Table 2

List and features of case-control studies included in qualitative analyses

Source [reference]EthnicityPD (N)Age at inclusion (years)*Prevalence of diabetes (%)Control subjects (N) and matching variablesOR** (95%CI) for casesAdjustment variablesPotential estimation bias and other observations
Leibson et al. 2006 [14]U.S. (Olmsted County; Minnesota)197 incident cases70 (11)9.1197 subjects matched for age (± 1 year), sex, and geographical location0.7 (0.4–1.4)NonePresence of other neurologic disease (e.g., stroke or dementia) was not an exclusion criterion. A trend toward a higher prevalence of stroke and dementia was present in the PD group. BMI was not included among the potential confounders and adjusting variables.
Powers et al. 2006 [15]U.S. (Group Health Cooperative database; Washington)352 newly diagnosed cases of idiopathic PD without cognitive impairment69 [35–88]7.4484 subjects matched for age (in decades), sex, year of enrollment, and geographical location0.62 (0.32–1.01)Age, ethnicity, education, and smoking habitMedical conditions were self-reported. BMI was not included among potential confounders and adjusting variables.
Scigliano et al. 2006 [16]Italy157 newly diagnosed cases (duration of PD <6 months) of idiopathic PD58.1 (11.4)3.4533 subjects matched for age (± 3 years) and sex0.30 (0.13–0.72)Age and sexControl subjects were recruited in a hospital setting, and a higher prevalence of vascular risk factors (diabetes, hypertension, or dyslipidemia) might have occurred. BMI, although similar in both groups, was not included among potential confounders and adjusting variables. Finally, in stepwise multivariable analysis, diabetes was no longer associated with reduced risk of PD.
Becker et al. 2008 [7]U.K. (General Practice Research Database)3,637 new drug-free cases of PD (90% with an age at onset >60 years)Not reported (90% aged >60 years)8.0%3,637 subjects matched for age (same year of birth), sex, and general practice0.95 (0.80–1.14)BMI, smoking, and multiple comorbidities (several neurologic disorders, hypertension, cardiovascular diseases, and dyslipidemia)Detection bias deriving from increased medical surveillance related to some medical conditions could not be excluded at all.
D'Amelio et al. 2009 [8]Italy (Italian region of Sicily)318 newly diagnosed cases66.7 (-)4.1318 subjects matched for age (± 2 year) and sex0.4 (0.2–0.8)Age, sex, education, BMI, occupational status, alcohol and coffee consumption, and smoking habitAscertainment of diabetes was based on self-reported data.
Miyake et al. 2010 [17]Japan (Osaka, Kyoto, and Wakayama Prefectures)249 cases with a disease duration <6 years68.5 (8.6)4.0368 in-patients and out-patients not, individually or in larger groups, matched to cases0.38 (0.17–0.79)Age, sex, smoking, area of residence, BMI, education, leisure-time exercise, intake of energy, cholesterol, vitamin E, alcohol and coffee, and the dietary glycemic indexAscertainment of diabetes was based on self-reported data. Although control subjects were recruited in a hospital setting, and higher rates of comorbidities could be expected, prevalence of diabetes was comparable with that of the general population in the same area. Moreover, PD cases were thinner (P = 0.01) and older (P = 0.006), but multivariable models accounted for the effect of these confounders.
Schernhammer et al. 2011 [18]Denmark1,931 PD cases admitted to hospital with a first-time diagnosis and identified through the nationwide Danish Hospital Register72.2 (10.5)6.59,651 free-living individuals (5 for any case) selected from the Central Population Registry, matched for age (same year of birth) and sex1.36 (1.08–1.71)Age, sex, and chronic obstructive pulmonary disease (lagged 5 years as surrogates of smoking)Although cases were identified through the Danish Hospital registry, only those who were registered for the first time with a primary diagnosis of PD were included. This criterion should have significantly reduced the risk of higher rates of vascular factors (diabetes, hypertension, or dyslipidemia) in hospitalized patients. However, BMI was not included among potential confounders and adjusting variables. In sensitivity analysis performed after the exclusion of cases and control subjects diagnosed with dementia or cerebrovascular disease 2 years before the indexing, risk estimates were changed only minimally. Finally, in analyses restricted to PD cases aged >60 years at diagnosis, diabetes was no longer associated with increased risk of PD (OR = 1.16 [0.85–1.57]).
  • *Data are presented as median [range] or as mean (SD);

  • **OR (95% CI) for the association between PD and preceding diabetes.