Prevalence and Factors Associated with Drooling in Parkinson's Disease: Results from a Longitudinal Prospective Cohort and Comparison with a Control Group

Introduction Drooling in Parkinson's disease (PD) is frequent but often goes underrecognized. Our aim was to examine the prevalence of drooling in a PD cohort and compare it with a control group. Specifically, we identified factors associated with drooling and conducted subanalyses in a subgroup of very early PD patients. Patients and Methods. PD patients who were recruited from January 2016 to November 2017 (baseline visit; V0) and evaluated again at a 2-year ± 30-day follow-up (V2) from 35 centers in Spain from the COPPADIS cohort were included in this longitudinal prospective study. Subjects were classified as with or without drooling according to item 19 of the NMSS (Nonmotor Symptoms Scale) at V0, V1 (1-year ± 15 days), and V2 for patients and at V0 and V2 for controls. Results The frequency of drooling in PD patients was 40.1% (277/691) at V0 (2.4% (5/201) in controls; p < 0.0001), 43.7% (264/604) at V1, and 48.2% (242/502) at V2 (3.2% (4/124) in controls; p < 0.0001), with a period prevalence of 63.6% (306/481). Being older (OR = 1.032; p = 0.012), being male (OR = 2.333; p < 0.0001), having greater nonmotor symptom (NMS) burden at the baseline (NMSS total score at V0; OR = 1.020; p < 0.0001), and having a greater increase in the NMS burden from V0 to V2 (change in the NMSS total score from V0 to V2; OR = 1.012; p < 0.0001) were identified as independent predictors of drooling after the 2-year follow-up. Similar results were observed in the group of patients with ≤2 years since symptom onset, with a cumulative prevalence of 64.6% and a higher score on the UPDRS-III at V0 (OR = 1.121; p = 0.007) as a predictor of drooling at V2. Conclusion Drooling is frequent in PD patients even at the initial onset of the disease and is associated with a greater motor severity and NMS burden.


Introduction
Sialorrhea, commonly referred to as drooling, is defned as excessive saliva beyond the margin of the lip. Drooling can be a complication of Parkinson's disease (PD) and is one of the most prevalent complaints of patients, but it is often underrecognized and undertreated [1]. A wide prevalence range has been reported in the literature, ranging from 10 to 84%, with no signifcant variation across ethnic groups [2][3][4][5][6][7][8][9][10][11][12][13][14][15]. However, when studies compared PD patients with controls, drooling only occurred in 6-15% of people without PD [5,6,15,16]. Te broad range in PD patients is likely due to the lack of a standard defnition of and diagnostic criteria for sialorrhea and the diferences in the PD population studied and the methods used. Despite these obstacles, drooling has still been found to negatively impact the quality of life (QoL) of both patients and caregivers [5,12,13,[17][18][19]. Sialorrhea may bring repercussions for the psychosocial health of the person who drools and added burden for the caregiver as well (e.g., odor, stained clothes, constant wiping, restricted social life, etc.) [1]. Moreover, drooling is associated with an increased risk of dry mouth, impact on bolus formation, loss of antibacterial efects of saliva, perioral dermatological changes, ulceration, tooth decay, gingivitis, dehydration, candidiasis, halitosis, and increased speech difculties [20][21][22]. Drooling in PD patients appears to be primarily related to reduced swallowing efciency and not to an increase in saliva production [20,23], as dysphagia is the strongest factor associated with drooling [7,12,23]. Other reported factors associated with drooling are orofacial rigidity/hypomimia, lingual bradykinesia, aging, male gender, cognitive impairment, hallucinations, nontremor dominant PD phenotype, longer disease duration, and more advanced disease stage [3-8, 11-14, 23-26].
