SECURITY ANALYSIS OF THE GERMAN ELECTRONIC
HEALTH CARD’S PERIPHERAL PARTS
Ali Sunyaev, Alexander Kaletsch, Christian Mauro and Helmut Krcmar
Chair for Information Systems, Technische Universität München, Boltzmannstraße 3, 85748 Garching, Germany
Keywords: Security Analysis, Electronic Health Card, Health Care Telematics.
Abstract: This paper describes a technical security analysis which is based on experiments done in a laboratory and
verified in a physician’s practice. The health care telematics infrastructure in Germany stipulates every
physician and every patient to automatically be given an electronic health smart card (for patients) and a
corresponding health professional card (for health care providers). We analyzed these cards and the
peripheral parts of the telematics infrastructure according to the ISO 27001 security standard. The
introduced attack scenarios show that there are several security issues in the peripheral parts of the German
health care telematics. Based on discovered vulnerabilities we provide corresponding security measures to
overcome these open issues and derive conceivable consequences for the nation-wide introduction of
electronic health card in Germany.
1 INTRODUCTION
During the next years in Germany the present health
insurance card will be replaced by the new
electronic health card (eHC) (Sunyaev et al., 2009).
The introduction tends to improve the efficiency of
the health system and the patients’ rights (Bales,
2003, p.5). In order to reduce costs in the public
sector and to create a homogeneous communication
basis a nationwide system is created – the health
care telematics infrastructure (TI). The eHC will not
only contain administrative data but also detailed
information about the patient and his treatments.
These pieces of information, covered by the
obligation of secrecy in the physician-patient
relationship and highly protected by law (Berg,
2004, pp.412-413), will now be stored in central
databases in order to improve services for the
patients.
Digitizing this information bears risks (Mandl et
al., 2007). Insurance companies, banks, employers
or marketing firms are only a few of several
organizations highly interested in health data (Huber
et al., 2008, p.1). Getting to know people’s state of
health, etiopathology or congenital diseases could
give them a remarkable competitive advantage. Each
individual whose data are stolen could get into
serious trouble (Blobel, 2004). As a consequence
patients could possibly get significant issues when
taking out a loan or trying to find insurance
(Anderson, 2001). Furthermore, one’s reputation
could get tarnished when the wrong pieces of own
sensitive medical information becomes publicly
accessible (Schneider, 2004).
This paper is based on extensive laboratory
experiments and on a detailed review of gematik’s
specifications (detailed information about health
care telematics specifications can be found at the
organization`s website - http://www.gematik.de).
Based on ISO 27001 for Information Security
Management Systems Standard and BSI Security
Guidelines (BSI, 2004), we focus on security issues
in the peripheral parts of the telematics system and
verify them in practice. These concerns are
categorized and possible solutions are presented in
this paper.
After the introduction of the German health care
telematics and its peripheral parts, the configurations
of the laboratory and the physician`s practice are
described in section 4. The results of the performed
security analysis and possible consequences are
presented in sections 5 and 6. Section 7 summarizes
our key findings and provides recommendations for
future work in this area.
19
Sunyaev A., Kaletsch A., Mauro C. and Krcmar H. (2009).
SECURITY ANALYSIS OF THE GERMAN ELECTRONIC HEALTH CARD’S PERIPHERAL PARTS.
In Proceedings of the 11th International Conference on Enterprise Information Systems - Information Systems Analysis and Specification, pages 19-26
DOI: 10.5220/0001854000190026
Copyright
c
SciTePress
2 THE GERMAN HEALTH
TELEMATICS
INFRASTRUCTURE
As requested by law (SGB V, 2007, § 291b) the
business organisation gematik (Gesellschaft für
Telematikanwendungen der Gesundheitskarte mbH)
was created in order to lead the introduction of the
electronic health card in Germany. gematik created
all specifications used in the health care telematics
infrastructure.
A nationwide telematics project was set up in
order to introduce the eHC in Germany. The health
care telematics infrastructure is divided in the central
part, which consists of data centres with central
databases and the peripheral parts, which are at the
different renderers of service – e.g. in physician’s
practice, hospitals or pharmacy (gematik, 2008e,
p.8). Both will be connected via a VPN tunnel. At
the client’s side the connection is established from
the so-called connector and at the central part
accepted by the VPN concentrator (gematik, 2008e,
p.8). The connector allows the primary systems and
the card reader to interact, which are both
components of the peripheral part as well.
