Journal of Neurology, Neurosurgery, and Psychiatry 1992;55(Suppl):19-25

Clinical audit and neurology

Anthony Hopkins

AbstractMedical audit has been defined as "asystematic critical analysis of the qualityof medical care, including theprocedures used for diagnosis and treat-ment, the use of resources, and theresulting outcome for the patient". InBritain, recent reforms of the HealthService increase the need for neuro-

logists to undertake audit. The basicprinciples of audit in relation to themanagement ofcommon conditions suchas headache and epilepsy are described.Audit must consider not only efficiencybut also effectiveness, but the difficultyof developing valid outcome measures

should not be underestimated, especiallyin chronic disabling conditions.

Research Unit, RoyalCollege of Physicians,St Andrews Place,London, UKA Hopkins

Medical audit was defined in the White Paper"Working for Patients"' that preceded theNational Health Service and Community CareAct of 1990 as "a systematic critical analysis ofthe quality of medical care, including theprocedures used for diagnosis and treatment,the use of resources, and the resulting out-come for the patient." I believe this to be an

excellent description of the principles ofmedical audit, and has not been bettered inspite of the explosion of interest in audit sincethe White Paper was published early in 1989.

I have drawn attention elsewhere to some

specific words in the White Paper definition,as I believe that they inform our thinkingabout audit.23 Neurologists may object to theword audit, believing that it has overtones ofaccountancy and costs. The etymology of theword, however, is derived from the Latinaudire-to hear, simply because it was

customary to give accounts orally, forexample, in the parable of the five talents.4Neurologists may also object to the term"critical" in the White Paper definition, butthey may relax when they look at MatthewArnold's definition of criticism "a dis-interested endeavour to learn and propagatethe best that is known and thought in theworld".' No physician would quarrel withcriticism in that sense when applied to his or

her professional work. Then there are difficul-ties in defining what is meant exactly by "thequality of medical care", as used in the WhitePaper definition. A quality car usually means a

more expensive car, such as a Rolls Royce or a

Mercedes. However, one outcome of theprocedure for which the expensive car isused-a journey from A to B-is accom-

plished just as well by an inexpensive Fiat

Uno. Quality, as judged by the achievement ofthis successful outcome, is independent ofcost. If, however, the defined outcome ofpossessing a Mercedes is different, forexample, arriving more speedily, or moresafely, or with greater satisfaction, then costdoes become a relevant factor.We are all aware of the constraint of

resources available for health care. Medicalcare of good quality is the best possible careprovided within the resources available. It isnot feasible to arrange for widespreaddiffusion of high technology procedures. Thecare of a patient with an extraduralhaematoma cannot be as good if he is located150 miles from the nearest neurosurgicalcentre as it will be if he lives next door. Whatdetermines the delivery of good quality care inthis context is the provision not only of goodneuroradiological and neurosurgical facilities,but also good access. Attention to means ofaccess has resulted in excellent neurosurgicalservices being provided to the west ofScotland in spite of the considerable distancesinvolved.This example illustrates the point that care

of good quality has a number of dimensions.An American physician, Donabedian, firstidentified the dimensions of quality as struc-ture, process and outcome.6 By structure, hemeant the capital resources available, such as,in the United Kingdom, the monies allocatedfrom the Department of Health to a HealthRegion, the number of hospital beds available,and the number of trained staff. Good processof care has been well defined by Brook andKosecoff,7 "The performance of specificactivities in a manner that either increases orat least prevents the deterioration of healthstatus that would have occurred as a functionof a disease or condition." Such a specificactivity must clearly be effective. Brook andKosecoff go on to identify that quality of careconsists of two components-the selection ofthe right (effective) activity or task, and theperformance of those activities in a mannerthat produces the best outcome.

In a small book already cited,2 I drew uponthe work of Donabedian,6 and also Maxwell8to construct figure 1. As the figure shows, caremust also be not only appropriate and effec-tive, but ethical, equitable, relevant, co-ordinated and socially acceptable to bothpatients and the providers of care. The wholedirection of the process of care is to achieve adefined, favourable outcome, as indicated inthe last box. As described later,' there areparticular difficulties in defining outcomes inneurology.

