Fix it: see and solve the problems of digital healthcare 1st edition thimbleby - The full ebook vers

Page 1


https://ebookmass.com/product/fix-it-see-and-solve-theproblems-of-digital-healthcare-1st-edition-thimbleby/

Instant digital products (PDF, ePub, MOBI) ready for you

Download now and discover formats that fit your needs...

How to Get the Most from Your Home Entertainment

Electronics: Set It Up, Use It, Solve Problems Michael Jay Geier

https://ebookmass.com/product/how-to-get-the-most-from-your-homeentertainment-electronics-set-it-up-use-it-solve-problems-michael-jaygeier/ ebookmass.com

Pressure cooker: why home cooking won't solve our problems and what we can do about it Bowen

https://ebookmass.com/product/pressure-cooker-why-home-cooking-wontsolve-our-problems-and-what-we-can-do-about-it-bowen/

ebookmass.com

The Only Purple House in Town (Fix-It Witches 4) 1st Edition Ann Aguirre

https://ebookmass.com/product/the-only-purple-house-in-town-fix-itwitches-4-1st-edition-ann-aguirre-2/ ebookmass.com

Chemtrails existieren doch - Teil 3 - Chemische Kondensstreifen im 'Kampf' gegen den erlogenen Klimawandel

Frank Hills

https://ebookmass.com/product/chemtrails-existieren-dochteil-3-chemische-kondensstreifen-im-kampf-gegen-den-erlogenenklimawandel-frank-hills/ ebookmass.com

The Boss's Baby Rescue: A K9 Handler Romance (Disaster City Search and Rescue, Book 28) Brandt

https://ebookmass.com/product/the-bosss-baby-rescue-a-k9-handlerromance-disaster-city-search-and-rescue-book-28-brandt/

ebookmass.com

Moteris juodais drabužiais Susan Hill

https://ebookmass.com/product/moteris-juodais-drabuziais-susan-hill/

ebookmass.com

Three Finger Joe William Heyliger

https://ebookmass.com/product/three-finger-joe-william-heyliger/

ebookmass.com

Trajectories

of Governance: How States Shaped Policy Sectors in the Neoliberal Age Giliberto Capano

https://ebookmass.com/product/trajectories-of-governance-how-statesshaped-policy-sectors-in-the-neoliberal-age-giliberto-capano/ ebookmass.com

Curse of the Night Witch Alex Aster

https://ebookmass.com/product/curse-of-the-night-witch-alex-aster-2/ ebookmass.com

Machine Learning for Planetary Science Joern Helbert

https://ebookmass.com/product/machine-learning-for-planetary-sciencejoern-helbert/

ebookmass.com

FixIT

FixIT

Seeandsolvetheproblems ofdigitalhealthcare

HAROLDTHIMBLEBY

GreatClarendonStreet,Oxford,OX26DP, UnitedKingdom

OxfordUniversityPressisadepartmentoftheUniversityofOxford. ItfurtherstheUniversity’sobjectiveofexcellenceinresearch,scholarship, andeducationbypublishingworldwide.Oxfordisaregisteredtrademarkof OxfordUniversityPressintheUKandincertainothercountries

©HaroldThimbleby2021

Themoralrightsoftheauthorhavebeenasserted

FirstEditionpublishedin2021

Impression:1

Allrightsreserved.Nopartofthispublicationmaybereproduced,storedin aretrievalsystem,ortransmitted,inanyformorbyanymeans,withoutthe priorpermissioninwritingofOxfordUniversityPress,orasexpresslypermitted bylaw,bylicenceorundertermsagreedwiththeappropriatereprographics rightsorganization.Enquiriesconcerningreproductionoutsidethescopeofthe aboveshouldbesenttotheRightsDepartment,OxfordUniversityPress,atthe addressabove

Youmustnotcirculatethisworkinanyotherform andyoumustimposethissameconditiononanyacquirer

PublishedintheUnitedStatesofAmericabyOxfordUniversityPress 198MadisonAvenue,NewYork,NY10016,UnitedStatesofAmerica

BritishLibraryCataloguinginPublicationData Dataavailable

LibraryofCongressControlNumber:2021934818

ISBN978–0–19–886127–0

DOI:10.1093/oso/9780198861270.001.0001

PrintedandboundintheUKby TJBooksLimited

LinkstothirdpartywebsitesareprovidedbyOxfordingoodfaithand forinformationonly.Oxforddisclaimsanyresponsibilityforthematerials containedinanythirdpartywebsitereferencedinthiswork.

