IT-structures and Algorithms for Quality Assurance in the Health

Insurance Medical Advisory Service Institutions in Germany

Vera Ries

, Klaus-Peter Thiele

, Martin Schuster

and Reinhard Schuster

Project Coordinator and Assistant to the Medical Director of the Medical Advisory Service

Institution of the Statutory Health Insurance in North Rhine (MDK Nordrhein), 40212 D

usseldorf, Germany

Medical Director of the Medical Advisory Service Institution of the Statutory Health Insurance in North Rhine

(MDK Nordrhein), 40212 D

usseldorf, Germany

Faculty of Epidemiology, Christian-Albrechts University Kiel, 24105 Kiel, Germany

Chair of Department of Health Economics, Epidemiology and Medical Informatics,

Medical Advisory Service Institution of the Statutory Health Insurance in Northern Germany (MDK Nord),

23554 L

ubeck, Germany

Keywords:

Quality Assurance, Statutory Health Insurance, Medical Advisory Service Institution, Communication

Structures between Different IT-systems, Server Data Structures, Data Protection, Script Programming, Client

Ofﬁce Answers using Perl Modules, Integer Linear Programming.

Abstract:

The 15 Regional Medical Advisory Service Institutions of the Statutory Health Insurance in Germany (MDK)

create medical expertises (sozialmedizinische Gutachten) on behalf of the Statutory Health Insurance Funds

in the ﬁelds of inpatient and outpatient treatment, incapacity of work and other ﬁelds. The process of internal

quality assurance within the local advisory service institutions as well as between them plays an increasing

role and got new impulses by actual national legislation. The assessment process was established in 2004 and

covers only one single indication: long-term care. It consists in a paper-based procedure focusing on a manual

distribution process performed by staff of the central quality assurance bureau. We will present organizational

concepts of the new standardized regional and nationwide peer review process that will cover the multitude

of all existing indications of health care. It is completely digitalized using mathematical IT-based procedures

not only for randomized sampling, but as well to achieve an equal distribution of the medical expertises to be

assessed by the peer Medical Advisory Service Institutions. These peer reviews are supposed to be distributed

among the institutions depending on occasion groups and further subtypes of medical expertises, posing a

constraint satisfaction problem that needs to be solved. Therefore we discuss models that address this kind

of problem type and present possible solutions for the concrete distribution problem mentioned above. We

further present our technical framework that will support the workﬂow needed for peer review distribution,

data collection and ﬁnal result analysis. The 15 regional medical service institutions with different IT-system

have to be connected, while data protection concerns have to be taken into account. Finally, the statistical

distribution of the review results is analyzed.

1 INTRODUCTION

In Germany, 15 Regional Medical Advisory Ser-

vice Institutions are mandated by the German So-

cial Code to assist the Statutory Health Insurance

(Krankenversicherung) as well as the Statutory Long-

term Care Insurance (Pﬂegeversicherung). Overall,

nearly 10.000 employees are working within the Med-

ical Advisory Service Institutions, providing medical

knowledge by case management consultancy and by

medical expertise recommendation services in several

ﬁelds of long-term care and of healthcare.

1.1 Preexisting Quality Assurance Plan

For long-term care there was a preexisting qual-

ity assurance plan for medical expertises. About

3.500 specialized nurses are employed by the Medi-

cal Advisory Service Institutions in those both assess-

ment ﬁelds about long-term care. A national qual-

ity assurance plan was mandated by law since 2004

(Schmacke, 2016). It consists of the exchange of

paper printed medical expertises that are distributed

manually by staff of the central quality assurance bu-

reau. For two years now, the assessment forms are no

Ries, V., Thiele, K., Schuster, M. and Schuster, R.

IT-structures and Algorithms for Quality Assurance in the Health Insurance Medical Advisory Service Institutions in Germany.

DOI: 10.5220/0008912303530360

In Proceedings of the 13th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2020) - Volume 5: HEALTHINF, pages 353-360

ISBN: 978-989-758-398-8; ISSN: 2184-4305

353

longer paper-based assessment sheets but can be ﬁlled

out using a web-based data platform as well.

1.2 Starting a New Quality Assurance

Plan in 2020

In 2020 a new quality assurance plan for medical ex-

pertises covering the multitude of all health care ser-

vice ﬁelds provided by the Statutory Health Insurance

Funds in 2020 will be started. For health care, the top-

ics covered by the Medical Advisory Service Institu-

tions’ consultancy and expertise services are diverse.

