Type 2 Diabetes Risk among Incident Parkinson’s Disease Patients: A 20-Year Retrospective Population-Based Cohort Study

Abstract

Introduction

Parkinson’s disease (PD) and type 2 diabetes mellitus (T2DM) are common chronic conditions that may share biological mechanisms such as inflammation and insulin resistance. While T2DM is associated with increased PD risk, evidence for T2DM risk in newly diagnosed PD patients remains limited. This study was designed to estimate the risk of T2DM among incident PD patients and assess the effect of T2DM on PD patient mortality.

Methods

A population-based retrospective cohort study was conducted in 7,976 incident PD patients without prior T2DM according to a validated registry in Maccabi Health Services medical data (2000–2019). Cox regression models, adjusted for age, sex, and comorbidities, were used to estimate hazard ratios (HRs) for both T2DM onset and the effect of T2DM on mortality risk among PD patients.

Results

During a mean follow-up of 7.1 years, 666 patients (8.4%) developed T2DM. Male sex was significantly associated with a reduced risk of T2DM onset (HR = 0.77; 95% CI: 0.66–0.89) compared to females. Furthermore, patients with T2DM but without complications had a significantly lower risk of mortality than those without T2DM (HR = 0.70; 95% CI: 0.62–0.78).

Conclusion

The risk of developing T2DM was relatively low in male PD patients without pre-existing diabetes. Moreover, PD patients with uncomplicated T2DM had a lower mortality risk than nondiabetic individuals. These findings highlight the importance of considering the timing of comorbidities and suggest sex-specific mechanisms linking PD and T2DM that warrant further investigation.

Introduction

Type 2 diabetes mellitus (T2DM) and Parkinson’s disease (PD) are two of the most common chronic conditions in aging populations, both of which contribute substantially to morbidity and to the healthcare burden [1–4]. The global age-standardized incidence of T2DM is estimated at approximately 285 per 100,000 persons, with notable regional variation and a continuing upward trend. Between 1990 and 2019, the global incidence increased at an annual rate of 1.25% (95% CI: 1.19–1.31) [5–7]. The global age-standardized incidence of PD is about 15.6 per 100,000 person-years, with higher rates among older adults and men [8]. Both diseases contribute significantly to disability, mortality, and healthcare costs, representing a growing global challenge in the context of population aging and longer life expectancy [1–5, 8].

Epidemiological studies and meta-analyses have consistently demonstrated an association between T2DM and an increased risk of developing PD, with relative risks ranging from 1.19 to 1.41 and an even greater risk observed in individuals with long-term diabetes or with diabetes-related complications [9–12]. Recent meta-analyses and population-based studies indicate that the risk is up to 1.54 times higher in those with diabetes complications and increases further with longer disease duration (hazard ratio [HR] up to 1.62 for ≥5 years) [9–11]. This association remains robust across diverse populations, including large cohorts from Europe and Asia, and persists after adjustment for major comorbidities and confounders such as cardiovascular, cerebrovascular, and chronic kidney diseases [9–12]. Findings from systematic reviews, Mendelian randomization analyses, and focused reviews reinforce the reproducibility and consistency of this relationship between T2DM and increased PD risk [1, 9, 13–15].

Mechanistic studies have identified several convergent pathogenic pathways linking T2DM and PD. Central among these is insulin resistance, where impaired insulin signaling in the brain promotes α-synuclein aggregation, dopaminergic neuronal loss, and neuroinflammation, hallmarks of PD pathology [1, 16–18]. Mitochondrial dysfunction, driven by oxidative stress and disrupted energy metabolism, contributes to neuronal vulnerability and progressive neurodegeneration in both conditions. Chronic inflammation, particularly activation of the NF-κB/NLRP3 inflammasome axis, connects peripheral metabolic dysfunction in T2DM with central neuroinflammatory processes in PD. Impaired autophagy and defective protein handling further exacerbate neuronal loss through shared disruption of the ubiquitin-proteasome system and endoplasmic reticulum stress [16, 19–21]. Clinical studies also suggest that poor glycemic control, as evidenced by elevated HbA1c levels, is associated with greater neuroaxonal damage and more severe motor and non-motor manifestations in PD patients [22, 23]. More recent evidence also implicates ferroptosis, an iron-dependent form of cell death, as a shared mechanism in which oxidative stress and reactive oxygen species drive both β-cell and dopaminergic neuron degeneration [2]. These converging metabolic and neurodegenerative mechanisms have recently been conceptualized under the term “metabolic PD,” highlighting the role of systemic metabolic dysfunction in PD pathogenesis [24].

