When this chapter was written, Drs Roger Glass and Eric Noji were medical epidemiologists at the Centers for Disease Control and Prevention in Atlanta Georgia. Dr. Glass is currently Director of the Fogarty International Center at the National Institutes of Health in Bethesda Maryland. Dr. Noji is currently Chief Executive Officer of Noji Global Health and Security LLC based in Washington DC and Geneva Switzerland
2. 14
Epidemiologic Surveillance
Following Disasters
Roger I. Glass and Eric K. Noji
Disasters are extruoi dinary nntui al or munrnude catastrophes that de
stroy properly, disrupt society, and cause significant morbidity and
mentality, and which can therefore overwhelm the capacity of the
affected population 10 function normally. Over the past two decades,
natural disasters VOl ldwide have killed more than 3 million people ami
adversely affected the lives of 80n million.' In recent years. more 11.. 11 .
)0 million refugees. primarily in Asia and Africa, have been displaced
when Ilecmg violence, civil wui , political unrest. OJ' f.uninc.2 Due to the
massive impact or drsnstcrs on hUIIl.1Il health, the United N,llioJ1'>
General Assembly has declared the I'J91h to he the lntcrnntionul Decade
of Natural Disaster Reduction, and has called for a global ellm I to
reduce the impact of these untoward events.' If this effort is to succeed,
it will rcquii e considerable involvement by cpulcmiologists to deret minc
how such reductions can be achieved.
The term "disaster" covers a range of events that drustlcally niter
our natural environment. fmm geologic activity (e.g .• cunhqunkes and
volcanic eruptions) or severe weather (e.g., droughts, floods, and luuIi·
canes) 10 manmade calumnies (e g.. tires, chemical accidents. or war s].
They can occur at one moment in time (e.g., tornados. eurthquakcs) 01
continue for long periods (e.g .• III oughts, radiation accidents). Some can
be predicted well in advance (e.g., lloods, hurncanes), while others cannot
[c.g., em thquukcs) Some occur wuh gr cul Ircqucncv (e .g.• lor lliH!O!o"
hun icuncsj while others me extremely nu e (c.g .. reactor meltdown}. Fi
nally, some uffec! millions of people (e.g .. Iamine, earthquakes! while
others can affect relatively few (e.g., bridge collapse). The unifying feature
of disasters is the catasu ophic effect these events C.,11 have <III (hose that
* Public Health Surveillance. New York: Van Nostrand
Reinhold. 1992: 195-205.
3. have experienced them ami 011 a sudety's ability to respond. The health health events. optimize the relief response,
monitor Ihe effectiveness ofthe
impacl will vary greutly, as will Ihc population's needs[0 return the society relief effort, IIIIlI provide rccummcndutions
10 decrease the consequences
to working order. 4 The epidemiologist CUll provide I irnely assessments of from future disasters." The luw mortality associated with recent disasters
Ihe health problems related 0 the disaster in order to assist in mounting an such as hurricanes Gilbert (1988) and Hugo 11989), the San Francisco
clfc I.:Iivc, appropriate relief response. amlin prevent similar consequences earthquake (199(l). and tornados in North America represent the SllCCCSS
1'1'0111 luture disasters.' uf programs uf weather forecasting and warning. disaster preparedness.
emergency medical response. public awareness. and aseismic huildiug
codes: some or these efforts likely resulted from knowledge gnlhercd
OBJECTIVES
through epidemiologic studies conducted as a consequence ul' previous
EpiticlIliologit: surveillance "ncr .. ttisastcl' involves lhe rapid assessment disasters.
of Ihe distribution lind dctcrminnnt« ofdisaster-related deaths. illnesses,
nud injuries ill the population affected, witlt the aim or determining their
IlIo~1iuuncdiutc problems und uuuchiug th ese with a planned anti appro SPECIAL CONSIDERATIONS FOR
printc response." Fphlcll1illlogists playa vital role ill developing reliable DISASTER EPIDEMIOLOGY
inluruuuiun Oil the health consequences of the disaster, conducting sur
veys Hilt! investigations where necessary. providing advice un health prob
The basic principles of epidemiologic surveillance alter Iisastcrs arc IIll
different Irom surveillance applied in other settings. lei However, in P1:1I..:
lcms (hilt may arise, cstuhlishing priorities for action, and in elllphasizing
tice, the timelramc is often reduced if 110 instuntaneuus. epidemiologic
lImt proper decision-making requires timely am] appropriate infonuutiun."
