Epidemiology research in the area of dementia has been carried out on the basis of cross-sectional observational studies and longitudinal cohort studies that have established the prevalence, incidence and mortality rates of the main subtypes of dementia. [1] These studies have also made it possible to identify risk factors and open new research lines. [2, 3] Nevertheless, this methodological approximation, due to its high economic cost and the logistic complexity that the follow-up of participants overlong periods of time represents, provides limited information on the impact of the dementia cases on the healthcare system in terms of needs and utilisation of resources. [4] Moreover, this type of studies does not allow to elucidate what the reality of detection, referral and diagnosis is in routine clinical practice of primary care services and specialised care in a specific territory. However, given the progressive ageing of the population, this information is essential for proper planning and allocation of health and social care services in the territory, in accordance with patient needs at mid and long term. Other epidemiological strategies such as a clinical registry of cases in a delimited geographical area could be a complimentary approach.

The creation of registries for certain diseases is standard procedure in some pathologies such as cancer or cardiovascular diseases but uncommon in the area of dementias certain. A number of initiatives do exist however such as the Consortium to Establish a Registry for Alzheimer's Disease (CERAD) [5] and others. [6–8] Nevertheless, these initiatives, of variable duration, have mainly been geared towards the creation of cooperative networks, the primary objective of which, by means of clinical procedures and standardised diagnoses, is to increase the statistical power of the clinical research. At present, there are only two dementia registries for which references can be found in the specialised literature [9–12], although these registries have non-comparable methodologies and registry scopes.

The objective of this paper is to present descriptive data for 2007 from the Registry of Dementia of Girona (ReDeGi), a clinical registry of dementia cases based on the guidelines of epidemiological surveillance. [13]

Dementia is a disease that fulfils the parameters proposed by the CDC in order to assess the importance and need for an epidemiological surveillance system in terms of public health: frequency and severity, differences and dissimilarities amongst population groups, high cost, potential of primary and secondary prevention, and public interest. [13] The ReDeGi sets out from a conceptual model of registry, which fulfil the CDC guidelines for evaluating a surveillance system and is adapted to the particular needs of dementia. [14] Following a 2-year long test in which the operating structure of a registry system took the lead with satisfactory results [15], 2007 represented the start of activity of the ReDeGi in the entire RSG area which extends for 5,517 km² and includes a population of 690,207 inhabitants. The results obtained are an innovation with regard to the existing dementia registries and can be evaluated from a point of view of methodology or from a practical utility perspective.

On the one hand, from a strictly methodology perspective, the ReDeGi has adopted a registry model different to those used by the South Carolina Alzheimer's Registry and by the New York State Dementias Registry. The first of these is based on the integration of computerised data originating from different sources of information. The information entered includes the diagnosis of dementia (ICD-9 criteria), other medical diagnoses and basic demographic data. The information that enables the elaboration of a registry is circumscribed to the stratification of dementia subtypes according to demographic and administrative characteristics such as gender, race or age, and to the origin of the particular case. Access to death certificates via a link also allows estimating the mortality rates of entered cases from a certain inclusion date in the registry onwards. [11, 12, 30] Even though the New York State Dementias Registry only admits cases of dementia (ICD-9 criteria) from discharge forms from all the hospitals in the state of New York, it does allow to carry out an estimate of the hospitalisations required by patients with dementia and their causes as well as the type of admission and destination at discharge. [9, 10] Unlike the previously mentioned registries, which are based on information from electronic databases, the ReDeGi is characterised by the compilation of data straight from the patient's medical chart by a ReDeGi technician. This method allows the standardised registration of clinical information available in the patient's medical history. Moreover, it is a procedure that permits the maintenance of a dementia subtypes registry in accordance with DSM-IV criteria and other diagnostic criteria specific to the subtypes of dementia that ICD-9 and administrative databases are unable to register. For example, in table 1 cases with the diagnostic category "Dementia caused by non-specified diseases" include a subcategory named "Lewy's bodies" that correspond to those cases that accomplished the diagnostic criteria for dementia with Lewy bodies, and the 23 cases under "dementia caused by Pick's disease" diagnostic category correspond to 18 cases that accomplished the diagnostic criteria for frontotemporal dementia and 5 cases the diagnostic criteria for primary progressive aphasia. In this sense, the data collection procedure implemented at the ReDeGI ensures greater validity of the information recorded in the registry related to the diagnosis.

On the other hand, from a strictly pragmatic point of view, the data obtained from the ReDeGi provide relevant information about the healthcare pressure brought about emerging dementia cases in a certain territory over a defined period of time. The ReDeGi provides evidence on the distribution of the dementia subtypes and the clinical and demographic characteristics of patients who contact specialist healthcare devices.