The electronic health card has the same
proportions as a normal plastic card, e.g. like a credit
card. On the front side there are individual-related
information, a picture of the insurant and the
microchip. Also some recognition features like
braille, the name and logo of providing insurance
company are placed there. On the back there is the
European health insurance card (EHIC) (Drees,
2007, p. 1).
The eHC is a smartcard, which means it has its
own microprocessor with its own instruction set
(Caumanns et al., p.343). This distinguishes it from
the present health insurance card in Germany, which
is only a memory card. Not only administrative data
about the insurant is stored on the card, but also
medical data like electronic prescriptions. The
insurant can decide whether information for medical
emergencies, pharmaceutical documentations,
insurants receipts and medical reports will be stored
and if whether directly on the eHC or on central
databases (Neuhaus et. al., p.1).
3 PRIMARY SYSTEMS IN THE
PERIPHERAL
INFRASTRUCTURE
Primary systems are types of software which offer
the eHC’s functionality to the renderers of service,
e.g. practice or hospital information systems. This
software is usually installed on normal personal
computers which are used in the reception and
treatment rooms. As these are standard PCs also
standard services and programs are offered, e.g.
email and internet.
These facts make the primary systems’
computers a highly interesting target for attackers
who want to achieve access to patients’ data
(Sunyaev et al., 2008b, p.3). In experiments and
reviews attacks were subdivided into three different
target categories: users, hardware and software.
Hardware can be stolen or hidden, keyloggers (e.g.:
http://www.keyloggersdirect.com/
index.php?products) could be attached. Users could
be blackmailed, corrupted or spied on. But the most
likely scenario is that software could be
manipulated. This could be done by trojans, viruses
or spyware which infiltrate systems by accessing
websites, emails or through other security
vulnerabilities (Sunyaev et al., 2008a).
In order to handle these issues detailed security
knowledge is needed, not only when setting up the
systems, but also when using them. Practice personal
has to be trained to use these systems securely
(Schneider, 2004).
A big issue is that there is no present standard for
secure practices. gematik shifts them into the service
consumer tier (SCT) and in this vein they place the
responsibility for the primary systems on the
renderers of service. It is defined that SCT’s systems
are not part of the telematics infrastructure, but only
use them (gematik, 2008d, p.71). This means that
there are no rules defined for them at gematik
(gematik, 2008c, pp.134-138) and there is no
separate security concept as well. Also gematik
states that it should not be a problem that primary
systems can be unsupervised for up to 30 minutes
(gematik, 2008b, p.22).
4 LABORATORY’S/PHYSICIAN’S
PRACTICE CONFIGURATION
The laboratory consists of three main components:
the connector, the card reader and the primary
system. This is a standard configuration which is
used in every physician’s practice in Germany.
The connector is the central component in the
peripheral part of the telematics infrastructure. If the
primary system is to access an electronic health card
placed into the card reader it has to call a
connector’s function in order to proceed. It is not
ICEIS 2009 - International Conference on Enterprise Information Systems
20
Figure 1: The laboratory’s configuration (Source: own Figure).
possible to establish a direct connection between
card reader and primary system. So if data have to
be transferred into the central part of the health care
telematics infrastructure only the connector can do
so. The connector used in this test is part of the
Futro S400 series by Siemens in version V1.07R4.5;
hpscV1.07R4_build_2493_R13198.
The card reader Cherry SICCT Terminal in
version 10037 is another component attached to the
laboratory’s network. At the card reader the
electronic health card and the health professional
card (HPC) can be inserted (Mauro et. al., 2008). It
also has a keypad where numeric codes can be
entered in order to gain access to these cards.
A usual personal computer was used as primary
system which had an AMD Opteron Processor 144
with 1.81 GHz and 2.5GB RAM. Windows XP with
Service Pack 2, DocConcept 8.2, DocConnect and
Siemens Trusted Viewer are installed on the
computer. With the practice software the connector’s
functions can be initialized in order to use the
electronic health card’s functionality. The functions
allow the user to read administrative data, electronic
prescriptions and emergency information stored on
the eHC.
