Factors associated with dream enacting behaviors among US farmers

Abstract

Introduction:

Dream enacting behavior (DEB) during REM sleep is a characteristic feature of REM sleep behavior disorder (RBD), the most specific prodromal symptom for Parkinson’s disease (PD) and related synucleinopathies.

Methods:

We screened for DEB among 20,591 male farmers in 2013–2015 using a validated question, and examined its association with pesticide uses and other potential risk factors reported about twenty years ago in 1993–1997. We reported odds ratios (OR) and 95% confidence intervals (CI) from multivariable logistic regression models.

Results:

A total of 1,623 (7.9%) farmers reported having had DEB. Farmers with DEB were more likely to report other nonmotor and motor symptoms of PD with age-adjusted ORs ranging from 1.9 to 3.0. DEB prevalence varied little by age, but was significantly associated with current smoking (adjusted OR: 1.4, 95%CI: 1.2, 1.6), daily alcohol drinking (OR: 1.4, 95%CI: 1.1, 1.6), a history of head injury (OR: 1.3, 95%CI: 1.2, 1.5), and being married (OR: 1.4, 95%CI: 1.1, 1.7). We identified significant associations for several pesticides, especially cyclodiene organochlorines and pyrethroids, with adjusted ORs ranging from 1.2 to 1.5. The results were similar after excluding PD cases or when farmers with at least three DEB episodes in life were considered as DEB cases.

Conclusions:

This study suggests that DEB are not rare among male farmers. Findings on potential risk factors for DEB are intriguing, and given the close link between RBD and PD, the associations should be further investigated.

Introduction

Rapid-eye-movement (REM) sleep behavior disorder (RBD) is characterized by the loss of muscle atonia during REM sleep and the presence of dream enacting behaviors (DEB). A clinical diagnosis of RBD requires videotaped polysomnographic (PSG) evidence of repeated episodes of sleep-related vocalizations and/or complex motor behaviors without muscle atonia during REM sleep that cannot be explained by other causes [1]. Although RBD is likely rare in the general population [2, 3], it is very common in patients with brain synucleinopathies such as Parkinson’s disease (PD), multiple system atrophy, and dementia with Lewy bodies, with prevalence estimates of 30–50%, 80–95%, and 50–80% respectively [4]. Importantly, RBD is a specific prodromal symptom of these conditions, as up to 80% of idiopathic RBD patients may eventually develop one of these neurodegenerative diseases [5]. Therefore, research on RBD and its risk factors will critically inform the etiology, natural history, and preventive strategies for synucleinopathies.

To our knowledge, only a few studies have specifically examined potential risk factors for RBD [6–8], including both PSG-based studies [7] and studies simply based on DEB symptom screening [6, 8]. Screeners were used mainly because PSG is costly and logistically prohibitive in large population-based studies. Nevertheless, these studies have generated interesting findings on head injury, smoking, alcohol consumption, education, and certain medical conditions, as well as on farming and occupational pesticide use. Notably, two studies reported a higher likelihood of RBD or the so-called screening-based “probable” RBD among pesticide users [7, 8]. This is interesting because pesticide exposure is also a known risk factor for PD [9, 10], although data on specific pesticides and PD are limited. In the Agricultural Health Study (AHS), we screened for DEB symptoms in a large population of licensed private pesticide applicators (hereafter referred to as farmers) and examined its associations with potential risk factors, especially uses of specific pesticides.

Methods

Study population

The AHS is a longitudinal cohort that recruited 52,394 farmers from Iowa and North Carolina to investigate impacts of pesticides and other farming exposures on health [11]. These farmers are predominantly men (97.4%) and whites (95.1%). Detailed information on AHS has been published previously [11]. In brief, 52,394 farmers (84% of the eligible) completed an enrollment questionnaire at pesticide training/licensing locations between 1993 and 1997. The cohort has since been followed up every 5–6 years with questionnaires and phone interviews. As few farmers were women, the current analysis is limited to male farmers. Of the 51,035 male farmers enrolled in the cohort, 23,478 further participated in the most recent follow-up in 2013–2015. This follow-up survey screened for several PD prodromal symptoms [12], including DEB, olfactory impairment, constipation, daytime sleepiness, depression, anxiety, and several motor symptoms such as tremor and small handwriting. All participants implied informed consent by returning study questionnaires and participating in phone interviews. The Institutional Review Boards at the National Institute of Environmental Health Sciences and the National Cancer Institute approved the study protocol.