19 on F

ebruary 21, 2022 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.55.Suppl.19 on 1 M

arch 1992. Dow

nloaded from

Hopkins

Figure 1 Some aspectsworth considering whenmeasuring the quality ofcare. (Reproduced bypermission from reference 2).

STRUCTURE PROCESS

Capital faci lities- Appropri ate,hospitals, doctors' effective, ethical,surgeries, relevant, equitable,Oambulances I coordinated and i. ~~~~~~~socially acceptable

LTrained personnel ca re

SYSTEM PATIENTCHARACTE RISTICS CHARACTE RISTlI

Accessibility (financial,geographical, timely)

In addition to the flow illustrated betweenthe three boxes of the system of care, it mustbe recognised that the system is modified bythe characteristics of the patient-mostnotably the severity of the illness suffered, andby his or her age and co-existing illnesses (co-morbidities). The patient has also to gainaccess to the system. Limitations on access

may be geographical, temporal, financial, ordue to limitations in knowledge of the patientor family doctor or specialist of what effectivecare is available.

I now use the system outlined byDonabedian, filled in by Maxwell, and illus-trated in fig 1 to show how the practice ofneurology may be audited.

The structure of neurological practiceLangton Hewer and Wood have accuratelyrecorded for the first time the numbers ofneurologists in different Health Regions, andthe amount of time (notional half days) thatthey work in the health service.9 They showedthat there are nearly 600 000 people per wholetime equivalent neurologist in Trent, nearlyfour times less than in the North East ThamesRegion. There is therefore a priori evidencethat, through deficiencies of structure (trainedpersonnel), neurological care cannot be as goodfor the whole population of Trent as in theNorth East Thames Region, however hard our

colleagues in Trent work.The natural response of physicians, public

health physicians and neurologists in theunder-provided regions may be to request theprovision of more consultants in order toapproach the population/neurologist ratio inthe better provided regions. However, from thedata of Langton Hewer and Wood it cannot besaid that there is necessarily an under-prov-ision of neurologists in Trent, it could be thatthere is an over-provision in North EastThames. Unless the purposes of the structureare defined exactly (the supply of trainedneurologists and their associated colleaguesand equipment), and the effects of their processof care upon the outcomes of care, no decisioncan be made about whether sufficient capital isemployed. The data suggest that the neurolog-ists in Trent, Wales, and other under-providedregions are hard pressed in looking after acuteneurological illnesses, and new presentation ofserious neurological disease such as multiple

Severity of illness, age,co-morbidity

sclerosis and motor neuron disease. It is un-

likely therefore that they can take a greatinterest in stroke and in rehabilitation, and inthe acute care of patients admitted under theduty general medical firm. If there be a nationalconsensus that neurology should take a muchgreater interest than before in patients withcerebrovascular disease, and in rehabilitationfrom stroke and cranial injury, then even thenumber of neurologists in North East Thamesis probably insufficient to allow this. However,that national consensus should be founded on

good research evidence that the outcomeachieved by an intervention by a neurologist issignificantly greater than that achieved by a

physician with less advanced training.Outcome measures are difficult to determine.

Until the effectiveness of neurological care can

be more clearly demonstrated, a worthwhileaudit is to discover exactly what neurologistsare doing now. Thirteen neurologicalcolleagues in the United Kingdom kindly kepta diary of all their new patient encountersduring one week. Figure 2 shows the number ofnew patient encounters made by theseneurologists in one week in 1986.10 Colleagueswere asked to record all new encounterswhether they were National Health Service(NHS) outpatients, private patients, or wardreferrals. Bearing in mind the difficulty ofkeeping such a log in a day to day practice, it isprobable that the numbers considerablyunderestimate the true total. The range ofvalues is striking (33-144; mean 74; median79). However, such a simple study illustratesthe weakness of simple performance indicatorsbased upon activity, which have been the onlyavailable audits nationally within the HealthService during the past four decades. Withoutbeing aware of the individual backgrounds tothe raw data presented in fig 2, it cannot bestated that the most active neurologist is neces-sarily working harder in the Health Service. Asit happened, inspection of the data showed thatthe most "active" neurologist was very active inprivate practice, and the least "active"neurologist was a distinguished professor ofneurology who was contributing extensively tobiological and neurological research. Thesecond least "active" neurologist had beenappointed only a few weeks before the survey,and had not built up his pattern of work to thatwhich he is now undertaking. Such activity