These components are normally connected by
LAN via a standard switch. But when doing special
network analyses this switch was replaced by a
repeating hub (Figure 1).
A laptop acting as an attacker joined to the
network was used for some experiments. It is
equipped with a Core 2 Duo T74002x 2.16GHz
processor and 2.0GB RAM. While running
Windows Vista Business with Service Pack 1 as
operating system, it has none of the tools which are
normally used to connect to the electronic health
card. But it has several tools installed which allow
analyzing the network’s traffic. Also an own client
for the connector was developed.
In order to validate the results of the experiments
tests were performed in a real physician’s practice.
A treatment room was used for the trials. Within
these an attacker accessed the LAN by using a port
behind a small commode. Figure 2 shows which
hardware was found by analyzing the network and
which of them were accessible.
As the practice has a well secured network, it
was not possible to break into the windows domain
which connected the practice computers, but it was
possible to access the telematics hardware. That
means that the attacker did not have access to any
PC with practice software, but it was able to control
the connector and with it every card reader in the
practice.
Figure 2: The Practice Network from the Attacker’s Point
of View (Source: own Figure).
5 THE NETWORK TRAFFIC
ANALYSES AND ITS
CONSEQUENCES
While analyzing the data sent over the network, all
components were connected via a repeating hub.
That means in contrast to a normal switch all data
are sent to every attached device. Now, the use of
tools like Wireshark (http://www.wireshark.org) or
EttercapNG (http://ettercap.sourceforge.net) makes
it possible to get a good impression of the network’s
dataflow.
The results show:
a) The connection between the card reader and
connector is fully encrypted.
SECURITY ANALYSIS OF THE GERMAN ELECTRONIC HEALTH CARD'S PERIPHERAL PARTS
21
Figure 3: Attack Tree (Source: own Figure).
b) There is no encryption between the primary
system and the connector.
The encrypted connection between the card
reader and the connector is enforced by gematik
(2008a, p.21). The missing encryption between
primary system and connector is caused by a lack in
the specification of gematik (2008b, pp.16-17;
2008c, p.297), which defines the use of a TSL
encryption between connector and primary system
as optional. As there is no encryption in the
implementation all requests and answers from the
primary system and the connector can be captured
and looked at in plain text by a third party. This
security issue is known at gematik (2008b, p.270),
but it is labelled as a residual risk. It is left to the
readers to decide if the possible theft of their private
data, which includes administrative data as well as
medical emergency information or electronic
prescriptions, is an acceptable threat.
In addition to the fact that there is no encryption
between the primary system and the connector there
is also no enforced authentication. That means
everyone can access the functions offered by the
connector. Using PHP (http://www.php.net) as
programming language in combination with the
principles of extreme programming (Beck, 2000) an
own client was implemented in order to fulfil
derived attack scenarios. The program can be
controlled via different interfaces, e.g. with a
command line tool or a web interface.
There are three modes offered by the program:
The “direct”-mode allows calling functions
instantly.
The “wait”-mode tries every three seconds to
find an attached card and then sends the request.
The “listen”-mode registers at the connector
and waits for an event, which is triggered when a
card is attached and then the request will be send
to the connector.
The client is able to call all functions at the
connector that could also be used by DocConnect
8.2. So it can act like a normal primary system and
keeps itself quite well covered.
Abusing the functions provided by the connector
leads to some strong attack scenarios.
6 ATTACKING THE GERMAN
ELECTRONIC HEALTH CARD
The attack tree shown in Figure 3 denotes that the
attacks can be classified into disrupting and spying
types. The utilisation can be interrupted when
permanently ejecting all cards, which get attached to
a card reader. The deletion of prescriptions stored on
the electronic health card is as possible as the
blocking or destruction of the card itself. An attacker
can also steal administrative data, prescriptions and
medical emergency information stored on eHCs.
All attacks are based on the following scenario:
The attacker can gain access to the physician’s
practice network, e.g. through hacking the WLAN or
just plugging into a socket. Also another common
procedure is needed in order to call some functions:
The physician unlocks his own health professional
card (HPC) with his personal identification number
(PIN) in the morning and locks the card in the
evening, which means the card is ready to use for the
whole day.
