scale
SCID-P:Structured Clinical Interview for DSM-IV-TR Axis I Disorders
SCOPA:Scales for Outcome in PD-Psychiatric Complications
SEND-PD:Scale for Evaluation of Neuropsychiatric Disorders in Parkinson’s Disease
SHaPS:Snaith-Hamilton Pleasure Scale
STAI:Spielberger State-Trait Anxiety Inventory
STN–DBS:Subthalamic nucleus deep brain stimulation
UPDR-S:Unified Parkinson’s Disease Rating Scale
ZCIB:Zarit Caregiver Burden Inventory.

Data Availability

The data used to support the findings of this study are available from the corresponding author upon request.

Ethical Approval

For this type of study, formal consent is not required.

Conflicts of Interest

RRL and RER received speaker fees and travel support from Teva, AbbVie, Zambon, BIAL, Italfarmaco, Biogen, Merck Serono, Novartis, Sanofi, Lundbeck, and Roche. RRL, RER, and JJGR received travel support and training from Medtronic, but not related to this study. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The authors have stated explicitly that there are no conflicts of interest in connection with this article.

Authors’ Contributions

Paloma Macías-Garcia and Raúl Rashid contributed equally to this study.

Acknowledgments

This work has been supported by the Spanish Ministry of Science, Innovation and Universities (MICINN) under grant (PSI2017-85951-R) and by the European Regional Development Fund through the Andalusian Ministry of Health and Families under grants (PI-0025-2017; PI-0034-2019). The medical writer for this manuscript mainly performed English editing and final check of drafting before submission, and it was funded by AbbVie in the context of a contract of service. AbbVie did not selected authors for participation in the manuscript. No payments were made to the authors for the development of this manuscript either. The authors maintained complete control over the manuscript content, and it reflects their opinions. AbbVie did not review the final manuscript draft for scientific accuracy, and was not involved in methodology, data collection and analysis, or drafting.

Supplementary Materials

Supplementary Table 1: excluded articles. (Supplementary Materials)