Parkinson’s disease and parkinsonism in a longitudinal study

Study population.

The eight centers involved in the ILSA were the entire municipalities of Casamassima (Bari Province), Fermo (Macerata Province) Impruneta (Firenze Province), and Rubano-Selvazzano (Padova Province) and well-defined city districts of Catania, Genova, Milano, and Napoli. On March 1, 1992, using the population registers, all of the inhabitants of the eight study centers aged 65 to 84 years, free-living or institutionalized, were stratified by 5-year age group (65 to 69, 70 to 74, 75 to 79, and 80 to 84 years) and gender. Using the equal allocation strategy, in each study center, 88 individuals of each gender were randomly selected and allocated to each age stratum, yielding a sample of 704. The total sample size across the eight centers totalled 5,632 individuals.

The parkinsonism-free cohort was identified within the overall ILSA sample using a cross-sectional study carried out in 1992 to 1993, which led to the detection of the prevalent cases of parkinsonism (baseline survey).9,12⇓ The parkinsonism-free cohort was followed for an average of 3 years. All of the cohort individuals were contacted again during the second cross-sectional study carried out in 1995 to 1996.

Case finding strategies.

Cases of parkinsonism were identified using the same two-phase procedure both at baseline and at follow-up. In each study center, the staff consisted of one or two interviewers, one clinical investigator, one neurologist (the neurologist from Impruneta was one of the authors, A. Di C.), and one geriatrician. All of the staff were centrally trained. In phase 1, trained interviewers administered a questionnaire about symptoms of parkinsonism (tremor of head, arms, or legs), about previous diagnoses of parkinsonism (either by a neurologist or by a primary care physician), and about antiparkinsonian drug treatments to each sampled individual. Moreover, a brief neurologic examination (check for elbow tone or resting tremor) was administered by the trained clinical investigator. All individuals who reported at least one symptom of parkinsonism, a diagnosis of parkinsonism, antiparkinsonian drug treatments, or who had at least one positive test result at the brief neurologic examination were considered to have screened positive and were extensively evaluated by the trained neurologists in phase 2. The structured clinical workup in phase 2 included the motor examination of the Unified Parkinson’s Disease Rating Scale,13 a standardized medical history, a brief summary of the natural history of the disease, and a review of previous clinical records (when available). Through phase 2, we confirmed the diagnosis of parkinsonism and defined specific types of parkinsonism. Two years after the baseline survey, the overall study population was followed up by telephone interview to determine the vital status. Information regarding individuals who either had died or were unable to answer the interview was obtained from the general practitioners working in the study communities and from the surviving relatives. Information about presence of parkinsonism, drug prescriptions, other diseases, hospitalization, and institutionalization was gathered from the general practitioner or the relatives. Hospital discharge diagnoses and death certificate diagnoses also were available for each subject who had died. When the time of parkinsonism onset for deceased individuals was unclear, we used the midpoint between the date of screening for the baseline survey and the date of death. Information was gathered with the same procedure for individuals who had died between the telephone interview and the second cross-sectional survey. The medical information available for deceased individuals was reviewed entirely by the adjudication panel (see later).

Incident cases of parkinsonism included all of the individuals who were free of parkinsonism at baseline but developed parkinsonism during the follow-up.

Diagnostic criteria.

Parkinsonism was diagnosed when at least two of four cardinal signs (rest tremor, rigidity, bradykinesia, and impaired postural reflexes) were present in an individual not receiving antiparkinsonian medication, or if one or more cardinal signs, documented by medical history, were improved by antiparkinsonian treatment. PD was defined among those affected by parkinsonism by exclusion of all other possible causes. Parkinsonism associated with other causes included the following: 1) parkinsonism in dementia: the onset of dementia clearly preceded the occurrence of parkinsonism (dementia is another disease investigated in the ILSA sample through a screening and a clinical confirmation, the latter by the same neurologist that ruled the parkinsonism diagnosis; therefore, we could provide reliable dementia onset); 2) parkinsonism with associate features (e.g., multiple system atrophy or progressive supranuclear palsy); 3) drug-induced parkinsonism: use of neuroleptics or other antidopaminergic drugs in the 6 months preceding the onset of symptoms and with no history of parkinsonism; 4) vascular parkinsonism: clear time relationship between a cerebrovascular event and the onset of atypical parkinsonism, preferably supported by neuroimaging, usually without tremor; and 5) parkinsonism unspecified: no clear time relationship with possible causes (e.g., dementia or antidopaminergic drugs, or more than one possible cause). Patients otherwise fulfilling the criteria for PD but who showed no disease progression over 15 years or who were not responsive to antiparkinsonian drugs were classified as unspecified. To increase the diagnostic reliability across the eight study centers and between patients who were living and those who had died, all putative cases were independently reviewed by an adjudication panel of senior neurologists who were kept unaware of the initial diagnosis made in each center. In case of disagreement, the diagnosis by the panel replaced the original diagnosis.

Data analysis.