About 2.500 specialized physicians, 90 other health

care professionals and 400 medical coders are work-

ing in the health care division of the Medical Advi-

sory Service Institutions. The main tops are

• hospital quality and billing control on behalf of

the health care insurance funds (Thiele K.-P.,

2018),

• out-patient case management consultancy and

medical expert opinion service (N

uchtern, 2008)

in multiple ﬁelds of medical treatment.

• To learn more about the German social system of

health care in detail, see (Altenstetter and Busse,

2005), (Cortina, 1993), (Machnik, 2009), (Saw-

icki and Bastian, 2008).

So, there are many different health care topics cov-

ered by the Medical Advisory Service Institutions’

consultancy and medical expertises. For a long time,

however, there was no monolithic quality assurance

plan established for the health care segment of con-

sultancy and medical expertises. The Regional Med-

ical Service Institutions performed a kaleidoscope of

singular quality assurance measures lacking a nation-

wide perspective and any public reporting, instead. In

2017, a nationwide task force was established by the

medical directors’ conference of all regional Medical

Advisory Service Institutions to design, to initiate and

to implement a total quality assurance plan.

The preexisting regional quality assurance initia-

tives were used as a draft to establish a uniﬁed contin-

uous quality circle (Deming, 1982), (Shewhart, 1931)

suitable to the diversity of medical topics, perform-

ing annual quality measurements and validations. The

total quality assurance organizational framework es-

tablished addresses both, the internal quality assur-

ance of every Regional Medical Advisory Service In-

stitution as well as a nationwide quality assurance.

The quality assurance method is based on peer re-

views, a common method in statutory social insur-

ance frameworks in Germany (DRV, 2018), (Polak,

2018), (Strahl et al., 2016), (Strahl et al., 2018). By

facilitating mutual learning, thereby enhancing con-

tinuous quality improvement, both regionally and na-

tionwide. Crucial to this nationwide initiative was a

commitment by all Regional Medical Advisory Ser-

vice Institutions of a joint and uniform conception of

quality. A prime nationwide reporting of the qual-

ity assurance results of all Medical Advisory Service

Institutions starting in 2020 will contribute to a high

degree of transparency between each other and fosters

mutual learning. What is more, it inspires reliability

and trustworthiness towards their clients, the Statu-

tory Health Insurance Funds, and towards society.

All nationally and internationally available infor-

mation on quality assurance in general and especially

in health systems and health and pension insurance

have been adapted to the newly established system

with the conceptional time ending 2019 and the role

out starting in 2020.

2 MATERIAL AND METHODS,

PROCESS ORGANIZATION

The Medical Advisory Service Institutions yearly

produce about 2.5 million medical expertises about

inpatient hospital treatments and documentations for

on behalf of the statutory health insurance funds. A

random sample of 0.5% will be used for regional qual-

ity assurance using peer reviews. 10 % of those med-

ical expertises randomly chosen get a second exter-

nal peer review by another advisory service institu-

tion generating the nationwide perspective. Thereby

we get 12,500 peer reviews for regional quality assur-

ance and 1,250 nationwide peer reviews.

Additionally there are about 650 thousand medi-

cal expertises in other ﬁelds classiﬁes in eight occa-

sion groups (inability of work, outpatient care, new

and unconventional treatment methods and drug treat-

ment, prevention and rehabilitation, remedies, den-

tistry, factual or putative medical treatment, claims

against third, other) with further subgroups. In anal-

ogy to industrial quality standards for production,

quality is deﬁned as a measure that describes the de-

gree of correspondence between the service provided

by an deliverer and the service expected by the cus-

tomer (Masaaki, 1986), (Gerlach, 2001), (Kamiske

and Brauer, 2011), (Institute of Medicine; Commit-

tee to Design a Strategy for Quality Review and As-

surance in Medicare, 1990), (Institute of Medicine,

2001), (Internationale Organisation f

ur Normung,

2015). The concept of nationwide and regional qual-

ity assessment is worked out and will be further

developed by the permanent commission of senior

physicians and quality managers. The questionnaire

HEALTHINF 2020 - 13th International Conference on Health Informatics

354

was developed by a task force based on an informa-

tion retrieval about preexisting regional quality assur-

ance concepts for the common set of quality crite-

ria followed by a modiﬁed Delphi method to shape

the different sets of indication-speciﬁc quality criteria

(Normand et al., 1998). Decisions are made by the

conference of medical directors conference (Leitende

Arzte) and the conference of the managing directors

(Gesch

aftsf

uhrer).