Table 1 shows the specifications of the functions
that will be abused in the following in order to attack
the German electronic health card.
ICEIS 2009 - International Conference on Enterprise Information Systems
22
Table 1: Connector’s Functions that can be abused.
Permanent-Card-Ejection
PIN needed: None
supported card types: All
Connector’s function: EjectCard
gematik’s
specification:
gematik 2008e, 200f; chapter
5.4.3.3.6
Delete or fill all prescriptions
PINs needed:
HPC practice’s PIN
Supported card types:
EHC
Connector’s function
names and their
gematik’s specification
ReadVO: gematik
2008h, 73ff; chapter 6.2
DeleteVO: gematik
2008h, 79ff; chapter 6.4
DispenceVO: gematik
2008h, 76ff; chapter 6.3
Block a Card
PINs needed:
PIN to change
Supported card types:
EHC, HPC
Connector’s function: ChangePin
gematik’s
specification:
gematik 2008e, 204f; chapter
5.4.3.3.8
Destroy a Card
PINs needed: PUK for locked PIN
Supported card types:
EHC, HPC
Connector’s function: UnblockPin
gematik’s
specification:
gematik 2008e, 209f; chapter
5.4.3.3.12
Steal private data from electronic health card
PINs needed: HPC practice’s PIN
Supported card types: EHC
Connector’s functions
and their gematik’s
specification:
ReadVSD: gematik
2008f, 52ff; chapter 7.1
ReadVO: gematik
2008h, 73ff; chapter 6.2
ReadNFD: gematik
2008g, 31ff; chapter 6.2
6.1 Permanent-Card-Ejection
The Attack. There are two ways to realize a
permanent ejection. On the one hand it is possible to
constantly call a function that ejects a card, e.g.
every three seconds. On the other hand the
registration to an event handler at the connector is
possible. Then an event is triggered and immediate
response can take place. Regardless of the way used
it is not possible to attach any card to a card reader
anymore (Figure 4).
A Possible Solution. As there is no additional
benefit created when ejecting a card via the network
this functionality could be easily removed. It is fully
satisfying when a card can only be ejected locally at
the card reader.
Figure 4: Activity Diagram - Permanent Card Ejection
(Source: own Figure)
6.2 Fill or Delete Prescriptions
Figure 5: Activity Diagram - Fill or Delete Prescriptions
(Source: own Figure).
SECURITY ANALYSIS OF THE GERMAN ELECTRONIC HEALTH CARD'S PERIPHERAL PARTS
23
The Attack. In order to delete or fill a prescription
its Object ID is needed which is received from the
connector in a first step. After the receipt the
function for deleting or fulfilling the prescription can
be called (Figure 5). As there is a maximum capacity
of eight prescriptions on every electronic health card
this procedure will repeat at most eight times then
every prescription will be dispensed or deleted.
A Possible Solution. For writing or changing a
prescription on an eHC the physician’s signature
personal identification number is needed. The usage
of this PIN while deleting or dispensing electronic
prescriptions would suppress a fully automated
function call as described above. Therefore, the
attack would not be possible anymore, because
every action had to be authorized by a human on the
card reader.
6.3 Lock a Card’s PIN
Figure 6: Activity Diagram - Lock a PIN (Source: own
Figure).
The Attack. There is a connector’s function which
enables the user to change a card’s PIN remotely. A
PIN can be entered wrong three times before it is
locked. This means abusing this functionality could
block the card (Figure 6). In an experiment it took
350ms to call this function. That means within about
one second a PIN can be locked.
A Possible Solution. As this function does not
generate additional value and would be probably
used very rarely, e.g. only initially, the function
should be removed. It would be sufficient to be able
to change the PIN only directly at the card reader.
6.4 Destroy a Card
Figure 7: Activity Diagram - Destroy a card (Source: own
Figure).
The Attack. When a PIN is locked, a function can
be called which allows the user to unlock the PIN.
As a parameter the personal unlocking key (PUK) is
needed. It can only be used ten times before it
becomes finally locked. So using this function with a
wrong PUK for ten times on a locked PIN would
lead to a locked PIN and a locked PUK. Not being
able to unlock the PIN means that the card cannot be
used anymore. The card is destroyed (Figure 7).