Screening for DEB, and other nonmotor and motor symptoms of PD

The AHS 2013–2015 follow-up survey asked about six nonmotor symptoms that often occur in prodromal PD as well as about typical motor symptoms (Supplementary Table 1). For DEB, we asked participants: “Have you ever been told, or suspected yourself, that you seem to ‘act out dreams’ while sleeping? For example, punching or flailing arms in the air, shouting, or screaming while asleep.” [13]. For farmers who answered ‘yes’ to the screening question, we further asked about the frequency of symptoms (<3 times in life, <1/month, 1–3/month, once a week, or > 1/week). We previously reported that AHS farmers who reported DEB were eight times more likely to report a PD diagnosis with a clear dose-response relationship for symptom frequency [12].

Data on other motor and nonmotor symptoms were also collected using established questions and instruments as detailed previously [12]. In brief, we defined olfactory impairment as a self-reported loss of or a significantly decreased sense of smell, constipation as ≤3–4 bowel movements per week or use of laxatives, excessive daytime sleepiness as feeling sleepy most of the day for 6–7 days per week, depression as a score ≥ 3 on the 2-item Patient Health Questionnaire or use of prescribed antidepressants, and anxiety as a score ≥ 3 on the 2-item Generalized Anxiety Disorder scale. For motor symptoms, we used questions similar to the 9-item PD screening questionnaire [14] to identify tremors, smaller handwriting, softer voice, shuffling gate, and trouble rising from a chair. Potential PD cases in the AHS are also identified and confirmed according to established protocols [12]. Briefly, AHS surveys asked participants to report whether they had ever been diagnosed by a physician with PD and their age at diagnosis, followed by collection and evaluation of medical information on PD diagnosis, symptoms and treatment from patients and their treating physicians for diagnostic validation. This approach has been used in several other large prospective cohort studies consistently with an 80% or higher agreement rate between self-reports and information from their treating physicians or review of medical record [15,16].

Pesticide use and other data collection at enrollment

The enrollment questionnaire asked participants to report whether they had mixed or applied 50 specific pesticides, including 22 insecticides, 18 herbicides, six fungicides and four fumigants (http://aghealth.nih.gov/collaboration/questionnaires.html). We examined associations of DEB with ever-use of 49 of these specific pesticides and four functional and seven chemical classes of pesticides with at least 10 exposed cases. In addition, the enrollment questionnaire collected data on demographics such as age, sex, race, education, and marital status, as well as lifestyle risk factors such as smoking status and alcohol consumption in the past year. The enrollment questionnaire did not ask about history of head injury; that information was obtained from a more detailed take-home questionnaire at baseline and from the first follow-up survey in 1999–2003.

Statistical analysis

We used multivariable logistic regression models to estimate odds ratio (ORs) and 95% confidence intervals (CIs). The overall analytic sample included 20,591 farmers after excluding 378 who had missing data on DEB and 2,509 with proxy respondents because the proxy questionnaire did not ask about PD prodromal symptoms. Sample sizes for individual analyses varied due to missing exposures or covariates. We first examined associations of DEB with other nonmotor (n=19,460) and motor (n=20,038) symptoms in all eligible participants as well as in participants without a PD diagnosis, adjusting for age. We then evaluated demographic and lifestyle factors reported at cohort enrollment in relation to DEB that was reported approximately two decades after enrollment (n=17,855), mutually adjusted as covariates. We focused on baseline data as potential confounders because baseline information was more likely to reflect experience prior to RBD onset. In the analyses of specific pesticides, we adjusted for baseline age, smoking, alcohol consumption, marital status, education, state of residence, and a history of head injury. For pesticides that showed significant associations with DEB, we conducted additional analyses that further adjusted for other (outside that pesticide’s parent class) functional/chemical classes of pesticides that were also associated with DEB.