In Germany the regions differ considerably in size

and population density. So, questions of equity are a

strong point in any discussion. Therefore the reviews

start with a minimal size setting feasible for any par-

ticipant.

The peer review process is organized quarterly. If

at least one medical expertise is transferred to every

other local advisory service institution, 14 expertises

have to be included per quarter or 56 per year. If we

demand, that at least the double size have to be in-

cluded for local quality assurance, we come to ask

for 112 expertises in each ﬁeld. This leads to 11,660

medical expertises in total on the regional level and

5,300 medical expertises on the nationwide level in

peer review evaluations for all smaller occasion ﬁelds.

This number of medical expertises leads to a personal

and ﬁnancial effort which should not be exceeded.

For reasons of practicability, we reduce the number

of medical expertises given to an external review pro-

cess for the smaller ﬁve medical service institutions

to a half (7 instead of 14 medical expertises for the 15

existing regional medical service institutions).

The internal number of peer reviews for hospital

expertises nearly equals the total number of all other

occasion groups, thereby is no need to enlarge the

number of hospital peer reviews.

The peer reviews for medical expertises are based

on pdf ﬁles and related meta data. Additional med-

ical treatment information is not used by reasons of

practicability (for instance large hospital treatment

records). The use of additional medical data in re-

gional quality assurance would create a bias towards

the nationwide quality assurance because the addi-

tional data like clinical case logs are unable to be

transfered for data protection and practicability rea-

sons. What is more, the medical expertises aims to be

perspicuous and understood without additional medi-

cal documentation data available.

Meta data contain the type (occasion) and date of

release of medical expertise, subregion of the local

advisory service institution and possibly additional

process parameters. Pdf ﬁles and meta data are re-

lated by a regional unique identity key.

With respect to m=9 occasion types (starting with

m=2) and further subtypes of medical expertises

the pear review process yields an n-dimensional (n

around 30 in different groups) quality vector with

components of k outcome types (actually used: ”cor-

rect”, ”improvement potential with no result rele-

vance”, ”incorrect” and ”not relevant”), cf. Figure

2. Only a minority of questions depend on the oc-

casion type. The questions for the peer reviews have

been tested with positive results in a reliability anal-

ysis which will be repeated periodically, cf. (B

uhn

et al., 2017), (Wirtz and Caspar, 2002), (Eisinga et al.,

2012).

ServerMDK x MDK 1

MDK 2

consensus

down

Figure 1: Communication between Medical Advisory Ser-

vice Institutions (MDK) and the QA-server.

The central IT-system is a linux sever (Debian)

which is located in an external data center which

we will call QA-server (quality assurance). It uses

MySQL as database system, perl as the main script

language including the perl module writeexcel in or-

der to generate excel ﬁles to give overviews about

data and aggregate informations for medical review

service institutions. The communication uses a role

concept for administrative persons, medical experts,

peer reviewers and members of consensus confer-

ences. The server management uses the ssh protocol

with key pairs and a passphrase. The communication

between the server and the regional advisory service

institution use the https protocol. Every user needs

a personal password. All transfer actions are logged.

Data backups are made daily.

Every communication step is a pair of upload and

download exchange streams between advisory service

institutions with individual IT-structure and the QA-

server or between other participants as the consensus

conference and the server. It is a star topology with

the QA-server in the center, cf. Figure 1.

In order to start the quarterly local quality assur-

ance process, a matrix which contains all local medi-

cal expertise identities (pdf ﬁle names) with the men-

IT-structures and Algorithms for Quality Assurance in the Health Insurance Medical Advisory Service Institutions in Germany

355

tioned meta data concerning a period of a month or

quarter has to be uploaded. The respective download

answer gives the random sample for regional qual-

ity assurance. A second random sample for external

quality assurance is determined, but not transmitted to

the local advisory service institution. As a next step

the pdf ﬁles related to the sample have to be uploaded

to the QA-sever as single pdf ﬁles or as zipped ﬁles in

individual parts or totally. The server answer consists

in a ﬁrst ﬁle structure analysis. The regional identity

is converted into an external unique ﬁle identity, the

transformation is stored in a MySQL table.