Though most research has focused on the risk of PD among individuals with pre-existing T2DM, emerging evidence suggests that the relationship between these diseases may be bidirectional. Patients with PD may have an increased likelihood of developing T2DM due to reduced physical activity, autonomic dysfunction, metabolic consequences of dopaminergic therapies, and the shared pathogenic mechanisms described above. The coexistence of T2DM in PD patients is associated with more severe motor and non-motor symptoms, faster progression of both motor and cognitive decline, greater loss of independence, and elevated risk of depression and gait impairment. Recent longitudinal and clinical studies indicate that diabetes in PD is associated not only with a worse symptom burden but also with accelerated disease progression and increased mortality, highlighting a clinical relevance beyond disease risk alone [14, 16, 25–27]. Observational and genetic studies support a causal role of diabetes in PD risk and progression, while meta-analyses confirm that comorbid diabetes accelerates the disease trajectory and diminishes quality of life in PD patients [4, 9, 13–15]. Importantly, even prediabetes and impaired fasting glucose have been linked to a modestly increased risk of PD, particularly among younger individuals and women, suggesting that a continuum of metabolic dysfunction may contribute to neurodegeneration. Given the high prevalence of T2DM in aging populations, a substantial proportion of PD patients may benefit from targeted metabolic screening and intervention, including the potential repurposing of antidiabetic medications with neuroprotective properties [1, 4, 9, 15, 16, 22, 25].

Despite growing evidence that diabetes influences PD risk, severity, and progression, data concerning the incidence, timing, and determinants of T2DM following PD diagnosis remain limited and inconsistent across populations [27–29]. Population-based cohort studies are required to clarify the temporal dynamics, risk factors, and clinical implications of T2DM in incident PD, with the goal of informing integrated management strategies and improving patient outcomes. Accordingly, this population-based retrospective study was designed to estimate the incidence of T2DM among newly diagnosed PD patients and to evaluate the effect of T2DM status on mortality risk.

Methods

Study Design and Source Population

We conducted a population-based, retrospective cohort study of PD patients during the period between January 1, 2000, and December 31, 2019. The source population comprised members of Maccabi Health Services (MHS), the second-largest health maintenance organization (HMO) in Israel, covering approximately 2.15 million individuals (25.4% of the national population).

PD assessment: PD diagnosis was identified using a validated drug-tracer algorithm based on pharmacy dispensing records [28]. The algorithm incorporates 17 generic anti-parkinsonian drugs (APDs) available in Israel during the study period and classified as dopaminergic agents under ATC code N04B. Diagnosis was based on patterns of medication use, and data included age at first purchase, purchase frequency, and treatment duration. The algorithm was validated against clinical diagnoses from the Movement Disorders Unit in Tel Aviv Sourasky Medical Center, a tertiary referral center, and exhibited high sensitivity and positive predictive value [29, 30]. The date of first APD purchase was used as a proxy for PD diagnosis. Patients in the PD cohort were followed from the date of PD diagnosis until the earliest occurrence of one of the following: T2DM diagnosis, leaving MHS, death, or the end of the study (December 31, 2019).

T2DM assessment: PD cohort members were cross referenced to the MHS Diabetes Registry. T2DM diagnosis was based on American Diabetes Association criteria, including fasting plasma glucose >125 mg/dL, or random plasma glucose ≥200 mg/dL. Additional inclusion criteria were at least two purchases of hypoglycemic medications or a single insulin purchase within 6 months, or HbA1c ≥7.25%. Patients with HbA1c >6.5% were included only if they had a documented clinical diagnosis of diabetes. Patients with pre-existing PD who had purchased insulin before the start of the study period were excluded.

Demographic information of PD cohort members included year of birth, sex, and MHS status (active, deceased, or transferred to another HMO). Clinical data, including hypertension, chronic obstructive pulmonary disease (COPD), cancer, myocardial infarction (MI), congestive heart failure (CHF), and stroke, as based on the relevant ICD-10 diagnostic codes, were obtained from validated disease-specific registries. The date of registry entry for each condition was recorded.

Statistical Analysis

Descriptive statistics for continuous variables are presented using the mean and standard deviation, and categorical variables are presented as frequencies and percentages (Table 1). Kaplan-Meier curves were used to assess the cumulative probability of incident T2DM from the time of PD diagnosis, stratified by sex. Cox proportional hazards regression was applied to estimate adjusted HRs and 95% confidence intervals (CIs) for T2DM onset among the incident PD patients during follow-up, adjusting for age, sex, and comorbidities, including cancer, MI, CHF, COPD, and stroke (Fig. 1).

