Occurrence of Depression and Anxiety prior to Parkinson’s Disease

Introduction

It is well-established that depression and other psychiatric disorders such as anxiety occur at higher rates in patients with Parkinson’s disease (PD) than in the general population [1, 2]. However, reported frequencies of depression in patients with PD vary widely with estimates from 7% to 70% [3]. A recent review of 36 studies [4] reported the mean depression prevalence to range from 17% to 35% among US and European PD patients, while a lifetime depression prevalence of 16% and a 12-month prevalence of about 7% have been documented among the general US population[5].

Depression and anxiety are significant non-motor symptoms in PD as they may be associated with a more rapid deterioration in cognitive and motor functions. [6] These psychiatric disorders are also major factors in health-related quality of life in PD [7–9]. Given the implications for clinical and quality of life outcomes, it is important to improve recognition and understanding of these disorders in PD.

Some studies have proposed that both depression and anxiety disorders precede PD development and therefore could be risk factors for PD.[10] However, it remains unclear whether depression and anxiety are independent risk factors for PD or early manifestations of the disease. Several studies have found that PD patients have more frequent histories of depression, anxiety and nervousness than controls and that these psychological symptoms were part of the prodromal phase that developed 3 to 6 years before the onset of motor dysfunction [11, 12]. Therefore, if depression and anxiety are risk factors, we would expect that PD patients, referred to as cases in the following, would be more likely than controls to have both a distant (before development of PD) and recent history of depression or anxiety. If they are early manifestations, we would expect that PD cases would not be more likely to have a distant history, but would be more likely to have a recent history of depression or anxiety than controls.

Most studies concluding that depression and anxiety are risk factors for PD did not differentiate between recent and distant diagnoses [13–18]. Instead, they included all diagnoses of depression or anxiety that occurred prior to PD diagnosis. Therefore, they failed to determine whether these psychiatric symptoms might have been prodromal symptoms of PD or distant occurrences that could be considered risk factors [19]. One record-based study explored this hypothesis by examining antidepressant use separately in those who initiated treatment >2 years and ≤2 years prior to diagnosis date. However, this study was limited to a maximum of 6 years of follow-up time after initiation of anti-depressant medications and did not have the benefit of having all PD cases examined by a movement disorder specialist.[20] Another study that investigated depression/anxiety diagnoses up to 20 years before onset of PD was limited by a small sample size (196 cases total, with only 10 cases and 8 controls affected by depression up to 20 years before index date) [21]. Here we investigate the role of depression and anxiety prior to PD diagnosis by examining the distant and recent past separately.

Methods

Results

Study participants were predominantly Caucasian, over the age of 60, and without a family history of PD (Table 1). Cases were slightly older than controls, were more often male, and had completed fewer years of education. Cases were also more likely to have never smoked or to have quit smoking. Male cases were more likely to have ever received a diagnosis of depression or anxiety than male population or sibling controls, but female population controls were more likely to have ever received a psychiatric diagnosis than female cases. Study participants who had received a psychiatric diagnosis and had been treated with psychotropic medications generally used more than one class of drug over their lifetime. Selective serotonin reuptake inhibitors (SSRIs) were most commonly reported. Less frequently reported were other anti-depressants, serotonin-norepinephrine reuptake inhibitors (SNRIs), benzodiazepines, tricyclic antidepressants, and other psychotropics.

PD cases had a slightly (40–50%) higher odds of having been diagnosed with depression and/or anxiety prior to index date than controls, but were not more likely to have reported both a psychiatric diagnosis and psychotropic medication use compared to either population or sibling controls (Table 2). However, male cases had higher odds of having been diagnosed and treated for depression or anxiety than population or sibling controls, which was mostly due to events within a 2 to 5 year period prior to PD diagnosis. Female PD patients seemed somewhat less likely to have been diagnosed and treated than population controls, but all confidence intervals included the null value of 1. Analyses stratified by age at PD onset (<=60, >60 years) and education (<12 years, 12 years, 13–16 years, >16 years) did not reveal stratum-specific differences (results not shown). Results of depression only analyses did not differ from the results reported for depression and anxiety together (results not shown).

Finally, we calculated mean GDS scores after grouping subjects according to their reported current (at time of interview) psychotropic medication use (Table 3). We found that both cases and controls who reported current medication use scored higher on the GDS than their un-medicated counterparts (t-test p-value = 0.03 for currently medicated cases vs population controls; p-value < 0.0001 for currently un-medicated cases vs. population controls). Among currently medicated participants, more cases than population or sibling controls were moderately and severely depressed.

In order to determine whether the difference in GDS scores may be due to PD motor symptoms that overlap with depression items from the GDS, we dropped GDS questions #2 (dropping activities and interests) and #9 (preference to stay at home, rather than going out and trying new things) and re-calculated mean GDS scores, since we hypothesized that positive responses to these questions may reflect cases’ motor symptom status rather than depressive symptoms. After excluding these two questions, we found that the differences in mean GDS scores between cases and controls persisted (t-test p-value = 0.04 for currently medicated cases vs population controls; t-test p-value <0.0001 for currently un-medicated cases vs population controls) suggesting that these differences reflect actual differences in depression status rather than differences due to motor symptoms.