As only 46% of the male farmers initially enrolled participated in the 2013–2015 survey, we used inverse probability weighting to account for the loss of participants and for missing covariates to make inferences about all male farmers who enrolled in the study [17]. Briefly, we used logistic regression models to calculate stabilized weights to account separately for the loss of participants and for missing covariates, which were then multiplied to obtain overall stabilized weights. Final models on pesticide in relation to DEB were adjusted for potential confounders and employed the overall stabilized weights, with the weights truncated at the 1^st and 99^th percentiles. The 95%CIs were estimated using robust estimates of standard errors [17]. We performed three sensitivity analyses: (i) excluding participants with a possible PD diagnosis (n=179); (ii) restricting cases to those with ≥ 3 lifetime episodes of DEB; and (iii) adjusting for all correlated pesticides (correlation coefficient ≥ 0.30) that were also associated with DEB. We used AHS data releases AHSREL20150600, P1REL201209_00, P2REL20120900, P3REL20120900, and Final_06172015. Statistical analyses were performed using SAS software v.9.3 (SAS Institute Inc., Cary, NC).

Results

The average age of study participants at the 2013–2015 follow-up was 65.5±11.0 years. Of the 20,591 male farmers, 1,623 (7.9%) reported having had DEB and 1,001 further reported experiencing symptoms ≥ 3 times. As expected, farmers with DEB were more likely to report other PD nonmotor and motor symptoms (Table 1). Results were little changed when we excluded PD cases or defined DEB as ≥ 3 episodes.

Table 1:

Cross-sectional associations of DEB with other nonmotor and motor symptoms in the 2013–2015 survey

	All participants			Excluding PD cases	Repeated DEB episodes ≥ 3
	No DEBa	DEBa	OR (95% CI)b	OR (95% CI)b	OR (95% CI)b
Nonmotor symptoms
Loss of smell
No	17155	1268	Ref	Ref	Ref
Yes	1708	347	2.8 (2.5, 3.2)	2.5 (2.2, 2.9)	3.3 (2.8, 3.8)
Infrequent bowel movementc
No	15631	1189	Ref	Ref	Ref
Yes	3042	417	1.9 (1.7, 2.1)	1.8 (1.6, 2.0)	1.9 (1.6, 2.2)
Excessive daytime sleepinessd
No	18059	1475	Ref	Ref	Ref
Yes	618	126	2.5 (2.1, 3.1)	2.4 (1.9, 2.9)	2.6 (2.1, 3.4)
Depressione
No	17340	1316	Ref	Ref	Ref
Yes	1143	257	3.0 (2.6, 3.5)	2.9 (2.5, 3.4)	3.1 (2.6, 3.7)
Anxietyf
No	17984	1456	Ref	Ref	Ref
Yes	623	145	2.9 (2.4, 3.5)	2.8 (2.3, 3.4)	2.8 (2.2, 3.6)
Motor symptoms
Shake or tremble limbs
No	17405	1297	Ref	Ref	Ref
Yes	1491	320	2.9 (2.6, 3.4)	2.6 (2.2, 3.0)	2.9 (2.5, 3.4)
Smaller handwriting
No	17171	1326	Ref	Ref	Ref
Yes	1606	287	2.5 (2.1, 2.8)	2.1 (1.8, 2.4)	2.7 (2.3, 3.2)
Softer voice
No	16644	1246	Ref	Ref	Ref
Yes	2059	357	2.5 (2.2, 2.8)	2.2 (1.9, 2.5)	2.6 (2.2, 3.1)
Feet shuffle when walking
No	17514	1336	Ref	Ref	Ref
Yes	1318	274	2.9 (2.5, 3.4)	2.5 (2.1, 2.9)	3.3 (2.7, 3.9)
Trouble rising from chair
No	15491	1080	Ref	Ref	Ref
Yes	3365	536	2.5 (2.2, 2.8)	2.3 (2.0, 2.6)	2.6 (2.3, 3.1)

^an differs due to missing values;

^badjusted for age;

^cever taken medication, or bowel movement frequency of ≤ 3–4 times/ week;

^dsleepy 6–7 days per week;

^ecurrently taking medicines for depression, or Patient Health Questionnaire-2 score ≥3;

^fDefined by the Generalized Anxiety Disorder Questionnaire-2 score ≥3

DEB: Dream enacting behaviors; PD: Parkinson disease; OR: odds ratio; CI: Confidence Interval

Table 2 presents associations between population characteristics at enrollment and DEB prevalence about twenty years later. Age was not associated with DEB prevalence. Farmers with DEB were more likely to be daily alcohol drinkers, current smokers, ever married/living with partner, and to have a history of head injury. Although 1–3 years of post-high school education was associated a slightly higher prevalence of DEB, no clear pattern with education level was observed.