Therealdataofsafetyarestories.

—JamesReason

Themostpowerfulpersonintheworldisthe storyteller.Thestorytellersetsthevision, values,andagendaofanentiregeneration thatistocome.

TheCorporation’sgalaxy-widesuccessis foundedontheirsystems’fundamentalflaws beingcompletelyhiddenbytheirsuperficial designflaws.

—DouglasAdams

Theproblemsoftherealworldareprimarily thoseyouareleftwithwhenyourefuseto applytheireffectivesolutions.

—EdsgerDijkstra

…andnow [1977] whenthecomputerpeople moveinandthenon-medicalpeoplemovein, theycanhardlybelievewhattheysee.And thereisacrisisofmajorproportions.

—LarryWeed

Digitalhealthcareismuch riskierthanwethink,butit canbemadefarmore effectiveandmuchsafer.This booksplitsuptheactioninto storiesofproblems,the solutions,andthenthebetter futurewecanreach.

Howtoreadthisbook

Healthcarehasbeenaroundforthousandsofyears,certainlysincelongbeforetheHippocraticOathtodonoharm (figure1.1).Incomparison,digital isverynew,hardlyablinkofaneye.Unsurprisingly,healthcareanddigital technologyhaven’tyethadthetimetoworkouthowtoworkwell together

FixIT:Seeandsolvetheproblemsofdigitalhealthcare isabookabout digitalhealthcareandhowithasanimpactonallofus,bothpatientsand healthcareprofessionals.Theuniquecontributionofthebook FixIT isto show,withlotsofpowerfulstories,howsurprisinglyriskydigitalhealthcare is.Oncewestarttobeshockedbyitsproblems,it’seasytoseehowdigitalhealthcarecanbemade muchsaferforeveryone’sbenefit,forpatients andtheirfamilies,aswellasforstaff.Digitaltechnologiescancertainlybe improvedtomakehealthcaremoreeffective,butso,too,couldhealthcare changetomakeiteasierfordigitaltohelpit.Itshouldbeacollaboration, notaone-waystreet.

FixIT isdividedintothreeparts:

Part1 ⋄ Diagnosis ⋄ Riskierthanyouthink —Iwantyoutoseethe unnoticedrisksofdigitalhealthcare,andtheseriousproblems thatarisewhendigitalismisunderstoodandmisapplied.

Part2 ⋄ Treatment ⋄ Findingsolutions —Iwantyoutoseethat digitalhealthcare’sproblemsarefixable.Therealsolutionsaren’t justaboutgettingnewerormoreexcitingstuff;thesolutionsare aboutthinkingmoreclearlytounderstandwhatweneed,and howtoinnovate,design,andimplementdigitalhealthcaremore reliably.

Part3 ⋄ Prognosis ⋄ Abetterfuture —thereisapossible,much better,safer,andfarmoreeffectivedigitalhealthcareforallofus. Thefinalpartofthebooksketchestherealdigitalpromise.

Figure1.1. Healthcareisasoldashumanity,butthinkingclearlyabouthealthcarecamelater.Someoftheearliest“modern”writingonhealthcarewasbyHippocrates.1 AlthoughHippocrateslivedaround400 BC,thisistheoldestsurviving HippocraticOath,writtenonafragmentofthePapyrusOxyrhynchusdatingfrom around300 AD.Thinkingclearlyaboutdigitalhealthcareisalreadylongoverdue.

