Background: Research on gender in Parkinson’s disease (PD) frequently focuses on clinical differences between men and women. Socioeconomic factors such as race, ethnicity, insurance type, and level of educational attainment, have not been extensively examined in relation to gender differences in PD. The goal of this study was to identify differences in PD presentation in men and women, and identify socioeconomic factors that may confound such differences. Methods: A movement disorder patient database containing 445 patients with idiopathic PD was analyzed for gender differences in motor symptoms and disease complications using linear or logistic regression. Socioeconomic variables were then evaluated as possible confounders. Results: A greater proportion of women were non-white (p < 0.05). Univariate analysis of gender, and multivariate analysis controlling for age at diagnosis and socioeconomic factors were concordant in demonstrating increased frequency of motor fluctuations and dyskinesia in women (p < 0.05). Controlled, multivariate analysis, when compared to univariate analysis, uniquely demonstrated that women were less likely to experience dementia and autonomic dysfunction, relative to men (p < 0.05). Conclusion: Women with PD are susceptible to motor fluctuations and dyskinesia, but may be relatively protected against dementia and autonomic dysfunction. Women and men with PD in our population had different socioeconomic profiles, which may have confounded some gender-associated differences.
Parkinson’s disease (PD) is a neurodegenerative disorder whose incidence and clinical features are observed to differ between the sexes. In many populations, men are more frequently affected than women, with epidemiologic studies in Western nations reporting age-adjusted male to female ratios between 1.35 and 2 [
PD also presents with clinical differences between men and women. Studies in Dutch and Norwegian populations demonstrate that men are significantly younger at PD onset than women by about 2 years [
This analysis examines a population of PD patients at an urban safety net hospital in Boston, MA to describe gender differences in PD within the context of socioeconomic factors such as racial and ethnic background, educational attainment, and insurance status. Often, clinical differences between men and women with PD are presumed to be a byproduct of sex, which refers to biological differences of men and women. Gender, in contrast, refers to the sociocultural factors by which men and women differ [
Patients were recruited from a movement disorders clinic in the neurology department at Boston University Medical Center (BUMC) in Boston, Massachusetts between 2007 and 2012. Patients entered into the database for this analysis were eligible if diagnosed by a movement disorder specialist at the clinic with idiopathic PD. New and existing patients without overlap were included in this analysis. The Institutional Review Board (IRB) at BUMC gave approval for this research.
Clinicians collected data using a standardized clinical characteristics data form. The data was then entered into a de-identified clinical database.
The data collection form included sections for demographics and clinical information. The demographics information included year of birth, sex, ethnicity, highest educational level completed, and insurance type. Clinical characteristics included the patient’s diagnosis and date of diagnosis, as well as the date of symptom onset, as reported by the patient. Disease severity was characterized for each patient using the modified Hoehn and Yahr (HY) scale, both ON and OFF medication. This scale stages PD based on severity of motor symptoms, with Stage 1 representing unilateral motor symptoms only, and Stage 5 representing a wheelchair-bound or bedridden state [
Four hundred and forty-five (445) patients were included in this analysis, whose demographic characteristics are outlined in
Patient Demographics | Female | Male | Whole Population | Sample Size | p Value* |
---|---|---|---|---|---|
Number of patients (%) | 184 (41.3) | 261 (58.7) | 445 | - | - |
Age in years (SD) | 68.6 (11.8) | 68.2 (10.4) | 68.3 (11.0) | 441 | 0.712 |
Age at onset in years (SD)† | 59.2 (12.6) | 57.4 (10.7) | 58.2 (11.6) | 355 | 0.147 |
Age at diagnosis in years (SD)† | 61.1 (10.7) | 59.1 (10.9) | 59.9 (10.9) | 369 | 0.08 |
Years since diagnosis (SD) | 7.58 (6.7) | 8.39 (7.2) | 8.01 (7.0) | 371 | 0.274 |
Years between onset and diagnosis (SD) | 1.65 (1.7) | 1.45 (2.4) | 1.55 (2.1) | 332 | 0.378 |
Educated beyond high school | 67.7% | 71.6% | 70.0% | 373 | 0.428 |
Public insurance† | 69.5% | 60.6% | 64.3% | 403 | 0.067 |
Non-white† | 18.4% | 11.0% | 14.1% | 433 | 0.03 |
SD = standard deviation, *α = 0.05 was threshold for statistical significance, while potential confounders of sex were identified using α = 0.2, †demarcates variables identified as potential confounders of sex.
