Jump to: Page Content, Site Navigation, Site Search,
You are seeing this message because your web browser does not support basic web standards. Find out more about why this message is appearing and what you can do to make your experience on this site better.
D W Chadwick a IS Institute, University of Salford, Salford M5 4WT, b Computer Sciences, University
of Salford, Salford M5 4WT, c Department of Diabetes and Endocrinology, Hope
Hospital, Salford M6 8HD
Correspondence to: J P New john.new{at}virgin.net
Effective programmes for management of chronic disease are
invariably supported by information technology. These have typically been developed within secondary care, with limited access available to
primary care. The British government's white paper The New NHS clearly states that a dedicated NHS network service (NHSnet), linking information systems in primary and secondary care, will be
working by 2002.1 Although this could improve the flow of information between primary and secondary care, thereby improving patient care,1 there are, quite rightly, grave concerns
about the security and confidentiality of patients' data both in terms of who has legitimate access and who has illegitimate access to the
data.
2 3
Furthermore, it cannot be assumed that all
general practitioners will be willing to spend the time and money
necessary to connect to NHSnet and to conform to its code of
connection,4 especially as more than one in eight general
practitioners are already connected to and familiar with the
internet5 and, increasingly, internet service providers
are free of charge (apart from telephone charges). Users of NHSnet,
however, will be charged for connection and use, the costs of which are
not yet known. These are major concerns for general practitioners and
may be a reason for the poor uptake of NHSnet by general
practice.5
In addition, although NHSnet could increase hospitals' and
general practices' access to patients' data, this information will still not be available within the patients' homes. This is exactly where general practitioners, seeing sick patients whom they may not
know, need immediate access to this information. Internet technology
could also allow patients access to their own data, thereby empowering
them to allow appropriate healthcare professionals access to their data.
We suggest that the internet will be as good, if not better, an
integrating network as NHSnet, providing that the issues of security
and reliability can be adequately addressed. Using our existing
diabetes information system (Westman Medical Software, Manchester), we
have developed a secure, encrypted internet connection that allows
general practitioners, diabetes nurse specialists, and, potentially,
patients immediate access to the diabetes information system (see
figure). We describe the problems, and solutions, that we encountered
in ensuring the confidentiality of the data and in retaining the
security of the hospital's private intranet (see
table).
The diabetes information system was introduced in 1992 to
facilitate diabetes care within Salford and is used by all local general practitioners and hospital diabetes services. This system is
used by 35 districts in Britain, facilitating diabetes care for about
200 000 people. For all patients it contains data, based on the UK
diabetes dataset,6 that are updated and verified during
their annual review. The diabetes system prompts these annual reviews
in both primary and secondary care. For those patients receiving
primary care, the diabetes system generates a single page form
containing a summary of the data that is sent through the post to their
general practitioner. This system is inflexible; if a patient were to
attend his or her general practitioner before the annual review was due
this information would be unavailable. Furthermore, after the review
the completed form is returned to the central hospital, where the data
are entered on to the diabetes system, often several weeks later. An
internet connection with the central hospital would allow appropriate
healthcare professionals real time access to the diabetes information system.
As with all information systems, it is essential that the
people accessing the data are who they claim to be. Traditionally, computer systems have asked for passwords to validate users. On a
"stand alone" computer or secure private networks (such as NHSnet) this is usually adequate, but not for the internet. The ready availability of "packet sniffers" (programs that sit in a computer network and copy data that are later analysed to extract usernames and
passwords) and "cracker programs" (which attempt to find a user's
password) make it relatively easy for unauthorised people to obtain
passwords. Consequently, passwords that are not encrypted cannot be
used for reliable authentication across the internet. We therefore used
an encryption technology known as public key encryption (Entrust PKI,
Entrust Technologies, Ottawa, Canada) to ensure users were who they
claimed to be (table).
Asymmetric encryption relies on two keys that work together as a
pair Careful management of public keys is essential. If a hacker
successfully substitutes his public key for that of a genuine user he
could masquerade as that user and gain access to the hospital diabetes
system. The keys are managed by a certification authority, a trusted
third party, that validates users and issues digitally signed public
keys (certificates) so that recipients can be assured that the owner of
the public key is genuine. The Entrust system allows the management and
use of public key certificates to be centrally controlled by the
trusted third party. Such rigorous authentication procedures ensure
that hackers cannot masquerade as a legitimate user.
