clinical risk management: experiences united states · clinical risk management:experiencesfromthe...

Quality in Health Care 1995;4:90-96

Clinical risk management: experiences from theUnited States

Don Harper Mills, G E von Bolschwing

While a major goal of clinical risk managementtoday is to improve patient care, the conceptand development of risk management generallyhave evolved out of the litigation process andcontinue to do so. Hoxx should injuries thatlead to litigation be compensated, managed orprevented, or both? The answers depend onthe risks involved: What is the risk of injury?How severe are injuries when they occur? Canthey be modified once they occur? Do themodifications impart cost savings? Can theinjuries be prevented in the first place? Doesthe cost of prevention exceed the value of theinjuries themselves? Do social costs have to beconsidered? These questions apply to anyindustry in which injury litigation is a recurringtheme. Health care is a prime example. Studiesin California' and New York state have shownthat injuries and adverse outcomes from healthcare occur often enough to warrant concerneven without litigation: about 40/ of patients inacute care hospitals incur injuries and adverseoutcomes of medical and clinical managementsufficient to cause substantial financial impactthrough prolonged hospital stay, readmission,surgery, reoperation, special clinical treatment,or even death. About a quarter of these out-comes occur under circumstances that couldlead to successful litigation in the United Statesshould the affected patients or their relativesdecide to sue. That only a small percentage ofthem actually sue (about 10%) is of littlesolace, because the cost of their lawsuits isalready a social burden, and the situation couldget worse. Since we cannot always change orcontrol social structure, including the tendencyto sue, solutions wt ill have to be found inantecedent factors, hence the need for riskmanagement.

Published technical books describing stan-dard risk management structure and functionabound. We will not duplicate those efforts;rather, we want to tell of our personal experi-ences as we have participated in the growth anddevelopment of risk management related tohealth care in the United States. Most ofthese experiences have not previously beenpublished, and we feel they may be particularlygermane to the United Kingdom at a timewhen risk management is just beginning tobecome an important factor in health care.

History of risk management in the UnitedStatesWe found that the focus of risk management atany time depends greatly on the content ofcurrent lawsuits. Before the 1 960s most liti-gation concerning medical injury involved com-plaints against hospitals and their nursing staff

(as opposed to their medical staff). Doctorswere rarely the primary targets because thelegal profession was not sufficiently sophis-ticated to confront the judgmental decisionmaking that forms the basis of clinical manage-ment. Thus, when a risk management pro-gramme was developed for the hospital industryin the mid-1950s attention was directedtowards patients' falls, treatment errors, patientmisidentifications, retained operative sponges,and the like.' Since the largest hospitallitigation losses at that time arose from retainedsponges, initial efforts were devoted to thatproblem. Analysis indicated that the methodsof counting sponges during surgery wasineffective, even though universally followed. Athird sponge count at the time of skin closurewas recommended and adopted and x asfollowed bv a precipitous drop in retainedsponges and resultant claims. Interestingly, asimilar analysis of instrument counting pro-cedures concluded that, although adding aterminal count might prevent retention ofsome instruments, prolonging surgery undulywould be counterproductive. Balancing risksand benefits therefore became a key element inthe development of clinical risk management.

INC 1)ENI RI'PORIIN(, SYS TIMSExcept for reported claims, there was no data-base to help to determine where the manage-ment problems lay and which of these wasimportant. One of the first incident reportingsystems was inaugurated for the CaliforniaHospital Association. It merely asked nurses toreport unusual events occurring to and aroundpatients - that is, events that were inconsistentwith good nursing management. The resultingdatabase disclosed that patients' falls out ofbed were much too frequent. Bed rails werepurchased, but these proved more of a hazardthan a solution. Patients determined to get outof bed would do so anyway, only to find thatthe drop to the floor was much farther thanbefore, producing more severe injuries.Lowering the beds so that the falls would beless severe only aggravated the nurses, who hadto care for patients in stooping positions.Nurses' complaints of back strain led to thecreation of high-low beds, which allowed thenurses to crank up the beds for attending topatients and then crank them down to avoidexcessive injuries from falls. This developmentwas the forerunner of the electrically controlledbeds we have today.Another problem at the time was losing

patients when they were transported toradiology, surgery, or some other specialisedcare unit. Patients became separated from their