Allchaptersin FixIT havestoriesthat’llbeofinteresttopatientsandto healthcareprofessionals.Allthematerialusedinthisbookiseitherinthe publicdomain(andfullycitedinthebook’snotes)orhaspermissionfrom thepeopleinvolved.Thisopenapproachisessentialtotheintegrityof Fix IT,andthereasoningbehindthisopennessisdiscussedlaterinthebook.a Thedigitalindigitalhealthcarecannotbeavoided.Thereare,therefore, afewslightlytechnicalchaptersinthisbook,whichwillbeofspecialinterest toprogrammers,developers,andregulators—thisbookwillbecomeauseful referenceforthem.Thesechaptersarehighlightedwitha1960scomputer chip(amodernonewouldbesmallerandhardertosee),bothinthetableof contentsandinthemarginsofthechaptersthemselves,drawnlikethis:

There’salotofjargonbothinhealthcareandindigitaltechnology,oftenmakingthingshardertounderstand.Sometimesit’shardtoknowwhen whatappearstobeanordinaryeverydayphrasehasaspecialistmeaning.So, whenIintroduceaspecialistterm,it’sbeenhighlightedin bold toavoidany confusion.

a SeeChapter 35:Healthcareopennessandacknowledgments,page 553 →

1Howtoreadthisbook 1

Digitalhealthcareismuchriskierthanwethink,butitcanbe madefarmoreeffectiveandmuchsafer.Thisbooksplitsup theactionintostoriesofproblems,thesolutions,andthenthe betterfuturewecanreach.

PartI Diagnosis ⋄ Riskierthanyouthink

2Wedon’tknowwhatwedon’tknow 15

Forthousandsofyears,healthcarewasheldbackbecausewe couldn’tseeanddidn’tunderstandthegermsmakingusill. Today,healthcareisbeingheldbackbecausewedon’tsee computerbugs,andwedon’tunderstandtheriskscausedby them.

3CatThinking 25

CatThinking explainsourloveofallthingsdigital.Our hormone-drivenloveoftechnologyoverridesobjective thinking.Thinkingthatcomputersarewonderful,wefeelwe don’tneedtoworryaboutlookingforrigorousevidencethat theyaresafeandeffective.

4Dogsdancing 33

Lookcarefullyforthem,andyou’lluncoverlotsofstoriesof digitalhealthcarebugs.Thischapterhaslotsofexamplesof buggydigitalhealth.

5Fataloverdose

DeniseMelansondiedafteracalculationerrorthatledtoa drugoverdose.Whatcanwelearnfromtheincident?

6SwissCheese

SwissCheesefamouslyhasholes,whichcanrepresentthe holesandoversightsthatleadtoharm.The SwissCheese Model hasbecomeapowerfulwaytohelpthinkmoreclearly abouterrorsandharm.

7Victimsandsecondvictims

Whenpatientsareharmed,staffoftengetblamed— especiallywhennobodyrealizeshowdigitalsystemscango wrongandcreatetheproblems.

8Sideeffectsandscandals

WeacceptthatmedicalinterventionslikedrugsandX-rays havesideeffects.Itmakesalotofsensetothinkofdigital healthcareashavingsideeffectstoo,andthereforeitshould beevaluatedandregulatedascarefully.

9Thescaleoftheproblem

Wedon’tknowhowmanypeoplearedyingorbeingharmed fromerrorsinhealthcare,letalonethosecausedbydigital errors.Whatarethefacts,andwhatcanwedoaboutit?

10Medicalappsandbugblocking

Medicalappsareverypopular,buttheyareaspronetobugs asanyotherdigitalsystem.Thischaptergivessometypical examplesandbeginstosuggestsolutions.Likealldigital healthcare,appscouldbedesignedtoblockbugsandavoid theharmsthatfollow.

49

61

69

81

109

121

11Carsaresafer 137

Thecarindustryhasmadecarsmuchsafersincethe1960s. Whatcanwelearnfromcarsafetyandfromwhycarsafety improvedtohelpimprovethesafetyofdigitalhealthcare?

