ABC of health informaticsReferral or follow-up?Frank Sullivan, Jeremy C Wyatt

When patients ask their doctors if a preventable problem couldhave been avoided by earlier investigation or referral, thedoctors can be in an unenviable position. Given the informationavailable at the time, the response will often be a qualified, “yes.”It must be a qualified response because the aspects of theproblem considered during earlier encounters with patients areoften unknown. The matter is further complicated by issues oftrust, professional ethics, and the law.

This article discusses information flows that may havereduced the risk of Ms Smith (see box opposite) developingsymptomatic renal impairment. The risk could have beenreduced at three different points.x If her underlying vesicoureteric reflux had been diagnosedand fully investigated in childhoodx When her chronic pyelonephritis was discoveredx During the intervening period when no follow-up wasarranged.

Early detection of underlying problemsChildren aged ≤ 7 years with urinary symptoms, fever, orseveral non-specific symptoms and signs should be tested forurinary infections because, in some circumstances, prophylaxiscan prevent recurrence. Guidelines are available, but theresearch that underpins the advice was published too late forMs Smith. Were Ms Smith a young girl today, any primary careor emergency clinician who saw her would probably have accessto this evidence base as part of their clinical software, orthrough access to guidelines on the internet.

Undergraduate education, postgraduate training, andcontinuing professional development are more traditionalroutes of knowledge transfer. Unfortunately, traditional sourcesof knowledge are relatively inefficient: our stores of knowledgedecay over time, and our brain’s working memory may becomeoverloaded. Prompts and reminders at the point of care areuseful adjuncts to an overworked human brain for certain tasks.Some doctors worry that use of such electronic aids may reducepatient trust, but the evidence is to the contrary.

Ensuring appropriate investigationFifteen years ago, when Ms Smith’s chronic pyelonephritis wasdiagnosed, her investigation would probably have been directedby a consultant whose experience would have ensured theappropriate level of expertise was achieved. In the informaticsage, some of this expertise can be represented in protocols. Ifthe protocols are followed, investigation in primary care mayavoid referral or identify the nature of the problem quickly andclearly. In some health systems, referring clinicians may be givenshorter waiting lists if the referrals have been preceded byappropriate first line investigation.

Arranging referralHealth maintenance organisations in the United States, whichprovide integrated primary and secondary care, can bookelectronic appointments routinely. More complex referralsettings may have difficulty doing this. Delays can occur at any

PRODIGY (Prescribing RatiOnally with Decision Support) guideline oninvestigation of urinary tract infection in children

Working memoryInterpretation

RearrangementComparison

StoragePreparation

Perceptual filterUndergraduate

studiesOther experience

Scientific evidenceWider literature

Proposed actions

Examination

History

Incominginformation

Learning/feedback loop

Long termmemoryBranchedor scripts

Store

Retrieve

A model of human clinical information processing

Steps in the NHS process for non-urgent referralx During a consultation the general practitioner (GP) considers if

referral is appropriate

x Decision is negotiated, to a greater or lesser extent, with patient

x Decision and relevant clinical information is communicated toconsultant or other secondary care provider, usually by letter

x Letter is posted or faxed to hospital

x Consultant prioritises referral

x Outpatient administrator allocates an appointment depending onthe level of priority and the availability of appointments

x Appointment is communicated to patient, usually by letter

x Patient attends outpatient clinic and sees consultant or a memberof their team

This is the eighth in a series of 12 articlesA glossary of terms is available at http://bmj.com/cgi/content/full/331/7516/566/DC1

Ms Smith is a 58 year old florist with a 15 year history ofrenal impairment caused by childhood pyelonephritis.She has hypercalcaemia. She remembers being told in thepast that she had “a slight kidney problem,” and asks herrenal physician whether anything could have been donethen to prevent the current problem developing

Clinical review

1072 BMJ VOLUME 331 5 NOVEMBER 2005 bmj.com

stage between the decision to refer and its realisation, evenwhen an appointment is available.

Arranging follow-upAt the end of a hospital outpatient visit a decision is madeabout whether hospital, a GP, or shared care is mostappropriate for the patient. Unless an arrangement is made thepatient may have a “collusion of anonymity.” This occurs whenpersonnel at the hospital think that staff at the primary carepractice are providing follow-up and vice versa. In reality,neither are doing so. To avoid such errors, healthcare systemshave developed ways of integrating multiple service providersand proactive measures (see chronic care model opposite).