Table 1.

General and clinical characteristics of the incident PD cohort (n = 7,976)

Age, years
Mean (SD)	70.0 (10.6)
Sex (male), n (%)	4,277 (53.6)
Follow-up time, yearsa
Mean (SD)	7.1 (4.5)
T2DM, n (%)b	666 (8.4)
Cancer, n (%)c	2,267 (28.4)
MI, n (%)c	610 (7.6)
CHF, n (%)c	681 (8.5)
COPD, n (%)c	627 (7.9)
Stroke, n (%)c	812 (10.2)
Diabetes retinopathy, n (%)b	12 (0.1)
Diabetes nephropathy, n (%)b	12 (0.1)

^aFrom PD diagnosis to T2DM diagnosis, leaving MHS, death, or end of study.

^bIncidence.

^cPrevalence.

Fig. 1.

Kaplan-Meier curves of time to T2DM onset among incident PD cohort members, by sex (n = 7,976).

A Cox regression model was used to estimate adjusted HRs and 95% CIs for the effect of T2DM on mortality risk among incident PD patients. Survival time was calculated from the date of first APD purchase until death, exit from the HMO, or end of follow-up. The model was adjusted for age, sex, and comorbidities. T2DM was defined as a categorical variable reflecting the presence of diabetes-related complications: retinopathy, nephropathy, or both (Tables 2, 3). All statistical analyses were performed using R statistical software (version 4.1.3; R Foundation for Statistical Computing, Vienna, Austria).

Table 2.

Adjusted HRs and 95% CIs for T2DM onset among incident PD patients (N = 7,976)

Variables	HR (95% CI)	p value
Sex (male)	0.77 (0.66–0.89)	<0.001
Age at PD diagnosis	0.99 (0.99–1.00)	0.101
MI	1.15 (0.86–1.50)	0.350
CHF	2.18 (1.72–2.77)	<0.001
COPD	0.92 (0.69–1.22)	0.551
Stroke	1.36 (1.08–1.73)	0.010

Table 3.

Adjusted HRs and 95% CIs for the effect of T2DM on mortality among incident PD patients (n = 7,976)

Variables	HR (95% CI)	p value
Sex (male)	1.44 (1.35–1.53)	<0.001
Age at PD diagnosis	1.09 (1.08–1.09)	<0.001
T2DM
No T2DM	1	​
T2DM (without any diabetes complications)	0.70 (0.62–0.78)	<0.001
T2DMa	0.39 (0.05–2.77)	0.346
T2DMb	1.25 (0.56–2.79)	0.581
T2DMc	0.98 (0.14–6.96)	0.984
MI	1.09 (0.98–1.21)	0.097
CHF	1.21 (1.10–1.33)	<0.001
COPD	0.94 (0.84–1.05)	0.257
Stroke	1.14 (1.05–1.24)	0.002

^aType 2 diabetes mellitus+ – diabetes retinopathy.

^bType 2 diabetes mellitus++ – diabetes nephropathy.

^cType 2 diabetes mellitus+++ – diabetes retinopathy + nephropathy.

Results

General Characteristics of the Incident PD Cohort

The study cohort included 7,976 patients with incident PD who were free of T2DM at the time of initiating anti-parkinsonian treatment. The mean age at PD diagnosis was 70.0 years (SD = 10.55), and 53.6% of the cohort were male. Over a mean follow-up period of 7.1 years (SD = 4.5), 666 patients (8.4%) developed T2DM with an average time to diabetes onset of 4.34 years (SD = 3.58). Common comorbidities included cancer (28.4%), MI (7.6%), CHF (8.5%), COPD (7.9%), and stroke (10.2%). Diabetic microvascular complications were rare in this group, with retinopathy and nephropathy diagnosed in only 0.1% of patients. Table 1 summarizes the demographic and clinical characteristics of the study cohort.

Risk of T2DM Onset among the Incident PD Cohort

Figure 1 presents Kaplan-Meier curves showing time to incident T2DM over the 19 years of study period, stratified by sex. Among the 7,976 patients without T2DM at the start of PD treatment, 3% and 8% of males developed T2DM after 10 and 20 years, respectively, compared to 4% and 10% of females at the same time points. The difference in cumulative incidence between sexes was statistically significant (log-rank p = 0.003), indicating a higher risk of T2DM among PD females.