Mortality, perceived health

Morbidityhealth status

Defined end pointsPatient satisfaction

OUTCOME

ICS

20 on F

ebruary 21, 2022 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.55.Suppl.19 on 1 M

arch 1992. Dow

nloaded from

Clinical audit and neurology

analyses are useless unless they take account ofthe complexity ofthe work undertaken, and thecontributions made by participants to otheraspects ofmedicine. Corrections for such crudedata could be inserted into fig 2 to take intoaccount the varying number of notional half-days, for example.Menken, Dr Friese and I asked our

colleagues for a provisional diagnosis made atthe time of the first new patient encounter, andthese diagnoses were subsequently coded(fig 3). No analysis was made of the type ofwork undertaken by any individual neurolog-ist, as the numbers were too small. However,when the diagnoses were pooled, our findingsagreed with previous reports,'2 that headaches,migraine, and "funny turns" (that may or maynot be epilepsy), are the commonest diagnosticgroupings for first presentation to a neurolog-ist. More important, and more difficult is totake into account the complexity and severity ofillness-and these variables are not the same.For a patient who first attends outpatients witha slowly progressive hemiplegia, and who hasmild papilloedema on examination, the correctoutpatient management is to take an immediatedecision to arrange a chest radiograph andcranial CT Scan or MRI and review the patientlater that day. All of us who are experienced inclinical practice know that such decisions aboutsevere disorders can be taken within five min-utes. However, a patient with recurrentmigraine who has many questions to ask, andwho wishes to explore dietary advice andreview his or her medication may well takeconsiderably longer to manage. That is to say,the time taken by the activities which aneurologist undertakes, and the intensity ofthose activities are not readily revealed bysimple diagnostic coding. In resource man-agement terms, one needs to distinguish case-mix groups which consume the same resources("iso-resource") from "iso-diagnosis" andfrom "iso-severity". Neurologists musttherefore be very chary of any attempts toimpose any crude analysis of activity as an auditmeasure of their work. It is not uncommon inthe United States for large private clinics to laydown contracts with their neurological staff onhow many patients they should "see" duringthe course of a year, and continuedemployment and promotion in the privatesector depends upon such activity. This issimplistic, indeed foolish.We need to be cautious about advising the

Figure 2 Distribution ofnumbers of new patientsseen by 13 neurologists inone week in 1986.(Reproduced by permissionfrom reference '°).

A record of patient encountersin neurological practicein the United Kingdom

50

40

.0 30E

z 20

10

0HFT hnUn-nn

Department of Health that more neurologistsare needed if, by creating a further supply, weincrease the demand for referrals for headachesand migraine, which might be better lookedafter in primary care practice. It would beproper to make such a demand if it could beshown that a neurological consultation had amajor impact upon the natural history ofsomeone with headaches and migraine, a topicto which I shall return later, when consideringthe measurement of outcomes.With reference to fig 1, I have so far

considered in detail only the structure in termsof the number of neurologists. The next aspectto consider is the access of patients to thatstructure. Until 1 April 1991, access in theUnited Kingdom had not been limited byfinancial constraints. That is to say, any familydoctor could refer patients to a neurologist if aneurological opinion was necessary. In the1960s and 1970s, it was not unusual to try andrestrict such access by running what wastermed "closed clinics". It was thought thatneurological time was so precious that patientswith neurological complaints should first bereviewed by a general hospital physician(specialist in internal medicine), and thenreferred on to a neurologist only if thisphysician thought it to be appropriate. Suchclinics have now almost entirely disappeared,probably due to better postgraduate educationof general practitioners and growing trustbetween hospital physicians and neurologistsand family doctors. The full implications ofwhat may happen under the National HealthService and Community Care Act of 1990 havenot yet been realised, but some scenarios mayoccur. Under the Act, certain large generalpractices have been allocated their own budgetsfrom which the care of their patients can bepurchased. Towards the end of the financialyear, it is probable that such budgets will runlow, and a family doctor will then have to makea decision as to the "worth" of a neurologicalreferral for a patient with headache or migraineor funny turn, compared to the "worth" ofsending another patient for a badly needed hipreplacement.Within the hospital community, the creation

of new neurological consultant posts, or thereplacement of those neurologists who are dueto retire, may depend upon the family doctor'scontinuing awareness of their value to thequality of the patients' lives. It is essentialtherefore that neurologists set in place work toestablish the "value-added" by their work.Individual neurological departments will alsobecome increasingly financially dependentupon providing a specialist service that DistrictHealth Authorities other than their own wish topurchase. For example, some neurologicaldepartments with the necessary expertise maywell find it worth their while setting up special-ised epilepsy clinics, and marketing theirservices with some sort of special deal. Suchdeals may involve appropriate imagingprocedures and video monitoring for patientswith intractable epilepsy. Neurological depart-ments without specialised expertise may findthemselves attracting few patients outside the