12SafetyTwo 145

Focusingonthebadstuffisthetraditional SafetyOne approach.SafetyOneisunconstructive.Instead, SafetyTwo meansfocusingondoingmoregood.SafetyTwoemphasizes doingmoregoodthingsandthereforesqueezesoutthebad things.

13ComputationalThinking

There’salotmoretodigitalhealththanbeingexcitedabout digitalcomputing.Weneedtolearnhowtothinkcomputationallytotakefulladvantageofdigital. Computational Thinking isthematurewaytothinkaboutcomputing—and digitalhealthcare.

(Don’tforgetthatthecomputerchip meansthatthisisa moretechnicalchapter.)

14Riskycalculations

Drugdosesandotherformsofpatienttreatmentrequire detailedcalculations.Calculationerrorsareoneofthemost commontypesoferrorandtheycouldbereducedinmany ways.Calculatorsthemselvesignoreerrors,andtheyshould befixediftheyaregoingtobeusedinhealthcare.

15Who’saccountable?

Softwarewarrantiesgenerallydenyallliabilityforproblems. Manufacturersanddevelopersshouldberequiredtobemore accountable.Everyoneneedstobeconstantlycuriousabout improvingsystemsandreportingproblems.

151

177

193

16Regulationneedsfixing

Digitalhealthcareneedsmuchbetterregulation—and regulationneedstokeepupwiththeuniqueissuesofdigital healthcare.BetterregulationisaSafetyTwoapproach: regulateforbetterprocessestostopthingsgoingwrong.

17Safeandsecure

Cybersecurityisaseriousproblemforallcomputersand digitalsystems.Inhealthcare,patientsafetyisparamount, butinthewiderworld,securityhasahigherprofilethan safety.Bothhaveproblemscausedbypoorprogrammingand allthedesignanddevelopmentprocessesthatprecedeactual coding.

201

211

18Whoprofits? 225

Whoisprofitingfromourdata?IsArtificialIntelligence(AI) thesolutiontobetterhealthcare?

19Interoperability

Interoperability —or,rather,lackofinteroperability—isa besettingprobleminhealthcare.Weneeddigitaltowork seamlessly—tointeroperate—acrossallspecialties, disciplines,andhealthcareinstitutions(takingdueaccountof privacy,cybersecurity,andsoon).Itrequiresnewthinkingto getthere.

20HumanFactors

Understandinghowhumansmakemistakesinpredictable waysisthefirststeptowardsmakingfewermistakes.This appliestoclinicians,andmostespeciallytoprogrammers— whosemistakesendupasbugsaffectingthousandsofusers.

21ComputerFactors

Understandinghowcomputerscanavoidbugsandmistakesis thefirststeptowardprogrammingsaferandmoredependable systems.Thischapterintroducessomeimportantsoftware engineeringideasthatcanhelpmakesaferdigitalsystems.

245

259

277

22UserCenteredDesign

Howevergoodacomputersystemis,itstillneedstodowhat’s needed—notwhatwethinkisneeded. UserCentered Design findsouthowpeoplereallyusesystems,andhowto improvetheirexperienceandreliability.

23IterativeDesign

UserCenteredDesignmeansfindingouthowsystemsare usedwiththeirrealusersdoingrealtasks.Theinsightsfrom workingwithusersleadstodesigninsightsandwaysto improvethesystems.Theseideasareformalizedinthe importantideaof iterativedesign

24WedgeThinking

Developersandprogrammersneednoqualificationsto developdigitalhealthcaresystems.Weneedtodevelopa qualificationstructurefordigitalhealthcare,anddomuch moreresearchondigitalsafety.Bothwillhaveahugeimpact onfrontlinesafety.

25Attentiontodetail

Whyispoor-qualitysoftwaresowidespread?Simplebugs mightseemtrivial,buttheyareverycommonanddon’thelp patientsafety—theymakeeveryoneinefficientand error-prone,ifnothingelse.Healthwouldimproveifwepaid attentiontodigitaldetails.

26Planesaresafer

Aviationsafetyreliesongettingverycomplexengineering right,andit’sgettingsaferandsafer.Whatcandigital healthcarelearnfromaviationandaviationengineering?