Professional RiskManagement Group,Long Beach, CA 90815,United StatesDon Harper Mills,directorG E von Bolschwing,attorneyC orrespondence to:Dr D H Mills,Professional RiskManagement Group,)1I Studebaker Road,Suite 250, PO Box 1539Q8,ILong Beach, C A 90815,United States

90)

on October 12, 2020 by guest. P

rotected by copyright.http://qualitysafety.bm

j.com/

Qual H

ealth Care: first published as 10.1136/qshc.4.2.90 on 1 June 1995. D

ownloaded from

http://qualitysafety.bmj.com/

Clinical risk management: experiences from the United States

charts (medical records), and, because theywere often sedated no one could determinewhere they belonged. The frequency of lostpatients was recorded by the incident reportingsystem and led to the introduction of wristband identifiers. But patients going to surgery

still became lost occasionally because anaes-

thetists would cut off the wrist bands tofacilitate vascular access. The bands were

subsequently loosened to avoid that necessity,and anaesthetists were counselled on the needfor retaining them.The incident reporting system also became

a product of litigation in the United States,serving as an early warning system for potentialclaims, as well as for finding problems beforethey were bad enough to cause claims. Injuriessustained by patients falling out of bedprompted lawsuits and solutions. Losingpatients in the hospital rarely caused injuriesbut needed to be avoided; similarly, treatmenterrors rarely caused injuries, but they tooneeded to be avoided just to prevent an

unusual toxic or allergic reaction. Fromunpublished data in our files for 1980-2, theincidence of treatment errors was 0 54/100patient days in participating California hospitalswhereas that of actual treatment reactions was

only 0-03/100 patient days. According to themedical insurance feasibility study inCalifornia in 1977, 85% of drug reactions weretransient in nature.'As malpractice litigation increased during

the '60s and early '70s, attorneys representingpatients became more experienced in managingissues of clinical care and the emphasis on

liability shifted from the hospital and nursingstaff to primarily doctors. But during thistransition, risk management programmes

lagged behind, continuing to address problemsassociated with hospitals and nursing and rarelyfocusing on those associated with doctors.Doctors allowed hospitals to accumulate dataabout non-clinical issues but often resisted thedevelopment of information about their own

clinical conduct, even though they had func-tioning tissue committees and surgical case

review committees. (These committees ofmedical staff review appropriateness ofmedication for surgery and complications.)It was not until the mid-1970s, with theCalifornia medical insurance feasibility study,'that the development of databases involvingclinical problems became a reality. The con-

cept that generic screens could flag adverseoutcomes made this transition possible. Thisstudy measured the frequency of patients'injuries and adverse outcomes arising fromhealthcare management, much as did the NewYork study in the late '80s.2 To facilitate thesearch for these injuries and outcomes specialcriteria were developed to screen patients'charts. Of the 20 original criteria, only 11

proved necessary to identify virtually all of theinjuries and outcomes: criterion 1, admissionin the previous six months; criterion 3, ad-mission for conditions suggesting prior failureor adverse result of treatment; criterion 4,trauma incurred in hospital; criterion 8, returnto surgery; criterion 10, unplanned removal of

an organ or part during surgery; criterion 11,acute myocardial infarction during thisadmission; criterion 12, wound infection;criterion 13, neurological deficit occurringduring admission; criterion 14, death inhospital; criterion 15, length of stay exceedingthe 90th percentile for the region; and criterion20, any other unlisted complication of clinicalmanagement.

