[Quality assurance in coding expertise of hospital cases in the German DRG system. Evaluation of inter-rater reliability in MDK expertise]. / Qualitätssicherung in der Kodierungsbegutachtung von Krankenhausfällen im G-DRG-System. Prüfung der Interrater-Reliabilität bei MDK-Gutachtern.

AIM: The purpose of this study was to evaluate differences in the D-DRG results of a hospital case by 2 independently coding MKD raters. Calculation of the 2-inter-rater reliability was performed by examination of the coding of individual hospital cases. The reasons for the non-agreement of the expert evaluations and suggestions to improve the process are discussed. METHODS: From the expert evaluation pool of the MDK-WL a random sample of 0.7% of the 57,375 expertises was taken. Distribution equality with the basic total was tested by the χ² test or, respectively, Fisher's exact test. For the total of 402 individual hospital cases, the G-DRG case sums of 2 experts of the MDK were determined independently and the results checked for each individual case for agreement or non-agreement. The corresponding confidence intervals with standard errors were analysed to test if certain major diagnosis categories (MDC) were statistically significantly more affected by differing expertise results than others. RESULTS: In 280 of the total 402 tested hospital cases, the 2 MDK raters independently reached the same G-DRG results; in 122 cases the G-DRG case sums determined by the 2 raters differed (agreement 70%; CI 65.2-74.1). Different DRG results between the 2 experts occurred regularly in the entire MDC spectrum. No MDC chapter in which significant differences between the 2 raters arose could be identified. CONCLUSION: The results of our study demonstrate an almost 70% agreement in the evaluation of hospital cost accounts by 2 independently operating MDK. This result leaves room for improvement. Optimisation potentials can be recognised on the basis of the results. Potential for improvement was established in combination with regular further training and the expansion of binding internal code recommendations as well as exchange of code-relevant information among experts in internal forums. The presented model is in principle suitable for cross-border examinations within the MDK system with the advantage that further trends could be uncovered by more variety and larger numbers of the randomly selected cases.

[Strategic controlling in the consideration of social demographic changes in social medical services]. / Strategisches Controlling zur Berücksichtigung des soziodemographischen Wandels in einem sozialmedizinischen Dienst.

Annually the Medizinische Dienst der Krankenversicherung Westfalen-Lippe (MDK-WL) performs approximately 120,000 nursing care assessments according to the 11th social statute book (SGB XI). A prospective reflection on the amount of expert assessments and the spectrum of services for the Nursing Care Insurance until 2020, was performed in order to establish a long term strategic controlling. The insured party makes the request for nursing care according to their personal estimation when the need for assistance is increasing. To predict the future amount of expert assessments you have to take into consideration the social background of the families in addition to age and gender (there is a clear correlation between age and the need of assistance). The database from nursing care assessments in 2001 was projected on a demographic model for the region of Westfalen-Lippe. The amount of requests correlates as expected with the patients age and increases exponentially. The incidence in the need of care shows relevant gender differences, but taken into consideration the very different age structure, the overall incidence is very similar. Against the background of the current nursing care law, the MDK-WL has to deal with an annual increase in assessments of 2 %. The requests for single persons are extremely often without foundation (55 % not substantially in need of care vs. 35 %). Looking at the requests of couples, it shows that the ones for women are more often unfounded then the ones for men (39 % vs. 32 %). It is necessary to take the development of the amount of single living persons into consideration to achieve more accurate predictions for the amount of assessments.