The analysis of the data shows that AD accounted for 60.1% of the entered cases, mixed dementia (AD + vascular component) for 10.7% and vascular dementia (VaD) for 4.6%. These percentages are similar to those from the South Carolina Alzheimer's Disease Registry, in which the AD cases entered in the period 1988–2003 accounted for 64% of the total and VaD cases accounted for the remaining 14%. In relation to the place of residence of the patients, the differences between the South Carolina Alzheimer's Disease Registry and the ReDeGi can reflect two distinct sociocultural and healthcare models. In the former model, 57% of the cases live in the community and 38% in assisted centres, whereas in the latter 83.1% live in the community (their own home or the home of a family member) and 6.6% live in long-term care centres. Data on the mean time between the onset of symptoms and clinical diagnosis indicate that for most dementias the mean time from onset to diagnosis is approximately 2 years. In this sense, it is hard to establish a precise date of dementia onset, although the estimate based on the data collected by the informant carer has proven to be a reliable source of information. [31]

The diagnostic coverage of the registry has been estimated to be 61.3% of the total expected cases in the territory, with differences being observed amongst the different healthcare sectors. However, there are certain limitations that should be taken into account when estimating the diagnostic coverage on the basis of expected number of dementia cases per year in the catchment area. Formally, in an epidemiological study, the incidence rate would be calculated dividing the number of cases detected in specified period by the free-of-disease inhabitants by time unit. However, the cases entered in the registry correspond to patients who were referred to a specialised diagnostic unit, some of whom presented moderate to severe severity and a mean time of 2 years from onset of symptoms to time of diagnosis. These are new cases of dementia for healthcare devices and consequently the results of the estimated diagnostic coverage are not representative of the real incidence in the general population. These results can however be reliably used to ascertain the incidence of new cases that contact specialised health services. A possible explanation for the number of cases not registered (38.7%) could be hypothetized that they correspond to mild or advanced cases of dementia. It may be that the first group did not attend specialised consultations due to the low impact of the symptoms of dementia in the early stages and the second group did not attend consultations because the benefit of a diagnostic could be more limited. It is interesting to note that we have observed a great variation in the diagnostic coverage across hospitals. These differences might due to different variables and may help to identify factors related to the referral and diagnosis in clinical practice. For example, these differences may be due to differences between types of specialist physicians (neurology or geriatrics) or due to the existence of specific devices for dementia diagnostic. However, it is important to emphasise that, for the ReDeGi, more than achieve a full diagnostic coverage, it is more important to register all cases that have been diagnosed in the 7 RSG hospitals. Thus, the ReDeGi will become a powerful information device in few years when it provides cumulative information about the pressure of dementia cases and its evolutionary course.

The ReDeGi illustrates the application of epidemiologic surveillance to the setting of a chronic condition such as dementia and for which the available information usually comes from such dispersed sources as vital statistics, health surveys, hospital discharges, etc.

The ReDeGi is a valid, representative and updated source of information of great utility for the planning and allocation of the health and social care services needed by patients with dementia and of great special interest to healthcare administrators. The main limitation is that actually, the ReDeGi can't provide dementia incidence rates for its geographical area. However, as the ReDeGi continues registering new dementia diagnosis, in few years it will be possible to identify all cases with a reported onset during each calendar year, and to obtain population rates by age for variables of interest, such as rates of illiteracy or other. Future plans of the ReDeGi encompass an extension of the influence area of primary care devices at short term and an annual follow-up of the cases included in the registry. The proposed annual follow-up of registered cases will allow us to gain further insight into the course of the disease and to ascertain the percentage of cases that continue to live at their homes, the number of cases that are admitted into long-term care centres and the number of cases that die, as well as the clinical and demographic variables associated with such changes. The maintenance and extension of the registry over the next few years will allow us to describe how the pressure on healthcare services evolves and to determine the clinical features and course of the cases diagnosed with the disease. Although actually mild cognitive impairment (MCI) is a controversial diagnostic category [32], a potential application of the ReDeGi in the future may be to register cases with this diagnostic category from the neurology or geriatrics outpatient consultation offices using the same methodology. However, due to the low validity and reliability of the actual diagnostic criteria for MCI, this possibility will be considered in the future. An other potential application of the ReDeGi may be to conduct etiologic case-control studies by selecting comparable control subjects from primary care databases, to determine, for example, the odds of dementia between diabetic and non-diabetic patients. Moreover, this application will contribute to a better interpretation of the clinical characteristics of dementia patients respect the general population.