6

HAZARD SURVEILLANCE
David H. Wegman

Epidemiologic surveillance is intended to provide early warning that serve to guide

timely intervention and prevention activities. By nature the systems utilize methods
distinguished by their practicality, uniformity and frequently, their rapidity, rather than by
their complete accuracy. The main purposes are to detect and follow changes in trends or
distributions of health related events in order to initiate control measures.
To consider the full potential of health surveillance it is worth looking closely at two
common definitions. These are that surveillance is
The on going , systematic collection, analysis, and interpretation of health data
essential to the planning, implementation, and evaluation of public health practice,
closely integrated with the timely dissemination of these data to those who need to
know.
and
Surveillance implies the continuing observation of all aspect of the occurrence
and spread of disease that are pertinent to its control
The use of the phrases health data and all aspects of the occurrence and spread
provide the context to think as broadly as possible about helath related events, not just a
disease, as objects for routine surveillance. This chapter is directed at considering the benefits
to be derived from surveillance of hazards as a complement to surveillance of disease.
Hazard surveillance is the assessment of the occurrence of, distribution of, and the
secular trends in levels of hazards (toxic chemical agents, physical agents, biomechanical
stressors, as well as biologic agents) responsible for disease and injury. In a public health
context, hazard surveillance identifies settings or individuals exposed to inappropriate or
controllable levels of specific hazards.

There are already a number of a familiar and important hazard surveillance activities
which guide health policy. Several examples are illustrative :
1. Each summer, residents in urban areas who live near ocean beaches are informed of
the safety of swimming at those beaches through published reports of coliform counts.
2. Health department direct attention to the control of food poisoning outbreaks by their
inspections of kitchens in public restaurants, which focus on a survey of hazardous
condition rather than cases of food poisoning.
3. In many large cities, published reports of daily variation in ozone levels are useful in
guiding health protective behaviors for those who are susceptible to pulmonary
disease.
4. Over a number of years, the monitoring of seat belt use and its association with
injuries and fatalities in automotive accidents led manufacturers and the government
to develop and introduce passive automotive restraints to protect the large number of
nonusers
5. A particularly effective hazard surveillance system follows the pattern of type specific
salmonella isolates reported by bacteriology laboratories nationwide. Routine
examination of the results from these laboratory tests have proved highly successful in
focusing early attention to point source outbreaks of salmonella epidemics.
6. A recent striking example of the impact of monitoring for hazards was the
measurement of benzene contamination in a commercial bottled water. The finding
led to a major product recall, and elimination of the source of benzene from the
identified water supply.
What is about hazard surveillance that makes it different from the better known practice
of disease surveillance ?
BENEFITS OF HAZARD SURVEILLANCE
In public health, there are several potential benefits gained from hazard surveillance that
expand on or complement those already provided by disease surveillance. These include :
ease of measurement of the condition under surveillance, attention to the proper target of
control, higher frequency of events and avoidance issues of privacy.

Ease of Measurement

When considering infectious disease, the measurement (i.e., reliable diagnosis and
reporting) of disease has been relatively straightforward. Routine reporting of measles,
mumps, chickenpox, rubella and so forth, has proven effective due, in large part, to the
reliable diagnosis of the characteristic symptoms and disease course. As a result, the
surveillance of childhood disease has proven to be an effective public health tool, particularly
in the targeting of efficient/effective immunization programs.
In contrast, the measurement of noninfectious disease, when applied to the purpose of
disease surveillance, has generally been disappointing. This results from the fact that many
noninfectious disease have multiple cause, which often requires relatively specific linking of
cause with disease before a report is useful for surveillance.
Rutstein et al. developed the concept of sentinel health even for surveillance by
directing attention to disease or conditions which, by their occurrence alone, provided
evidence of a preventable disease, disability or death. The list they developed included the
common childhood infections as well as events such as maternal death. The underlying
principle was that each condition, simply by its occurrence, provided enough evidence of
preventable causes that public health action was indicated. However, when this concept was
later adapted for use in recognizing and preventing occupational disease, the occurrence of
the disease had to be conditioned on (linked to) the occurrence of appropriate exposure
conditions as well. For example, exposure to manganese is a recognized cause of Parkinsons
disease. A public health effort to eliminate this cause of Parkinsons disease, however, would
be poorly served by surveillance of the disease alone. Manganese is only one cause, and a
proportionally small one compared with other, less clearly delineated causes. Thus, to
accomplish the goal of preventing manganese related Parkinsons disease would require
surveillance of both disease and exposure simultaneously or possibly, as will be described,
the surveillance of Manganese exposure alone.
Proper Focus of Control
A second benefit from focusing on hazards for surveillance is that items under
surveillance are also those to be targeted for primary prevention. For example, surveillance of
the problem of asbestos related lung cancer could focus on lung cancer in asbestos workers.
However, a not insubstantial amount of lung cancer in such as population would be do to
cigarette smoking either endependent of, or interacting with, asbestos exposure. On the other
hand, surveillance of asbestos exposure could provide information on the levels and patterns