patients (6.5%), and Hispanic patients (3.7%) comprised the next largest groups, respectively. Remaining racial and ethnic groups included Asian, Native American, Native Hawaiian or other Pacific Islander, and mixed race. These remaining groups altogether comprised 3.96% of the patient population. Level of educational attainment was also simplified for statistical analysis, dividing patients into those who had completed at least some college, and those with a High School diploma or less education. Each patient’s insurance plan was classified as private or public (as of 2007 all patients were required by Massachusetts law to have insurance).
Among the socioeconomic variables, there were some missing data. Of the 445 patients included in this analysis, 90 had no age at onset, 76 had no age at diagnosis, 4 had no available age, and for 113, the number of years between onset and diagnosis could not be calculated. Seventy-two (72) patients were missing level of education information, 42 had no available insurance information, and 12 had no racial or ethnic information. Generally, data were missing because the study neurologists sometimes completed data collection forms after the patient encounter, and were unable to clarify the missing information once the patient had already left. After excluding those with missing data, 329 patients were included in the multivariate regression analyses.
Men and women were first compared using univariate analyses. T-tests and Chi-Square tests were used for continuous and categorical variables, respectively. For the analysis of socioeconomic variables, α was set to 0.2, in order to screen out socioeconomic variables that were unlikely to be confounders, while remaining sensitive enough to identify variables that could confound the effect of sex. Since potential confounders were included in a multivariate regression model, it was optimal to limit the number of variables included, to achieve acceptable statistical power. For univariate analyses on the clinical variables of interest, as well as the multivariate regression analyses, α was set to 0.05. Linear regression was used for continuous outcomes, with β coefficients and standard error (SE) reported, while logistic regression was used for dichotomous outcomes, with odds ratios (OR) and confidence intervals (CI) reported. JMP statistical software (version 11) was used to run all statistical tests.
Women were more likely to be non-white compared to men (18.4%, 11.0%, respectively, p = 0.03). Men and women differed according to the confounder criterion p < 0.2 on age at disease onset, in which women were on average 1.8 years older at onset (p = 0.147). A difference by this criterion was also observed for age at diagnosis, in which women were, on average, 2 years older at diagnosis than men (p = 0.08), and for insurance, in which women were more likely to be on public insurance than men (69.5%, 60.6% respectively, p = 0.067). These tests established age at onset, age at diagnosis, insurance type, and race as possible confounding variables to include with sex in a regression model. Age at onset was excluded from the regression model, as it was too tightly linked to age at diagnosis to include both variables.
On univariate analysis, women were found to have greater disease severity OFF medication on the HY scale, by 0.38 points (p = 0.0024). No difference, either ON or OFF medication, was observed for this measure on multivariate regression analysis. Univariate analysis also demonstrated that, among clinical complications, women more frequently experienced dyskinesia (55.4% vs. 31.4%, p < 0.0001) and motor fluctuations (60.3% vs. 44.1%, p = 0.0007), while men more frequently experienced freezing (23.4% vs. 13.0%, p = 0.0055). In comparison, multivariate regression analysis showed that women more frequently experienced dyskinesia (p < 0.0001, OR = 2.92, CI = 1.78 - 4.85) and motor fluctuations than men (p = 0.004, OR = 2.07, CI = 1.26 - 3.42). Women were less likely to experience dementia (p = 0.005, OR = 0.436, CI = 0.234 - 0.786) and other autonomic dysfunction (p = 0.038, OR = 0.513, CI = 0.263 - 0.964) compared to men.