Table 2:

Characteristics of male farmers in 1993–1997 in relation to DEB reported in 2013–2015

	All participants			Excluding PD cases	Repeated DEB episodes ≥ 3
Characteristics	No DEB (n = 16441)	DEB (n = 1414)	OR (95% CI)a	OR (95% CI)a	OR (95% CI)a
Age at enrollment
≤ 45 years	8107	700	Ref	Ref	Ref
46–55 years	4566	378	0.9 (0.8, 1.1)	0.9 (0.8, 1.0)	0.9 (0.8, 1.1)
56–65 years	3089	268	1.0 (0.9, 1.2)	1.0 (0.9, 1.2)	1.0 (0.9, 1.3)
> 65 years	679	68	1.2 (0.9, 1.6)	1.2 (0.9, 1.6)	1.3 (1.0, 1.9)
Education
≤ High school graduate	8501	701	Ref	Ref	Ref
1–3 years beyond high school	4365	421	1.2 (1.0, 1.3)	1.2 (1.0, 1.3)	1.3 (1.1, 1.5)
College graduate or more	3575	292	1.0 (0.9, 1.2)	1.0 (0.9, 1.2)	1.1 (0.9, 1.3)
State
Iowa	11855	1001	Ref	Ref	Ref
North Carolina	4586	413	1.1 (1.0, 1.2)	1.1 (1.0, 1.3)	1.0 (0.9, 1.2)
Alcohol (Number of drinks/month)
0	5277	410	Ref	Ref	Ref
1–10	6901	598	1.1 (1.0, 1.3)	1.1 (1.0, 1.3)	1.2 (1.0, 1.4)
11–30	2391	202	1.1 (0.9, 1.3)	1.1 (0.9, 1.4)	1.1 (0.9, 1.4)
> 31	1872	204	1.4 (1.1, 1.6)	1.4 (1.2, 1.7)	1.5 (1.2, 1.9)
Smoking status
Never smoker	9559	743	Ref	Ref	Ref
Former smoker	5006	460	1.1 (1,0, 1.3)	1.1 (1.0, 1.3)	1.1 (1.0, 1.3)
Current smoker	1876	211	1.4 (1.2, 1.6)	1.4 (1.2, 1.7)	1.4 (1.1, 1.7)
Marital status
Never married	1553	99	Ref	Ref	Ref
Ever married/Living as married	14888	1315	1.4 (1.1, 1.7)	1.3 (1.1, 1.7)	1.4 (1, 1.8)
Head injuryb
No	12898	1030	Ref	Ref	Ref
Yes	3543	384	1.3 (1.2, 1.5)	1.4 (1.2, 1.6)	1.5 (1.3, 1.7)

^aCovariates are mutually adjusted for each other.

^bfrom both enrollment take home questionnaire and first follow-up in 1999–2003; DEB: Dream enacting behaviors; PD: Parkinson disease; OR: odds ratio; CI: Confidence Interval

Tables 3 and 4 present results for occupational pesticide use; 20 specific pesticides were modestly associated with DEB, with ORs ranging from 1.2 to 1.5. After adjustment for other pesticide classes that were also associated with DEB, statistically significant but attenuated associations (ORs 1.2 to 1.3) were seen for two chemical classes (cyclodiene organochlorines and pyrethroids) and 14 specific pesticides (five organochlorines, the carbamate insecticide carbofuran, three organophosphates, the insecticide permethrin (poultry/livestock), two fumigants, and two herbicides). The results were similar in analyses that excluded PD patients or that defined DEB cases as reporting symptoms ≥ 3 times in life (Supplementary Table 2). When adjusted for all correlated pesticides that were also associated with DEB, ORs remained about the same for all pesticides, except for organochlorines. Only the organochlorine heptachlor remained statistically significant (data not shown).

