Risk of Dementia in Patients with Depression or Parkinson's Disease: A Retrospective Cohort Study

Background This study aimed to clarify whether Parkinson's disease (PD) and depression were independent risk factors or with synergic effects in dementia. Methods Newly diagnosed PD (n = 1213) patients and control subjects (n = 4852) were selected from the Taiwan National Health Insurance Research Database from January 2001 through December 2008. Follow-up ended in 2011 with an outcome of dementia occurring or not. This cohort was divided into controls with or without depression, PD only, and PD with depression. The incident rate of dementia and hazard ratio (HR) using Cox's regression analysis were calculated for each group. Results When compared with controls without depression as HR 1.00, the adjusted HR for dementia was 3.29 (p < 0.001) in the PD only group, 2.77 (p < 0.001) in the PD with depression group, and 1.55 (p=0.024) in the depression only group. The incident rate of dementia was 29.2 (per 1000 person-years) in the PD only group and 13.2 in the PD with depression group. The effect of PD on dementia in the depression group produced a HR of 0.97 (p=0.905). Conclusions Parkinson's disease served as a risk factor for dementia. By comparison, depression was not a risk factor for dementia in PD patients, although it did act as a risk factor for dementia.


Introduction
In addition to the motor symptoms such as bradykinesia, rigidity, tremor, and postural instability, the cumulative risk of dementia in Parkinson's disease (PD) has been confirmed by researchers [1][2][3]. Depression is also conceived as a good predictor of dementia, and the available evidence indicates shared risk factors and biological changes between dementia and depression [4,5]. Age, duration, and severity of PD are risk factors of dementia in PD [6], but whether depression and PD demonstrate a synergic effect in the causes of PD dementia (PDD) remains unclear [7]. To answer this query, we have compared the risk of dementia between the general population and PD patients with or without depression.

Study Design and Population.
In Taiwan, around 99% of the population participate in National Health Research Insurance (NHRI); hence, the data from the National Health Insurance Research Database (NHIRD) are representative of epidemiology in Taiwan [8]. We randomly selected subjects from Longitudinal Health Insurance Database 2005 who were newly diagnosed with PD from January 2001 to December 2008. e definition and search methodology were based on the International Classification of Diseases, 9 th revision, and clinical modification (ICD-9-CM) code 332.0. We took a previous study as a reference for inclusion and exclusion criteria [9]. e subjects were included only with a diagnosis made in three or more consecutive visits at outpatient clinics or with a diagnosis made during hospitalization to increase the validity. e patients who had dementia before PD was diagnosed, had risks for secondary or atypical Parkinsonism, took antidopaminergic medicine more than three times within 3 months before the diagnosis of PD, or never took antiParkinsonian medicine after the diagnosis of PD were excluded. e non-PD control subjects were selected randomly from the database, and they were matched 4 : 1 with PD subjects for age, sex, Charlson comorbidity index (CCI), and the index date (the date of initial diagnosis of the PD subjects). ose who had a diagnosis of dementia before the index date and had a diagnosis of Parkinsonism were excluded from the non-PD control group. e PD subjects and the non-PD subjects were divided based on depression. Depression was defined based on the ICD-9-CM codes, including 296.2 (major depressive disorder, single episode), 296.3 (major depressive disorder, recurrent episode), 300.4 (dysthymic disorder), 309.0 (adjustment disorder with depressed mood), 309.1 (prolonged depressive reaction), and 311 (depressive disorder, not elsewhere classified). For clinical validation, inclusion was also confined to diagnoses made in three or more consecutive visits at outpatient clinics or made during hospitalization.
We followed up with the subjects from the diagnosis of PD until December 2011. e diagnosis of dementia was the primary clinical outcome. Dementia was diagnosed by ICD-9-CM codes (290, 331.0, 331.2, and A210) in three or more consecutive visits at outpatient clinics or during hospitalization.
e usage and release of the data were approved by the ethics committee of NHRI. Because of anonymous analysis, there was no written consent from participants. e study complied with the Declaration of Helsinki.

Statistical
Analysis. Stata version 11.1 was used to perform all statistical analyses. Between PD and non-PD subjects, we used Student's t-test and chi-squared test/ Fisher's exact test to compared age, gender, Charlson comorbidity index (CCI), depression, hypertension, diabetes mellitus, hyperlipidemia, and ischemic heart disease. e incident rate of dementia was calculated as the number of events divided by person-years. Hazard ratio (HR) was calculated as the risk of dementia between PD and non-PD subjects using Cox's regression analysis. HRs in different groups (depression or not), genders, and ages (<50, 50-64, and ≥65 years old) were calculated to elucidate the impact of PD in each group. Based on previous studies [9][10][11][12][13], adjusted HRs were acquired for age, gender, hypertension, diabetes mellitus, hyperlipidemia, and ischemic heart disease because metabolic syndrome is broadly considered as having an association with PD, depression, and dementia. p values< 0.05 and confidence intervals (CIs) for HRs, which excluded value 1.00, were statistically significant. Table 1 shows the characteristics of the study population. Six thousand and sixty-five subjects were included in the study, 1213 of whom were newly diagnosed PD subjects. e prevalence of depression in all of the patients was 11.3%. e occurrence of depression was significantly higher in the PD group than in the non-PD group (27.4% vs. 7.3%, p < 0.001). Significantly, more patients had hypertension in the PD group than in the non-PD group (37.3% vs. 33.5%, p � 0.013). Between the two groups, there were no significant differences in age, gender, Charlson comorbidity index (CCI), or the presence of diabetes mellitus, hyperlipidemia, or ischemic heart disease. Table 2, the adjusted HR for dementia was 1.10 (95% CI 1.09-1.11, p < 0.001) with aging, 1.64 (95% CI 1.30-2.07, p < 0.001) in subjects with diabetes mellitus, 0.70 (95% CI 0.52-0.93, p � 0.014) in subjects with hyperlipidemia, and 1.23 (95% CI 1.00-1.51, p � 0.047) in subjects with hypertension. In terms of genders and ischemic heart disease, no statistically significant differences were detected for the HRs of dementia. When comparing to the subjects who had neither PD nor depression, the adjusted HRs for dementia were 1.55 (95% CI 1.06-2.25, p � 0.024) in the PD only group, 3.29 (95% CI 2.66-4.06, p < 0.001) in the depression only group, and 2.77 (95% CI 1.88-4.08, p < 0.001) in the group with both PD and depression.