Hospital follow-upWhen a consultant decides that a patient’s problem needshospital resources the flows of information are straightforward,but potential exists for errors and omissions. Often, patients areasked to book their next appointment as they leave the clinic.Alternatively, one of the clinic team may make the arrangementon the patient’s behalf, and inform them at the time or by post.

GP follow-upIf the hospital team decide that the patient requires medicalsupervision, but no other hospital resources, the primary careteam may be asked to resume sole responsibility for care. This isthe simplest option for hospitals because it only needs adischarge letter to be sent. Most practices in the UnitedKingdom and other industrialised countries have thetechnology and systems to support a call-recall system forscreening. Although this can be extended to support GPfollow-up of chronic diseases, few practices are able to harnesssuch systems to long term clinical care. This will probablychange in countries like the United Kingdom, where achievingtargets is increasingly important.

Shared careAlthough shared care seems the most complex of the threefollow-up options, done properly, it may be the best for thepatient. An integrated service takes responsibility for all patientswith the problem it is set up to deal with. Specialists ensure thathealthcare services are configured to respond effectively topatients with problems, and to support clinicians working in thecommunity.

Antenatal care is an example of this approach. Other areasof care, such as chronic diseases, are following suit, withexcellent results seen in the care of people with diabetes andcardiovascular disease. Good, but often asynchronous,communication between colleagues with complementary skillsis vital. In some systems, records may be seen by cliniciansirrespective of where they are working.

Patient empowerment“Expert patients” have always been with us, but some doctorswere not aware of it, or would not acknowledge it. No matterwhat arrangement is made for follow-up by the healthprofessions, patients with chronic illnesses must deal with itevery day. Better use of information helps, and consultationsshould be patient centred (to deal with patients’ ideas, concerns,and expectations), but also extend beyond the visit.

1. CommunityResources and policies

2. Health systemOrganisation of health care

3. Self-managementsupport

4. Deliverysystemdesign

5. Decisionsupport

6. Clinicalinformationsystems

Functional and clinical outcomes

Prepared, proactivepractice team

Informed,activated patient Productive interactions

Overview of the chronic care model. Adapted from Wagner EH. EffectiveClin Pract 1998;1:2-4

Sharing information across health systems. Clinical data (for example, dataon prescribing or blood pressure) in one part of the health system that havebeen recorded in primary care can be made available to other users, such ashospital clinicians, on a “need to know” basis

If follow-up at hospital is needed thendirect booking of the next visit avoidssome potential difficulties

Many doctors give patients an audiotape of theirconsultation, written material about their problem, orwebsite addresses that provide further information.Others (hospital consultants, for example), may copyletters sent to GPs as text messages to the patient’smobile phone or send the letters to the patient as emailattachments

Clinical review

1073BMJ VOLUME 331 5 NOVEMBER 2005 bmj.com

Clinical governanceAs a result of apparent failures to ensure adequate patient care,society has demanded that arrangements for the supervision ofclinical services are improved. The days of autonomy andpaternalism are being replaced by rigorous inspectionprocedures and publication of results. Clinical teams need toshow that they are working to the highest standards. Thisdepends on their access to the best evidence about the criteriaof good care and the standards that can be attained. Data, oftenfrom patient records, are then collected to confirm whetherstandards are being met, or if there are any defects to treasure.Failure to hit the target (for example, to offer annual bloodpressure and renal function tests to Ms Smith), is anopportunity to improve the service. Electronic records makemost of the service automatic, provided that patients agree to(or at least do not refuse) the secondary use of their personaldata. Clinical teams can concentrate on providing a service, andusing the information that has been captured (and processed)electronically to improve patient care. In the United Kingdom,the quality and outcomes framework of the new generalmedical services contract for GPs relies heavily on the electronicprocessing of Read coded data in clinical systems.

SummaryAchieving effective data transfer and electronic continuity ofcare between different parts of a health service is not essentiallya technical challenge, rather it is a cultural and political one. It islargely about reconfiguring workflow. In 2004, the VeteransHealth Administration showed that integrating clinical recordsacross geographically and clinically diverse sites is feasible andvaluable. Linking individual electronic patient records fromdifferent locations into a single electronic health record willprobably transform the quality of health services over the nextdecade.