of exposure (Jobs, processes, or industries? Where the poorest exposure control exists. Then,
even without an actual count of lung cancer cases, efforts to reduce or eliminate exposure
would be appropriately targeted.
Frequency of Events
A further potential benefit from hazard surveillance is the much higher frequency of
events. Exposures above a certain level can all be treated as an event on each occasion an
exposure survey is undertaken. The higher level of detail regarding, patterns and trends, along
with the opportunity for limited, more targeted, investigations as needed, is in sharp contrast
to that provided by the counts of a disease that occurs only once in an individual, and not
even in all exposed individuals.
To take advantage of the higher frequency of hazard events compared with disease
for surveillance, depends, however, on a central underlying assumption. Since hazard
surveillance is directed at nondisease events, its use in guiding public health intervention
requires that a clear exposure outcome relationship has been established. With such a
relationship established it can then be assumed that a reduction in exposurewill result in
reduced disease, obviating the need to document reduction in disease or adverse effects.
Although a quantitative exposure response relationship is not necessary for hazard
surveillance, its existence provides more effective guidance. For example, a well described
quantitative relationship between asbestos exposure and lung cancer allows an intervention to
focus on the need to control exposures well below the level commonly associated with the
disease. Once done, the disease can be assumed to be controlled without further
documentation.
Note how the need for a clear exposure disease relationship limits the potential
applications of hazard surveillance. For example, in the case of repetitive trauma disorders
(RTDs), epidemiologicstudies have begun to show the increased prevalence of RTDs with
awkward and highly repetitive motions. However, the development of good measures of
these types of exposure and related exposure outcome relationships is still in its infancy. At
present, the lack of quantification of the exposure diseaseassociations, as well as only a crude
understanding of the relative risk of different degrees of the exposures, limit the creation of
an effective hazard surveillance activity for RTDs.
Protection of Privacy

A particularly important advantage of hazard surveillance is that data collected for this
purpose do not infringe on an individuals privacy. Confidentiality of medical records is not
at risk and possibility of stigmatizing an individual with a disease label a avoided. This is a
particularly important advantage in industrial settings, where persons job may be in jeopardy
or a potential compensation claim my affect a phisicians choice of diagnostic options.
Use of Existing System
There are a variety of existing systems that can be adapted for or directly utilized in
hazard surveillance. This benefit of hazard surveillance parallels similar opportunities present
for disease surveillance. Examples of ongoing collection of hazard information that already
exist include such items as registries of toxic substance use or hazardous material discharges,
registries for specific hazardous substances and information collected for use in compliance
activity. An example of thus last activity, workplace monitoring information collected during
regulatory inspections, is further described later in this chapter.
On unusual example of the potential recognized in an existing system concerns the
possible use of routine corporate financial reports. Studies of the airline and trucking
industries have suggested that financial stability relative to payroll size, proportion of
expenses devoted to maintenance, and other similar factors predict future accident
experience. Specifically, as the proportion of funds used for repairs and maintenance
decresease there is evidence that there is an increasing incidence of accidents within the
trucking industry. A variety of financial indicators are used by accountants to routenly
measure and predict the financial behavior and health oh a business. These preliminary
studies suggest that opportunity exist to examine similar economic features of a business as a
hazard surveillance tool to predict safety and possibly health experience.
Using existing records rather than information from a system specifically designed for
a surveillance function typically provides more general than detailed results. Their advantage
however is that no new expenditures or reporting is required
GENERAL ISSUES IN UNDERTAKING HAZARD SURVEILLANCE
The potential use of a hazard surveillance activity varies both by the latency period of the
condition of interest and by a consideration of available alternatives.
Disease of Long Latency