Generic screens were intended to serve asindicators for the possibility that adverse out-comes from medical or clinical managementhad occurred. That is, if a patient was returnedto surgery, that chart needed to be examinedto see what adverse outcome, if any, wasresponsible (such as postoperative infection,postoperative haemorrhage, or some othercomplication related to the previous surgicalmanagement). Also, the transfer of a patientfrom a non-intensive unit to an intensive careunit creates the presumption that somethingmight have gone wrong; therefore, the chartshould be examined to see what diverse out-come, if any, was responsible. That a patient'schart was flagged by a generic screen, there-fore, did not mean that an actual adverseoutcome had occurred; it merely meant thatcertain outcomes might be present and thechart needed to be examined for thosepossibilities. Yet a tendency developed in theUnited States to regard return for surgery andunanticipated transfers to intensive care unitsas ultimate adverse outcomes.4 In our opinion

Doctors ... often resisted thedevelopment of information about

their own clinical conduct ...

this was accumulation of ineffective infor-mation. These criteria identified 16 486 occur-rences (79%) in 20 864 charts; yet, only 7% ofthese occurrences were shown by peer reviewto represent clinically caused injuries andadverse outcomes. Such a high sensitivity/lowspecificity may be good enough for research toestablish baselines, but it is not efficientenough for prospective risk management. If apatient returned to surgery for peritonitis afteran appendicectomy the return would bereported first on the generic screen, laterfollowed by an examination of that chart to seeif an adverse outcome was responsible. Takingthe case that it was a burst appendiceal stump,the case should be added to other similar caseswhich could later be reviewed as a group. If thishospital had only one burst stump over 12months, that episode would not be consideredimportant in terms of quality management.However, if there had been six such occur-rences a surgical problem probably existed anddeserved attention with a view to prevention.The development of this scenario requiredthree stages: a flagging system, a chart analysisfor adverse outcomes, and the audit of thegroup of adverse outcomes. Today, with theutilisation ofmore refined indicators by depart-ments, the data gathering can be simplified bycombining the first and second stages into a

91



j.com/

Qual H


ownloaded from


2Mills, Bolschzwi'ig

single outcome measuring system. The initial

flagging system should have identified the burststump directly.

OVERC(OMING R15SIST[ANCE TI() REPORTINGx

We encountered resistance in expanding riskmanagement from hospital and nursing care to

include outcomes of clinical management.Doctors not only became protective when theirown information was being screened for qualityanalysis and management but they also tendedto find many reasons for not reporting theirown adverse outcomes. We therefore chosenurses to do this reporting. They were alreadyproficient in reporting non-clinical problemson their old incident report, and since clinicalreporting did not involve assessment, cause, or

conduct they were fully capable of identifyingcurrent adverse outcomes sufficiently to complywith our reporting requirements. But knowingthat doctors themselves tended not to reportthese outcomes, nurses complained that theywere being required to do the work of others.They felt that doctors should do their own

reporting. So we had to meet with doctors andnurses together to try to solve the riddle of datagathering on a reasonably concurrent basis. Weestablished hotlines for reporting incidents bytelephone. Doctors preferred this method towriting reports, particularly when the out-comes were serious. We found that doctorsused the hotline primarily to find out what theyshould do. Thus we piggybacked the datagathering system on requests for information.Of course, additional time and effort are

required for hospitals to maintain adequatelytrained staff to respond to such hotline calls,but where this has been done risk managementhas improved, both for the care of injuredpatients, and for developing information forprotecting yet uninjured patients. Once thenurses realised that doctors were participatingin the reporting system, albeit only orally, theybecame less reticient in complying with thewritten reporting requirements.

Managing injured patientsAn injured patient is one who has experiencedsome form of injury or adverse outcome eitheras a result of custodial or nursing managementor as a result of clinical practice. Once a

mechanism to identify such patients is in place,the next action is to minimise the injuries. Forexample, if a patient fell and sustained a

fracture we need to evaluate that fracture andensure appropriate management; if a patientdeveloped a postoperative deep venous throm-bosis and a pulmonary embolism we need toensure that appropriate anticoagulation has beenundertaken to avoid the next, and possiblyfatal, embolism. Someone should monitor theprocess of identifying and managing theseadverse outcomes and, when necessary, obtainclinical consultation to ensure that appropriateaction is taken. Many hospitals in the UnitedStates did not have ongoing systems to carry

out these functions. As early as 1963 a review

of 1000 malpractice cases disclosed that at least50 of adverse outcomes were worsened bysubsequent problems with management (such

as inadequate management of postoperativewound infections).' These additional com-plicating factors have to be avoided.