27Storiesfordevelopers

Weshouldprogrambettersothatdigitalhealthcaregets safer,whichisaSafetyTwoapproach. FormalMethods is widelyusedinsafety-criticalindustries,butnotoftenenough inhealthcare.Here’swhyFormalMethodsisneeded,and howitworks.

301

313

325

337

347

367

28Findingbugs 383

Althoughithelpseverythingelse,FormalMethodsisn’t enoughonitsown.Thoroughtestingisessentialtoensure thingsreallyworkwell,especiallywhenthingsaregoingtobe usedincomplexenvironmentslikehealthcare.

29Choosesafety 401

Let’shaveareliablewayofclearlyseeinghowsafesystems are,sothatwecanmakechoicesbasedonevidenceand improvesafety.

PartIII

Prognosis ⋄ Abetterfuture

30Signsoflife 417

We’veemphasizedproblemsandsolutionstoproblems,but ofcoursedigitalcandosomefantasticthingstoo.This chaptercollectssomepositivestoriesaboutdigitalsuccesses andhowdigitalcantransformlives.

31Thepivotalpandemic? 437

ThehorrificCOVID-19pandemichasforcedhealthcare systemstoinnovateindigitalhealth.Somechangeshave beenamazing,liberatingpatients,andprotectinghealthcare staff—butsomehavebeenratherworrying.Whatcanwe learn?

32Livinghappilyeverafter

There’safutureworldwheredigitalhealthcareworks,and workswell.Here’showtogetthere.

33Goodreading

Thisbookisn’ttheendofthestoryaboutdigitalhealthcare, anditsproblemsandsolutions.Thischapteron recommendedreadinggiveslotsofsuggestionstohelptake yourthinkingfurther.

34Notes 497

Supportingthisbookareover500notesandreferencesondigitalhealthcareandpatientsafetyincidents.Thesenotescover mediastories,peer-reviewedcutting-edgeresearch,aswellas nationalandinternationalreports.

Notesin bold areespeciallygoodsourcesforfurtherreading.

35Healthcareopennessandacknowledgments 553

Traditionalhealthcareviewsofconfidentialityarechallenged bydigitalhealthcare.Thischapteralsoincludesheartfelt thankstoallthepatientsandhealthcarestaffandothers who’vetoldtheirstoriesandbroughtthisbooktolife.

HaroldThimbleby 557

Fontsforcancer 559

Index 561

Boxes

Box2.1Malware,Trojans,Bugs,Viruses… 19

Box2.2FixIT—thebigpicture 21

Box3.1Successbias 29

Box4.1Bugsareoftenobviousinmanuals 45

Box5.1Usingacalculator 51

Box6.1Programmingwithbettercheese 64

Box6.2Designerrorscauseuseerrors 65

Box7.1TheBlameGame 75

Box8.1BRAN:Benefits–Risks–Alternatives–doNothing84

Box8.2CalculatingdrugdosageonaGrasebysyringedriver87

Box8.3Designtrade-offs 88

Box8.4SideeffectsandthePrincipleofDualEffect 90

Box8.5Designawardsignoresafety 96

Box9.1Risksofmakingcomputers“easiertouse” 112 Box9.2WHO’sglobalfactsonpatientharm 114