According to the adjusted Cox regression model, male sex was associated with a 23% lower risk of developing T2DM compared to female sex (HR = 0.77; 95% CI: 0.66–0.89). CHF and stroke were significantly (p < 0.001) associated with an increased risk of T2DM (HR = 2.18 and HR = 1.36, respectively). In contrast, age at PD diagnosis, MI, and COPD were not statistically associated with T2DM onset in this cohort. Table 2 summarizes the adjusted HRs for all variables.

Effect of T2DM on Mortality Risk among PD Patients

According to the multivariable Cox regression model (Table 3), patients with T2DM but no complications had a significantly lower risk of mortality than patients without diabetes (HR = 0.70; 95% CI: 0.62–0.78; p < 0.001), although there was no significant relationship between mortality and the presence of diabetes complications. In addition, male sex was significantly associated with a higher mortality risk than that seen in females (HR = 1.437; 95% CI: 1.348–1.533; p < 0.001). Age at PD diagnosis was also associated with a higher risk of mortality (HR = 1.088; 95% CI: 1.084–1.093; p < 0.001). In addition, the comorbidities, CHF (HR = 1.209; 95% CI: 1.100–1.328; p < 0.001) and stroke (HR = 1.138; 95% CI: 1.047–1.235; p = 0.002), were significantly associated with increased mortality risk, but there was no correlation with MI and COPD.

Discussion

This study utilized a large, population-based, retrospective cohort comprising 7,979 incident PD patients to examine the development of T2DM following PD diagnosis. To our knowledge, this is the first study to explicitly address the chronology between PD onset and subsequent T2DM diagnosis, with a follow-up period spanning two decades. During the 20-year study period, 8.4% of incident PD patients developed T2DM. Notably, this incidence demonstrated significant sex-related differences, with males exhibiting a 23% lower risk of developing T2DM than females.

Our findings are supported by a previous population-based study using the UK General Practice Research Database, in which Becker et al. [28] reported a lower incidence of T2DM among PD patients compared to matched controls. The protective effect was most pronounced among patients treated with levodopa, while those not using levodopa showed no significant difference in risk. Importantly, that study did not find significant sex-related differences in T2DM risk. Additionally, Scigliano et al. [31] reported a lower frequency of diabetes and other vascular risk factors in newly diagnosed PD patients compared to controls, suggesting a possible protective or inverse association between diabetes and PD risk.

Notably, most large-scale epidemiological studies have consistently failed to detect sex-based differences in the relationship between T2DM and PD [32]. Nevertheless, population-level data suggest that biological factors such as sex hormones, genetic predisposition, and differences in body composition and metabolic reserve may contribute to sex-specific diabetes risk. These mechanisms may be even more important in the context of neurodegenerative disease and treatment [4].

The observed associations between CHF and stroke and increased risk of T2DM in PD patients align with prior evidence implicating vascular comorbidities in diabetes pathogenesis. Mechanistically, these conditions may promote insulin resistance and hyperglycemia through reduced physical activity, autonomic dysfunction, and systemic inflammation [9, 21]. In contrast, factors such as age at PD diagnosis and the presence of MI and COPD were not significantly associated with incident T2DM in our cohort. This finding is in accordance with previous reports that these variables do not predict new-onset T2DM after multivariable adjustment [28].

The results of our present study identify several important factors associated with mortality risk among individuals with incident PD in the context of comorbid T2DM. As demonstrated previously, men with PD exhibited a significantly greater risk of mortality than females. This may be attributable to a combination of biological, behavioral, and comorbidity-related factors, as well as potential differences in disease phenotype and progression [33–36].

Older age at PD diagnosis was also associated with higher mortality, with an approximate 9% increase in risk per year. This finding is well supported by multiple large-scale studies and meta-analyses, which have consistently identified age at onset as a key prognostic factor [33, 35, 36]. Patients diagnosed at an older age tend to have shorter survival, probably because of a combination of greater frailty, higher comorbidity burden, and more rapid disease progression [36].

Most importantly, we found a significantly lower risk of mortality for PD patients with T2DM but no diabetes-related complications than for nondiabetic individuals. It should also be noted that data on specific anti-diabetic drug use were not collected, and therefore the potential influence of these medications on the observed mortality patterns could not be evaluated. In the absence of direct literature evidence, it is plausible that this counterintuitive finding reflects the impact of the more frequent medical surveillance and routine cardiovascular and metabolic monitoring associated with diabetes care. Such monitoring may facilitate earlier detection and management of comorbidities and complications, ultimately contributing to improved survival outcomes [1, 9, 10, 13, 14, 17, 25, 26, 37].