The certification authority server and the public key
directory server used to publish all the certificates are housed in a
securely fortified room in the University of Salford. Participants have
a component of the Entrust software (Direct Client) installed on their
computer, which generates the user's pair of keys. It communicates
with the servers in order get the user's public key certified, and to
download the certified public keys of other users. The certification
authority server will issue certificates only to registered
participants, so certification is tightly controlled. In addition to
the Entrust infrastructure, a user's private key is stored either on
the computer's hard disk or on a personal smartcard (ActivCard Gold,
Activcard SA, France). Storing the private key on a smartcard provides
increased security, allows users to be authenticated on different
machines, and provides for sequential access by many users from the
same machine. In order to access the private key a personal
identification number (PIN) must be entered, so an impostor would need
to both steal a user's smartcard and know his or her personal
identification number in order to masquerade as the user. This
technology and infrastructure enables strong authentication of users
accessing the diabetes system over the internet.
Finally, it is essential that healthcare professionals are confident
that they are accessing data from the diabetes system and not from a
false source. With the same public key infrastructure, a certificate is
issued to the diabetes system. Consequently, a user receives a
digitally signed message from the diabetes system, providing mutual authentication.
Before a user can be authenticated by the hospital diabetes
system, the digitally signed message must be given access to the hospital's private intranet. It is essential to maintain the integrity of the hospital intranet and avoid unauthorised entry via the internet
while allowing authorised traffic to gain access. This is a difficult
balancing act: too strict a policy will deny authorised people
legitimate access to and from the internet, but too lax a policy will
allow intruders into the hospital intranet. This security was achieved
by installing a firewall (Checkpoint Software Technologies, Israel)
between the hospital intranet and the internet and requiring users to
be strongly authenticated to the Entrust Direct Server in the firewall.
This connection complies with the NHSnet code of
connection.4
Sending patients' information across the internet, which
can be openly accessed and is therefore subject to eavesdropping or
interference, requires strong encryption to make it extremely difficult
to interpret or interfere with the contents of a message. This was
achieved using 128 bit, strong encryption (CAST 128 algorithm) of all
the data transmitted across the internet. Any attempt to decrypt such a
message by working through every possible key combination would,
allowing one attempt every microsecond, take an average of
5.4×1024 years (longer than the earth has been
in existence). The encryption is automatically initiated after user
authentication. This ensures the encryption of all data before they
traverse the internet and enables users to access the data unaware of,
and not inconvenienced by, background encryption.
All the healthcare professionals who used the diabetes information
system were familiar with the paper version of the diabetes form, and
most were familiar with using web browsers (Netscape Navigator
(Netscape Communications, California, USA) or Microsoft Internet
Explorer (Microsoft, Redmond, USA)). We therefore developed a web based
interface to the diabetes system that closely resembled the existing
form. The web interface uses hypertext mark-up language (HTML) to
present the form to users. The form is generated dynamically from the
diabetes server by Microsoft's Internet Information Server. The
diabetes server is queried by means of structured query language (SQL),
which is generated by computer graphics interface running from the
Internet Information Server. Consequently, healthcare professionals can
access the diabetes information system in a format similar to the one
they are already using via a web interface that they are already
familiar with.
To simplify these processes, user verification, encryption, and
starting the web browser automatically start after the Entrust Direct
Client is launched and the user's smartcard and personal identification number have been inserted. This ensures that all data
are encrypted and digitally signed. The Entrust Direct Server then
decrypts the message and sends it to the web server, which is oblivious
to this intervening security layer. The Entrust Direct Server was
located in the hospital firewall system, thereby ensuring that users
were authenticated before they accessed the hospital intranet (see table).
The existing diabetes system already had simple access controls,
with users divided into two groups (general practitioner or consultant)
and each group having different privileges to allow them access to
their own patients' data. A security table within the diabetes system
holds and enforces these privileges.