Box10.1Alwayssay“useerror”not“usererror” 123

Box10.2Usingwilddataasaworkaround 134

Box11.1Theproblemsofbuyinglemonsandsellingpeaches141

Box12.1Whatencouragessuccess? 146

Box12.2Theorange-wiretest 148

Box12.3Nevereventsandalwaysconditions 149

Box13.1TheBritishEDSACcomputer 157

Box13.2Reproducibilityisessential 161

Box13.3Medical“algorithms”aren’tdigitalalgorithms 173

Box14.1Handhelddevicesmayhaverealbugs 178

Box14.2Austeritybyspreadsheet 185

Box14.3Solvingahindsightproblem 188

Box16.1Drugregulationandfakedrugs 206

Box17.1Ransomwareandcyberextortion 213

Box18.1Simpson’sParadox 231

Box18.2Problemswithspellingcorrection 237

Box18.3Digitalandpharmaceuticaldevelopmentcosts 241

Box19.1Addingvalueormanagingrisk? 251

Box19.2Interoperabilityisn’tjustadigitalproblem 254

Box20.1Magicmakesdigitalhealthcaresafer 262

Box20.2SimplisticHumanFactorsbackfires 268

Box20.3Risksofinternationalcodeandpoorcodereview 273

Box21.1Theetymologyandentomologyofbugs 278

Box21.2Goodprogramshaveassertions 291

Box21.3NewerComputerFactors 298

Box22.1Howmanyusersisenoughforsafedesign? 304

Box22.2Cloneddocumentation 310

Box22.3ExternalizingUserCenteredDesign 311

Box23.1Howdoyouknowwhennothinghappens? 314

Box23.2DonnaMeyer’spersona 322

Box24.1ParallelswiththeGermanEnigma 329

Box25.1Digitalinternationalization 344

Box26.1ThescandalofAlternativeSummaryReporting 353

Box26.2Anaviationanalogy 355

Box26.3TheDunning-KrugerEffect 358

Box26.4Legoworkshops 359

Box27.1Epic’sdaylightsavingandY2K 373

Box27.2Nevereventsandgoodprogramming 376

Box28.1Don’tusebadprogramminglanguages 385

Box29.1Exampleevaluationcriteria 409

Box30.1Earlycomputerdiagnosis 424

Box31.1The1918SpanishFlu 444

Box31.2Simpleethicalquestions? 450

Box32.1IntroducingtheIEC61508standard 460

Box33.1DigitalchaosnearlyhadadoctorremovedbySecurity493

Forthousandsofyears, healthcarewasheldback becausewecouldn’tseeand didn’tunderstandthegerms makingusill.Today, healthcareisbeingheldback becausewedon’tsee computerbugs,andwedon’t understandtheriskscaused bythem.

Wedon’tknow whatwedon’tknow

WorkinginViennaGeneralHospital,backinthe1840s,IgnazSemmelweis noticedthattwomaternityclinicshadverydifferentdeathrates:onedeath ratewasdoubletheother,andhis,unfortunately,wastheworseone.2 Semmelweisstartedtostudyeverythingtotryandworkoutwhatthereasonswere.Manymothersweredyingofthehorribleandusuallyfatalpuerperalfever.3 Hediscoveredthattherewerelowerdeathratesinthesummer. Thenhenoticedthestudentdoctorswentdowntowninthesummer—but inwintertheypreferredtostayinthewarmhospital.Thenhenoticedthat whenthestudentswereinthehospitaltheyattendedpost-mortems.

Hegraduallycametotheconclusionthatthings,whichhetentatively called“cadaverousparticles,”werebeingcarriedbythestudentdoctorsfrom thepost-mortemsaroundthehospital.Thestudentdoctorsexamineddiseasedbodiesinthemorgueandthenwalkedovertoseepatientsonthe wards.Today,wewouldcallthatprocesscross-infection,butSemmelweis hadnosuchmodernconceptstounderstandwhatwasgoingon.Nevertheless,heinstitutedhandwashingtostoptheparticlesgettingaround (figure 2.1).Thedeathratedulyratewentdown.

Beforehandwashing,thematernaldeathratehadaveragedabout10% (deathsperbirths),andwassometimesover30%inwintermonths.After institutinghandwashing,Semmelweiseventuallygotthedeathratedownto zeroforacoupleofmonths,despitehaving537birthsinthesameperiod.

UnfortunatelySemmelweislosthisjob—hiscolleaguesfoundhimirritating.Thesuccessdidnotcontinue,anddeathratesroseagain.Semmelweisfinallydiedinignominy.Today,however,heisahero,especiallyin midwiferyandstatistics.It’sinterestingthathisveryearlyuseofstatisticsin healthcareuncoveredthecauseofaproblemandhelpedfindasolution,yet withouthiseverunderstandingtheinvisiblemicrobesbehindhisdiscovery.