In addition to minimising injuries, we needto tell patients what happened and what will bedone. Whether an adverse outcome is a resultof bad clinical care or merely a calculated riskof modern medicine, doctors tend to becomesilent and defensive when it occurs. They havebeen told over the years not to confess topatients for fear of litigation, and therefore theytend not to talk to patients at all about adverseoutcomes. This is counterproductive. In theUnited States many injured patients go tolawyers simply to find out what happened - anunnecessary and costly investigative approach.It would be much better for doctors to sit withtheir patients and talk about what happenedand what will be done medically and surgically.If liability is involved, now is the time to beginto consider some form of compensation. Butthe purpose of this communication is not forlitigation, it is primarily to satisfy patients'clinical uncertainty. Yet, communicationssuch as this will go far to preclude expensivelitigation.

M\AN'A( ING( MNA LPRACT'lICEt^ C-IA1NS

Decades ago all malpractice claims weredefended with determination. Few claimantssucceeded because attorneys representing themwere incapable of ferreting out the real issues;obtaining medical advice and experts; andproving negligence, causation, and damages;but they have learnt well subsequently. Totalresistance is no longer an appropriate manoeuvrefor claims or risk management, particularly forpublic entities (county or state hospitals in theUnited States). These institutions may be

Total resistance is no longer anappropriate manoeuvre for claims

or risk management

embarrassed by such tactics taken against thevery population they serve. Serious andmeritorious claims can and should be resolved,often without entering into formal litigation.Managing claims before or during litigation

requires risk managers to utilise in houseconsultations. For instance, cases involvingadverse outcomes from fractures will almostalways be decided by what the radiographsshow. These should be gathered and deliveredto the chief of the orthopaedic department fora review to describe the initial fractures, todetermine how well they were managed, and toascertain whether the outcomes should havebeen different. Although the chief might havean interest in protecting his or her staff,utilising this help in evaluating claims allowsfor both departmental interaction and theimplementation of whatever preventive meansmight surface from the evaluation. Thismethod therefore entails both awareness andparticipation of the institution in the claimsmanagement process. If the chief feels the caseis "clean" the claims manager or risk manager

92



j.com/

Qual H


ownloaded from



may still seek independent consultation thatavoids conflicting interests to ensure that theassessment has been appropriate. This processshould not be put in the hands of solicitors.They are not trained for this purpose and theymay not know how to ask appropriatequestions. Solicitors and barristers are trainedto conduct litigation; they may not be goodclaims managers.

Risk managers need to view medical andlegal issues in malpractice litigation in a waythat makes sense to doctors. To talk aboutduty, breach of duty, and proximate causeoften leaves doctors bewildered. Thus we havelearnt to approach these concepts differently.In clinical negligence there are two types ofadverse outcomes.The first type of adverse outcome is a new

abnormal condition (such as an allergicreaction) caused by doing something to or forthe patient (such as prescribing a drug.) A setof medical issues pertain to this type of out-come, a breakdown in any one of which mayresult in liability for the doctor: (1) Was thedrug or procedure calculated to be effective forits intended use for this patient (technologyassessment, for example, North AmericanSymptomatic Carotid Endarterectomy TrialCollaborators7)? (2) Was it adequately indicatedat this time (appropriateness, for example,McGlynn et al8). (3) Was the doctor competentto perform the procedure (credentialing, forexample, Langsley9)? (4) Was valid consentobtained from the patient? (5) Was theprocedure or prescription properly performed?

Lack of valid consent ... is afrequent allegation in clinical

negligence cases ....