21

on February 21, 2022 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.55.S

uppl.19 on 1 March 1992. D

ownloaded from

Hopkins

Diagnosis andICD codes

307-Headache, migraine 346-

784-0

Epilepsy 345

Peripheral nerve disorders(includes Bell's palsy, 350-trigeminal neuralgia and 357entrapments)

Cerebrovascular disease 433-457

Multiple sclerosis 340

298Anxiety, depression and 300personality disorders 301

311

Spine and disc problems 720-724

Giddiness, fatigue

Trauma

Parkinson's disease,extrapyramidal syndrcmes

Myasthenia, myopathy

Neoplasia

Motor neurone disease

Syncope

Dementia

780-4780-7800-806850-854

332-333

358-359191225-237-5

335-1335-2

780-2

290

All other codes

Percentage of 411 new patient consultations by 13United Kingdom neurologists in one week in autumn 1986

0 5 10

Figure 3 Distribution of diagnoses of 411 new patients seen by 13 neurologists in oneweek in 1986. (Reproduced by permission from reference 10).

geographical area of their own District HealthAuthority.

Apart from the probable limitations onaccess for financial reasons to neurologists as aresult of the National Health Service andCommunity Care Act of 1990, and the likelychanges that the Act will have upon some

neurological departments, there remains therunning problem of geographical access. Is itbetter (that is, more value added to outcome) tohave neurologists congregated in regionalcentres, together with neurosurgeons, neuro-

radiologists, neurophysiologists, neuro-

pathologists, physiotherapists, and othercapital intensive resources, or is it better tohave a neurologist primarily in a district, with a

less intense relationship with the high tech-nology centre? I shall not discuss thearguments for and against each system of care,as I have reviewed the development of thepattern of neurological services in the UnitedKingdom elsewhere.'3 14 My impression is thatover the last ten to fifteen years a reasonablecompromise has been reached between the twopossible ways of providing neurological care,and the National Health Service and Commun-ity Care Act of 1990 may de-stabilise thissituation through the mechanisms suggestedabove.

Ifpatients are offered high technology care ofproven efficacy at some distance from theirhomes, they may well choose to opt for the lessefficacious but local care. This has arisen inrelation to radiotherapy services for cancer in

the North East Thames Region, in which atleast one local population pressed for the reten-tion of their technically less well equipped localtreatment centre, rather than make the journeyto a linear accelerator in central London.As for temporal access, most hospitals run

such audits as part ofmanagement. The timingof available appointments in outpatientsshould, as far as possible, reflect the reality ofthe neurologist's time of arrival and their othercommitments, in order to minimise thepatient's waiting time. However, an audit ofthedays or weeks spent by new patients waiting tosee a neurologist after their referral from theirfamily doctor is another matter. Personalexperience suggests that if strenuous efforts aremade to reduce the mean waiting interval, thenmore neurological referrals from familypractice may be made, so that the waiting listextends again, and the neurologist is morepressed by even larger numbers of referrals.This may be no bad thing (apart from possiblydestroying the health, sanity and marriage ofneurologists!) but returning to the point madepreviously-a neurological intervention mustadd value over and above the care of a primaryfamily practitioner to make the referralappropriate.There is another problem related to access to

care which I term "covert inaccessibility." Bythis, I mean that a referral to a neurologist is notmade because a family practitioner or otherphysician chooses not to do so, believing thatthe best has already been done for his or herpatient. A common example is older patientswith Parkinson's disease who can be signifi-cantly further helped by the appropriateprescription of levodopa, a medication whichmany family doctors still seem anxious aboutusing. Or a patient after a stroke may not beprovided with a simple aid which a neurologistcould instantly have recommended as beingeffective simply because the family doctormade the decision that "nothing could bedone". Covert inaccessibility by its very naturecannot easily be measured, but a notableexample in another field is that of Hampton etal'5 who found that in primary care practicethere were patients who would have benefittedfrom cardiological intervention and had notbeen referred.The central box in fig 1 refers to the process