Although many studies have analyzed the frequency of drooling in PD, there is less information about its prevalence and associated factors in early PD patients and how it impacts QoL and change over time. Some studies have reported a prevalence of about 20% in de novo and untreated PD patients and that prevalence increases in the long term [27,28]. Our hypothesis was that the prevalence of drooling in early PD patients would be high and would negatively impact QoL. Te aim of the current study was to examine the prevalence of drooling, and its progression, in a PD cohort and assess its impact on QoL. Furthermore, we compared the frequency of drooling in PD patients with a control group and analyzed all these aspects in a subgroup of patients from the cohort with a short disease duration of ≤2 years since the onset of the symptoms. Moreover, we identifed in both groups, the entire cohort and the subgroup with early PD, factors associated with not only drooling but also drooling severity as well.

Materials and Methods
PD patients who were recruited from January 2016 to November 2017 (baseline visit; V0) and evaluated again at a 2-year ± 30-day follow-up (V2) from 35 centers in Spain from the COPPADIS cohort [29] were included in this study. Te methodology of the COPPADIS-2015 study can be consulted in https://bmcneurol.biomedcentral.com/ articles/10.1186/s12883-016-0548-9 [30]. Tis is a multicenter, observational, longitudinal prospective, and 5-year follow-up study designed to analyze disease progression in a Spanish population of PD patients. All patients included were diagnosed according to the UK PD Brain Bank criteria [31].
Information on sociodemographic aspects, factors related to PD, comorbidity, and treatment were collected. Motor status, nonmotor symptoms (NMS), QoL, and disability were assessed at V0 and at V2 using diferent vali-  [30]. In patients with motor fuctuations, the motor assessment was made during the OFF state (without medication in the last 12 hours) and during the ON state. Te assessment was only performed without medication in patients without motor fuctuations. Te same evaluation as for the patients, except for the motor assessment, was performed in control subjects at V0 and at V2 (2 years ± 1 month). Furthermore, motor (H&Y, UPDRS-III, and UPDRS-IV) and nonmotor assessment (NMSS and ADLS) was conducted in PD patients at 1 year ± 1 month (V1) [30]. LEED was calculated based on the literature [32].
Subjects were classifed as with or without drooling according to item 19 of the NMSS [33]. Tis item is one of the 30 items on this scale and is included in domain 6 (gastrointestinal tract). Tis question asks about drooling: "Does the patient dribble saliva during the day?." Te score range is from 0 (without the symptom) to 12 (the most frequent and severe). Subjects with an NMSS-item 19 score � 0 were considered "without drooling," whereas subjects with an NMSS-item 19 score ≥1 (from 1 to 12) were considered "with drooling." Drooling was identifed at V0, V1, and V2 in patients and at V0 and V2 in controls. Te drooling burden was also calculated for PD patients. Te score at V0, V1, and V2 and the sum of the score from the three visits (NMSS-Drooling V0+V1+V2 , from 0 to 36) were calculated. Patients reporting drooling during the three visits were defned as patients with "persistent drooling." Te same method was used to defne dysphagia (item 20 of the NMSS) [34] and hypomimia (item 19 of the UPDRS-III during the OFF state) [35].

Statistical Analysis.
Data were processed using SPSS 20.0 for Windows. For comparisons between PD patients in the control group and PD patients with and without drooling, the Student's t-test, Mann-Whitney U test, chi-square test, or Fisher test were used as appropriate (distribution for variables was verifed by one-sample Kolmogorov-Smirnov test).
Binary and linear regression models were used for determining independent factors associated with drooling (drooling as the dependent variable) and drooling severity (NMSS-Drooling V0+V1+V2 score as the dependent variable), respectively. Variables with univariate associations with p values <0.20 were included in a multivariable model, and a backward selection process was used to remove variables individually until all remaining variables were signifcant at the 0.10 level. For exploring the association between drooling and QoL, linear regression models were used with PDQ-39SI (health-related QoL) and EUROHIS-QOL8 (global QoL) as dependent variables. Te total domain scores of the PDQ-39 were expressed as a percentage of the corresponding maximum possible score, and a summary index was obtained as an average of the domain scores (PDQ-39SI). Te efect was controlled by age, gender, disease duration, LEDD, comorbidities (total number of non-anti-Parkinsonian drugs [36]), motor (H&Y, UPDRS-III, UPDRS-IV, and FOGQ) and nonmotor (NMSS) status, cognitive function (PC-CRS total score), dysphagia, hypomimia, and autonomy for ADL (ADLS), which were included as covariates in the model [36].