The incident reporting system should disclosewhat procedures or drugs are causing undueproblems, and the case analyses should identifywhich issues are involved. Groups of cases canthen be subjected to clinical audit for qualityimprovement.Lack of valid consent from the patient is a

frequent allegation in clinical negligence casesin the United States.'0 It does not fit into theusual category of clinical negligence in that itis not a professional judgment or performanceissue. Rather, it is a doctor-patient com-munication problem. Did the doctor disclosesufficient information for the patient to makea personal decision to proceed? Not only is thedoctor required to balance risks and benefitsin deciding to recommend the procedure(involving categories (1) and (2) above) but heor she must then allow the patient to make asimilar judgment, though at a non-clinicallevel. The usual lawsuit in the United Statesstems from experiencing a complication of theprocedure which the patient alleges he or shewould have refused had the risk of thatcomplication been adequately disclosed.Doctors in the United States have been slowto realise their obligations of disclosure, findingit difficult to switch from paternalism to

accepting patients' autonomy. The duty todisclose the risks of a beneficial proceduresometimes seems antithetic to the desire tohelp the patient. Fortunately, we have beenable to defend most lack of consent cases byemphasising the need for the procedure(benefit to the patient) and by establishing thatmost reasonable patients would have acceptedthe risks had these been disclosed. But we havenot been able to cope successfully with consentcases based on misinformation (as opposed tolack of information) - that is, if doctors down-play the risks, they place themselves in analmost indefensible position should the patientsue, even though the sole intent was to benefitthe patient.The second type of adverse outcome is a

worsening or prolonging of the patient'sdisease or condition caused by a lack of timelyintervention (that is, a failure to do something).The medical or legal issues for such adverseoutcomes are different. (1) Was the lack ofdiagnosis or misdiagnosis reasonable? (2) Didthe doctor accumulate enough informationbefore making a diagnostic or therapeuticdecision? If the database was adequate thedoctor may have a right to be wrong, but if itwas insufficient the misdiagnosis is probablynot defensible, medically or legally. Examplesare (a) not performing a rectal examination inacute appendicitis (the patient went on toexperience rupture of the appendix and diedfrom sepsis) and (b) not performing a spinalfluid examination in acute meningitis (thepatient developed severe, permanent neuro-logical sequelae because of delayed treatment).We found that the audit for problems of mis-diagnosis may be more fruitful if it is focusedon the status of the database, rather than on themisdiagnosis alone.What we have just described are forms of

system errors that may be remedied if therehas been adequate data gathering, analysis, andaction. But episodic errors also exist. These areinstances of clinical negligence that occurregularly but seem to defy prediction andtherefore prevention. The problem has notbeen lack of analytical ability, but the failure toaccumulate enough information about theoccurrence of episodic errors. Lawsuits alonedo not supply enough information to permitanalysis of these episodes, so we have had tofold in the cumulative data from incidentreports to establish trends and to permitauditing, which seems to be satisfactory, butonly when the incident reporting system isequal to the task.Few doctors know how to analyse clinical

negligence cases well. They are not trained toappreciate the credibility factors or the types ofevidence most meaningful to judges. Nor arethey conversant with the types of system errorswe have been talking about. Solicitors andbarristers may have similar shortcomings unlessthey are extraordinarily skilled in this field. Therisk or claims manager, therefore, has a role tooversee the evaluation of claims and to ensurethat consulting doctors are focused appro-priately. It is inefficient and insufficient for asolicitor or claims manager simply to provide

93



j.com/

Qual H


ownloaded from


Mills, Bolschwing

a medical record to a consulting reviewer with-out guidance. A factual determination of whathappened in the case must be carried outbefore asking consulting experts to evaluate thedefendant's conduct. This preparation oftenincludes the use of in house consultation withdepartmental chiefs.