Table 3:

History of insecticide use reported in 1993–1997 and DEB in 2013–2015

	N exposed cases	OR (95% CI)a	OR (95% CI)b
Any insecticides	1350	1.2 (0.9, 1.7)	-
Organochlorines	807	1.2 (1.0, 1.4)	1.1 (1.0, 1.3)
Cyclodienes	633	1.3 (1.1, 1.4)	1.2 (1.1, 1.4)
Aldrin	338	1.3 (1.1, 1.5)	1.2 (1.0, 1.4)
Chlordane	448	1.3 (1.1, 1.4)	1.2 (1.0, 1.3)
Dieldrin	137	1.4 (1.1, 1.7)	1.3 (1.0, 1.6)
Heptachlor	299	1.4 (1.2, 1.6)	1.3 (1.1, 1.5)
DDT	410	1.3 (1.1, 1.5)	1.2 (1.0, 1.4)
Lindane	324	1.1 (0.9, 1.2)	-
Toxaphene	233	1.2 (1.0, 1.4)	-
Carbamates	991	1.1 (1.0, 1.3)	-
Aldicarb	152	1.2 (0.9, 1.5)	-
Carbaryl	842	1.1 (1.0, 1.3)	-
Carbofuran	480	1.3 (1.1, 1.5)	1.2 (1.1, 1.4)
Organophosphates	1297	1.2 (0.9, 1.5)	-
Chlorpyrifos	670	1.2 (1.0, 1.3)	1.1 (1.0, 1.2)
Coumaphos	153	1.1 (0.9, 1.3)	-
Diazinon	497	1.1 (1.0, 1.2)	-
Dichlorvos	223	1.4 (1.2, 1.6)	1.2 (1.0, 1.5)
Fonofos	364	1.2 (1.1, 1.4)	1.2 (1.0, 1.4)
Malathion	1042	1.0 (0.9, 1.2)	-
Parathion	240	1.1 (1.0, 1.3)	-
Phorate	555	1.3 (1.1, 1.4)	1.2 (1.0, 1.3)
Terbufos	596	1.1 (1.0, 1.3)	-
Pyrethoids	428	1.3 (1.1, 1.5)	1.2 (1.1, 1.4)
Permethrin (poultry/livestock)	272	1.4 (1.2, 1.6)	1.3 (1.2, 1.6)
Permethrin (crops)	217	1.2 (1.0, 1.4)	-

^aAdjusted for age, smoking, alcohol consumption, marital status, education, state, and head injury;

^bConducted for pesticides with statistical significance in the first analysis; further adjusted for functional/ chemical classes that were also significantly associated with DEB

DEB: Dream enacting behaviors; OR: odds ratio; CI: Confidence Intervals;

DDT, Dichlorodiphenyltrichloroethane

Note: bold-face font indicates statistically significant association (p < 0.05)

Table 4:

