We conducted a mono-centered cross-sectional study at a tertiary care facility. This study has been carried out in accordance to STROBE as general guideline for observational studies26 and in particular STROSA for studies analyzing secondary data27.

Population

We included all adult (≥ 19 years) cases admitted and discharged between 2014 and 2018 in the University Hospital Carl Gustav Carus, Dresden. We excluded children/ adolescents, cases with prevalent PU, psychiatric treatment and length of stay < 2 days.

Outcomes and covariates

The outcome/dependent variable was case-specific incident PU. To correctly identify prevalent and incident PU, a consistent assessment beginning at admission is essential. Especially in nursing home residents, it is not always clear whether a pressure ulcer was already present on admission. Our in-house standard requires a pressure ulcer assessment for high risk cases (internistic treatment, intensive care and surgery) within 24 h from admission. Every PU detected within this timeframe has been marked as prevalent and excluded from our analysis.

We grouped the independent variables into case- and care-related characteristics.

Case-related characteristics include age, (male) sex and comorbidities. To define comorbidities (based on ICD-10) appropriately, we followed the German inpatient quality assurance program. The German inpatient quality assurance indicator for PU adjusts for Diabetes mellitus type 2, BMI ≥ 40, underweight and/ or malnutrition, dementia and/ or vigilance disturbance, infections, other severe diseases, mobility and incontinence. The ICD-10-based definitions are provided in Supplement S828.

Care-related characteristics include admission reasons (emergency case, transfer from another hospital), (length of) surgical anesthesia, number of wards involved in care and intensive care with or without ventilation.

We did not include the Braden score as predictor in the models since it was used for preventive PU screening in the hospital. This implies that likely cases of PU indicated by the Braden score may have been prevented and do not occur in our data. Accordingly, estimating relationships between observed PUs and the Braden score would induce misleading results. Some literature also adds length of hospital stay in risk-adjusted analyses for pressure ulcer on the one hand29,30. On the other hand, several studies showed that pressure ulcers extend the length of hospital stay31,32,33. This feedback effect causes endogeneity of length of hospital stay as a predictor of pressure ulcer and could seriously bias the results of our risk factor analysis. Therefore, we decided not to consider length of stay as part of the main analysis. However, we included length of stay as a predictor for sensitivity analysis (Supplement S9). In the main analysis, case complexity was captured by a wide set of variables such as comorbidities, anesthesia, reason for admission, intensive care treatment and ventilation.

Data sources

We used four data sources:

  1. I.

    internally standardized and routinely collected PU screening for the detection of incident PU,

  2. II.

    legally (§21 Krankenhausentgeltgesetz) required and prespecified accounting data for age, sex, comorbidities, intensive care treatment, ventilation and admission reasons,

  3. III.

    case-based surgery protocols for length of surgical anesthesia (induction to awakening)

  4. IV.

    case-based ward stays for the number of involved hospital wards per case

Study participation, privacy, and ethics

We analyzed pseudonymized routine datasets in a mono-centered setting. If reasonably justified, the legislation of the federal state of Saxony (§35(1–3) “Sächsisches Krankenhausgesetz”) does not require individual consent for large pseudonymized and mono-centric routine datasets. The legal justification in the federal state of Saxony is based on the principle of in-house research by the specific providers. We have integrated these data privacy relevant conditions and justifications into our study protocol. The Institutional Review Board (IRB00001473 and IORG0001076) of the Medical Faculty of the TU Dresden reviewed and approved the study protocol.

Patient and public involvement

It was not appropriate to involve patients or the public in the design, or conduct, or reporting, or dissemination plans of our research. This is a non-interventional cross-sectional analysis based on observational data, predefined outcomes and covariates.

Statistical methods

Descriptive statistics in case of categorical variables were provided as absolute and relative frequencies. Continuous variables were described by the median and the 1st and 3rd quartile. We used Bayesian Additive Regression Trees (BART) to predict pressure ulcers and estimate predictive relationships between pressure ulcers and risk factors10. Generally, BART is based on regression trees, which may be used when associations between independent and dependent variables cannot be described linearly. The advantage of regression trees over, e.g., logistic regression is the ability to handle non-logistic associations and interactions. Regression trees build homogeneous groups to identify relationships between the outcome and covariates. At a certain degree of heterogeneity in the groups, the groups are separated to achieve higher homogeneity (splitting). BART combines multiple trees in a “sum-of-trees” model, which facilitates more accurate and stable out-of-sample predictions than single regression trees. This ability led us to prospectively predict incidences of PU in addition to associations between dependent and independent variables. In this regard, it is noteworthy that a high/low predictive power of a model does not necessarily imply accurate/inaccurate estimation of relationships between outcome and covariates34.

We used data from 2014 to 2017 to fit the BART model. The number of trees (50, 75, 100) served as tuning parameter in tenfold cross validation. We assessed the predictive performance of the selected model based on a confusion matrix and area under the curve (AUC) using data from 2018. An AUC of 0.5 suggests no discrimination (i.e., ability to predict cases with and without incident PU), 0.7 to 0.8 is considered acceptable, 0.8 to 0.9 is considered excellent, and more than 0.9 is considered outstanding35. In addition to confusion matrices, we analysed performance indicators sensitivity, specificity, positive predictive value, negative predictive value, precision, recall, F1, prevalence, detection rate, detection prevalence, balanced accuracy (in case of high class imbalance) and accuracy. Subgroup analyses were performed for the full dataset, intensive care (yes/no) anesthesia (yes/no), ventilation (yes/no) and the different grades of PU. To assess the predictive performance of specific risk factors, we calculated variable importance as the proportion of times each risk factor was chosen for a splitting rule, i.e. to define a node in the sum-of-trees model. We calculated partial dependences to explore the influence of risk factors (e.g. age) on the predicted probability of pressure ulcers. We used 95%-credible intervals to assess the precision of partial dependence estimates. Statistical analysis was conducted using R 3.6.3 and the package bartMachine36. With respect to methodological rigor, the accuracy of BART predictions was compared with those based on multiple logistic regression, random forest, and LASSO (see Supplement S10 for a more detailed description).

By admin

https://muehlenbar.de/wp-includes/slot-online/

https://the-chef.co/sbobet/

Slot Gacor Terbaru

Link Slot Gacor

Sbobet

Situs Slot Gacor

Slot Gacor 2022

Login Sbobet

Daftar Sbobet

https://www.dcosmeticclinics.com.au/wp-includes/sbobet/

https://thetastesoflife.com/wp-includes/sbobet/

https://www.townshipofsugargrove.com/wp-includes/slot-gacor/

https://texasmamaboutique.com/wp-includes/slot-gacor/

https://bizu-me.com/wp-includes/slot-gacor/

https://tiketa.co.za/wp-includes/slot-gacor/

situs slot gacor

slot gacor gampang menang

slot gacor terbaru

daftar sbobet

slot bonus newmember

link sbobet

sbobet login