information is incomplete, dccision-muklng is by nature hasty, Ihe relief
The ulrimate goal of epidemiologic surveillance is to prevent or reduce the
response may be massive. visibility call be extensive. and chaos is ubiqui
adverse health consequences of the disaster itself as well as to optimize
tous. The surveillance cycle must tum many times: first with rapid. cursory
(he decisiou-making process associated with management of the relief
assessments or problems using the most rudimentary data collection tech
effort. These epidemiologic objectives can be simply defined as the surveil
lance cycle; thai is, the collection of data. analysis of data, and response
niques, followed by short-term assessments involving the establishment
to dala.7 or simple but reliable sources of data, and then with ungoing surveilluncc [0
The need for disaster epidemiology was apparent in many early disaster identify continuing problems and monitor the response to the interventions
relief operations. Managers and planners with no public health expertise chosen, Finally. in the aftermath of a disaster, focused analytic surveys
and no reliable information on the health of the population struck by can be used to compare victims with survivors and learn what could be
disaster were Iorced to mount major relief efforts. In the absence of an done to prevent the human loll of morbidity in subsequent disasters, The
adequate field assessment. their response was often dictated by the relief success of the epidemic investigation of a disaster can be measured directly
and medical assistance made available by donors. This led to the arrival by how rapidly data collected and analyzed can identity prevention strate
on the disaster scene of outdated ur inappropriate drugs. medical and gies. and how effectively these strategies can then he implcmcutcd by
surgical teams without proper support, and relief programs thai did not decision makers to direct relief and decrease ongoing nH.~I:bidity. This effurt
address immediate local needs. These problems were all compounded in requires active coordination bet ween the epidemiologist. who gathcrx the
(he vacuum created by the disaster, including the lack or communication, data and identifies the issues or strategies. and the decision maker. who
transportation, local supplies and support, and It decision-making struc must understand the data and strategies and implement the required poli
lure. Since these relief operations were oftcn conducted under the watchful cies. In the rapid evolution of a disaster relief program, major decisions
eye of the media, medical relief efforts were often pejoratively called "the regarding relief are made early on. hastily. and often irreversibly; thus,
second disaster. the need for reliable early data tll assist in making these dccisiuus is crucial.
In recent years, epidemiologic techniques have been effedively intro A decision maker who questions a proposed intervention nr donation can
duceil as a basic component in many disaster relief operations. Epidemiol also use his epidemiologic capability to provide an independent and rapid
ogists have been able to define quickly the nature ami extent of the health assessment of need.
problems. identify groups in the population at particular risk for adverse Field surveillance methods vary by disaster setting and the personnel
4. ami lime i vuilablc. Eat Iy Iicld stu vcys 1111151 lulluw the KISS princi been implemented, and 1he pOln/lat ions thus Icmnin ruor c vulnerable 10
ple-Keep II Simple, StllPid-and addl"css the essential, 111051 baste adverse health consequences !'IOtII natural di~ilSfet s.
quest ions Iequiling immediate answer .. thai will directly prevent loss of
life III injury. Subsequently, SUI VCYS can mllh C'iS issues such as the
avail:lbiliryor IIlctliclll CHIC, assessment of the need fur specific intcrven EPIDEMIOLOGIC ISSUES FOR
lions. und epidemic cunn ol (e.g .. establishment of n rumor clearinghouse DISASTER SURVEILLANCE
to IC(eiVC illlol'lilalioll on epidemics of diseuse Hllll provide Ior their
Famine Relief
1imely invexrlgntion}, each of which demands more careful investigation.