BMJ 2005;331:1072–4

UK general practice contract quality points for managementof hypertension

Indicator Coverage PointsRegister of patients Yes/no 9Smoking status 25-90% 10Smoking advice 25-90% 10Blood pressure recorded in past 9 months 25-90% 20Blood pressure ≤ 150/90 mm Hg 25-70% 56

Further readingx Williams GJ, Lee A, Craig JC. Long-term antibiotics for preventing

recurrent urinary tract infection in children. Cochrane Database SystRev 2001;4:CD001534

x The diagnosis, treatment, and evaluation of the initial urinary tractinfection in febrile infants and young children: www.guideline.gov/summary/summary.aspx?doc_id = 1838&nbr = 1064&ss = 6(accessed 19 October 2005)

x Sullivan FM, MacNaughton RJ. Evidence used in consultations:interpreted and individualised. Lancet 1996;348:941-3

x Hunt DL, Haynes B, Hanna SE, Smith K. Effects of computer-basedclinical decision support systems on physician performance andpatient outcomes. A systematic review. JAMA 1998;280:1339-46

x Wagner EH. Chronic disease management: What will it take toimprove care for chronic illness? Effective Clin Pract 1998;1:2-4

x NHS Confederation. General Medical Services contractnegotiations: www.nhsconfed.org/gmscontract/ (accessed 19October 2005)

x Perlin JB, Kolodner RM, Roswell RH. The Veterans HealthAdministration: quality, value, accountability, and information astransforming strategies for patient-centered care. Am J Manag Care2004;10:828-36

Frank Sullivan is NHS Tayside professor of research and developmentin general practice and primary care, and Jeremy C Wyatt is professorof health informatics, University of Dundee.

The series will be published as a book by Blackwell Publishing inspring 2006.

Competing interests: None declared.

One hundred years ago

Superstitions about pregnancy and parturition

Dr. Isambert of Tours has collected notes about localsuperstitions still prevalent amongst French countrywomen,notwithstanding the general distribution of education on modernprinciples throughout the Republic from Calais to thePyrenees.. . . Certain curious customs, based on an idea that themale has influence on the fetus beyond the date on which hebegot it—an idea not unknown amongst uncivilized races—stillprevail amongst the French peasantry. Dr. Marignan ofMarsillargues attended a woman who put on her husband’s hat tohasten labour, preferring it to ergot. Dr. Lalanne repeatedlyobserved a similar practice in the Landes, the parturient womanturning her husband’s hat inside out and then putting it on herhead. The idea is carried further in remote districts in Lorraine,the husband’s entire clothing being donned as an oxytocic. In thesame districts, and in parts of the country round Toulouse, themarital cotton night-cap, a familiar object on the head of theFrench rustic, is used as a pad tied against the vulva to prevent athreatened abortion. The gravest of all lying-in-roomsuperstitions receives much attention from Dr. Isambert, for hedwells on the strange tendencies in certain districts to prefer dirty

to clean bed linen. In the canton of Ligueil in Touraine, themother takes the greatest care to collect and soil as muchsheeting as possible when she is nearing term, and after beingdelivered between dirty sheets the supply of foul linen isfrequently changed during the puerperium. In one case ofpuerperal fever Dr. Isambert had the greatest difficulty in gettingthe soiled sheets removed under his own superintendence, andthough his orders were apparently obeyed, he suspects that thedirty linen was replaced in the patient’s bed directly his back wasturned, to be temporarily put aside at his visiting hour. In theDepartment of the Var dirty linen is always used to clean thewoman after delivery, the washing itself not being undertaken forseveral hours. The belief in which these insanitary practicesoriginated remains obscure.. . . In conclusion Dr. Isambertdeclares that some of the methods employed to hasten labour inhumble cots far from the ignoble strife of the madding crowd ofthe boulevards are enough to make the modern obstetrician turnpale, yet somehow fine children are born and bred in those parts.

(BMJ 1905;ii:345)

Clinical review

1074 BMJ VOLUME 331 5 NOVEMBER 2005 bmj.com