Univariate Analysis | Multivariate Analysis | |||||
---|---|---|---|---|---|---|
Hoehn and Yahr Scores | Female | Male | p Value* | β Coefficient | Standard Error | p Value*† |
ON medication (SD) | 0.755 (1.48) | 0.705 (1.33) | 0.713 | 0.006 | 0.077 | 0.934 |
OFF medication (SD) | 2.45 (1.27) | 2.07 (1.36) | 0.0024 | 0.103 | 0.071 | 0.145 |
Disease Complications | Female (%) | Male (%) | p Value* | Odds Ratio‡ | 95% CI | p Value*† |
Compulsive behavior | 5.4 | 7.7 | 0.351 | 0.932 | 0.373 - 2.22 | 0.875 |
Dementia | 17.9 | 23.4 | 0.164 | 0.436 | 0.234 - 0.786 | 0.005 |
Depression | 23.9 | 19.5 | 0.269 | 1.33 | 0.789 - 2.26 | 0.281 |
Dyskinesia | 55.4 | 31.4 | <0.0001 | 2.92 | 1.78 - 4.85 | <0.0001 |
Freezing | 13 | 23.4 | 0.0055 | 0.552 | 0.292 - 1.01 | 0.055 |
Hallucinations | 26.1 | 28.4 | 0.597 | 0.74 | 0.435 - 1.24 | 0.257 |
Motor fluctuations | 60.3 | 44.1 | 0.0007 | 2.07 | 1.26 - 3.42 | 0.004 |
Orthostatic hypotension | 13 | 12.3 | 0.807 | 1.15 | 0.568 - 2.32 | 0.691 |
Other autonomic dysfunction | 12.5 | 17.2 | 0.167 | 0.513 | 0.263 - 0.964 | 0.038 |
Psychosis | 3.8 | 2.7 | 0.507 | 1.62 | 0.468 - 585 | 0.44 |
SD = standard deviation, CI = confidence interval, *α = 0.05, † = p value for sex, adjusted for age at diagnosis, insurance type, and race, ‡ = compares women to men.
We have demonstrated several sex-associated socioeconomic and clinical differences in this PD patient population. Men were more numerous in our cohort: the male to female ratio was 1.42, a distribution consistent with what is reported in Western societies [
Our data suggest that age at diagnosis, as well as socioeconomic factors such as insurance type and race, may confound gender-associated differences in PD. In particular, univariate analyses demonstrated that women experience greater disease severity OFF medication, and experience less freezing than men; however there was no difference with respect to these features when controlling for age at onset, insurance type, and race in a multivariate regression analysis. Therefore, we assume that one or several of these factors may confound the effect of gender on these outcomes. Further study is needed to delineate the individual contributions of these confounders. Univariate and multivariate regression analyses were concordant in showing that motor fluctuations and dyskinesia occurred more frequently in women, suggesting a true biologic difference, which has been demonstrated repeatedly in the literature [
Controlling for confounders, multivariate analysis demonstrated that women are less likely to experience dementia and non-orthostatic type autonomic dysfunction. Studies on the relationship of gender with dementia in PD have been equivocal, showing increased frequency of dementia with PD in men in some populations [
In this analysis, we observe that men and women with PD have different socioeconomic profiles, and that such differences may confound some of the clinical differences that associate with gender. In our cohort, age at onset, race, ethnicity, and insurance status were identified as potential confounders of gender in PD. Future research should examine these relationships, to identify any predictive value they may have for disease features. Our analysis also validates prior knowledge about sex differences in PD that, in particular, women are susceptible to levodopa-induced motor fluctuations and dyskinesia, and may be relatively protected from dementia and some autonomic disturbances. This analysis is particularly valuable for examining a large population at a safety net hospital, whose health outcomes are often associated with socioeconomic factors. Our analysis was limited by some missing socioeconomic data, lack of longitudinal follow-up, and that disease severity was measured solely by the HY scale, which describes only motor symptoms. More nuanced scales, such as the UPDRS, have been developed that evaluate a broader array of disease manifestations, such as mood, behavior, and functional capacity. A strong follow-up study could examine a diverse population of PD patients over time, to identify socioeconomic factors that may associate with poorer outcomes.
The authors thank Boston University School of Medicine and the Medical Student Summer Research Program for providing guidance and funding for this project.
RobertMcInnis,WilliamCavanaugh,JaniceWeinberg,Marie-HélèneSaint-Hilaire,SamuelEllias,SamuelFrank,Anna DePoldHohler, (2015) Exploring Gender-Associated Socioeconomic Differences in Parkinson’s Disease. Advances in Parkinson's Disease,04,84-89. doi: 10.4236/apd.2015.44010