Surveillance I11l1sthe sensitive 10 monitor the impact of relief on the In 1957. Sayler and Gordon, in one uf the earliest review .. of Ihe rule of
health problems or the population. and to determine whether the effort epidemiologic assessment after nat ural {lis"..tcl S. com pal cd disasters 10
i~having u tangible impact 011 the population or if new strategies me epidemics nnd suggested that they could be described in clus sic epidell1io
needed. Surveillance becomes an iterative. cyclical process in which logic terms uf time., place. and person." This ubsci vation WHS sunila: 10
simple health outcomes [II e constantly monitor ed and interventions our change in thinking about accidents. which arc 110 longer fell 10 he
cuntiuually assessed Ii.H elllcacy. I<IHJOIll events but have been renamed unintentional injut ics SUhjCL1 Itl
I he attention span Ior I elief after 1I1usl natural disasters is short, mea epidemiologic scrutiny. This concept Was applied in the lute I%n...o as<;isl
t
SUrill!,! weeks to muuths , whereas the consequences for the populution the massive international relief operation moun led In respond til [umine
tilreeled can measure year s 01 decades," The Him of I elief must he tu help conditions associated with the Nigerian civil w(lr.~ Epilicl1ltolpgists dcvcl
Ihe population quickly reuu n to their predisaster state while investing oped new SUI vey tools and survey methods torapidly as<;c~s the 111111jl ioual
I clicf money uud aid in ways to CIlMII e the gl eatest long-term effect. 12 In status or huge displaced populations so lhal rehcf could he tlHl!elcd III
lhe early phase or rel lef', basic need" Hr water, food. clothing. sheller. anti those gl oups in greatest need, Subsequently. SUI vcillance was CI iticul to
medical cut e must he mel. nltcr which the longer-term process of rebuild monitor the nutritiunal status of (he pupululion in I CSPOIlSC 10 the 4t1anlil y
illg proceeds. Relief aid can often be squandered early on overreacting
by (IIIti types or louds deliver ell, Rapill epidemiologic HS~C<';~I1IClit pi ovcd
10 millOl problem .. when excitement is great, needs ale extensive, and invaluable 10 optimize food distubution practices ill the face of rupully
"l runny hy the media is umnipi cscnt. (1 Longer-term goals are often over changing conditions of health ami relief', Since then. IlUlllli)Il!l1 ;0;111 veil
looked even Ihtlligh they al e generally more difficult 1111[.1 costly to achieve lance has become a routine parI of teller wurk In famine atea" arnl i'l
and their impact can be longer lustiu]; Epidemiologic assessment. priorlti refugee populatlons ami is essential to rationalizing pr oblents of 10m]
zat inn of needs, lind planning an appl orl iate response can have a major lIistl lbution.
hcnclki,,1 elfcct 011 the ability uf a community to return to normalcy ill
both Ihe shutt and the longer leI m.
1 repealed ohset valiou [rom many I CCCIII disasters is Ihal [he health EPIDEMIC CONTROL: A RUMOR CLEARINGHOUSE
consequences of these e vents full most heavily 011 people living in devel
Epidemiologists have subsequently becomc il1Vlllvc'd In other :I.~pcch or
oping eOllntries.14 For example, cmthquakes measllring 6-7 on the RichIeI'
Scnle le~1lu massive loss uf life ill PerU ~197tl), Nicaragua (1912). Guate· postulsastcr assessment The Biblical fcar or epidclllk tii"!eile Illiowillg.