ICEIS 2009 - International Conference on Enterprise Information Systems
24
Figure 8: Activity Diagram - Spy Personal Information (Source: own Figure).
A Possible Solution. The case of a locked PIN
should be an exception. So there is no real need to
implement a function like this in the connector and
the primary systems. Furthermore, it would be
sufficient if the PIN can be unlocked directly on the
card reader.
6.5 Spy Personal Information
As the thefts of administrative data, electronic
prescriptions and emergency information work the
same way they are all combined in Figure 8.
The Attacks. When a card is inserted, connector’s
functions can be called in order to steal
administrative data, electronic prescriptions and
emergency information (Figure 8). As response a
XML file will be provided which can be easily
parsed and saved. The collection of these data (see
Table 2) gives the attacker private information about
the patient and detailed knowledge about his state of
health.
Table 2: Summary of Content that can be spied out.
administrative
data
insurant id
given name, family name
birthday, sex
full address
information about insurance coverage
and the insurance company
electronic
prescriptions
date of issue
patient’s and physician’s administrative
data
information about the prescription
name of medication
name of pharmacy
usage information
emergency
information
(past) diseases
medication (and incompatibility)
attending physician
persons to be notified
other notices (free text)
A Possible Solution. As these functions are
essential they cannot be removed. Protecting them
with a separate PIN would delay practice work. So
only a secure connection between the primary
system and the connector, which is encrypted and
authenticated, will solve the problem.
7 CONCLUSIONS AND
OUTLOOK
Table 3: Attacks tested in Laboratory and Practice.
Attack Consequences
Permanent-Card-
Ejection
Practice system cannot be used during an
attack. This results in a work delay.
Delete all
prescriptions
Patient looses all his prescriptions; this is
very annoying, especially if he has had
prescriptions form different physicians
on the card.
Block Card Unlocking with PUK is possible, but at
first the PUK has to be send to the
insurant by mail.
Destroy Card The insurant has to order a new card at
his insurance company.
Spy
administrative
data
Name, address, birthday and insurance
data get stolen.
Spy
prescriptions
Data which gets stolen can be used to
deduce the recent state of health.
Spy emergency
information
Information about medication
intolerance, previous diseases and other
highly private data gets stolen.
In the course of the present work critical security
issues in the German electronic health card’s system
have been discovered. These have been tested in a
laboratory and have been verified in a real
physician’s practice (Table 3). Therefore the
following statements must be made:
SECURITY ANALYSIS OF THE GERMAN ELECTRONIC HEALTH CARD'S PERIPHERAL PARTS
25
Patient’s private and very sensible data stored
on the German electronic health card are not
secure and it is possible to steal them, because
the card is used in an unsecured environment.
Also manipulations of the eHC and health
professional card are feasible.
Possible solutions to these security treatments
have been given. It is undeniable that the connection
between the connector and primary system must be
encrypted and authentication has to be enforced. The
given solutions are not extremely expensive or
complicated. Also, it should be possible to
implement them in time. In this regard rules and
scenarios for primary systems should also be
included in the specification in order to create a
nationwide standard for practice information
technology.
For future work, the components should also be
exposed to further penetration tests. E.g. man in the
middle attacks should be launched between the TLS
encrypted network parts and hardware manipulations
should be tried.
REFERENCES
Anderson, R. J., 2001. Security Engineering: a Guide to
Building Dependable Distributed Systems. 1st. John
Wiley & Sons, Inc.
Bales, S., 2003. Die Einführung der Telematik im
Gesundheitswesen als Herausforderung für die
Weiterentwicklung der Patientenrechte in
Deutschland. [Talk] Bonn: gematik. Available at:
http://www.dimdi.de/dynamic/de/ehealth/karte/downlo
adcenter/veroeffentlichungen/vortraege/bagh-bonn-
bal-031107.pdf [Accessed 9 September 2008].
Beck, K., 2000. Extreme programming eXplained:
embrace change. Reading, MA: Addison-Wesley.
Berg, W., 2004. Telemedizin und Datenschutz.
Medizinrecht, 22 (8), pp. 411-414.
Blobel, B., 2004. Authorisation and access control for
electronic health record systems. International Journal
of Medical Informatics, Vol. 73 Nr. 3, pp. 251-257.