Hazard Ratios for Dementia with PD in Different
Subgroups. Table 3 reveals a higher incidence of dementia in PD than in the control (24.2 vs. 8.3/1000 person-years). e crude hazard ratio (HR) for dementia in PD was 2.91 (95% CI 2.39-3.53, p < 0.001), and the adjusted HR was 3.05 (95% CI 2.51-3.70, p < 0.001). In terms of gender, hazard ratios were 2.92 (95% CI 2.19-3.89, p < 0.001) and 3.20 (95% CI 2.45-4.17, p < 0.001) in female and male groups, respectively. In groups divided by ages, all the groups had a HR greater than 1.00 for dementia in PD. e highest incident rate of dementia was in PD subjects older than 65 years  Table 4 reports that, without depression, the adjusted HR for dementia with PD was 10.08 (95% CI 4.67-21.76, p < 0.001) in the younger group and was 2.99 (95% CI 2.39-3.74, p value < 0.001) in the older group. In contrast, HRs did not reach statistical significance in each group with depression (adjusted HR 7.06; p � 0.074 in the younger group and 1.55; p � 0.12 in the older group).

Discussion
A higher prevalence of depression was observed in PD patients in our study which is similar to the findings of other studies [7,10]. However, the incident rate of dementia in the PD without depression group was higher than in the PD plus depression group. is finding implies that depression might not be a risk factor in dementia of PD. is supports the result of a previous cohort study, where the authors concluded a later age of onset of PD, increased disease severity of PD, and hallucination were predictive factors for dementia development, but depression was not [14]. In contrast, some studies have suggested that depression can precede dementia in PD patients and might be a risk factor for dementia in PD [6,15]. ese opposite conclusions might be caused by different study designs or the difficulty in determining the effect of depression upon dementia in PD due to shared symptoms, such as apathy, psychomotor problems, and sleep disturbances, between depression and dementia. In our study, having depression was associated with a higher rate of dementia; in the discussion about the incidence of dementia with or without PD in the groups having depression, having PD did not make much of a difference in the incidence rate of dementia in this study. is revealed that depression plays an important role in dementia. Previous studies had also revealed depression to be a risk factor of dementia [4,16]. e specific types of dementia associated with depression continue to be debated. It has been proposed that depression might precede the onset of Alzheimer's disease or vascular dementia [17,18]. Adjusted HR: adjusted for age, gender, hypertension, diabetes mellitus, hyperlipidemia, and ischemic heart disease. c "No" � without depression; "yes" � with depression. We also divided the PD subjects by age, and it reveals that having PD does increase the risk of subsequent dementia, irrespective of early-onset PD or older-onset PD. However, different characteristics of incidence rates and HRs of dementia were found between these groups. A higher incidence rate detected was in the older-onset group, but HRs were higher in the early-onset group.
is finding is consistent with the different progression of pathology between early-onset PD and older-onset PD proposed by others. Older-onset PD is associated with shorter survival and a complex disease course. More rapid cognitive decline is also found in this phenotype associated with higher Lewy-related pathology and additional neuropathologies, such as Alzheimer's disease [19].
ere are some limitations to this cohort. First, the data were collected from the Taiwan National Health Insurance Research Database. erefore, the validity of the diagnosis might be questioned. In a Taiwanese study, the authors randomly selected patients who coded PD, and these patients were examined by experienced neurologists. e sensitivity, specificity, positive predictive value, and negative predictive value were 97.6%, 92.3%, 98.8%, and 85.7%, respectively [20]. To increase the validity, we also set strict selection criteria and enrolled a sufficient sample size for the study. Second, we did not subtype depression according to severity, symptoms, or onset time. Some researchers have supposed that differences in the onset time of depression might have different risks for dementia [21,22]. ird, we did not subtype dementia in the general population. Furthermore, additional studies about the biological network of PD, depression, and dementia are necessary to identify the relationships and causalities among these conditions.

Conclusion
PD and depression both serve as important risk factors for dementia. However, depression was not a risk factor for dementia in PD, despite the high prevalence of depression in PD.

Data Availability
e data are available from Taiwan's National Health Insurance Research Database. Disclosure e funders had no role in this study.

Conflicts of Interest
All authors declare that they have no conflicts of interest.