Discussion
Te present study represents one of the largest cohorts of PD patients in whom the prevalence of drooling was reported using a validated global NMS scale. We observed that drooling was common in PD patients, clearly much more frequent than in the control group, and was associated with the male gender, older age, and a greater motor and nonmotor severity. In addition, patients with drooling had a worse global and health-related QoL, although the efect of drooling on QoL was not signifcant after adjusting to other covariates. Importantly, we observed that drooling was also a very frequent symptom at the beginning of the disease, as seen in the very early PD patients, suggesting the clinical importance of asking for the presence of drooling at the beginning of the patient's follow-up. About 2 out of every 3 patients from the Spanish cohort COPPADIS reported drooling over a 2-year follow-up. Tis cumulative prevalence is in line with the previously published data [11]. However, due to the lack of a standard defnition and criteria for diagnosing drooling in PD patients, estimates of its prevalence vary considerably with a wide range from 10% to 84% [2][3][4][5][6][7][8][9][10][11][12][13][14][15]. Tis is partly due to diferent tools such as the UPDRS-II, SCOPA-AUT (Scale for Outcomes in Parkinson's disease for Autonomic Symptoms), PD-NMSQuest (Parkinson's Disease Nonmotor Symptoms Questionnaire), NMSS, or diferent types of screening questionnaires have been used to screen drooling in PD cohorts with diferent characteristics also [1,3,11]. Some specifc scales to assess drooling have been designed, but they have been poorly used in studies with PD patients [37]. Using the NMSS-item 19 for detecting drooling like us, van Wamelen et al. [23] detected in a cohort of 728 PD patients with a mean disease duration of 5.6 years a prevalence of 37.2% at the baseline and 40.1% after a mean follow-up of 3.3 years (range 0.5-7.2 years). In many crosssectional studies, the prevalence of drooling in PD is between 40% and 50% [2, 6-8, 11, 23, 24, 38, 39], which is in agreement with our fndings. An interesting fnding is that like in other studies [5,13], drooling was not related to disease duration and in fact, the prevalence at each year (from 35% to 50%) and the cumulative prevalence after the 2-year follow-up (65%) was similar in those patients with no more than 2 years since symptom onset compared to the whole cohort. Drooling is frequent even in de novo patients. Erro et al. [27] reported in 61 de novo PD patients a frequency of drooling of 19.4% at the baseline and 15.3% after a 2-year follow-up, whereas Picillo et al. [28] found in 86 men and 48 women de novo PD patients a frequency of drooling at the baseline and after a 2-year follow-up of 23.3% and 25% and 10.4% and 4.1%, respectively. Te Picillo study, in addition to ours and other studies, suggests that drooling could be more frequent in males [3,11,28,40]. Although specifcally well-designed studies to analyze the prevalence of drooling using specifc validated scales [37,41] in large cohorts are required, all these data suggest a recommendation to rule out drooling in PD patients at the beginning and throughout follow-up since its presence is associated with a worse QoL and it is potentially treatable. Consideration is especially valid in elderly men for which the prevalence of drooling is more frequent. Despite this, drooling is an underrecognized and undertreated symptom in PD [1]. Of note, no patient from our cohort was receiving botulinum toxin injections. In addition to male gender, many other variables have been associated with drooling in PD patients such as dysphagia [1,42], dysarthria [1,43], hypomimia [14,17], lingual bradykinesia [14,17], cognitive status [15,24], hallucinations [5], aging [3,23], more advanced disease stage [16,24], orthostatic hypotension [1], camptocormia [44], and a history of using antidepressants [6]. Drooling in PD patients can be in part due to the inability to maintain saliva in the mouth (i.e., hypomimia, abnormal fexed posture, etc.) and impairment of salivary clearance (i.e., lingual bradykinesia, oropharyngeal dysphagia, and upper esophageal dysmotility), as dysphagia is the strongest factor associated with drooling [7,12,23]. We identifed dysphagia as a factor that doubles the probability of drooling independently of other variables, even though it was measured through patientreported outcomes. Moreover, not only dysphagia burden but also hypomimia burden correlated with drooling burden in the entire cohort and the very early PD group as well. On the