Preventing the next injuryIn working on preventing adverse outcomes,risk management becomes a subset of qualityassurance. As we found out in the mid-1970s,most adverse outcomes are not related toclinical negligence. In the sentinel Californiastudy 4-65% of patients in acute care hospitalsdeveloped adverse outcomes from theirmanagement, of which only 17% were causedby negligence.' In real numbers, out of threemillion admissions annually in California,140 000 adverse outcomes occurred, only23 000 of which were due to clinical negligence.Efforts at prevention therefore focused on a

much greater population than the liabilitycases. But even so, this comprises less than 5%of admissions; this is an outlier population, andsome people complain that focusing on thisgroup is an inefficient way of improving qualityof care." Although we agree that continuousquality management is an appropriate route toimprove the quality of care, we have to realise

Prediction should be high in thevocabulary of risk managers.

that when particular adverse outcomes lead tolitigation they exact substantial expense. Thisalone deserves attention. Even if adverse out-comes are not a major quality problem theyare certainly a cost control problem. Someepisodes are eggregious enough to demandinstant reaction whereas others may wait fortrends to be identified. The degree of clinicalresponsiveness often depends on the points ofview of the institution's administrator. InCalifornia some hospital administrators willhesitate to pay damages for an injury clearlynegligently caused unless steps are in place toprevent such an episode from recurring. Butthe way to prevent recurrence may not beapparent from solitary episodes. The riskmanager is therefore in a position of having tosatisfy both the administration and the doctors.It is not often a good idea to delay settling a

damage case just because we cannot immedi-ately produce methods to prevent recurrence.

PREDICTING PROBLEMS

Prediction should be high in the vocabulary ofrisk managers. They should be clinically awareof the progress in medical science to predictwhen and where problems may arise. Onceclosed chest cardiac massage proved effectivein the early '60s, we began to worry about theconcept of coronary care units, where patientscould be salvaged from cardiac electricalfailures. The first units were small and not allhospitals had them. We were concerned aboutthe survival of patients with heart attack

admitted to hospitals without such units (thatalone prompted the rapid development ofcoronary care units in many hospitals). But wewere more concerned about the small unitsthat were in operation. How would decisionsbe made to discharge patients prematurely tomake room for new patients who had greaterpriorities? Rather than wait to allow theseproblems to be decided ad hoc, werecommended the development of protocolswhich identified known criteria for admission,retention, and discharge from these units.Most hospitals developed these protocols early,and they proved effective. Although somepatients who were excluded from continuouscare owing to limited space in the unitsdeveloped fatal cardiac arrhythmias inunmonitored rooms, we had very few lawsuits.Attorneys who represented the families ofdeceased patients admitted that the protocolscontributed to the decisions for refusing theirrepresentation.

Prediction was again effective when theimmune globulin to prevent maternal sensi-tisation in mother/fetus Rh incompatibility wasdeveloped. We knew this globulin was to beuniversally available in southern California by15 June 1968; therefore, we flooded themedical media to prepare physicians andhospitals for this eventuality. Thereafter, onlya few children developed erythroblastosisfetalis from lack of administration of immuneglobulin. 12

Occasionally ethical dilemmas are depositedat the risk manager's doorstep. In the UnitedStates Jehovah's witnesses have the right torefuse blood transfusion, even if that refusalentails a risk of immediate death, and even ifthe patient is a pregnant woman. But somedoctors feel morally and ethically impelled tooverride these refusals if transfusions wouldalter the outcomes. Other doctors feel that itis only ethical and moral to abide by thepatient's wishes, with today's emphasis onautonomy. Neither group can be condemned,nor should they be. But the mere fact that bothare "correct" prompts the recognition of aspecial obligation of that group which feelscompelled to treat over objection. Since thereare others who feel just as strongly the otherway, the group that "must treat" needs to referthese patients elsewhere for management. Riskmanagement is involved in developing lists ofphysicians who will accept patients under thesecircumstances.