History of uses of other pesticides reported in 1993–1997 and DEB in 2013–2015

Pesticides	N exposed cases	OR (95% CI)a	OR (95% CI)b
Any herbicides	1392	1.3 (0.7, 2.5)	-
Alachlor	813	1.1 (1.0, 1.3)	-
Butylate	519	1.2 (1.1, 1.4)	1.1 (1.0, 1.3)
Chlorimuron ethyl	548	1.1 (0.9, 1.2)	-
Dicamba	780	1.0 (0.9, 1.2)	-
EPTC	297	1.0 (0.8, 1.1)	-
Glyphosate	1143	1.3 (1.1, 1.5)	1.2 (1.0, 1.4)
Imazethapyr	655	1.0 (0.9, 1.2)	-
Metolachlor	678	1.0 (0.9, 1.2)	-
Paraquat	339	1.1 (0.9, 1.3)	-
Pendimethalin	629	1.0 (0.9, 1.2)	-
Petroleum oil	759	1.2 (1.1, 1.4)	1.1 (1.0, 1.3)
Trifluralin	767	1.0 (0.9, 1.2)	-
Phenoxy herbicides	1178	1.2 (1.0, 1.4)	1.1 (0.9, 1.3)
2,4 D	1153	1.2 (1.0, 1.4)	1.1 (0.9, 1.3)
2,4,5 T	378	1.3 (1.1, 1.4)	1.2 (1.0, 1.4)
2,4,5 T P	164	1.2 (1.0, 1.4)	-
Triazine herbicides	1165	1.0 (0.9, 1.2)	-
Atrazine	1083	1.1 (0.9, 1.3)	-
Cyanazine	661	1.1 (1.0, 1.3)	-
Metribuzin	711	1.2 (1.0, 1.3)	1.1 (0.9, 1.2)
Any fumigants	357	1.2 (1.0, 1.4)	-
Carbon tetrachloride/ Carbon disulfide (80/20 mix)	117	1.5 (1.2, 1.8)	1.3 (1.1, 1.7)
Aluminum phosphide	103	1.4 (1.1, 1.7)	1.3 (1.0, 1.6)
Ethylene dibromide	59	1.2 (0.8, 1.6)	-
Methyl bromide	202	1.0 (0.8, 1.2)	-
Any fungicides	533	1.2 (1.0, 1.4)	1.1 (1.0, 1.3)
Benomyl	143	1.2 (1.0, 1.5)	-
Captan	199	1.2 (1.0, 1.4)	1.1 (0.9, 1.3)
Chlorothalonil	117	1.3 (1.0, 1.6)	-
Maneb	138	1.1 (0.9, 1.4)	-
Metalaxyl	308	1.1 (0.9, 1.3)	-
Ziram	22	0.9 (0.5, 1.6)	-

^aAdjusted for age, smoking, alcohol consumption, marital status, education, state, and head injury;

^bConducted for pesticides with statistical significance in the first analysis; further adjusted for functional/ chemical classes that were also significantly associated with DEB

DEB: Dream enacting behaviors; OR: odds ratio; CI: Confidence Interval;

2,4 D, 2,4-Dichlorophenoxyacetic acid; 2,4,5 T, 2,4,5-Trichlorophenoxyacetic acid; 2,4,5 T,P, 2-(2,4,5-trichlorophenoxy) propionic acid; EPTC, S-Ethyl dipropylthiocarbamate.

Note: bold-face font indicates statistically significant association (p < 0.05)

Discussion

To the best of our knowledge, this is the first epidemiologic study on potential risk factors of DEB in farming populations. Approximately 7.9% of these farmers reported DEB symptoms. As expected, DEB was strongly associated with other motor and nonmotor symptoms of PD, even among participants without a clinical PD diagnosis. Of the potential risk factors examined, we found significant positive associations with current smoking, daily alcohol consumption, and a history of head injury, but not with age. Being ever-married was also associated with DEB, but this association might be due to the fact that those with a bed partner were more likely to be told of abnormal sleep behaviors. Further, among these farmers, DEB was modestly associated with the use of some specific pesticides, especially cyclodiene organochlorines and permethrin. As data on potential risk factors were collected approximately 20 years before DEB screening, these associations are unlikely to be due to reverse causation or recall bias.

In large population-based studies like ours, overnight videotaped PSG is often infeasible and we thus choose to screen for DEB symptoms. Similar screeners have been used in other population-based studies [6, 8, 18, 19]. However, while these RBD screeners have been validated in clinical studies [13, 20], it may lead to substantial false positives when used in the general population. For example, these symptom-screening studies have reported prevalence of probable RBD in the range of 2.7% to 7.7% [6, 8, 18, 19], much higher than the ~1% reported in the few PSG-based studies [2, 3]. For this reason, we cautiously used the term DEB rather than probable RBD in the current study. Nevertheless, such symptom-based studies [6, 8, 18], along with PSG-based clinical studies [7], have generated some much-needed preliminary data about RBD and its potential risk factors.