of care, which must be appropriate, effective,ethical, relevant, equitable, coordinated andsocially acceptable. I have explained themeaning of these terms elsewhere2 and withparticular reference to epilepsy,'6 but here Iwrite specifically about coordinated care andeffective care.Those with a chronic illness, including many

neurological disorders, need to obtain healthcare from a variety of providers. For example,someone who is paraplegic and incontinentwith multiple sclerosis will need help from thelocal housing authority, the Department ofSocial Security, neighbours, meals-on-wheels,community nurse, community physio-therapist, local incontinence adviser, familypractitioner, and neurologist. I have writtenelsewhere of the difficulties that arise in co-

22 on F

ebruary 21, 2022 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.55.Suppl.19 on 1 M

arch 1992. Dow

nloaded from

23Clinical audit and neurology

ordinating such activities,'7 but the aim of theprovision of our services should always be"seamless care." That is to say, the patientshould not be aware of the administrativeaspects of coordinating the various providers ofcare, but rather be the recipient of coordinatedpatient-centred care. Often, there are unfor-tunate stories of failures of communicationbetween the various providers of care. Thenthe patient, or their family, has to take on therole of attempting to integrate the team ofproviders. A useful example of audit in thisfield is the provision of services to patients withmultiple sclerosis in Glasgow by Elian andDean.'8 A neurological department could wellaudit a group of patients with a common

disorder or disability in a similar way. Such a

small scale project would be suitable for a

medical student to undertake under super-

vision.

An effective intervention is one that achievesan outcome that significantly enhances thequality of the patient's life, or at least preventsthe deterioration in quality that would haveotherwise occurred. Without a defined out-come, it is impracticable to measure effective-ness. Before moving to a discussion of out-comes, it is important to distinguish betweenefficacy, effectiveness and efficiency. Efficacy isthe ability of a medical or surgical interventionto produce the desired outcome in a definedpopulation under ideal conditions. It is what ismeasured in randomised controlled trials.However, an efficacious procedure may well beless effective when introduced into everydaylife, simply because the profession uses theefficacious treatment in a different patientpopulation (for example, older than those in thetrials, or with co-morbidities) and because theavailability of medical skills is not necessarilythe same in everyday practice as is available inthose centres which commonly undertaketrials. Also the follow up of patients in trialsmay increase compliance with the interventionunder study, and compliance in everyday lifemay be so much less that the procedure is lesseffective than the efficacy suggested in the trial.

Efficiency is the extent to which resources are

consumed by the interventions relative to theireffectiveness. If two procedures are equallyeffective, one should choose the more efficient(cheaper) one.The United Kingdom led the world in the

design of randomised controlled trials, and inthe proof of the efficacy of various inter-

ventions. Unfortunately, the lack of any infor-mation-gathering strategy about outcomes inthe Health Service has not resulted in usefulinformation about the effectiveness of many

procedures that we undertake. The data avail-able on MediCare magnetic tapes in the UnitedStates means that, for this client population,there is an increasing body ofknowledge aboutthe outcomes after various interventions. Forexample, Wennberg et al'920 have shown thatafter prostatectomy, the 30 day mortality rate,

the incontinence rate, and the impotence rate

are all very much higher than that suggestedfrom trials emanating from tertiary centres

specialising in urology. In general, we are not

so concerned with surgical interventions of acurative sort in neurology, (although the effec-tiveness of carotid endarterectomy has recentlybeen well studied.2' Neurology lacks basicstudies of the efficacy of a number ofcommonlyused interventions, particularly those relatingto therapies used in rehabilitation, althoughthere are some notable exceptions.2223How can we measure the outcomes of