s
mallll 1976). Tallgshan, Chill:! (1976), and At"menia (1988), wherea" quakes disasters led many decision makel s to enlist cpidcllllologj"1 s 10 inVC.,1l!(a1 c
of 'iimi!:lr magnitude ill California weI c associaled with few health conse~ tumors of cpidemics that werc frequently repOl ted. For example. following
(luenccs despite considerable loss of properly, Clearly, industrialized disasters in I,[eveluping clluntric:o;, any disruption of the Wt1ter .,upply 01
cOllnlries :lIe buffeled from disuslers by their "bilily to forecast severe sewHge trcatmcn! was usually accolllpanied by rumors of olllhreak~ or
slnrms, enforce sll ict codes fm aseismic and flrcpl'uur conSlruclion, ulilize cholera or typhoiJ. Such rumors may weB IliIvc rcl1ecteu p:-,ychlliogical
cOlllmunication networks tu bW[lUCOlstdi:mster warnings and alerts. pro~ fears and anxieties :lbout a disastrous evcnt rather than tile true perception
vKlc emergency medical ~CI vices, mlll engage ill contingency rlanning 10 or an imminent prublem. Such epidemics are IlnCOIlIlll(ln in the wake or
prcpare the populalioll and Pllhlic in'itilutinn" for pos ..ible disastcl s. In Ilalural disaslers bul can ucellr in settings such <ISIl'fU{!CC call1p~, whele
developing Cllllntl ies. stich lIleasUl'cs 1IIe eilher nol aVllilable l)r huve nol large pupullliion" l~rtli~pJacell pel"~illll~ (If e crowded logcll,cl' ;1IIl) ~h;J(c
5. Ullsanilary t~lIldiliomi m contnuunmcd wntcr.!" Nonetheless, u clearing The second HIIHof survcillnnce was 10 momtor mOltalily lIlIt! IlHHhidil'
house call SCI'v'C I he irnpm tan! fundioll of monitoring such runun s as they 10 nsccuain whether the relief effort was effective, In lite absence (II'
:11iSl:. illvc~ligalilJg those th:lI have 'lleril ill it lilllely i;]"hinn, d;srellillg epldcnuologlc rlnln, many IIIcdi'l Ieprese niativcs Iksu lhctll he rci'lIgcc ~ Wi
'IIOSC Ihal ale «bviously false. aud Inforllling (he public I)f Iwzall.ls wher c a living in "death camps," and associated this cnlllliliun with a relief cfhlll
respollse i.'i required. "his concept has been helpful nul only in developing lhal was f;,iling by nol immediately preventing deaths.
vountrics but also in disasters UCCUI'I ing ill urban settings or industrialized Epidemiologic surveillance rapidly provided dHIa on the nile,> nldcuth,
couun ies. identifled mulnria as the PI incipal cause or
death and serious huvpitulizn
tlon, ami leu ID specific sinucgies for Ihe ;I!:!"!lc!i.~ivcrrcuuueut ofccr chrul
Surveillance for preventable illnesses, deaths malaria, the primary cause of death. The swift decline ill 1J100Ialily lht! illg
Ihe linit weeks of the effort was directly linked to a relief elTull Illat had
and injuries
correctly Imgeled prevention issues. The collection or simple llala 11 the
'I he health prohlems associated with lll<~iOl' disaster s arc usually more daily number and presumed cause or deaths and admissions to the Iw<;pilal.
extensive than fears of epidemic disenses alone [Jut ate often measured in use of basic field surveys targeted to the specific questions ortelief', aJlJ
counts of rcople who died were severely injured, or became ill. The
, preparation of" brief weekly surveillance report made the relief enol!
epidemiologist mus! identify the most severe health consequences that responsive [0 the prim ity of health needs in the camp and p. uvidcd •ciiallie
can still be ptevcnterl by active. well-targeted interventions. and develop information both for donor orgunizations and the press. The lise or epidc
priori lies to address them for the decision makers. These priorities likely miologic teams to coiled data. identify priorities, ami monitor the cllcc
diller for each disaster. challenging the epidemiologist 10 an ive rapidly liveness or the relief effort has become an inlegl'l.Il part of many intcr na
with 1111 upnrop: iate nlan. FUI eX:II11r1e, because most deaths from earth tional relief and assistance groups." 19
quakes occur dill ing the initial impact. the prevention of subsequent rnor
lality ami severe injury requires early treatment of the lIjured or rapid
ntl'.,cliOI1 of 1l1O~eentruppcd in collapsed buildings." At the same lime, Surveillance of health care needs
nuention must be given 10 earthquake-associated destruction of shelter, In disasters associated with substantial numbers uf victims with seve: c
[lOU and water supplies. road') aud conuuuuication networks, <Inti prob injuries (e.g., explosions. Illnlmlo.<;1 or IlIlIesses (c.g., nuclear accidents.