It is also possible that the absence of T2DM complications serves as a proxy for overall better health status. However, there is a general consensus that T2DM, especially when complicated, is associated with increased morbidity and mortality in PD [38], and the observed protective association in this cohort warrants further investigation.

Regarding comorbidities, both CHF and stroke were significantly associated with increased mortality. These results are consistent with large-scale cohort and meta-analytic data indicating that cerebrovascular events (including both ischemic and hemorrhagic stroke) and heart failure are among the strongest predictors of mortality in PD, often exceeding the influence of other cardiovascular comorbidities such as MI and COPD. Notably, after adjustment, we did not detect significant associations between these conditions and mortality in this study [35, 36, 39]. The differential impact of the examined comorbidities suggests that cerebrovascular and heart failure-related pathophysiology may interact more directly with the neurodegenerative and functional decline seen in PD, leading to greater vulnerability and worse outcomes. Similarly, a stroke can exacerbate motor and cognitive impairment, while heart failure may compound autonomic dysfunction and frailty. In contrast, MI and COPD, although important in the general population, do not independently predict mortality in PD after accounting for other risk factors [36, 40].

This study has several limitations. First, the PD cohort includes only individuals insured by Maccabi Healthcare Services, which covers approximately 25% of the Israeli population. As such, the findings may not be fully generalizable to the entire PD population in Israel, or to other healthcare settings. Second, the dataset lacked detailed clinical information about diabetes management, including the use of specific anti-diabetic medications, which could influence both disease progression and outcomes. Similarly, data on anti-parkinsonian medications were not available, and given the frequent changes and combinations of treatment regimens among PD patients, we could not account for potential effects of these drugs on disease course or mortality. Additionally, important metabolic and lifestyle variables such as the level of HbA1c, body mass index, and smoking status were unavailable. These unmeasured factors could potentially confound the associations observed between T2DM, comorbidities, and mortality in PD.

Future Recommendations

Future studies should further elucidate the mechanisms underlying the observed associations between PD, T2DM, and mortality. In particular, prospective studies incorporating detailed clinical parameters for glycemic control, diabetes duration, and the use of specific anti-diabetic medications are needed to clarify whether pharmacological factors or disease management strategies contribute to the apparent survival advantage observed among PD patients with uncomplicated T2DM. Additionally, future research should examine the potential influence of sex, metabolic status, and vascular burden by incorporating more granular biological and lifestyle data, including body mass index, smoking status, and laboratory measures such as HbA1c. Finally, studies conducted in other healthcare systems and populations will be essential to assess the generalizability of our findings and to determine whether targeted surveillance and management of metabolic comorbidities may improve outcomes in individuals with PD.

Statement of Ethics

This study protocol was reviewed and approved by the Institutional Review Board (IRB) of MHS-Assuta Medical Center (No. 0055-18-BBL) and by the Tel Aviv University Ethical Committee (No. 0000315-1). Personal ID numbers, used to cross reference the different databases, were encrypted prior to delivery to the investigators to ensure the anonymity of participants. The study involved no direct interaction with patients; therefore, informed consent was not required by the IRBs.

Conflict of Interest Statement

Dr. Elad Provisor and Dr. Violetta Rozani: none. Prof. Chava Peretz is a Review Board Member at Neuroepidemiology. Prof. Nir Giladi served as a Review Board Member for Neuroepidemiology and passed away prior to the submission of this manuscript. All remaining authors take full responsibility for the integrity and accuracy of the work and agree to the authorship of this manuscript.

Funding Sources

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Author Contributions

E.P.: methodology, formal analysis, and writing – original draft preparation. V.R.: conceptualization, supervision, methodology, and writing – original draft preparation. N.G. (deceased): conceptualization, study design, and data interpretation. The authors acknowledge with gratitude N.G.’s invaluable contribution to this work. C.P.: conceptualization, supervision, methodology, and writing – review. E.P., V.R., and C.P. read and approved the final manuscript.

Funding Statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data Availability Statement

The data that support the findings of this study are not publicly available due to Institutional Review Board (IRB) restrictions. De-identified, aggregated data gathered for this study may be available from the corresponding author (V.R.) upon reasonable request and subject to institutional approval.