other hand, some recent studies comprehensively evaluating many features of the disease found an association between drooling and late onset of the disease, a higher LEDD, fuctuations, depression, higher motor scores, and a greater NMS burden [13-15, 23, 45]. In this Spanish cohort, we identifed a greater motor severity (UPDRS-III) and a greater NMS burden (NMSS) as independent factors associated with drooling and/or also predictors of drooling after a 2-year follow-up. Specifcally, a worse status in terms of motor and NMS predicted a greater drooling severity as well. Tis could explain why drooling was associated with a worse QoL but was not an independent predictor of it. Karakoc et al. [46] reported drooling in 65% of 63 people with PD but no independent signifcant correlation of drooling severity with QoL. However, as in our case, they measured the latter from the total PDQ-39 score, rather than with a tool that measures drooling impact. In contrast, when we used the PDQ-39 domains, we identifed drooling as an independent factor associated with a worse autonomy for ADL (PDQ-39 domain 2) and communication (PDQ-39 domain 7). Psychosocially, PD droolers had worse QoL and had more difculty speaking, eating, and socially interacting compared to PD nondroolers [3,5,11]. In addition, drooling patients afect their caregivers by increasing their burden, depression, and anxiety and reducing their QoL [47]. For all these reasons, therapeutic options should be evaluated more intensively in patients with PD and drooling [1,11].
Te present study has some important limitations. Drooling was considered based on an answer to a simple clinical question from the NMSS and not after using a specifc scale [37,41]. However, this methodology is the most frequent in most studies [2,3,6,7,11,24,40,47,48]. Te sample size in the group of PD patients with no more than 2 years since the onset of the symptoms was small and clearly smaller than that of the entire cohort. In the 2-year follow-up group, there was a 30% loss in participants, although this has been observed in other cohorts, with retention rates of 71% [23], 67% [27], or 67% [28]. Te logistic regression models used to identify the independent factors associated with drooling and predictors of drooling only explain 20-30% of the variance in our analysis, but it was either also low or not provided in other studies [6,7,13,23]. For some variables, the information was not collected in all cases. Instead of a specifc tool for assessing comorbidity, like the Charlson index or others, the total number of non-anti-Parkinsonian medications was used as a surrogate marker of comorbidity [36], and the role of possible comorbidities inducing drooling was not considered. Finally, our sample was not fully representative of the PD population due to inclusion and exclusion criteria (i.e., age limit, no dementia, no severe comorbidities, no second-line therapies, etc.) [49]. Nonetheless, the strengths of our study include a very thorough assessment, a prospective longitudinal follow-up design, and the extensive clinical and demographic information recorded. Data about drooling severity and PDQ-39 domains are novel.
In conclusion, this study observes a high prevalence of drooling in PD patients, clearly much more so than in control subjects, and that this feature is frequent even at the frst stages of the disease as well. Dysphagia is associated with drooling, and a higher motor score and a greater NMS burden are predictors of drooling. PD patients with drooling have a worse QoL, and drooling is also an independent factor associated with communication problems. Tus, drooling screening and therapeutic options should be considered in clinical practice.      14 Parkinson's Disease review and critique and recruitment and/or evaluation of participants. Kurtis M. conducted review and critique and recruitment and/or evaluation of participants. de Fábregues O. conducted review and critique and recruitment and/or evaluation of participants. González Ardura J. conducted review and critique and recruitment and/or evaluation of participants. Alonso Redondo R. conducted review and critique and recruitment and/or evaluation of participants. Ordás C. conducted review and critique and recruitment and/or evaluation of participants. López Díaz L. M. conducted review and critique and recruitment and/or evaluation of participants. McAfee D. conducted review and critique and review of English style. Martínez-Martin P. conducted review and critique and supervision. Mir P. executed review and critique and recruitment and/or evaluation of participants.