As a next step one or several peer review man-

agers distribute the medical expertise ﬁles e

with

meta data m(e

) to peer review experts r

as a func-

tion f : e

→ r

. This may be done using the detailed

occasion reasons or regional features due to local con-

ﬁgurations. It also can be done completely automatic.

The server provides a radio buttom formula and the

answer arranges the combination of review experts

and review documents.

In order to carry out the review, the review expert

ﬁrst has to download the pdf ﬁle of the related medical

expertise. After that there is the possibility to use a

QA-server form online or to use a pdf form for ofﬂine

review. In both cases the server will give the result

of a formal answer check (all questions are answered

and for all incorrect points reasons are stated).

The result of the peer review is transfered to the

author of the medical expertise for feedback. The

medical expert gives notice if she/he agrees or dis-

agrees with the peer review. This happens primarily

anonymously. However, the medical expert may ask

for a personal contact to the peer reviewer. Further

quality management actions are up to regional advi-

sory service institution decisions.

Alternatively, the regional advisory service insti-

tution may carry out the regional peer review proce-

dure in his own IT structure. In this case, the service

institution quality manager has to upload the results

in csv or xls-format again with a related answer check

procedure.

review

pdf

#01 #02 #03 #04 #05 #06 #07 #08 #09

1 3 3 1 1 3 3 2 2 3

2 1 1 1 1 2 1 1 1 1

3 1 1 1 1 3 1 3 1 1

4 1 1 1 1 3 3 3 1 1

5 1 1 1 1 1 1 1 1 2

6 1 1 3 2 3 3 3 1 1

7 1 1 1 2 3 1 3 2 1

8 1 1 1 1 2 3 2 3 1

9 1 2 3 1 3 2 2 2 1

10 1 1 1 1 3 3 3 1 1

11 1 1 1 1 3 1 1 1 1

12 1 1 2 2 2 1 3 1 1

review question

Figure 2: Upload regional review results with meaning

green: correct, yellow: improvement potential with no re-

sult relevance, red: incorrect.

Using the upload, a random sample is chosen for

the nationwide peer review. The related identities are

given as the server answer in order to select the related

pdf ﬁle for an upload. All identities are transformed

to QA-server identities. One regional extension of the

quality assurance is consisting in the demand, that ev-

ery physician who is included in the creation of med-

ical expertises shall be included in the peer review

process at least two times a year in each relevant oc-

casion group. If this is not reached by the random

process, additional expertises have to be included in

the regional peer review process. If this enlargement

is done in the last quarter, the number of added ex-

pertises remain as low as possible, but it results in an

unequal workload. The expertises of the enlargement

shall not be included in the random selection for exter-

nal peer reviews, because of the fact that it would in-

crease the possibility of physicians with a small num-

ber of expertises to be in the external peer review pro-

cess.

The distribution of the regional medical expertises

to external medical service institutions shall satisfy

the following conditions:

• Every advisory service institution shall do as

much peer reviews as it has given medical exper-

tises to the nationwide quality assurance using the

case numbers stated above

• Every other advisory service institution shall be

included in order to get large range of opinions

• Boards with more medical expertises shall get

more peer reviews (with possibly ”smaller” vio-

lations of this rule)

• The service institutions with the largest number of

medical expertises shall get as few peer reviews as

possible in order to avoid concentration of special

opinions in one spot

We will discuss the ILP based method to solve this

problem later on. Only for the hospital treatment the

distribution differs essentially from an equal distribu-

tion.

Before the medical expertises are forwarded to a

the external peer reviewers is realized, a further ran-

dom process is performed. The distribution of the

medical expertises to the peer reviewers is done by

an analogous procedure described for internal peer re-

viewers. Every peer reviewer may propose medical

expertises for a discussion in the consensus confer-

ence. For each occasion group there shall be two con-

sensus conferences each year. The external peer re-

viewer does not know the results of the internal peer

review. If the regional valuation differs from the na-

tionwide valuation, the internal reviewer is asked, if

he accepts the result of the external peer review. If the

HEALTHINF 2020 - 13th International Conference on Health Informatics

356

difference persists, the expertise has to be discussed

in the consensus conference. In order to organize this

comparison, every internal peer reviewer has to con-

sider differing results to his own peer review results

after peer reviews are completed. The comparison is

generated automatically by the QA-server.