Excluded from the incident cohort were prevalent patients with parkinsonism at baseline and nonresponders to the follow-up examination. Rates per 100,000 person-years were estimated per 5-year age groups (65 to 84 years) and gender. Age- and gender-specific incidence rates were calculated as the number of new cases divided by the person-years at risk. The 95% CI were based on the Poisson distribution. Person-years for individuals who did not develop parkinsonism were calculated as the time between the screening test for the baseline survey and the screening test for the follow-up survey or death. For subjects who developed parkinsonism, person-years were calculated as the time between the screening test of the baseline survey and the reported parkinsonism onset, or the midpoint between the baseline screening test and the follow-up screening test (when the onset could not be defined clearly) or death.

We considered the different types of parkinsonism as competing causes of parkinsonism.

A forward stepwise Cox proportional hazards model14 was used to estimate the relative risks of parkinsonism associated with age and gender. Age and gender were entered as covariates. Age was entered into the models as a continuous variable (1-year increment). Using the same covariates, another forward stepwise Cox proportional hazards model was used to estimate the relative risk of PD. Individuals who died were treated as censored from their date of death. Individuals who either did not die or did not develop parkinsonism were treated as censored observations from the time of their follow-up screening test.

Our analyses were run on a personal computer using the 1995 release of Statistical Package for the Social Sciences (SPSS) statistical software (SPSS, Inc., Chicago, IL).15

Refusals.

The 882 subjects who refused to participate in the follow-up examination had a median age of 75.2 years (range 65 to 85) and were older than the participants (median age 72.4 years; range 65 to 85). Women refused the follow-up visit more often than men (55% versus 45%). Results of a logistic regression model, where being a refusal was the dependent variable, indicated that the individuals who refused to participate in the follow-up examination were significantly older, more frequently women, and more educated than participants. The odds ratios were 1.04 (95% CI, 1.02 to 1.06) for 1-year age increments, 0.75 (95% CI, 0.65 to 0.89) for male gender, and 1.03 (95% CI, 1.01 to 1.05) for 1-year increments of schooling.

Deceased subjects.

Subjects who had died were older (mean age 77.8 ± 5.2; median age 79.2 years; range 65.3 to 85 years) than survivors (mean age 73.9 ± 5.6; median age 73.3 years; range 65 to 85). Men died more frequently than women (59.2% versus 40.8%). Again, these results were confirmed by a logistic regression model, where being dead was the dependent variable. The odds ratios were 1.13 (95% CI, 1.11 to 1.15) for 1-year age increments, 1.64 (95% CI, 1.36 to 1.98) for male gender, and 0.97 (95% CI, 0.95 to 0.99) for 1-year increments of schooling. All of the 596 death certificates were retrieved. We obtained adequate information from relatives and general practitioners for 221 deceased individuals (37.1%). Among these 221 individuals, we identified eight incident cases of parkinsonism, of whom four had PD. The 221 dead individuals did not differ significantly regarding age and gender distribution from the remaining 375 dead individuals without reliable medical information.

Among both the 2,863 screened individuals and the 221 deceased individuals with reliable information, a total of 68 individuals were identified as incident cases of parkinsonism. Forty-two (62%) individuals met the criteria for PD, 7 (10%) had drug-induced parkinsonism, 8 (12%) had parkinsonism in dementia, and 8 (12%) had vascular parkinsonism. The remaining three cases (5.8%) were classified as parkinsonism, unspecified. PD was the most common type of parkinsonism in both men (67.4%) and women (52%). Drug-induced (16%) parkinsonism and parkinsonism in dementia (16%) were the second most common types in women; by contrast, vascular parkinsonism was more common (16.7%) than drug-induced parkinsonism (6.8%) and parkinsonism in dementia (9.1%) in men.

Table 1 shows the age- and sex-specific average annual incidence rates of parkinsonism (new cases per 100,000 person-years). Rates increase with advancing age both in men and women; however, the incidence in women also was high in the age class 65 to 69 years. Rates for men were higher than for women in every age group except the younger one (65 to 69 years). However, in each age stratum, the 95% CI overlapped, possibly because of small number of cases.

The average annual incidence rate for parkinsonism (per 100,000 person-years) in the population aged 65 to 84 years, adjusted to the 1992 Italian population, was 529.7 (95% CI, 400.5 to 658.9).

The age- and sex-specific incidence rates for PD are reported in table 2. PD rates parallel those of parkinsonism in age and sex patterns: the incidence rates increase with age in both men and women, and men had higher rates than women in every age group. However, men and women had similar rates in the age group 65 to 69 years. In each age stratum, the 95% CI for the incidence rates overlapped, possibly because of the small number of cases. The average annual incidence rate of PD (per 100,000 person-years) in the population aged 65 to 84 years, adjusted to the 1992 Italian population, was 326.3 (95% CI, 224.1 to 427.5).

We used the Cox regression analysis to summarize the pattern of incidence rates seen in the stratified analyses; results are shown in table 3. The risk of becoming affected by parkinsonism was 1.09 (95% CI, 1.05 to 1.15) for every increasing year of age; the risk for developing PD was 1.10 (95% CI, 1.04 to 1.16) for every increasing year of age. Men had a relative risk for parkinsonism of 1.66 (95% CI, 1.02 to 2.7) and a relative risk for PD of 2.13 (95% CI, 1.11 to 4.11).