Riskierthanyouthink

Figure2.1. AromanticizedpictureofIgnazSemmelweisoverseeinghandwashing inhishospitalward,sometimearound1840.

Allcredittohim,Semmelweis’sobstetriciancolleague,BernhardSeyfert, decidedtodoanexperimenttoo.Seyfertfoundthatwhenhegothisstaffto washtheirhands,thefrequencyandseverityofdiseasedidnotimprove.

Why?

Seyfert’sdoctorswereonlygoingthroughthemotions:theywereonly dippingtheirfingersinthewater.But,crucially,theyhadallbeenwashingtheirhandsin thesamewater—andafterafewdaysofuseithadbecomeopaque!WenowunderstandSeyfert’sproblemeasily:asmuchashe mighthavethoughthisdoctorswerewashingtheirhands,theywereactually cross-infectingeveryone.Eventhedoctorswhodidn’tgotopost-mortems werenowgettinginfected,probablythroughhandcontactwiththosewho did“wash”theirhands,orwiththingsthatwerealreadycontaminated.

Today,wetakecleanwater,sinks,taps,andwashinghandsforgranted. Wetakecleaningsurfacesforgranted.ButSeyfertdidn’tevenhaverunning tapwater.Seyfert’soriginalhygienemusthavebeendreadful,seeingashis experimentapparentlydidn’tincreasedeathrates.

Semmelweishadshownthattherewasarigorousinterventionthatsaved lives,basedonevidence.Itwasonlylater,withthedevelopmentofLouis Pasteur’s germtheory,thattherewasagood explanationforwhythein-

terventionworked.Germs,whetherbacteriaorviruses,causediseases,but ifyoudon’tknowaboutbugs,thentheinterventionsdon’tmakesense.

Wearestillmakingprogresswithmorecuresforbugs,andwearestartingtorealizetheproblemsofover-prescribingantibiotics,whichcausebacteriatoevolveandgethardertotreat.Ofcourse,nearlytwohundredyears later,handwashingisoneofthefirstlinesofdefenseagainstspreadingthe COVID-19pandemic.

Inthis,thetwenty-firstcentury,wearestartingtoseethatthereare other invisiblebugsthatalsoaffecthealth.Wedon’tunderstandcomputerbugs, letalonetheircures,andpeoplearebeingharmedandsomearedyingunnecessarily.

AbitlikeSemmelweis’swell-meaningcolleagues,wetendtohangonto ourloveoftheoldwaysratherthanfaceuptothefactthatmaybewecould bedoingbetter.Weneedtorecognizethefactthatdigitalhealthcarehas bugs,andthesedigitalbugsmakehealthcarerisky innewways.Digitalis everywhere;digitalbugsareeverywhere.

Untilwegraspthat,andtakeamorematureapproachtomanaging“digitalhygiene”weareasgoodaswashingourbuggyhandsinthesamewater aseveryoneelseandjustmakingthingsworse.

Somepeoplegetfussyoverwhatacomputerbugis,andsayprogramsare onlybuggywhentheyfailtodowhatwasspecified.Inthislight,bugsare programmingerrors—weknewwhatweweresupposedtodo,butsomehow thingswentawry.InthisbookIwanttotakeabroaderview.Whencomputersdothewrongthings,wecallthosethings bugs.Strictly,thebugsare theerrorsthatmakethecomputerdothewrongthings.Thewrongthings themselvesarethesymptomsofthebugs,butit’sstraightforwardtocallthem bugstoo.

Imagineadigitaldevice,likeyourphoneoradruginfusionpumporyour laptopcomputer,anditjuststopsworkingandbecomesunresponsive.It lookslikeit’scrashed.

Thisisclearlyabug.Oftenifyouswitchiton,oroffandonagain,it’ll resetitselfandyoucancarryon.Resettingitclearsthedevice’smemory, andhopefullyremoveswhateverproblemsthebugcausedorwasconfused about.

