Administrative data may help registry reliability: Experience of Lombardy Registry


Large administrative health care data sets are an important source of information for health care services about epidemiological measures of many diseases. Many papers recommend a development and expansion of a direct access to administrative databases as a priority in epidemiological researches. (Scott D-2010 [1](Peter S Millard-2012 [2] (full text)) Aim of the present study was to evaluate incidence and prevalence of CKD patients on RRT from administrative datasets in Lombardy region compared with data from RLDT census.


All subjects resident in Lombardy Region (Italy) recorded in the administrative databases with the first dialysis treatment in the period January 1, 2011- December 31, 2011 were selected and observed for the first 12 months after dialysis and for the 24 months before it. Were analyzed data from data set concerning

i) hospital admission and discharge card (SDO) 

ii) diagnostic/therapeutic procedures and outpatient episodes of care, 

iii) drug prescriptions. 

Considering the aim of this analysis, patients with acute kidney injury, and patients who recoverd renal function were excluded from the analysis. As data originated from Lombardy regional administrative databases, the perspective of the present study was retrospective and observational (Figure 1).


Over a population of more than 9,700,000 inhabitants, 1,682 patients incident to dialysis were identified; among them, 365 died during the 12 months following dialysis entrance, 28 received renal transplantation. Of them, in the first 12 month of dialysis, the 82% received only hemodialysis (HD), the 13% only peritoneal dialysis (PD) and the 5% both treatments. The number of these incident patients detected by administrative dataset is quite similar (3% difference) to the 1740  incident patients detected by Registry census among Lombardy dialysis centers. We consider the administrative data more accurate concerning Lombardy resident population, owing to the certain exclusion of patient coming from neighboring regions and patients who recovered from RRT. Of these 65,5% are males and median age is 70,2 yrs. More than 50% of incident patients are older than 70 yrs. (Figure 2) The value incidence in the region was 168 pmp, with some difference among the 15 local health district of the region (range 100 – 224 pmp) (Figure 3).

The higher values are recognized in districts with older general population. Dialysis access documented by hospital admission on ordinary or day-surgery modality, were 2081 (Figura 4), with a 1.2 hospital admission per patient. Prevalent 2011 patients were 8316 (63% males ) (857 pmp) with a median age of 71 yrs and 52% older than 70 yrs. (Figure 5). Prevalence values were normally distributed among local health districts without significant difference. Main cause of hospitalization are shown in (Figure 6) and account for about 30% of overall hospital admission.


The advantage of administrative data is the close link with the registry office with the possibility of the exact identification of each patient starting a RRT program. The reimbursement system of Lombardy region, based on fares for each modality treatment provides a clear picture of patients on dialysis treatment from the beginning to the outcome. So through this first study guidance we could fix precisely the values of incidence and prevalence of patients with CKD on RRT. Further analysis on dataset of following years will confirm the usefulness of this method, and the possibility to link administrative databases with clinical observation from nephrologists.