In the case of chronic conditions with long latency and without accessible, early
indicators, hazard surveillance would appear to have special advantages over disease
surveillance. There are at least five other advantages of hazard surveillance which
complement those provided by disease surveillance. First, identifying hazard events is usually
much easier than identifying occupational disease events, particularly for diseases such as
cancer that have long latency periods. Second, a focus on hazards (rather than illnesses) has
the advantage of directing attention to the exposures which ultimately are to be controlled.
For example, surveillance of lung cancer might focus on rates in asbestos workers. However,
a sizeable proportion of lung cancer in this population could be due to cigarette smoking,
either independently of or interacting with the asbestos exposure, so that large numbers of
workers might need to be studied to detect a small number of asbestos-related cancers. On the
other hand, surveillance of asbestos exposure could provide information on the levels and
patterns of exposure (jobs, processes or industries) where the poorest exposure control exists.
Then, even without an actual count of lung cancer cases, efforts to reduce or eliminate
exposure would be appropriately implemented.
Third, since not every exposure results in disease, hazard events occur with much
higher frequency than disease events, resulting in the opportunity to observe an emerging
pattern or change over time more easily than with disease surveillance. Related to this
advantage is the opportunity to make greater use of sentinel events. A sentinel hazard can be
simply the presence of an exposure (e.g., beryllium), as indicated via direct measurement in
the workplace; the presence of an excessive exposure, as indicated via biomarker monitoring
(e.g., elevated blood lead levels); or a report of an accident (e.g., a chemical spill).
Conditions with Biomarkers
Condition with long or short latency that have useful measures of biologic dose or
subclinical effects provide a different opportunity. Biological monitoring is sometimes
included as a screening procedure under occupational health surveillance. However, the
purpose of biological monitoring is to detect the presence of a toxicant or metabolite in a
biological sample (an indicator of exposure) rather than detect an early health effect. Hence,
according to our definitions, it fits in more with hazard surveillance instead of health
surveillance.
The term biological effect monitoring is used to refer to some early indicator of a
health effect, for example, a detectable change in a biochemical parameter. Unlike biological

monitoring which indicates the extent of exposure, biological effect monitoring shows an
early effect, and hence rightly belongs under health surveillance. For example, the use of
blood lead (Pb) levels for biological monitoring serves to determine the extent of exposure to
lead in an exposed person. Biological effect monitoring, for example, measuring free
erythrocyte protoporphyrin (FEP), and urinary delta amino laevulinic acid (-ALA) detects
an early effect. These measure the direct effect of lead on haem synthesis. Other examples of
biological effect monitoring are the measurement of serum cholinesterase levels in workers
exposed to organophosphate pesticides, and the detection of specific low molecular weight
urinary proteins among cadmium exposed workers.
Acute Exposure or Disease Events
There are a number of available data sources that may prove useful for hazard
surveillance of carbon monoxide in ambient air. Carbon monoxide emissions data that are
available from the U.S. Environmental Protection Agency (EPA) National Emission
Inventory database and concentrations from air monitoring station data that are available
from the EPA Air Quality System database are readily available. There is a growing body of
literature that shows an ecological association between increased levels of ambient air carbon
monoxide and adverse CVD, stroke, and birth outcomes that supports the need to investigate
long-term effects of chronic low dose exposures. Few areas in the United States, however, do
not attain federal ambient air standards for carbon monoxide levels. Two sources of potential
hazard data on carbon monoxide levels in indoor environments, currently largely untapped,
are data obtained by utility companies and fire departments. Utility companies are required,
in most states, to take and investigate calls about potential gas leaks in homes, communities,
and workplaces, and they frequently conduct environmental sampling for carbon monoxide.
Fire departments respond to similar calls as well as to carbon monoxide detector alarms and
also take carbon monoxide measurements. While these data may represent a rich source of
hazard information, their availability is often limited by the absence of electronic data
systems and/or the willingness of the data owners to share information with health
departments. Hazard surveillance can also be conducted on known risk factors for carbon
monoxide exposures. For example, population-based surveys of housing and housing
conditions may identify geographic areas and demographic characteristics of residents
experiencing heating and electricity loss and who are dependent upon supplemental heating
sources

