doktorska disertacija
Petra Klanjšek (Avtor), Majda Pajnkihar (Mentor), Nataša Marčun-Varda (Komentor), Petra Povalej (Komentor)

Povzetek

Izhodišča: Neprepoznana podhranjenost pri hospitaliziranih otrocih in mladostnikih lahko vodi v kronično podhranjenost, otežuje zdravljenje osnovne bolezni ter poslabša klinične izide. Z rutinskim presejanjem tveganja za podhranjenost ob hospitalizaciji se olajša pravočasno prepoznavanje podhranjenosti, z ustreznimi prehranskimi intervencijami se preprečijo trajne posledice podhranjenosti, zmanjšajo se stroški zdravljenja in skrajša se hospitalizacija otrok in mladostnikov. Priporoča se uporaba presejalnega orodja, razvitega v kliničnem okolju za točno določeno populacijo hospitaliziranih otrok in mladostnikov ter kliniko. Namen doktorske disertacije je bil razviti model za ugotavljanje tveganja podhranjenosti pri hospitaliziranih otrocih in mladostnikih z metodami podatkovnega rudarjenja in neinvazivnimi kazalniki. Metode: Izvedli smo presečno opazovalno raziskavo z uporabo zaporednega eksplorativnega načrta mešanih metod na populaciji hospitaliziranih otrok in mladostnikov, starih od 1 meseca do 18 let. V kvalitativnem delu smo podatke zbrali s pregledom, analizo in sintezo literature ter jih analizirali z induktivnim generiranjem kategorij spremenljivk, ki so bile vključene v obrazec z vprašanji. V kvantitativnem delu smo podatke zbrali z obrazcem z vprašanji, zdravnikovo poglobljeno oceno prehranskega tveganja, klasifikacijo prehranskega stanja Svetovne zdravstvene organizacije, antropometričnimi meritvami in anketiranjem staršev otrok oz. mladostnikov. Podatke smo analizirali z uporabo deskriptivne in inferenčne statistike ter inteligentnimi metodami podatkovnega rudarjenja. Rezultati: Od 180 otrok in mladostnikov jih je v učni skupini sodelovalo 142 in v testni 38. Od 277 zbranih spremenljivk, vključenih v zbiranje podatkov, smo v razvoj modelov vključili 30 statistično značilnih, kot so: izguba telesne mase, izguba mišične ali maščobne mase, prehranski vnos, gastrointestinalni simptomi. Razvili smo 3 statistične modele in 10 modelov podatkovnega rudarjenja. Najboljše rezultate testiranja ima model GP (AUC = 1, 95 % IZ 1, 1), med statističnimi pa model Logistična regresija (AUC = 0,977, 95 % IZ 0,922, 1). Ujemanje modela GP s poglobljeno oceno prehranskega tveganja je popolno (κ = 1, 95 % IZ 1, 1). Ujemanje modela Logistična regresija s poglobljeno oceno prehranskega tveganja je prav tako skoraj popolno (κ = 0, 837, 95 % IZ 0,659, 1,014) s Se 93,3 %, Sp 91,3 %, PPV 95,5 % in NPV 87,5 %. Ujemanje s SZO klasifikacijo prehranskega stanja je pri obeh modelih precejšnje (κ = med 0,73 in 0,78). Nobeden od razvitih modelov se ne razlikuje statistično značilno od poglobljene ocene prehranskega tveganja in SZO klasifikacije prehranskega stanja. Model, razvit z inteligentnimi metodami, je v primerjavi s statističnim modelom uspešnejše ugotavljal podhranjenost pri hospitaliziranih otrocih in mladostnikih, prav tako v primerjavi s SZO klasifikacijo prehranskega stanja. Razprava in zaključek: Vseh 13 razvitih modelov presejanja je dokazano veljavnih z visoko napovedno vrednostjo ugotavljanja tveganja za podhranjenost. Priporočamo testiranje modelov na večji populaciji hospitaliziranih otrok in mladostnikov v ostalih pediatričnih zdravstvenih institucijah v Sloveniji. S tem bi modele modificirali, dopolnjevali in prilagodili kliniki, v kateri bi jih uporabljali z namenom zagotavljanja kakovosti celostne zdravstvene obravnave otrok in mladostnikov. Uvajanje rutinskega prehranskega presejanja z razvitimi modeli predstavlja temelj sistematične obravnave klinične poti prehranskega presejanja.