Research on RBD risk factors is important because PD may take decades to develop, and by the time of diagnosis, the pathogenesis may be too advanced to decelerate or stop. Battle against PD therefore critically depends on a good understanding of the decades of disease’s prodromal period, during which many factors may come into play to initiate pathology or modify progression. Research on prodromal symptoms and their risk factors may therefore critically inform the etiology and natural history of PD. This is especially true for RBD as it is the most specific prodromal symptom for PD. Existing data on factors associated with RBD are limited, but there are a few interesting observations. Unlike PD, RBD or probable RBD showed little age-dependence. Although the above mentioned clinical study [7] reported a weak association between age and PSG-confirmed RBD, most studies using symptom screener, including ours, found no relationship [6, 8, 18, 19]. Also, in contrast to PD [9] and earlier clinical observations of clinical RBD patients [21, 22], recent data [6–8, 18, 19] seem to suggest little sex difference in the prevalence of RBD or the symptom-based probable RBD.

Preliminary findings on smoking and coffee drinking in relation to RBD or probable RBD are equally intriguing. Evidence to date suggests either a modest positive (reference [7] and the current study) or null [6, 8] association with smoking, and a null association with coffee drinking [6–8]. These are in clear contrast to the robust epidemiological findings that smokers and coffee drinkers have a lower PD risk [9]. While these data argue against neuroprotective effects of smoking or coffee drinking in PD development, other possibilities exist. For example, smoking and coffee drinking may selectively protect nigrostriatal dopaminergic neurons at later stage of prodromal PD but not the earlier Lewy pathology at extranigral structures [23], or they may differentially associate with PD subtypes as suggested by a recent study [24]. Future studies should follow up these paradoxical findings to clarify roles of these factors in RBD development and subsequent progression to clinical neurodegenerative diseases.

Our data, together with those from previous studies [6–8], suggest that head injury is associated with higher risk for RBD. These observations are consistent with the overall literature on head injury and PD risk [25, 26]. Further, two recent studies showed that head injury in early life was especially associated with a higher risk of PD [25, 26]. As RBD may develop years, if not decades, prior to PD diagnosis, it is likely that head injury plays an important role in early stages of PD development.

To the best of our knowledge, only two studies have examined the potential role of pesticides in RBD. In the above-referenced case-control study, PSG-confirmed RBD cases were more likely to report occupational insecticide or herbicide use [7]. A cross-sectional study in China reported non-significant higher odds of probable RBD with ever-use of pesticides [8]. Both studies assessed pesticide exposure retrospectively and neither examined specific pesticides. Our results find significant associations of DEB with several pesticides with modest ORs in the 1.2 to 1.5 range. Five of the seven organochlorine insecticides showed statistically significant ORs for DEB, although we observed attenuation of risk estimates after adjustment for other pesticides that were also associated with DEB, possibly due to over-adjustment for correlated organochlorines. Some studies have linked organochlorine insecticides to PD. For example, higher levels of DDT [27], dieldrin [27], and β- and γ- isomers of hexachlorocyclohexane [10] were found in the brains or blood samples of PD cases compared with controls. Further, a recent study reported that consumption of milk possibly contaminated with heptachlor epoxide was associated with reduced neuron density in the substantia nigra in the brains of deceased participants in the Honolulu-Asia Aging Study [28]. Our findings on DEB suggest that these organochlorines might also contribute to PD-associated synucleinopathy early in the disease process.

Compared to previous studies, the current study is much larger in sample size and collected risk factor data about 20 years before DEB screening. Further, the study was conducted among farmers, who are more likely to accurately report use of specific pesticides than the general population [29]. Our study also has several limitations. First, as discussed earlier, the use of DEB screening in this study has likely included substantial false positives. Future studies should explore the approach proposed by Postuma et al. [30] to minimize false positives in large population-based studies. Second, exposures were self-reported and therefore misclassifications are inevitable. However, as exposure information was collected about two decades before DEB screening, we expect that both exposure and outcome misclassifications are non-differential and would tend to bias risk estimates toward the null. Third, in a large study such as this one, we had to rely on self-report as the first-step in identifying potential PD patients. Although we conducted diagnostic confirmation by collecting and evaluating relevant medical information from patients and their treating physicians [12], ascertainment errors and diagnostic misclassification is inevitable. Fourth, the associations we identified were modest. Although we adjusted for several potential confounders, we could not exclude the possibility of residual confounding. Finally, as our study participants were male farmers with occupational pesticide exposure, the results may not be readily generalizable to other populations.