Discussion

Depression and anxiety seem to be pre-clinical symptoms prior to the onset of PD motor symptoms rather than etiologic risk factors in our study. We observed a positive association between depression and anxiety diagnoses and PD among males but not females. When we restricted the definition of affected participants to only those with a psychiatric diagnosis and a history of psychotropic medication use, results were again driven by males and we did not observe any differences between PD cases and controls more than 5 years prior to index date. The results observed in males suggest that PD cases are more likely than controls to have been diagnosed with depression or anxiety up to 20 years prior to PD onset. However, the more stringent Diagnosis + Medication Model confirmed differences between male cases and controls only for the period 5 years but not 10 or 20 years prior to diagnosis. These latter results are consistent with two previous studies which also examined the timing of psychiatric disorders prior to PD.[20, 21] The differences observed in our two models might be due to recall bias (male cases more frequently recalling early-life depression or anxiety diagnoses for which they did not seek treatment), or a reluctance of male PD cases to seek medical treatment for psychiatric disorders earlier in life while seeking medical care more often for non-motor symptoms and having more opportunities to be diagnosed and treated for depression in the years before PD diagnosis, or chance.

Several previous studies have reported that depression and anxiety are associated with an increased risk for PD [13–17, 21], but most of these studies did not account for the prodromal phase of PD. [13] Thus the observed associations could be a result of depression occurring within 2 to 5 years before PD diagnosis, as observed in our study. The two studies most comparable in their approach to ours, in that they accounted for the timing of psychiatric events prior to PD, support our findings that the association between PD and prior psychiatric events is mostly due to events within 2 to 5 years prior to PD diagnosis. [20, 21]

Our current study is based on the largest population-based sample of incident PD cases evaluated by a movement disorder specialist in a standard protocol to explore this hypothesis. Misclassification of PD diagnosis is expected to be minimal due to our study’s stringent diagnostic criteria and clinical examination (repeated if necessary) of every PD case by a movement disorder specialist. Although we recruited from a well-defined source population, there is a possibility of selection bias if history of depression is linked to participation in the study. We would expect that depression at the time of recruitment would influence participation due to apathy or other psychiatric symptoms in both cases and controls, and depressed controls may be even more reluctant than cases to enroll in a medical study. However, this pattern was not observed for females. Also, since we investigated lifetime depression, we do not expect that a history of past depression would influence study participation. We might have over-matched on genetic factors when comparing cases to sibling controls, which would be expected to result in greater similarity with regard to depression and therefore a weakening of the associations. Yet, the effect estimates from our case/sibling based analyses were similar to or stronger than the case/population control estimates, suggesting that prodromal PD rather than genetic predisposition is at play. Finally, because we assessed depression retrospectively and via self-report, misclassification of lifetime depression and anxiety diagnoses and psychotropic medication use is possible. Studies have demonstrated that self-reported age at onset of major psychiatric diagnoses is fairly reliable [25, 26]. Importantly, one might speculate that exposure recall of diagnoses and medication use decades prior to the interview is unreliable and may result in misclassification that is differential, since recently diagnosed PD cases may be more inclined to recall and report depression and anxiety, but this pattern was not observed in females. We investigated whether these differences are due to differences in education but did not find evidence supporting this hypothesis. Based on a national survey conducted from 2001 to 2003, the lifetime prevalence of depression in participants above age 65 was 5.6% in males and 13.3% in females, while that of anxiety was 9.6% in males and 16.6% in females.[27] Thus, while the rates in our male controls are similar to those of the national survey, the rates in our female controls are considerably higher. These high rates of psychiatric diagnoses in female controls could be due to over-reporting, truly increased rates, selection bias, or chance.

There are several early symptoms of PD, such as fatigue, that may be mistaken for depression. We account for this in part by lagging, i.e. removing from consideration depression diagnoses in the years immediately prior to PD diagnosis when fatigue may be present and mistaken for depression. We also believe that our more stringent diagnosis plus medication requirement lessens the possibility of misclassifying depression due to fatigue, assuming that symptoms of fatigue alone would not result in both a depression diagnosis and anti-depressant treatment.

History of psychotropic medication use was employed both to confirm psychiatric diagnoses and as a marker of severity of psychiatric illness. Medication use may indicate that depression is more severe and, according to our GDS scores, there may even be some under-treated cases in our population. This is consistent with other studies that suggest depression in PD is under-diagnosed and under-treated [28, 29]. Although males who developed PD had slightly higher odds of receiving psychiatric diagnoses up to 20 years or more before PD diagnosis, we did not observe differences in psychotropic medication use until 5 years before index date.

This study contributes to the growing body of literature examining the role that psychiatric disorders may play prior to PD motor symptom onset. There is evidence of a prodromal phase in PD from imaging, pathology, and clinical, epidemiological, and animal studies [11, 30–32]. Estimates of duration of the prodromal period vary widely, from 3 up to 20 years, and symptoms can include olfaction deficits, dysautonomia, and sleep and psychiatric disorders[11, 30, 32]. While we assumed an average duration of 5 years for prodromal symptoms in this study, age and disease etiology may be important factors determining the actual length of this phase [11, 30, 32]. Although the pathophysiology of depression and anxiety in pre-clinical PD is not fully understood, [33] its basis may be noradrenergic, dopaminergic or serotonergic [34, 35] and the neurologic structures involved may included the substantia nigra or coeruleus/subcoeruleus complex, where dysfunction occurs in Braak Stage 2, and has been estimated to begin ten years or more prior to motor symptoms.