Hondebrink, L.L.C. (2015) De diagnostische waarde van patiëntkarakteristieken en symptomen in het voorspellen van causale enteropathogenen bij acute diarree in de Nederlandse huisartsenpraktijk : Een Historisch Cohortonderzoek in combinatie met een Meta-Analyse. thesis, Medicine.
![[img]](https://umcg.studenttheses.ub.rug.nl/style/images/fileicons/text.png) |
Text
HondebrinkLLC.pdf Restricted to Registered users only Download (2MB) |
Abstract
Background: Gastroenteritis (GE) is one of the most common illnesses worldwide. In the Netherlands, 4.5 million people are living through an episode each year, whereof 220.000 people are visiting their GP anually. Hitherto, no other studies concerning a systematic view of diagnostic value of patient characteristics and accompanying symptoms regarding additional diagnostics in GE in primary care have been published. The aim of this study was to develop predictive models to select patients for additional diagnostics based on clinical predictors. Methods: Primarely, a meta-analysis has been performed to detect the correlation of clinical predictors and GE in previous studies, to determine candidate-predictors for our study. A historical cohort (database), including 855 patients who visited a GP affiliated to the Julius GP Network (JGPN) in 2013 because of GE and whereby molecular faecal testing is performed via PCR, is used for our diagnostic study. The predictors that appeared significant throughout the meta-analysis were extracted from each patient. Furthermore, based on clinical knowledge some other predictors were added. After completing missing data using multiple imputation, the correlation with causal enteropathogens was determined in SPSS 20.0 by logistic regression, whereafter development of predictive models took place. The models were evaluated by AUC and calibration. Results: 1.938 Articles were screened throughout meta-analysis, whereupon 10 articles (n=6.487) were eligible for extraction of 8 significant candidate-predictors (p<0.05); namely abdominal pain/cramps, blood admixture, mucus admixture, fever, vomiting, headache, myalgia and age. Based on clinical knowledge, nausea, belching/flatulence, comorbidity (malignancy, HIV/aids, chronic intestinal diseases), medication (proton pump inhibitors, immunosuppressants, anti-inflammatory drugs), duration and season were added. Predictive models were developed for viruses (0.747 + (-0.005*age) + (-2.346*malignancy)), bacteria (-0.542 + (-0.281*vomiting) + (0.549*fever) + (-0.326*sex)) and possible pathogenic parasites (-0.582 + (0.003*duration) + (-0.802*syst.corticosteroïds)). None of the variables was significant (p<0.2) for pathogenic parasites. The associated mean AUC-values are: 0.55 (viruses), 0.57 (bacteria) en 0.56 (possible pathogenic parasites). The Hosmer and Lemeshow test showed enough goodness of fit (p>0.2). Discussion: After evaluating through AUC, the prognostic value for all three models appeared insufficiently distinctive for daily practice. Possibly, a reliable model with a greater degree of distinctiveness can be developed through a new dataset with complete outcome measurement.
| Item Type: | Thesis (Thesis) |
|---|---|
| Supervisor name: | Facultair begeleider and Nijholt, Dr. I. junior-affiliatiecoördinator Isala Klinieke |
| Supervisor name: | Broekhuizen, Dr. B. D. L. assistent Schierenberg and Julius Centrum, Utrecht Afdeling huisartsgeneeskunde |
| Faculty: | Medical Sciences |
| Date Deposited: | 25 Jun 2020 11:00 |
| Last Modified: | 25 Jun 2020 11:00 |
| URI: | https://umcg.studenttheses.ub.rug.nl/id/eprint/2030 |
Actions (login required)
 |
View Item |