gms | German Medical Science

1st International Conference of the German Society of Nursing Science

Deutsche Gesellschaft für Pflegewissenschaft e. V.

04.05. - 05.05.2018, Berlin

Risk-adjusted quality indicators for nursing homes using multiple logistic regression

Meeting Abstract

  • presenting/speaker Kathrin Seibert - Institute of Public Health and Nursing Research, University of Bremen
  • Mathias Fünfstück - SOCIUM Research Center on Inequality and Social Policy, University of Bremen
  • Heinz Rothgang - SOCIUM Research Center on Inequality and Social Policy, University of Bremen
  • Stefan Görres - Institute of Public Health and Nursing Research, University of Bremen
  • Martina Hasseler - Ostfalia University of Applied Sciences, Wolfsburg, Germany
  • Sylvia Schmidt - Competence Center for Clinical Trials, University of Bremen, Bremen, Germany
  • Werner Brannath - Competence Center for Clinical Trials, University of Bremen, Bremen, Germany

Deutsche Gesellschaft für Pflegewissenschaft e.V. (DGP). 1st International Conference of the German Society of Nursing Science. Berlin, 04.-05.05.2018. Düsseldorf: German Medical Science GMS Publishing House; 2018. Doc18dgpO43

doi: 10.3205/18dgp043, urn:nbn:de:0183-18dgp0431

Veröffentlicht: 30. April 2018

© 2018 Seibert et al.
Dieser Artikel ist ein Open-Access-Artikel und steht unter den Lizenzbedingungen der Creative Commons Attribution 4.0 License (Namensnennung). Lizenz-Angaben siehe http://creativecommons.org/licenses/by/4.0/.


Gliederung

Text

Background and Purpose: In 2019, a new proceeding to measure and report nursing home care quality will be introduced throughout Germany, using resident-specific quality indicators (QI). To take case-mix characteristics into account, advantages of risk-adjustment (RA) by multiple logistic regression (MLR) have been described in comparison with alternative methods. We compared outcomes for QI adjusted by MLR or stratification.

Methods: In a prospective longitudinal design, data from 3.246 residents in 62 nursing homes were analysed. The stratified RA approach divided residents into two subgroups. The MLR approach first identified significant variables not influenceable by nursing care in a step-wise logistic regression. The selected model was extended by a second step-wise logistic regression with forward selection also considering partly influenceable variables. Prognostic quality of the tested models was assessed using the receiver operating characteristic. Extend of deviation in QI outcomes when using MLR RA compared with RA by stratification was also assessed.

Results: Prognostic quality was higher for all models adjusted by multiple logistic regression. MLR also contributed to changes in QI outcomes in at least 20 % of the observed nursing homes. The MLR approach to risk-adjustment has proven empirically meaningful and superior to the stratified approach.

Conclusions: Risk-adjustment by MLR yielded substantial changes in QI outcomes and highlighted the importance of MLR for a fairer comparison between nursing homes. The studied QI can thus contribute to the reporting of quality of care in German nursing homes, when implementing a statistical risk-adjustment by MLR and developing a suitable rating classification of nursing homes based on QI outcome.

Funding: The study was funded by the National Association of Statutory Health Insurance Funds (GKV-SV).