Agents Without Known Health Effects

In general the surveillance of hazards that are not yet associated with adverse health
outcomes in unproductive. In this sense, hazard surveillance is not useful for hypothesis
generation (one of the general goals of disease surveillance). While such information may not
be immediately useful, it my serve an important function in the future when matched with
information on disease as knowledge of helath effects develops.
POTENTIAL USER FOR HAZARD SURVEILLANCE
In describing the nature, advantages and limitations of hazard surveillance, a number
of possible applications have been noted or can be in ferred.
Focusing Intervention Action
To address these issues, we propose three types of surveillance for use in
environmental public health: hazard surveillance, exposure surveillance, and outcome
surveillance. If a clear link has been documented between a hazard and an adverse health
outcome, there is a route of exposure to the hazard, and the hazard can be readily monitored
in the environment, then hazard surveillance offers the best potential for early intervention
and prevention. If the hazard cannot be monitored readily but there is a marker for exposure
to the hazard, then exposure surveillance would provide information to inform the earliest
opportunity for intervention. Finally, if an important public health outcome has a suspected
(but undocumented) relationship to an environmental hazard, then outcome surveillance, in
combination with etiologic studies, is warranted. Data from such efforts can also provide
evidence of need for new or revised regulation for a specific hazard. hazard surveillance data
should prove valuable in planning epidemiological studies by identifying areas where such
studies would be most fruitful.
The data from these sources would also provide an idea of the common health
problems among workers, and can be used for the planning of workplace health promotion
activities. All this information is usually routinely collected, and few extra resources are
required to direct the data to the occupational health and safety authorities in a developing
country.
Program Planning

Programmes of hazard surveillance can have a variety of objectives and structures.

First, they permit focus on intervention actions and help to evaluate existing programmes and
to plan new ones. Careful use of hazard surveillance information can lead to early detection
of system failure and call attention to the need for improved controls or repairs before excess
exposures or diseases are actually experienced. Data from such efforts can also provide
evidence of need for new or revised regulation for a specific hazard. Second, surveillance
data can be incorporated into projections of future disease to permit planning of both
compliance and medical resource use. Third, using standardized exposure methodologies,
workers at various organizational and governmental levels can produce data which permit
focus on a nation, a city, an industry, a plant or even a job. With this flexibility, surveillance
can be targeted, adjusted as needed, and refined as new information becomes available or as
old problems are solved or new ones appear.
Levels of Focus
Hazard surveillance data can complement disease surveillance both for research to
establish or confirm a hazard-disease association, as well as for public health applications,
and the data collected in either instance can be used to determine the need for remediation.
Different functions are served by national surveillance data (as might be developed using the
US OSHA Integrated Management Information System data on industrial hygiene
compliance sample resultssee below) in contrast to those served by hazard surveillance
data at a plant level, where much more detailed focus and analysis are possible.
National data may be extremely important in targeting inspections for compliance
activity or for determining what is the probable distribution of risks that will result in specific
demands on medical services for a region. Plant-level hazard surveillance, however, provides
the necessary detail for close examination of trends over time. Sometimes a trend occurs
independently of changes in controls but rather in response to product changes which would
not be evident in regionally grouped data. Both national and plant-level approaches can be
useful in determining whether there is a need for planned scientific studies or for worker and
management educational programmes.
Resource Allocation
Hazard surveillance is the "assessment of the occurrence of, distribution of, and the
secular trends in levels of hazards (toxic chemical agents, physical agents, biomechanical

stressors, as well as biological agents) responsible for disease and injury." Exposure
surveillance is the monitoring of individual members of the population for the presence of an
environmental agent or its clinically inapparent (e.g., subclinical or preclinical) effects. In the
United States, decisions affecting public health policy and allocation of resources usually are
made yearly in conjunction with government budgets. Timely annual data summaries would
provide immediate estimates of the magnitude of a health problem, thus assisting
policymakers to modify priorities and plan intervention programs. These same data would be
useful to those assessing control activities and would help researchers establish priorities in
applied epidemiology and laboratory research. In addition, reviewing surveillance data
annually can facilitate the testing of hypotheses related to prevention and intervention efforts
(e.g., ocular injuries associated with fireworks). As intervention programs are evaluated and
priorities are set, policymakers must evaluate the effects of the programs on populations (e.g.,
protective measures to reduce the threat of lead toxicity in workplaces
Epidemiologic Studies
National and state maps were prepared that show the items surveyed, such as the
pattern of workplace and worker exposures to formaldehyde (Frazier, Lalich and Pedersen
1983). Superimposing these maps on maps of mortality for specific causes (e.g., nasal sinus
cancer) provides the opportunity for simple ecological examinations designed to generate
hypotheses which can then be investigated by appropriate epidemiological study.
Finally, hazard surveillance data should prove valuable in planning epidemiological
studies by identifying areas where such studies would be most fruitful. National
Occupational Exposure Survey (NOES). The US NIOSH carried out two National
Occupational Exposure Surveys (NOES) ten years apart to estimate the number of workers
and workplaces potentially exposed to each of a wide variety of hazards.