Every participant of the consensus conference will

be provided with detailed information about all ex-

pertises proposed to the consensus conference of the

respective occasion group. Every advisory service

institution sends out a member into this conference.

The conference members only get labels of the peer

reviews done by members of there home advisory

service institution in order to remain information as

anonymous as possible at the one hand and to get

additional information about reasons of review deci-

sions as deep as possible at the other hand. If the con-

sensus conference does not reach an agreement with

respect to a special expertise, the simple majority will

decide about the ﬁnal result. The responsible manager

of the consensus conference has to upload the ﬁnal de-

cisions to the QA-server, this can be done during the

conference.

The amount of uploaded data in relation to the

stated case numbers and the fraction of realized peer

reviews may be requested by the quality managers

with the related roles (regional or nationwide level).

Additionally frequency results of all parts of the re-

view process can be downloaded at every time. The

results are given as ﬁles in xls or xlsx format with

choosable evaluation contents. Every advisory ser-

vice institution can download all details about the ex-

pertises of his home service institution including the

external peer review results, but there are no labels

which external service institution gave the nationwide

evaluation. The overall amount of analysis is done via

background processing.

Aggregated results of the regional and the nation-

wide peer review process will be published. In or-

der to get information about the security of the results

conﬁdence levels are calculated. Tables and ﬁgures of

the results in the decided amount are produced by the

QA-server.

3 RESULTS, MATHEMATICAL

MODELS

A ﬁrst pretest in the ﬁeld of incapacity of work and

a second pretest outpatient care with the subgroups

DRG (diagnoses related groups) and PEPP (pay

groups in psychiatry and psychosomatic medicine)

has been used to reduce the variance and to increase

the kappa values between the reviewers with respect

to the questions and with respect to the medical opin-

ions. After improvement and clarifying of the ques-

tionnaire we got an intraclass correlation coefﬁcients

(average measures) of 0.923, 0.903 and 0.868 in the

mentioned ﬁelds and thereby a good reliability was

reached, cf. (Cortina, 1993), (Eisinga et al., 2012),

(Wirtz and Caspar, 2002). Let w(i, k) be the result

with value 1 (resp. 2 and 3) for correct (resp. im-

provement potential with no result relevance and in-

correct) medical opinions of the expert from the eval-

uation of the reviewer for medical opinion i and ques-

tion k. Then

d(x, y) =

∑

|w(i, x) − w(i, y)|

is the Manhattan distance of the opinion vector with

respect to the questions x and y. For each question

we look for the other questions with the smallest two

distances. We use the questions as the vertices of a

graph and the edges we get by the shortest distances.

Using the modularity method we get graph communi-

ties. For the calculations we have used Mathematica

by Wolfram Research. An example for the pretest is

shown in Figure 3. The results after a year of qual-

ity assurance will give further hints for improvement

of questions and further training opportunities for the

medical experts.

Figure 3: Graph Communities of questions Selection.

Let n(i, j) be the number of medical expertises

which are transfered from medical service institution

i the medical expert to the medical service institution

j of the reviewer and w(i) the number of expertises

from service institutions i given in the external review

process. The order of medical service institutions are

chosen such that w(i) ≥ w( j) for i > j. No expertise

remains in the same medical service institutions, so

the the matrix n(i, j) has an empty diagonal. If one

would use the proportional approach

n(i, j) = int(w(i)

w( j)

∑

k6= j

w(k)

)

IT-structures and Algorithms for Quality Assurance in the Health Insurance Medical Advisory Service Institutions in Germany

357

one would not get a balance between ingoing and out-

going expertises (rounding problems are only a minor

effect). One could implement a random process with

certain rules, but it would result in under- and over-

representation of certain medical service institutions

in most quarters.

We will look for solutions of equations and in-

equalities with integer solutions with a certain optimal

solution. As a balance demand and the demand the no

expertise stay in the same medical service institutions

we state

∑

n(i, j) = w(i)

∑

n(i, j) = w( j)

∑

n(i, i) = 0.

If we require the symmetry of the matrix n(i, j) it will

reduce the number of variables which may reduce the

machine time, but it may lead to insolvability of the

problem. Let m(i, j) the matrix we get from n(i, j

by deleting the main diagonal (delete in columns).

We could demand monotony in the columns and rows

without the main diagonal:

m(i, j) ≥ m(k, j) for i > j

m(i, j) ≥ m(i, k) for i > j

Already with w(l) < l (l number of medical service

institutions) the problem is unsolvable. So we will

take a weaker demand:

m(i, j) + d ≥ m(k, j) for i > j

m(i, j) + d ≥ m(i, k) for i > j

with d as small as possible for solvability.