Interpreting and implementing courtdecisionsRisk management in the United States mustoften interface between courts and the health-care system, and it has the obligation to be thecaregivers' legal interpreter. The rules for civilliability, which includes clinical negligence, arecontrolled by the legislatures and courts of theindividual states. Risk managers must thereforemonitor legislative enactments and the statecourt opinions to interpret these for thehospitals and doctors in their particular state.In California we learnt in 1958 that doctorsand nurses have the same direct obligation to

94



j.com/

Qual H


ownloaded from



serve their patients. 13 This should not seemearth shaking, but it required nurses to inter-vene to protect patients from doctors' mis-actions. The appellate court opinion arosefrom a postpartum maternal death caused byexsanguination from a cervical laceration. Thenurses were aware that the mother was notresponding to the doctor's ministration butfailed to intervene. The court held the doctorand the nurses responsible. This caseaccelerated the development of chains ofcommand that now exist in most hospitals.Time and again risk managers had to explainto nurses and doctors that knowledgeablehospital staff have the duty to act. They can-not sit by knowing a patient may be injured.Their duty to intercede exceeds the rights ofdoctors to make mistakes, even non-negligentmistakes.Many of the important court decisions

affecting doctors' obligations arise fromproblems of communication. As we statedpreviously, doctors are not prone to be par-ticularly informative to their patients. Courtstherefore accepted the task of developingduties of communication for them. In Trumanversus Thomas doctors were told that theyhad to warn patients of the downside riskswhen they refuse recommended procedures."4That case involved a woman who refusedsuccessive annual Pap (Papanicolaou) smearsand developed advanced cervical cancer.She complained that the doctor should havetold her why it was important not to refuse(and the court agreed). The impact of this casehas been far reaching. Diabetic patients withinfected foot ulcers must be told why theyshould not refuse admission for intensive care.Even patients who leave hospitals againstmedical advice must be warned of theconsequences.

Various court decisions have affecteddoctors' duty to warn patients of the danger ofdriving motor cars while taking sedative drugs'5or if they have brittle diabetes'6 and to warn anidentified potential victim of a psychiatricpatient's stated intent to harm that person.17The risk manager's task is not only to dis-seminate these principles but also to devisemeans to document warnings whenever given.Patients who drive under the influence ofsedatives or tranquilising drugs who becomeinvolved in motor car accidents may claim thatthey were unaware of the dangers. Doctorshave to prove that the warning was given, butthey complain that they are already over-burdened with menial recording tasks. How-ever, in this case a stamp would suffice("sedative warning given") if placed in anappropriate part of the chart at the time of thepatient's visit.One obligation for communication does

not include patients directly. The JointCommission on Accreditation of HealthcareOrganisations (JCAHO) requires doctors totalk with one another in evaluating andimproving clinical care systems and in carryingout credentialling processes. These functionsare enforced with varying degrees of success byhospitals seeking accreditation. Doctors have

worried about these communications, fearingthey may be developing information thatpatients' solicitors can use against them. Theseclinical care judgments are "built in" expertopinions that would greatly facilitate theprosecution of malpractice lawsuits. In the1950s and early '60s we thought that judgeswould understand that quality assuranceactivities were essential to improving careand that, since communications about theseactivities were not directly involved in anyparticular patient's day to day care, they wouldbe considered irrelevant and inadmissible in amalpractice case.'8 A California appellate courtproved us wrong, and the floodgates wereopened for patients' solicitors to go on fishingexpeditions in the entire clinical audit systemfor whatever evidence or opinions might bebeneficial to their claims.'9 Doctors threatenedto close down the audits unless they wereassured their activities would not reboundagainst them. There was a danger that medicalaudit would be reduced to examining non-threatening issues only. Risk managementinvestigations into individual, potential law-suits would not be impeded because ofthe attorney-client confidentiality and work-product rules; however, injury prevention andother programmes to improve clinical carewould be devastated. The California Legis-lature responded by reversing the appellatecourt's decision with section 1157 of theCalifornia Evidence Code. This new law madeproceedings and documents of quality carecommittees non-discoverable in legal pro-ceedings. It has withstood attacks by theplaintiffs' bar,20 and most other states have nowfollowed suit. These legislative changes arenow the mainstay of doctors' cooperativeventures into quality improvement.