Ključne besede

tveganje za podhranjenost;nedohranjenost;prehransko presejalno orodje;otrok;prehransko stanje;

Podatki

Jezik: Slovenski jezik
Leto izida:
Tipologija: 2.08 - Doktorska disertacija
Organizacija: UM MF - Medicinska fakulteta
Založnik: [P. Klanjšek]
UDK: 616-056.25-053.2:004.6/.9(043.3)
COBISS: 209381891 Povezava se bo odprla v novem oknu
Št. ogledov: 0
Št. prenosov: 2
Ocena: 0 (0 glasov)
Metapodatki: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Ostali podatki

Sekundarni jezik: Angleški jezik
Sekundarni naslov: Development of a model for identifying malnutrition in hospitalized pediatric patients with data mining methods and non-invasive indicators
Sekundarni povzetek: Background: Unrecognized malnutrition in hospitalized children and adolescents can lead to chronic malnutrition, complicate treatment of the underlying disease and worsen clinical outcomes. Routine screening for the risk of malnutrition at admission facilitates the timely identification of malnutrition and appropriate dietary interventions, prevents permanent consequences of malnutrition, reduces treatment costs, and shorten the hospitalization. The application of a screening tool developed in a clinical setting for specific population of hospitalized children and clinic is recommended. Therefore, the aim was to develop a screening model for identifying malnutrition in hospitalized children and adolescents with data mining methods and non-invasive indicators. Methods: A cross-sectional observational study using mixed methods of exploratory sequential design was conducted on a population of hospitalized children and adolescents aged from 1 month till 18 years. Data within the qualitative strand were gathered through the review, analysis, and synthesis of literature, afterward analysed by inductively generating categories of variables that were included in the question form. Data within the quantitative strand were gathered with a question form, a doctor's an in-depth nutritional risk assessment and the World Health Organisation's classification of nutritional status, anthropometric measurements, and survey of parents with children and adolescents. Data were analysed using descriptive and inferential statistics and intelligent data mining methods. Results: A total of 180 recruited hospitalized children and adolescents, 142 participated in the study group and 38 in the test group. Of total 277 collected variables included in the data gathering, 30 statistically significant ones were included in the development of the models, such as: weight loss, loss of muscle or fat mass, dietary intake, gastrointestinal symptoms. There were 3 statistical models and 10 data mining models developed. The GP model has the best test results (AUC = 1, 95 % CI 1, 1) and among the statistical ones, the Logistic Regression model (AUC = 0,977, 95 % CI 0,922, 1). The agreement between GP model and the in-depth nutritional risk assessment was perfect (κ = 1, 95 % CI 1, 1). The agreement between Logistic regression model and the in-depth nutritional risk assessment was also almost perfect (κ = 0.837, 95 % CI 0,659, 1,014) with Se 93,3 %, Sp 91,3 %, PPV 95,5 % and NPV 87,5 %. Both models agreed with the WHO classification of nutritional status substantially (κ = between 0,73 and 0,78). None of the developed models did not differ statistically significantly from the in-depth nutritional risk assessment or from WHO classification of nutritional status. The model developed with intelligent methods compared to the statistical model more successfully determined malnutrition in hospitalized children, equally so compared to the WHO classification of nutritional status. Conclusions: All 13 developed screening models have been shown to be valid, with a high predictive value for determining the risk of malnutrition. We recommend further testing of the models on a larger population of hospitalized children in other pediatric health institutions in Slovenia. This would allow models to be modified, supplemented, and adapted for the clinic in which they would be used to ensure the quality of integrated health care for children. The introduction of routine nutritional screening by the developed models represents the foundation of the systematic approach to the clinical pathway of nutritional screening.
Sekundarne ključne besede: risk of malnutrition;undernutrition;nutritional screening tool;child;adolescent;nutritional status;
Vrsta dela (COBISS): Doktorska disertacija
Komentar na gradivo: Univ. v Mariboru, Medicinska fak.
Strani: XX, 403 str.
ID: 15722999