The reviews should concentrate as little as possi-

ble to large service institutions and be as uniformly as

possible between the medical service institutions. To-

gether with the weak monotony demand we will use

n(1, 2) + n(2, 1) −→ Min!

as an optimization condition.

In the case w(i) = 14 for 1 ≤ i ≤ 10 (large medi-

cal service institutions) and w(i) = 7 for 11 ≤ i ≤ 15

(small medical service institutions) as an example for

two equal groups which will be applied for all occa-

sion groups except of hospital care a balance between

and within the groups can be reached by additional

constraints as

∑

i=11

n(i, j) = 2 for j = 11, ..., 15

∑

j=11

n(i, j) = 2 for i = 11, ...,15.

Thereby we have got an integer linear program (ILP).

This can by solved (if there is a solution at all) e.g. a

Python program and using CPLEX bindings. We also

could ﬁrst solve a linear program without the restric-

tion to integers and then do a post processing step to

obtain integer solutions by local changes. This can

be done with Mathematica by Wolfram Research or

also using Python and CPLEX, cf. (Shinano et al.,

2003). For ﬁnding a suitable distribution matrix for

distributing the reviews over the medical advisory we

use an integer linear programming approach. Inte-

ger linear programs (ILP) allow to formulate an op-

timization problem with an linear objective function

and constraints given as a series of linear inequalities.

Then, an integer valued assignment of the variables

is wanted that fulﬁlls all constrains. Although solv-

ing ILP problems is computationally hard (NP-hard

to be more speciﬁc), there are powerful ILP solvers

available that perform well on our concrete problem

formulations, see (Newman, 2006).

A solution with the smallest value d = 1 is shown

in Figure 4. In each quarter one should apply a ran-

dom permutation within the two groups and/or use an

other solution of the ILP system. One solution of the

ILP is shown in Figure 5.

BY NR BW HE WL NI NO RP BB SN ST TH MV SL HB sum

BY 1 2 1 1 1 2 1 1 2 1 1 14

NR 1 2 2 1 1 1 1 1 2 1 1 14

BW 1 2 2 1 1 1 1 1 1 1 1 1 14

HE 1 2 1 2 1 1 2 1 1 1 1 14

WL 1 1 1 2 2 1 1 1 1 1 1 1 14

NI 1 1 1 1 2 2 1 1 1 1 1 1 14

NO 2 1 1 1 1 2 1 1 1 1 1 1 14

RP 2 2 1 1 1 1 1 2 1 1 1 14

BB 1 1 1 1 1 1 1 2 2 1 1 1 14

SN 2 1 1 1 1 1 1 2 2 1 1 14

ST 1 1 1 1 1 1 1 7

TH 1 1 1 1 1 1 1 7

MV 1 1 1 1 1 1 1 7

SL 1 1 1 1 1 1 1 7

HB 1 1 1 1 1 1 1 7

sum 14 14 14 14 14 14 14 14 14 14 7 7 7 7 7

quarterly permutation

Figure 4: Distribution matrix of medical expertises to exter-

nal peer reviews for occasion group ”incapacity of work” of

medical service institutions.

The result of the review (i.e. probability of im-

provement potential) may be considered as binomial

distribution and one can calculate conﬁdence inter-

vals for yearly reports. Such conﬁdence intervals are

much more suitable values which contain much noise.

The review results may depend on further parameters

which can be used to optimize the quality assurance

process. An example of the comparison of the re-

sults of the considered medical service institutions is

shown in Figure 6, real results can be given after for

quarters of quality assurance, which will be started in

2020.