neurological care, and therefore chooseour effective procedures? Acute disorders suchas meningitis or cerebral abscess arecomparatively rare in everyday neurologicalpractice, although there remains concern thatacute infective disorders are not as wellmanaged as they might be, possibly because oftheir unfamiliarity.24 Mortality is clearly aninappropriate outcome measure for many of thedisorders with which we deal, such as multiplesclerosis, epilepsy, motor neuron disease andParkinson's disease. It would be difficult to usecrude death rates among those attending aneurological clinic with these disorders, orwithin a Health District or Region as a measureof the neurological care provided, because ofthe many confounding variables. However, theexample of epilepsy can be used to illustratehow an unexpected death can be used as ameasure of medical audit. Status epilepticus isa potentially remediable condition. I believethat every neurological team should review indetail all such deaths to improve the system ofcare. Such an audit in my own hospital someyears ago suggested a lack of readiness on thepart of the general physicians on admittingduty to appreciate the potential seriousness ofrepeated convulsions, resulting in a delayedreferral to a neurologist, and a delayed institu-tion of appropriate therapy in an intensive careunit.

Epilepsy can also be used to illustrate thepotential use of another type of adverse out-come - intoxication with drugs. Intoxicationwith phenytoin and carbamazepine is now rare,but certainly the records of any such casesshould be reviewed by a neurological team toassess if there had been avoidable errors inadvice about dosage, or in communication withthe family practitioner.Another adverse outcome is the development

of a pressure sore, now fortunately uncommonin people with neurological disease, but suchan outcome should trigger a review of wardpractice.

If health status is measured before and afteran intervention, then any change in healthstatus becomes an outcome measure. A wellknown example is the Kurtzke scale which iswidely used internationally in trials ofinterventions in multiple sclerosis. This is areliable and valid scale which measures changesin functional performance, and therefore theefficacy of an intervention, but it does notmeasure the quality of other neurologicalinterventions in multiple sclerosis. These aremore difficult to define and relate to support andencouragement and to the facilitation of auton-omy in patients with this progressive anddisabling illness. Such aspects of care areexceedingly difficult to measure, but in

on February 21, 2022 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.55.S

uppl.19 on 1 March 1992. D

ownloaded from

Hopkins

our type of practice are probably especiallyimportant.

It might be thought that it would be easy tomeasure outcome in epilepsy, to continue ourexample. However, the following illustratesvarious difficulties. At an individual level, allthe patient is concerned about is that he or shebecomes free of seizures, but if the intention isto audit the quality of care of a neurologist'sintervention upon his population of patients,then the severity of epilepsy with which hedeals becomes a pre-eminent factor. For exam-ple, the patients of colleagues working inspecial centres for those with intractableepilepsy would have very poor outcomes ifseizure frequency was taken as a measure oftheir neurological performance. Also, at anindividual case level, it is often hard to under-stand the meaning of changes in seizurefrequency, which sometimes appear randomand unrelated to changes in anticonvulsantmedication.25 It may be that patients referred tospecial epilepsy centres for monitoring couldprovide an audit of the quality of diagnosis inepilepsy, bearing in mind that approximately10-15% of patients with intractable (epilepsy)turn out to have pseudoseizures on suchreview. It is, however, the neurologists who aremost concerned about the quality of their carethat are most likely to refer patients to specialcentres for long-term monitoring. The lesssuccessful neurologist who may wrongly diag-nose epilepsy for pseudoseizures frequentlymay be supposed to fail to refer.Then there is the point that it is possible to

identify a number of outcomes for the samedisorder. A neurologist may do well on someoutcome measures and less well on others. Iillustrate this point with the work which Icarried out with Fitzpatrick on headache andmigraine about ten years ago.2128 The outcomemeasures that we used included the relief ofheadaches, using the dimensions of severityand frequency, the satisfaction of patients withtheir neurological consultation, the degree ofreassurance about organic illness, and thechange in frequency of consultation with theirfamily doctor one year following theneurological consultation. Some of theseoutcome measures were inter-linked. Forexample, 60% of the 109 patients when firstinterviewed expressed some anxiety about thepossibility that their headaches were due to aserious organic illness such as a brain tumour oran impending stroke. A further three weeksafter the examination showed that of those whohad been concerned, 60% had been completelyreassured by their consultation, at least on thisaspect, and a further 28% had been partlyreassured. Twelve per cent remained con-cerned, and this unfavourable outcome waslinked to another-those who were critical ofthe information received during the consulta-tion were less likely to be reassured(p < 0-01).26 Here there is an inextricable linkbetween properties of the patient, process andoutcome. Were the patients critical of informa-tion received because ofpoor information givenor did they remain unreassured due to somepersonal reason? Such an audit may encourage