lcms of access 10 health C;II e so thai the survivors can be spared 1'10111 epidenucs). the ability to prevent death or decrease severe 111111 bidity will
subsequent health problems." Since everyone in the disaster area will feel depend upon the provision of timely and adequate medical ell c, (]I 10 the
needs nnd experience loss. the challenge orthe early assessment is (0 triaging of victims 10 centers where such CHI e rs available. 211. 11 Rapid
decide which needs and alTcctcd mens will lIIosl benefit from early inter surveys of the number of victuns needing special attention and the nann e
ventiun, prcveutiug the gt eutest lu (,11' lire or 11l(lS severe morbidity.
ss of the injtn ies or illnesses will have a lIil'CL:L impact 011 the rCSr)lJII~C 1hill
As an example, {he usefulness of rapid epidemiologic surveillance to can be mounted, Again, identifying the need for aud-rnonitoriug the clfcct
lal gel a relief errorl involving deaths, injuries, and often severe illness was of the intervention ill c important epidemiologic [unctions.
deliwllstHlleti in 1979 when }o.ooo Cambodians mrivcd as refugees ill
Thailand.'9 E~C:lfli[1g from the Val. Ihi .. group aI rived in lhe Thai CHmps
nll1111.'ilcd i"WIIi figilling, ~llul uf [vuu. illjlllCU. llnd hCivily infected wilh Surveys to avoid unnecessary interventions
maim ia. 'illci, higJ. IIWlllllity was Visible 10 Ihe wurld wilen the ,"lcrlla Anct disasters, many agencics and donors olkr supplies, CllUiplltcnl, alld
lional media IcplJrled that dead bodies Were collccteu each morning for PCI sonnd for I clief thm
£II e nul always I cquired, 1"01 example, the 1)c1ive. y
hili iar. A nmssivc inlellmliollall elier openl1 ion was begun, but no informa of unllecesstl.Y, uUh.lated, ur unlabeled Jlugs tu alreclcd mcas llits been
li(lIl was available eal'ly on to determine whether telief e(folls should be doclIlJlented repeatedly following past disHsters alld is Jell illlcnCal 10 Ihe
1~1l .elcoJ ID chilthell or auuhs. or lo problems
g of malnutrition, immllni,.<1- relief errort, causing a diversion of personllel to iuentify I clcvant stJpplic'i
lioll, II calmelll of WHr i1uurjcs, or 4,:Ulllr-ol of malada (Inti oHler cplclcmic from II 111i1,~s ullnecessary IIlHlcrialY Vm.:cines rOI cholera and typhollf
of
(Iisca'ics. '1he immeuiate ajlll of SUI veillance was to idelltify preventable fever have never been needed or effectively used followill!,! iI ,Iisas(ci
causes of t1eath illullcdialely and 10 dccide Oil the lirst priorities for relief. but ale repeatetlly offered, placing politician'i amI It)cal pCI'~onnd ill Ihl.:
6. uucumlurtuhle hu! correct posilion uf saying "110." Disasters also often protection from the destructive forces of carl hquakcs. More analytic stud
prompt an altruistic urge among health profcssionuls. For example, no ies such as these are needed 10 test convcntlonal warnings tuul advisories.
fewer than 30,000 physicians anti nurses frum the United Stutes, Europe,
Latin America, nud Asia volunt ered to work with Cambodian refugees
e Anniversary Analysis of Relief Efforts
ill 1979- 198(l. The needs were limited in numhers , people with special
skills lind experience were required. and the efforts to select proper per The long-term health consequences of dism,lcrs have never been PHlpc.-Iy
assessed. No evaluations have been made five or ten years folluwing ..
sonnel were oltcn difficult. Depending upon the pressures perceived by major disaster to determine whether changes in epidemiologic or rei icf
dccisiou makers, the epidemiologists can often conduct surveys III assess practices. redirection uf relief funds 10 longer-term goals. or changes in
whether lntcrveutions being volunteered by donors with significant politi behavior or building patterns have had any long-term effect on a cuuuuuui
cal influence are in Iacl required. ty's lung-term response lu disaster. Nonetheless. many communities thut
have experienced disaster are more concerned about preparedness efforts
Analytic Epidemiology: Prevention of in the future.