A simple search facility was incorporated within the web interface to
enable users, according to their user privileges, to identify patients
and to view and modify their records. All modifications to patients'
record are applied in real time, immediately updating the patient
records in the diabetes system The use of information technology to improve health care is the
cornerstone of the government's strategy for the new
NHS.7 We have described a simple to use, yet effective,
method for securely connecting a hospital diabetes information system
across the internet, thereby providing patient specific information to
general practices or even patients' homes. We believe that such
systems could be generically applied to most, if not all, forms of
chronic disease management. Easy access to clinically relevant
information may improve patient care and, by reducing unnecessary
duplication of investigations, save money.
Although NHSnet will provide an NHS-wide private network,
clinicians should realise that it is not the only solution for
providing access to patients' information. We believe that our system
is as safe as, if not safer than, NHSnet. The sooner an NHS-wide standardisation for encryption and trusted third party verification is
defined the better. Hopefully, these standards will be seamlessly integrated into chronic disease management systems to enable clinicians to fully harness the benefits of improved availability of information without jeopardising patient confidentiality.
Summary points
As an alternative to NHSnet, the internet can be used to access
patients' medical records
Patient confidentiality is ensured by using strong encryption and
trusted third parties, these are stronger methods than those used
within NHSnet
The encryption and strong authentication have been integrated into
current browsers (such as Internet Explorer), producing an interface
that is familiar to most people
Improved access to patients' data should improve healthcare delivery

View larger version (33K):
[in a new window]
Organisation of internet access to Salford diabetes information
system showing location of firewalls, smartcard, encryption software,
and trusted third party applications
![]()
Salford diabetes information system
![]()
User authentication
![]()
Public key cryptography and digital signatures
an encryption key and a decryption key. If a user generates a
pair of keys and makes the decryption key public but keeps the encryption key private, then only the user can encrypt a message but
anyone with the public key may decrypt it and thereby be assured of the
sender's real identity (see appendix on the BMJ website for further details).
![]()
The security infrastructure
![]()
Entry into hospital network
![]()
Unauthorised data capture during transmission across the internet
![]()
Easy to use interface to the diabetes information system
![]()
Establishing access rights
a considerable improvement over the
current paper based system, where modifications may take weeks before
being available on the diabetes system. No data are actually deleted
from the diabetes system; mistakes can be deleted from the viewed
record but are retained within the diabetes system, being marked
"deleted." An audit trail identifies who has made any additions.
![]()
Summary
| |
Acknowledgments |
|---|
Contributors: JPN, TLD, and DWC had the original idea for the project. DWC designed the security for the application. PJC developed the internet interface to the diabetes information system. AJY integrated the diabetes information system and internet interface with the security features. DMMcD developed the diabetes information system. JPN and DWC wrote the manuscript.
| |
Footnotes |
|---|
Funding: The project was funded by the UK EPSRC (Grant No GR/L60548), the European Commission IV Framework Programme Trusthealth 2 Project (Contract No HC 4023), and Entrust Technologies (donation of software).
Competing interests: DMMcD owns Westman Medical Software.
Extra details about encryption of
computer files appears on the BMJ's website
| |
References |
|---|
| 1. |
Keen J.
Rethinking NHS networking.
BMJ
1998;
316:
1291-1293 |
| 2. | Department of Health. Report on the review of patient-identifiable information. London: Department of Health, 1997. (Caldicott committee report.) |
| 3. |
Anderson R.
NHS-wide networking and patient confidentiality.
BMJ
1995;
311:
5-6 |
| 4. | NHS Executive Information Management Group. NHS wide networking. London: NHS Executive, 1995. |
| 5. | Roscoe TJ. Two surveys of internet use in primary care. He@lth Information Internet 1998; 3: 2-3. |
| 6. | Vaughan NJ, Home PD. The UK diabetes dataset: a standard for information exchange. Diabetes Audit Working Group of the Research Unit of the Royal College of Physicians. British Diabetic Association. Diabet Med 1995; 12: 717-722[Medline]. |
| 7. | Department of Health. Information for health: an information strategy for the modern NHS 1998-2005. www.imt4nhs.exec.nhs.uk/strategy/index.htm, 1999. |
(Accepted 31 May 2000)
Read all Rapid Responses