us to pay more attention to communicatinginformation, but a failure to do so cannot belaid entirely at a neurologist's feet, any morethan can a failure of an oncologist to cure apatient with acute myeloid leukaemia. Fitz-patrick and I showed that patients who aredissatisfied with their neurological consultationhad a clear cluster of characteristics: inparticular, women who had initiated thereferral themselves, were rated anxious ordepressed, who had a clear view that theirheadache was migrainous in nature, and hadexperienced headaches for more than oneyear.2628A year later we followed up our sample of

patients with headache (we achieved interviewsin only 75 ofthem) and found that patients whohad been satisfied with their neurologicalconsultation were much more likely to rate theseverity of their headaches as improved.27However, only 17 patients enjoyed completefreedom from headaches. What was parti-cularly striking was that in spite of the continu-ing headaches (albeit with less severity and lessfrequently), the great majority decreased thenumber of visits to their general practitioner.As for reassurance one year later, nearly all(86%) attributed their reassurance to thehospital referral, whereas only 16% felt that thehospital doctors contributed to an understand-ing of their headaches, and even fewer (8%) feltthat they had learned anything from the doctorthat had allowed them to avoid or to managetheir headaches better.27 This range ofoutcomemeasures shows the diversity of dimensionsthat need to be considered in medical practice.Hospital information systems are not going togive us these data on a routine basis.

ConclusionHow can a neurologist begin to audit in thespecialty? Bearing in mind the importance ofoutpatient (ambulatory) practice in neurology,I believe that it is appropriate to begin there.All new outpatient visits should receive at leasta diagnostic code so that a department is awareof the range and quantity of its work. Inconsultation with the local family doctors, anddrawing upon the evidence of publishedresearch on suitable investigations in variousdisorders, the neurological department shouldlay down policies for the management andinvestigation of some of the more commondisorders, such as, headaches, migraine, andnew cases of epilepsy. Without undertakingformal research studies of the type describedabove for headache,2128 it would be worthinstituting local well designed surveys of theviews of the patients on the help that they hadreceived from a neurological consultation, also,at a more technical level, assessments could bemade on the outcome of therapy that theneurologist had suggested, for example, theprovision of a short course of physiotherapy torelease spasticity.The Research Unit of the Royal College of

Physicians is developing audit measures for theprocess of care of a number of common disor-ders.2' At an international level, researchendeavours must continue to find treatments

24 on F

ebruary 21, 2022 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.55.Suppl.19 on 1 M

arch 1992. Dow

nloaded from

Clinical audit and neurology

that are efficacious for our principal disorders.Where efficacious and effective treatmentsalready exist, such as in Parkinson's disease, itwould be helpful if the internationalneurological community could agree uponsome simple scoring system that would identifythe various dimensions of a patient's disability,such as rigidity, tremor and dyskinesia. Theonly advantage in collecting such informationon a routine basis would be to enable theselection of a group of patients for particularreview, for example, patients under 60 yearswith Parkinson's disease and severe dyskinesiamight be reviewed at a clinic to discover if allthe available therapeutic options had beenconsidered. There is absolutely no point inscoring patients on a routine basis for any otherpurpose, unless such scores are part of aparticular research project.The institution of medical audit is largely an

act of faith. By directing attention to themeasurement of some more accessible items ofprocess of care, and of outcome, it is hoped andbelieved that doctors will become more attunedto evaluating critically the quality oftheir work,so that aspects which are less easily measured,such as the facilitation of patients' autonomy,are also improved. The danger, however, is thatthe mechanical introduction ofnumerical auditmeasures may irritate and distract hard-pressed clinicians from time that could beusefully spent with patients. Neurologists mayalso rightly say that the quality of their care islargely restrained by the underprovision ofresources. To take a simple example, allneurologists would agree with the principlethat a family doctor should receive a dischargesummary within one week. Neurologists wouldgladly adhere to that principle if the secretarialresource was made available. It is importantnot to be distracted by the underprovision ofresources, of which this is a simple buteveryday example, and concentrate upondelivering the best quality care within the areawhich is within one's own personal profes-sional control, and use epidemiologically soundmethods of measurement of that quality.