Ormaybeit’sstoppedworkingbecauseithasaflatbattery.Isthatabug? Diditwarnyouthebatterywaslowsoyouhadachancetofixtheproblem? Itwasabugifitdidn’twarnyou,especiallyifyouwerelikelytolosework fromtheproblem.

Let’ssayyourechargethebattery,butthedevicehasstilllostyourwork. Yourworkwasprobablystoredinvolatilememory,whichislostwhenthereis nopower.Adifferentengineeringdesignchoicewouldhavehadeverything storedinnon-volatilememory(likeadisk)soitshouldneverdisappear. Ifyoumeettheprogrammers,theysaytheydidexactlywhattheywere

18 | CHAPTER2

toldtodo—theysaytheycorrectlyimplementedthespecification.Soif you’vegotaproblem,it’snotwiththemortheprogrambutwiththespecification,andthat’snottheirfault.They’darguethatsincetheyweresupposed toprogramit,ifitworksastheythoughtitshould,thenitcan’tbeabug!

Soalthoughwecanthinkoflotsofdifferentexplanations,theendresult isabug.Iftheusercan’ttellthedifference,orcan’tworkoutthecause, whetherit’sadesignfaultorasoftwarefault,let’scallitabug.Itshouldnot havehappened,andtheuserisinconvenienced.

Aspecialcaseofbugisnotamistakebutisa deliberatedeception.Like anyotherbug,themanufacturerhopesnobodynotices,butthereissome advantageforthemanufacturer(orforsomeonewhoworksthere).AnexampleofthisisPracticeFusion’ssystem.Theyhavebeenfined$145million becausetheydesignedinfeaturesintotheirsystemsoitincreasedprescriptionsforaddictiveopioiddrugs,eventhoughover-prescriptionsofopioids isawell-knownpublic-healthdisaster.Pop-upsthatsupposedlyprovided objectiveclinicaladviceweredesignedtonudgedoctorsintoprescribingspecificdrugs:theyweredesignedassubliminaladverts,notforgivingprofessionalclinicaladvice.ItisestimatedthatthePracticeFusionsystemdeceptivelyboostedopioidsalesforonedrugcompanyby$11.3million.4

ThePracticeFusionsystemwasusedbytensofthousandsofdoctors. Fromthedoctors’pointofview,thissubtlemanipulationisatotallyunwantedfeature—abug—thattheyhadnoideaaboutwhenthePractice Fusionsystemwasacquired.JamieWeismanhaduseditforfiveyearsbut doesn’trecallnoticingthemanipulativealerts.Shewasreportedassaying:

It’sevil.There’sreallynootherwordforit.Butifyouwantto modelelectronichealthrecordsasafor-profitsystemandnot regulatethemassuchandforcedoctorstobeonthem,it’s almostinevitablethatthey’regoingtobemanipulated.4

Thenthereare malware systems,whicharebuggysystemsdesignedby criminalhackersfordeliberatesabotage,tocausechaos,blackmail,ortosteal information.Thehackersmayworkforthemanufacturer,butmoreoftenthe hackersworkfarawaytotakeadvantageoftheinternettohackthroughinto hospitalsystems—see box2.1.Criminalhackersalmostalwaysgetmalware intoyoursystemsbyexploitingbugs,thoughsometimestheytrickusersinto takingafewstepsforthem,likeusingtheirpassword,whichthenallowsthe malwaretorun.We’lltalkaboutsomeexamples—likeWannaCry,ahuge malwareattackthataffectedmanyhospitalsworldwidea —laterinthebook.

ForthisbookI’mnotgoingtobepedantic.Bugsarethebitsofdigital systemsdoingthewrongthings,andwearen’tgoingtoworrywherethese thingsgowrong,becausewhereverbugshappeninhealthcaretheyneed fixing.Thisisamorerelaxeddefinitionofbugthanmanypeoplemightlike,

a SeeChapter 17:TheWannaCryattack,page 211 →

Turn static files into dynamic content formats.

Create a flipbook