Page 71 - S. Ličen, I. Karnjuš, & M. Prosen (Eds.). (2019). Women, migrations and health: Ensuring transcultural healthcare. Koper, University of Primorska Press.
P. 71
Working on Epidemiological Data: Problems Encountered and Lessons Learned
From Hospitalizations to the Patients
Since patients may repeatedly resort to hospital care, to estimate the num-
ber and type of user it is necessary to pass from data on admissions to a
database in which the unit of analysis is the patient. After recording in the
new database the number of admissions, it should be decided which other
information contained in the ‘demission’ database should be associated with
each patient. The procedure does not present any problems for the charac-
teristics who remain stable over time, which are however few because only
the place and date of birth certainly remain unchanged.
In case of multiple admissions, in addition to age, citizenship, residence,
marital status, department of demission, diagnoses, interventions, etc. may
also change. It is therefore necessary to decide whether to register, for exam-
ple, citizenship at first or last admission, but it may be advisable to register
both because, by comparing the two situations, one can obtain the useful
additional information of a possible change of citizenship.
Similarly, you can record the year of the first, last or all admissions but,
working in a longitudinal perspective, the choice between producing tables
in which patients are ‘anchored’ to the first, or last, year is not neutral. In the
first instance, there will be an artificial decline in patients over time: for ex-
ample, a patient ‘assigned’ to 2006 (first admission), will not appear in the
following years. On the contrary, assigning patients to the year of the last
hospitalization, going back in time we will be confronted with an artificial
rarefaction of patients.
In both cases, distortions do not allow an effective longitudinal analysis.
Therefore, it seems reasonable to propose a different solution: patients are
counted only once in case of multiple admissions in the same year and rein-
serted it in the database if admitted in a different year. Moreover, while ag-
gregating the hospitalizations carried out in any wider period needs some
kind of ‘justification’ (could be longer or shorter), year by year solution is sup-
ported by the Ministry of Health, using it for the Annual Report on Hospital Ad-
mission (Ministero della Salute, 2017) and hospital facilities too yearly allocate
resources, report activities, etc.
Even this solution, however, presents a criticality, because it does not put
all patients on the same level. In fact, about patients admitted at the begin-
ning of the year we know the subsequent ‘hospital history,’ but any previous
admission is not detected; about patients admitted at the end of the year we
know if they have already been hospitalized, but we cannot know what will
happen in the future. In practice, if you want to study the hospital history
69
From Hospitalizations to the Patients
Since patients may repeatedly resort to hospital care, to estimate the num-
ber and type of user it is necessary to pass from data on admissions to a
database in which the unit of analysis is the patient. After recording in the
new database the number of admissions, it should be decided which other
information contained in the ‘demission’ database should be associated with
each patient. The procedure does not present any problems for the charac-
teristics who remain stable over time, which are however few because only
the place and date of birth certainly remain unchanged.
In case of multiple admissions, in addition to age, citizenship, residence,
marital status, department of demission, diagnoses, interventions, etc. may
also change. It is therefore necessary to decide whether to register, for exam-
ple, citizenship at first or last admission, but it may be advisable to register
both because, by comparing the two situations, one can obtain the useful
additional information of a possible change of citizenship.
Similarly, you can record the year of the first, last or all admissions but,
working in a longitudinal perspective, the choice between producing tables
in which patients are ‘anchored’ to the first, or last, year is not neutral. In the
first instance, there will be an artificial decline in patients over time: for ex-
ample, a patient ‘assigned’ to 2006 (first admission), will not appear in the
following years. On the contrary, assigning patients to the year of the last
hospitalization, going back in time we will be confronted with an artificial
rarefaction of patients.
In both cases, distortions do not allow an effective longitudinal analysis.
Therefore, it seems reasonable to propose a different solution: patients are
counted only once in case of multiple admissions in the same year and rein-
serted it in the database if admitted in a different year. Moreover, while ag-
gregating the hospitalizations carried out in any wider period needs some
kind of ‘justification’ (could be longer or shorter), year by year solution is sup-
ported by the Ministry of Health, using it for the Annual Report on Hospital Ad-
mission (Ministero della Salute, 2017) and hospital facilities too yearly allocate
resources, report activities, etc.
Even this solution, however, presents a criticality, because it does not put
all patients on the same level. In fact, about patients admitted at the begin-
ning of the year we know the subsequent ‘hospital history,’ but any previous
admission is not detected; about patients admitted at the end of the year we
know if they have already been hospitalized, but we cannot know what will
happen in the future. In practice, if you want to study the hospital history
69