Consequences from Future Disasters
CONCLUSIONS
In some llisastcrs such as earthquakes. lunwllos. or hurricanes, a majority
of deaths or severe injuries occur at the moment of the disaster itself. For The epidemiologist involved ill Iisaslcr HSSC!;~IHCnl fal:I::S u number 01"
each or these disasters. prevention strategies 'are often recommended specific problems related 10 the political cnvlroruncnt and rapidly chanl!ing
Ihat have never been subject to epidemiologic scrutiny. In recent years. health profile, needs, and opportunities to intervene. Data musl be ":01-
cpidemlclogist« have focused 011 'assessing what strategies work best 10 lectcd rapidly under highly adver e cundiuons. Epidcuiiulogic inlurmatiou
s
prevent such disaster-related morbidlty." The questions raised follow a must be applied to n decision-making process since it can infhrence de
case-cuntrul design: Why did some people die (cases) while their neigh termiuing relief supplies, equipment. anti personnel needed 10 respond
burs. lruuily memhers, or ulhers survived (controls]? Risk factors for effectively. Standardized procedures fur collecting dula in di~aslcls llcl'd
survivalcan range Irom prior knowledge and heeding of disaster warnings to be developed that can be linked to operational deci ions ami actions.
s
(e.g .• tornado alerts), taking evasive action (e.g seeking shelter in a
.• A variety of epidemiologic methods has been denronstratcd tt' be of
stairwell), and availability of emergency medical care, to structural issues value before. during, and alter disasters. Before the disaster. energies
such as the building materials and codes adhered to for housing in an must be focused on delineating the populatiuns atrisk. and 011 asscssill~
earthqunkc-prune area. Such analyses ancr earthquakes and tornados the level olernergeucy preparedness. tile flexibility ofcxislillg surveillance
have each yielded new iulorumtion that has altered traditional thinking on systems, and the training of personnel. During impact, the healt h care
Iht: prevention of disaster-related mortalit y. For example. in the Wichita needs or the alfected populalion ami the needs 1"01' emergency services
Falls tOrtHldu ill 11i79. mnny people L1iedwhile fleeing from the tornado's have to he assessed quickly with the goul of preventing avoidable death.
(lath in vehicles, a recommendation promoted by the weather service at il~ury, or illness. In the post-impact phase. continuous monituring ami
surveillance of the health problems faced by the population arc required.
thai lilllc.2-I An epidemiologic analysis determined that people who were
as well all information on the effectiveness uf relief interventions. After
caught ill motor vehicles or motor homes had !l 10- to SO-fold greater risk
the disaster, epidemiologic methods cnn be used In evaluate in an iterative
of death ur severe injury than those who took shlter at home. Moreover,
e
fashion the effectiveness of e ach health intervention p rognuu. lncOI"plllH
110 severe jl~iurics occurred among people who took shelter in basements
tion uf epidemiologic surveillance and disa ster mnnagcment cau drnnuui
or dc:-;igllated shellers in large public buildings. National advisories for the
cully reduce the health consequences of these catastrophic events 01 the
prevention llf injury in tornados have been changed based upon these
affected population.
lilllling~. Slmiiurly, earthquake-related dCHlhs are directly linked to coo
st rue lion prncrices, cotll'irl1ling Ihe need for aseismic building codes, anti
Notes
10 bchllvioral practices of ncein~ buildings at the lime of preshucks or
1. National Research Council. 1987. Confronting Natural Disasters: an International
e:1I"1 ljuakcs. However. even indcvelupill~ counlries. simple construction
h Decade or Natural Disaster Reduction. Washington DC: National Academy Press
methods arc available lhilt have been epiuemiologic~'lIy associated wilh