BSI, Bundesamt für Sicherheit in der Informationstechnik,
2004. Studie zu ISO-Normungsaktivitten ISO/BPM -
Anforderungen an Information Security Management
Systeme.
Caumanns, J. et al., 2006. Die eGK-Lösungsarchitektur
Architektur zur Unterstützung der Anwendungen der
elektronischen Gesundheitskarte. Informatik-
Spektrum, 29 (5), pp. 341-348.
Drees, D., 2007: The Introduction of Health Telematics in
Germany. In: European Commission Directorate
General Information Society, Information Security
Solutions Europe/SECURE 2007 Conference. Poland,
Warsaw 25 27 September 2007. Vieweg: Wiesbaden.
Gematik, Gesellschaft für Telematikanwendungen der
Gesundheitskarte mbH, 2008a. Spezifikation eHealth-
Kartenterminal. Version 2.6.2.
Gematik, Gesellschaft für Telematikanwendungen der
Gesundheitskarte mbH, 2008b. Spezifisches
Sicherheitskonzept der dezentralen Komponenten -
Einboxkonnektor-Szenario. Version 0.9.0 Kandidat.
Gematik, Gesellschaft für Telematikanwendungen der
Gesundheitskarte mbH, 2008c. Übergreifendes
Sicherheitskonzept der Gesundheitstelematik. Version
2.3.0.
Gematik, Gesellschaft für Telematikanwendungen der
Gesundheitskarte mbH, 2008d. Gesamtarchitektur.
Version 1.4.0.
Gematik, Gesellschaft für Telematikanwendungen der
Gesundheitskarte mbH, 2008e.
Konnektorspezifikation. Version 2.8.0.
Gematik, Gesellschaft für Telematikanwendungen der
Gesundheitskarte mbH, 2008f. Facharchitektur
Versichertenstammdatenmanagement (VSDM).
Version 2.6.0.
Gematik, Gesellschaft für Telematikanwendungen der
Gesundheitskarte mbH, 2008g. Facharchitektur Daten
für die Nie Notfallversorgung (NFDM). Version 1.7.0.
Gematik, Gesellschaft für Telematikanwendungen der
Gesundheitskarte mbH, 2008h. Facharchitektur
Verordnungsdatenmanagement (VODM)
. Version
1.5.1.
Huber, M., Sunyaev, A. & Krcmar, H., 2008. Security
Analysis of the Health Care Telematics Infrastructure
in Germany. In: INSTICC, International Conference
on Enterprise Information Systems 2008. Spain,
Barcelona 12-16 June 2008.
Mandl, K.D. et al., 2007. Indivo: a personally controlled
health record for health information exchange and
communication. BMC Medical Informatics and
Decision Making, Vol. 7 Nr. 25.
Mauro, C. et al., 2008. A Proposed Solution for Managing
Doctor’s Smart Cards in Hospitals Using a Single
Sign-On Central Architecture. In: HICSS 41, Hawaii
International Conference on System Sciences. Hawaii,
Big Island 7-10 January 2008.
Neuhaus, J., Deiters, W. & Wiedeler, M., 2006.
Mehrwertdienste im Umfeld der elektronischen
Gesundheitskarte. Informatik-Spektrum, 22 (5),
pp.332-340.
Schneier, B., 2000. Secrets and lies: digital security in a
networked world. John Wiley: New York.
SGB V, 2007. Sozialgesetzbuch. Fünftes Buch. DTV-
Beck.
Sunyaev, A., von Beck, J., Jedamzik, S. & Krcmar, H.,
2008a. IT-Sicherheitsrichtlinien für eine sichere
Arztpraxis. Volume 1. Berlin: Shaker.
Sunyaev, A. et al., 2008b. Bewertung und Klassifikation
von Bedrohungen im Umfeld der elektronischen
Gesundheitskarte. In: Gesellschaft für Informatik,
Informatik 2008. Germany, Munich 8-13 September.
Gesellschaft für Informatik: Munich.
Sunyaev, A. et al., 2009. Analysis of the Applications of
the Electronic Health Card in Germany. In: WI 2009,
Proceedings of Wirtschaftsinformatik 2009, Austria,
Vienna 25-27 February 2009.
ICEIS 2009 - International Conference on Enterprise Information Systems
26