HEALTHINF 2020 - 13th International Conference on Health Informatics

358

BY NR BW HE WL NI NO RP BB SN ST TH MV SL HB sum

BY 7 6 5 6 5 5 5 5 5 1 1 51

NR 6 6 6 4 5 5 1 1 2 1 1 1 39

BW 6 6 6 5 4 2 2 1 2 1 1 1 37

HE 5 5 6 4 2 1 1 2 1 1 1 29

WL 6 4 5 3 3 1 1 1 1 1 1 1 28

NI 6 4 3 2 2 2 1 1 1 1 1 1 25

NO 5 5 3 1 1 1 2 1 1 1 1 1 23

RP 5 2 2 1 1 1 2 2 1 1 18

BB 5 1 1 1 1 1 1 2 1 1 1 1 17

SN 5 2 2 1 1 1 1 1 1 1 1 17

ST 1 1 1 1 1 1 1 7

TH 1 1 1 1 1 1 1 7

MV 1 1 1 1 1 1 1 7

SL 1 1 1 1 1 1 1 7

HB 1 1 1 1 1 1 1 7

sum 51 39 37 29 28 25 23 18 17 17 7 7 7 7 7

quarterly permutation

Figure 5: Distribution matrix of medical expertises to ex-

ternal peer reviews for for occasion group ”hospital care”

among medical service institutions.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

medical board

number of mistakes

Mistake in question 16 (hospital care)

internal external

Figure 6: Results and conﬁdence intervals for external and

nationwide failure medical expertises.

4 CONCLUSIONS

The quality assurance process in the sense of the con-

sidered PDCA cycle will be especially successful if

acceptance, practicability and transparency support

each other. In each Medical Advisory Service Insti-

tution has to reach each medical expert in order to

reach a continuous development of a common un-

derstanding ans progression of quality. The exter-

nal peer reviews shall reach a common understanding

of quality in each region with a transparent presen-

tation of the results to the public. In Germany there

is a considerable heterogeneity in size and number of

employees among the medical service institutions in-

volved. Questions of equity are a constant concern.

The discussed methods provide the necessary frame-

work that enables all medical advisory service insti-

tutions to take part and to shape trustful cooperation

among each other.

The distribution of the regional expertises to na-

tionwide reviews are subject to distributional fairness.

All regions have to be appropriately included on the

one hand and larger resources of larger services on the

other hand must be taken into account. There shall

be no avoidable regional concentrations of peer re-

views as well as no no overemphasis on certain cou-

pling patterns. It is realized as a solution of an integer

linear program which shall not only change slightly

from quarter to quarter. An important point is the

reliability and validity of the valuation results. It is

essential to communicate conﬁdence intervals for the

results instead of single values so that the public is

in a position to distinguish between random and se-

cured differences. In the practical implementation the

interaction of different IT systems used by the medi-

cal experts in terms of good manageability and great

security is a big challenge.

The implementation uses a Linux based MySQL

server which is reached by the regional medical ad-

visory institutions as well as for coordinating work-

ing groups by the https protocol. In the working pro-

cess are used predominantly perl scripts. Overviews

are generated on the server as excel tables. There are

ﬂexible regional design options through conﬁguration

tables. The described concept should enable a fast

nationwide development of a common understanding

and improvement of quality. Quality conferences in

each occasion group twice a year will discuss dif-

ferent views and will lead to a ﬁnal decision process

which has to be communicated in each region by the

quality managers.

REFERENCES

Altenstetter, C. and Busse, R. (2005). Health Care Reform

in Germany: Patchwork Change Within Established

Governance Structures. Journal of Health Politics,

Policy and Law, 30(1-2):121–42.

uhn, S., Mathes, T., Prengel, P., Wegewitz, U., Ostermann,

T., Robens, S., and Pieper, D. (2017). The risk of bias

in systematic reviews (ROBIS) tool showed fair relia-

bility and good construct validity to assess the risk of

bias in systematic reviews. Journal of Clinical Epi-

demiology, pages 121–128.

Cortina, J. (1993). What Is Coefﬁcient Alpha? An Exami-

nation of Theory and Applications. Journal of Applied

Psychology, 78 (1):98–104.

Deming, W. E. (1982). Out of the crisis. Cambridge: Mas-

sachusetts Institute of Technology.

DRV (2018). Qualit

atssicherung der sozialmedizinis-

chen Begutachtung. Manual zum Peer Review-

Verfahren. Berlin: Deutsche Rentenversicherung

Bund.

Uberarbeitete Fassung August 2018.

IT-structures and Algorithms for Quality Assurance in the Health Insurance Medical Advisory Service Institutions in Germany

359

Eisinga, R., Grotenhuis, M., and Pelzer, B. (2012). The

reliability of a two-item scale: Pearson, Cronbach or

Spearman-Brown? International Journal of Public

Health, 58 (4):637–642.

Gerlach, F. M. (2001). Qualit

atsf

orderung in Praxis und

Klinik: eine Chance f

ur die Medizin. Stuttgart:

Thieme.