1 Working for patients. London: HMSO, 1989.2 Hopkins A. Measuring the quality of medical care. London:

RCP Publications, 1990.

3 Hopkins A. Approaches to medical audit. JEpi Comm Health1991;45:1-3.

4 Matthew Ch 25:v19-30.5 Arnold M. Essays in criticism. London, 1865.6 Donabedian A. Evaluating the quality of medical care.

Millbank Memorial Fund Quarterly 1966;44(Suppl):166-206.

7 Brook RH, Kosecoff JB. Commentary: competition andquality. Health Affairs 1988;7: 150-61.

8 Maxwell RJ. Quality assessment in health. BMJ 1984;288:1470-2.

9 Langton Hewer R, Wood VA. A report on neurologicalservices in the United Kingdom-number and distribution ofconsultants in adult neurology-number and distribution ofCT head scanners. London: ABN, 1988.

10 Hopkins A, Menken M, DeFriese G. A record of patientencounters in neurological practice in the UnitedKingdom. J Neurol, Neurosurg Psychiatry 1989;52:436-8.

11 Perkin GD. Pattern of neurological outpatient practice:implications for undergraduate and postgraduate training.J Soc Med 1986;79:655-7.

12 Stevens DL. Neurology in Gloucestershire: the clinicalworkload of an English neurologist. J Neurol, NeurosurgPsychiatry 1989;52:439-46.

13 Hopkins A. The organisation of neurological services in theUnited Kingdom (Abstract). Proc joint meeting ABN andSpanish Society of Neurology, Spain, June 1990.

14 Hopkins A. Different types of neurologist. BMJ 1984;288:1733-6.

15 Yi J-j, Rowley JM, Hampton JR. Appropriate use ofinvestigations in cardiology. In: Hopkins A, ed.Appropriate investigations and treatment in clinicalpractice.London: RCP, 1989:31-9.

16 Hopkins A. Audit of the medical care of people withepilepsy. In: Chadwick D, ed. Quality of life and quality ofcare in epilepsy. London Roy Soc Med 1990;23:40-9.

17 Hopkins A. Practical help. Lancet 1984;1:1393-6.18 Elian M, Dean G. Need for and use of social and health

services by multiple sclerosis patients living at home inEngland and Wales. Lancet 1983;1:1091-4.

19 Wennberg JE, Roos N, Sola L, et al. Use of claims datasystems to evaluate health care outcomes: Mortality andreoperation following prostatectomy. J Am Med Assoc1987;257:933-6.

20 Fowler FJ, Wennberg JE, Timothy RP, et al. Symptomstatus and quality of life following prostatectomy. J AmMed Assoc 1988;259:3018-22.

21 MRC European carotid surgery trial: interim results forsymptomatic patients with severe (70-99%) or with mild(0-29%) carotid stenosis. Lancet 1991;337:1235-43.

22 Smith DS, Goldenberg E, Ashburn A, et al. Remedialtherapy after stroke: a randomised controlled trial. BMJ1981;282:517-20.

23 David R, Enderby P, Bainton D. Treatment of acquiredaphasia: speech therapists and volunteers compared. JNeurol Neurosurg Psychiatry 1982;45:957-61.

24 Sparrow OC. The importance of early detection ofintracranial suppuration. J Roy Soc Med 1991;84:187-9.

-25 Hopkins A, Davies P, Dobson C. Mathematical modelling ofpatterns of intervals between seizures. Arch Neurol1985;42:463-7.

26 Fitzpatrick RM, Hopkins A. Referrals to neurologists forheadaches not due to structural disease. J Neurol,Neurosurg Psychiatry 1981;44:1061-7.

27 Fitzpatrick RM, Hopkins A. Illness behaviour and head-ache, and the sociology of consultations for headache. In:Hopkins A, ed. Headache: problems in diagnosis andmanagement. London: WB Saunders, 1988:349-85.

28 Fitzpatrick RM, Hopkins A. Patients' satisfaction withcommunication in neurological outpatient clinics. Journalof Psychosomatic Research 1981;25:329-34.

29 Report of a Working Group of the Royal College ofPhysicians of London. Standards of care for patients withneurological disease. Royal College of Physicians 1990;24:90-4.

25 on F

ebruary 21, 2022 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.55.Suppl.19 on 1 M

arch 1992. Dow

nloaded from