Clinical reviewABC of health informaticsHow decision support tools help define clinical problemsFrank Sullivan, Jeremy C W...
Clinical reviewon (according to Feinstein) “the judgements of thoughtful                                                  ...
Clinical reviewfrom genomics, proteomics, and metabonomics. These newmodes of inquiry about patients’ underlying genetic s...
Upcoming SlideShare
Loading in...5

How decision support tools help define clinical problems


Published on

Published in: Health & Medicine
  • Be the first to comment

  • Be the first to like this

No Downloads
Total Views
On Slideshare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

How decision support tools help define clinical problems

  1. 1. Clinical reviewABC of health informaticsHow decision support tools help define clinical problemsFrank Sullivan, Jeremy C WyattThe patient, Mr Evans, presented with headaches and early This is the fourth in a series of 12 articlesmorning wakening (dealt with in article of 24 September) as the A glossary of terms is available at reason for his consultation. This article, however, discusses content/full/331/7516/566/DC1how informatics resources can be used to consider issues otherthan the presenting problem. The Stott and Davies model ofthe consultation indicates that three other areas of the Mr Evans is a 49 year old, recently unemployed,interaction should be considered. pharmaceutical company representative who hasx Management of continuing problems—The patient’s diabetes presented with low mood, poor appetite, and sleepmay be contributing to the overall picture. disturbance. He drinks two bottles of whisky per week,x Opportunistic health promotion—Ask screening questions but he does not volunteer this information initially. Heabout alcohol use and measure the patient’s blood pressure. has type 2 diabetes. A blood pressure check showsx Modification of help seeking behaviours—Discuss issues 178/114 mm Hg, and Mr Evans is asked to return to the practice nurse for follow-uprelevant to self care and when to attend for health checks forestablished or potential problems. Management of Modification of helpManagement of continuing problems presenting problems seeking behavioursAwareness of problems Management of Opportunistic continuing problems health promotionSometimes doctors and patients are not aware of relevantproblems. Issues that are apparent to one person may not beapparent to the other. In Mr Evans’s case the diabetes is known Exceptional potential in each consultation. Adapted from Stott NCH, Davies RH. J Roy Coll Gen Pract 1979;29:201-5to doctor and patient. The alcohol problem is, perhaps dimly,apparent to Mr Evans. The high blood pressure reading issomething that only the doctor is aware of initially. Neither Known to Not known patient to patientdoctor nor patient is aware of the depression at the beginningof the consultation, but information conveyed before, or during, Known tothe consultation may alter that. Open Blind clinician When a health professional realises that he or she is awareof an issue that the patient is not, the matter can be remedied. It Not known Hidden Unknown to clinicianis more difficult if the patient is aware of an issue that isrelevant, but is unwilling to divulge it. Even more difficult is asituation where neither patient nor doctor is aware of a The Johari Window shows situations where one or both individuals in the consultation may not be aware of all the relevant informationproblem that may be relevant to the patient’s problems (seeJohari Window). Electronic prompts to bring up such hidden Baseline Odds ratio Post-exposureissues are being incorporated into clinical systems, and are probability probabilityincreasingly effective. 0.01 0.99Problems underlying depression 0.02 0.98Depression is common and often associated with anxiety, 0.03 1000 0.97 500cognitive impairment, and substance misuse. 0.05 0.95 0.07 0.93 It is important to detect alcohol misuse because failure to do 100 0.1 0.9so may mean that treatment for the presenting problem is 50ineffective. Several screening tools with different characteristics 0.2 0.8 10for various clinical settings are available. When the CAGE 0.3 5 0.7questionnaire is used on its own in primary care, a positive 0.4 0.6response to two or more items on it has a sensitivity of 93% and 0.5 1 0.5 0.5a specificity of 76%. Different questionnaire screening tests for 0.6 0.4 0.7 0.3alcohol misuse, such as the fast alcohol screening test (FAST), 0.1may detect problems at an early stage, when intervention may 0.8 0.05 0.2be more effective than later on. Other clues can help the doctor, 0.9 0.01 0.1including comments from family members and the nature of 0.01 0.07 0.005past consultations—for example, injuries that were only partially 0.93 0.05explained. 0.95 0.001 0.03 When the baseline probability of a condition and the odds 0.97 0.02ratio of a modifying factor are known, then the effect of any 0.99 0.01new information can be calculated by using Bayes’ nomogram.Unfortunately, key items of information needed for such When estimates of the odds ratio and baseline probabilitycalculations are often unavailable. For the foreseeable future, are known, Bayes’ nomogram can be used to calculateinterpreting the results of most investigations still relies heavily post-exposure probabilityBMJ VOLUME 331 8 OCTOBER 2005 831