Institute of Medicine (2001). Crossing the Quality Chasm:

A New Health System for the 21st Century. Washing-

ton, DC, National Academy Press.

Institute of Medicine; Committee to Design a Strategy for

Quality Review and Assurance in Medicare (1990).

Medicare: A Strategy for Quality Assurance. Volume

II: Sources and Methods., volume 2. Washington

(DC): National Academies Press (US).

Internationale Organisation f

ur Normung (2015). DIN

EN ISO 9001:2015-11. Qualit

atsmanagementsysteme

- Anforderungen. Beuth Verlag.

Kamiske, G. F. and Brauer, J.-P. (2011).

Qualit

atsmanagement von A bis Z: Wichtige Be-

griffe des Qualit

atsmanagements und ihre Bedeutung.

Carl Hanser Verlag.

Machnik, W. (2009). 20 Jahre MDK Nordrhein. Ju-

bil

aumsschrift. MDK Nordrhein.

Masaaki, I. (1986). Kaizen: The Key To Japan’s Competi-

tive Success. McGraw-Hill Education.

Newman, M. (2006). Finding community structure in net-

works using the eigenvectors of matrices. Physical

Review E, 74.

Normand, S.-L., Mcneil, B., Peterson, L., and Palmer, R.

(1998). Eliciting expert opinion using the Delphi tech-

nique: identifying performance indicators for cardio-

vascular disease. International Journal for Quality in

Health Care, 10:247 – 260.

uchtern, E. (2008). Das professionelle Gutachten. Beson-

derheiten in der gesetzlichen Krankenversicherung.

MedSach, 105 (3):96–98.

Polak, U. W. (2018). Evaluation von Durchgangsarzt-

berichten mithilfe eines Peer-Review-Verfahrens.

Trauma Berufskrankh, 20 (Suppl. 4):237–240.

Sawicki, P. and Bastian, H. (2008). German Health Care: A

Bit of Bismarck Plus More Science. BMJ, 337:a1997.

Schmacke, N. (2016). Zur Positionierung des MDK in

der gesetzlichen Kranken- und Pﬂegeversicherung,

Gutachten im Auftrag des AOK-Bundesverbandes. In

IPP-Schriften, Eds. Bolte, G, G

orres, S. and Ger-

hardus, A., page Bd. 14. Institut f

ur Public Health und

Pﬂegeforschung (IPP) der Universit

at Bremen.

Shewhart, W. A. (1931). Statistical Method from the view-

point of quality control. (Reprint im Jahr 1986 einer

Ver

offentlichung des Department of Agricultuire Mil-

waukee 1931). New York: Dover Publication.

Shinano, Y., Fujie, T., and Kounoike, Y. (2003). Effective-

ness of Parallelizing the ILOG-CPLEX Mixed Inte-

ger Optimizer in the PUBB2 Framework. In Euro-Par

2003 Parallel Processing, pages 451–460. Springer

Berlin Heidelberg.

Strahl, A., Gerlich, C., Alpers, G., Ehrmann, K., Gehrke,

J., M

uller-Garnn, A., and Vogel, H. (2018). Devel-

opment and evaluation of a standardized peer-training

in the context of peer review for quality assurance in

work capacity evaluation. BMC Medical Education,

18:135–145.

Strahl, A., Gerlich, C., Wolf, H.-D., Gehrke, J., M

uller-

Garnn, A., and Vogel, H. (2016). Qualit

atssicherung

in der Sozialmedizinischen Begutachtung durch Peer

Review - ein Pilotprojekt der Deutschen Rentenver-

sicherung. Gesundheitswesen, 78:156–160.

Thiele K.-P., Kreuzer C., M. R. (2018). Krankenhaus-

landschaft in Deutschland: Zukunftsperspektiven, En-

twicklungstendenzen, Handlungsstrategien, chapter

Exkurs 3: MDK-Pr

ufung - Fluch oder Segen?, pages

89–94. Stuttgart: Kohlhammer.

Wirtz, M. and Caspar, F. (2002).

Beurteiler

ubereinstimmung und Beurteilerrelia-

bilit

at: Methoden zur Bestimmung und Verbesserung

der Zuverl

assigkeit von Einsch

atzungen mittels

Kategoriensystemen und Ratingskalen. G

ottingen:

Hogrefe.

HEALTHINF 2020 - 13th International Conference on Health Informatics

360