ConclusionWhat we have described here has been drawnfrom decades of firsthand experience inmedical and clinical risk management. We havenot always succeeded, but we have found mostof the essentials, as follows. (1) To succeed inthe larger arena of clinical risk management,doctors must be involved, both individuallyand as a group. They need to participate ingathering data and finding problems; they needto improve their skills in analysing evidence;and they need to communicate more readilywith their patients. (2) The risk manager hasfundamental obligations (a) to control lawsuitsin his or her jurisdiction (gathering evidence,seeking appropriate analysis, helping solicitorsand barristers, but always keeping control overeach claim); (b) to monitor injured patients,making every effort to minimise their injuriesand keeping them informed; (c) to help indata gathering and analysis for preventinginjury; and (d) to predict the clinical effectsof new medical and legal changes in healthcare. Risk management is thus a complexendeavour.Mr Jack J Fulton, hospital administrator and lawyer, wasthe first special representative for the California HospitalAssociation's risk management programme, beginning in themid-1950s. He was responsible for most of the early examplescited in this paper, and we appreciate his review and guidance.

95



j.com/

Qual H


ownloaded from


Mills, Bolschwing

1 Mills DH, ed. Report ott the medical iisurancefeasibility studs.San Francisco: Sutter Publications, 1977.

2 Brennan TA, Leape LL, Laird NM, et al. Evidence ofadverse events and negligence in hospitalized patients.NEnglJMed 1991 ;324:370-6.

3 Ludlam JE. A personal history of the California hospitalassociation insurance program. In: Dunlap HB, ed. Fiftssears. Glendale, California: Bright Publishers, 199 1:176-96.

4 Sanasaro PJ, Mills DH. A critique of the use of genericscreens in quality assessment. _7AA4A 1991;265:1977-81.

6 Mills DH. Medical lessons from malpractice cases. JAMA1963;183:1073-7.

7 North American Symptomatic Carotid EndarterectomyTrial Collaborators. Beneficial effect of carotidendarterectomy in symptomatic patients with high-gradecarotid stenosis. N EnglJ Med 1991 ;325:445-53.

8 McGlynn EA, Naylor CD, Anderson GM, eLt al.Comparison of the appropriateness of coronary angio-plasty and coronary artery bypass graft surgery betweenCanada and New York State. JAMA 1994;272:934-40.

9 Langsley DG. Medical competence and performanceassessment, a new era. JAMA 1991;266:977-80.

10 Mills DH. Whither informed consent? JAAIA 1974;229:305-10.

11 Kritchevsky SB, Simmons BP. Continuous qualityimprovement, concepts and applications for physiciancare. YAMA 1991;266:1817- 3.

12 Mills DH. Medical legal aspects of RhoGam. Mlaternal amd

Child Health 1968;1:18 9.13 Goff X Doctors General Hospital of San Jose. Pl~auf-

Reporter Second Senes 1958;333:29 (California).14 Truman v Thomas. Califoniola Reports 1980;165:308.15 Kaiser v Suburban. Pacific Reporter, Se)oss Scres 1965;

398:14 (Washington).16 Myers Nr Quesenberry. Cal/forsia Reportes 1983;192:583.17 Tarasoff v Regents. California Reporter 1976;131:334.18 Tissue committee of a hospital staff. Califomia Mlediciols

1961 ;95:264-6. Judd v Park Avenue Hospital. New YorkState Reporter, Second Sen'ies 1962;235:843.

19 Mills DH. Medical peer review: the need to organise a

protective approach. Health Matixs, J7oimnal of Law-Medicitle (Case Western Reserve University School ofLaw) 1991;1:67-76.

20 Mills DH. Protecting medical peer resiess, the case forconfidentiality. Los Asngelcs Ibuowesr 1993;15:23 5.

96



j.com/

Qual H


ownloaded from


clinical risk management: experiences united states · clinical risk management:experiencesfromthe...

Documents