  2. 2. Clinical reviewon (according to Feinstein) “the judgements of thoughtful Highest scoring diabetes indicators in UK GP Quality andpeople who are familiar with the total realities of human Outcomes Framework 2004*ailments.” Apart from depression, there are other situations in which Maximumharmful alcohol use may be important, and an electronic alert Indicator Points thresholdmay be useful in a consultation. Although it is not the main Recordsreason for consulting, Mr Evans also has type 2 diabetes and Practice can produce a register of all 6today’s consultation is an opportunity to deliver proactive care. patients with diabetes mellitus Ongoing managementProblems complicating diabetes Percentage of patients with diabetes in 16 50%The microvascular and macrovascular complications of type 2 whom the last HbA1c is 7.4 or less (ordiabetes need to be monitored regularly. Guidelines are equivalent test/reference rangeincorporated into local clinical governance structures to ensure depending on local laboratory) in pastthat all necessary care is given to patients. Organisations are 15 monthsresponsible for providing different elements of the care Percentage of patients with diabetes in 11 85%recorded in electronic patient records. Integrated services (such whom the last HbA1c is 10 or less (oras health maintenance organisations or the managed clinical equivalent test or reference range depending on local laboratory) in pastnetwork) share responsibilities, using electronic health records 15 monthsacross primary, secondary, and tertiary care. Percentage of patients with diabetes 5 90% Mr Evans’s only abnormal physical test result is a blood who have a record of retinal screeningpressure of 178/114 mm Hg. The raised blood pressure is in the previous 15 monthspotentially important, and the practice’s decision support Percentage of patients with diabetes in 17 55%software gives advice on what to do next. Most of the advice on whom the last blood pressure is 145/85checking for secondary causes of hypertension (such as mm Hg or lessexcessive alcohol ingestion and end organ damage) is familiar Percentage of patients with diabetes 6 60%to the doctor, as is the advice to repeat the examination on whose last measured total cholesterolseveral occasions before starting treatment. Grade 1 evidence within previous 15 months is 5 or lessfrom meta-analyses or large randomised controlled trials may *In total, 1050 quality points are available, of which 550 points are for clinical targets. The most important areas are coronary heart disease, hypertension, andbe available for straightforward clinical problems (for example, diabetes, which account for 325 (59%) of the 550 points for clinical indicatorsstarting antihypertensive drugs), but this is not always the case. Clinical decision support tools are being refined to providethe information that clinicians require without overloadingthem with unnecessary data. This is difficult as the amount of Revised grading system for recommendations in evidenceinformation needed and the sources from which information is based guidelines*obtained varies. Levels of evidence 1++ High quality meta-analyses, systematic reviews of RCTs, or RCTs with a very low risk of biasGuidelines 1+ Well conducted meta-analyses, systematic reviews of RCTs, or RCTs with a low risk of biasField and Lohr describe clinical practice guidelines as 1– Meta-analyses, systematic reviews or RCTs, or RCTs with a high“systematically developed statements to assist practitioners and risk of bias 2++ High quality systematic reviews of case-control or cohort studiespatient decisions about appropriate health care for specific or high quality case-control or cohort studies with a very low risk ofcircumstances.” One role of guidelines is to ensure that all confounding, bias, or chance and a high probability that therelevant issues are dealt with during clinical encounters. relationship is causalIndividual guideline organisations have their own websites and 2+ Well conducted case-control or cohort studies with a low risk ofother organisations, such as the Turning Research into Practice confounding, bias, or chance and a moderate probability that the(TRIP) database, integrate several guideline sources and other relationship is causalevidence based resources. 2– Case-control or cohort studies with a high risk of confounding, bias, or chance and a significant risk that the relationship is not Computerised guidelines provide evidence based causalrecommendations for, and can automatically generate 3 Non-analytic studies—for example, case reports, case seriesrecommendations about, the screening, diagnostic, or 4 Expert opiniontherapeutic activities that are suggested for a specific patient. Grades of recommendationsThe advantages of computerised guidelines over written A At least one meta-analysis, systematic review, or RCT rated as 1++guidelines are that they: and directly applicable to the target population orx Provide readily accessible references and allow access to A systematic review of RCTs or a body of evidence consistingknowledge in guidelines that have been selected for use in a principally of studies rated as 1+ directly applicable to the target population and demonstrating overall consistency of resultsspecific clinical context B A body of evidence including studies rated as 2++ directlyx Show errors or anachronisms in the content of a guideline applicable to the target population and demonstrating overallx Often improve the clarity of a guideline consistency of results orx Can be tailored to a patient’s clinical state Extrapolated evidence from studies rated as 1++ or 1+x Propose timely decision support that is specific for the patient C A body of evidence including studies rated as 2+ directly applicablex Send reminders. to the target population and demonstrating overall consistency of results or Extrapolated evidence from studies rated as 2++Knowledge from unfamiliar sources D Evidence level 3 or 4 or Extrapolated evidence from studies rated as 2+In the post-genomic world, clinicians will have to integrate their *Guidelines of the Scottish Intercollegiate Guidelines Network Grading Reviewunderstanding of patients’ phenotype with new information Group. RCT = randomised controlled trial832 BMJ VOLUME 331 8 OCTOBER 2005
  3. 3. Clinical reviewfrom genomics, proteomics, and metabonomics. These newmodes of inquiry about patients’ underlying genetic status helpto explain older, empirical observations. For example, therelative ineffectiveness of aspirin in preventingthromboembolic disorders in 25% of the population may becaused by several common gene variants that affect plateletglycoprotein function. The challenge to clinicians is to integratethis new knowledge into their diagnostic and therapeuticapproaches during consultations.Modifying help seeking behavioursSome patients with long term health problems do not attendreview appointments. This is a particular problem when theindividual has multiple comorbidities. A patient with depressionmay not think it is worthwhile spending scarce health serviceresources on themselves because they have low self esteem,which is often associated with depression. Electronic patientrecords summarise health problems and, potentially, promptwhen reviews have not been undertaken. Some services, likereview of the patient’s self monitoring, can be provided Definitionsimmediately. Others, such as retinopathy screening, may have tobe scheduled for another date and place. An electronic health x Electronic patient record—Records the periodic care providedrecord shared between colleagues in different professions and mainly by one institution. Typically, this information will relate to the health care given to a patient by an acute hospitalparts of the health services makes scheduling easier. x Electronic health record—A longitudinal record of patients’ health and health care: from cradle to grave. It combines the information about patient contacts with primary health care as well as subsets ofElectronic clinical information information associated with the outcomes of periodic care held insystems the electronic patient recordThe principal function of electronic clinical informationsystems is to facilitate patient care. This involves identifying,classifying, understanding, and resolving problems to thesatisfaction of the patients. Clinical records are also required torecall observations, to inform others, to instruct students, togain knowledge, to monitor performance, and to justify Further readingintervention. Electronic clinical information systems are x Brief description of Johari Window on practice:becoming integral components of healthcare services, and in industrialised countries they are replacing the established x Bodenheimer T, Grumbach K. Electronic technology: a spark topaper based system of records. Combining the electronic revitalize primary care? JAMA. 2003;290:259-64 x McCusker, MT, Basquille J, Khwaja M, Murray-Lyon IM, Catalan J.patient records of different organisations creates a single Hazardous and harmful drinking: a comparison of the AUDIT andelectronic health record. The challenge for many health services CAGE screening questionnaires. Quart J Med 2002;95:591-5is to provide “cradle to grave” information. Effective integration x Page J, Attia J. Using Bayes’ nomogram to help interpret oddsof records depends on establishing a workable unique patient ratios. Evidence Based Med 2003;8:132-4identification system such as the community health index. x Feinstein AR. The need for humanised science in evaluating medication. Lancet 1972;297:421-3 x Field MJ, Lohr KN, eds. Guidelines for clinical practice: fromSummary development to use. Washington DC: National Academy Press, 1992 x TRIP database: www.tripdatabase.comIndividuals in most industrial societies who are, or believe x Bray PF. Platelet glycoprotein polymorphisms as risk factors forthemselves to be, ill can turn to a variety of sources of advice thrombosis Curr Opin Haematol 2000;7:284-9 x Smith R. The future of healthcare systems. BMJ 1997;314:1495-6other than health professionals. However, these sources will x Wyatt JC. Clinical knowledge and practice in the information age: aprobably only help with the problems that a person deals with handbook for health professionals. London: Royal Society of Medicinethat day. A doctor is often needed to provide additional Press, 2001information, and to interpret and individualise advice for all the x Community Health Index (CHI) in Scotland. brought to the consultation by the patient, not just ehealth/the presenting problem.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.BMJ 2005;331:831–3BMJ VOLUME 331 8 OCTOBER 2005 833