EXAMPLES OF HAZARD SURVEILLANCE

Carcinogen Registry (Finland)
In 1979 Finland began to require national reporting of the use of 50 different
carcinogens in industry. The trends over the first seven years of surveillance were reported in
1988 (Alho, Kauppinen and Sundquist 1988). Over two-thirds of workers exposed to

carcinogens were working with only three types of carcinogens: chromates, nickel and
inorganic compounds, or asbestos. Hazard surveillance revealed that a surprisingly small
number of compounds accounted for most carcinogen exposures, thus greatly improving the
focus for efforts at toxic use reduction as well as efforts at exposure controls.
Another important use of the registry was the evaluation of reasons that listings
exited the systemthat is, why use of a carcinogen was reported once but not on
subsequent surveys. Twenty per cent of exits were due to continuing but unreported exposure.
This led to education for, as well as feedback to, the reporting industries about the value of
accurate reporting. Thirty-eight per cent exited because exposure had stopped, and among
these over half exited due to substitution by a non-carcinogen. It is possible that the results of
the surveillance system reports stimulated the substitution. Most of the remainder of the exits
resulted from elimination of exposures by engineering controls, process changes or
considerable decrease in use or exposure time. Only 5% of exits resulted from use of personal
protective equipment. This example shows how an exposure registry can provide a rich
resource for understanding the use of carcinogens and for tracking the change in use over
time.
National Occupational Exposure Survey (NOES)
National Occupational Exposure Survey (NOES). The US NIOSH carried out two
National Occupational Exposure Surveys (NOES) ten years apart to estimate the number of
workers and workplaces potentially exposed to each of a wide variety of hazards. National
and state maps were prepared that show the items surveyed, such as the pattern of workplace
and worker exposures to formaldehyde (Frazier, Lalich and Pedersen 1983). Superimposing
these maps on maps of mortality for specific causes (e.g., nasal sinus cancer) provides the
opportunity for simple ecological examinations designed to generate hypotheses which can
then be investigated by appropriate epidemiological study.
Changes between the two surveys have also been examinedfor example, the
proportions of facilities in which there were potential exposures to continuous noise without
functioning controls (Seta and Sundin 1984). When examined by industry, little change was
seen for general building contractors (92.5% to 88.4%), whereas a striking decrease was seen
for chemicals and allied products (88.8% to 38.0%) and for miscellaneous repair services
(81.1% to 21.2%). Possible explanations included passage of the Occupational Safety and

Health Act, collective bargaining agreements, concerns with legal liability and increased
employee awareness.
Inspection (Exposure) Measure (OSHA)
Inspection (Exposure) Measures (OSHA). The US OSHA has been inspecting
workplaces to evaluate the adequacy of exposure controls for over twenty years. For most of
that time, the data have been placed in a database, the Integrated Management Information
System (OSHA/IMIS). Overall secular trends in selected cases have been examined for 1979
to 1987. For asbestos, there is good evidence for largely successful controls. In contrast,
while the number of samples collected for exposures to silica and lead declined over those
years, both substances continued to show a substantial number of overexposures. The data
also showed that despite reduced numbers of inspections, the proportion of inspections in
which exposure limits were exceeded remained essentially constant. Such data could be
highly instructive to OSHA when planning compliance strategies for silica and lead.
Another use of the workplace inspection database has been a quantitative examination
of silica exposure levels for nine industries and jobs within those industries (Froines,
Wegman and Dellenbaugh 1986). Exposure limits were exceeded to various degrees, from
14% (aluminium foundries) to 73% (potteries). Within the potteries, specific jobs were
examined and the proportion where exposure limits were exceeded ranged from 0%
(labourers) to 69% (sliphouse workers). The degree to which samples exceeded the exposure
limit varied by job. For sliphouse workers excess exposures were, on average, twice the
exposure limit, while slip/glaze sprayers had average excess exposures of over eight times the
limit. This level of detail should prove valuable to management and workers employed in
potteries as well as to government agencies responsible for regulating occupational
exposures.

Priority Setting Using a National Database

In the USA, the Occupational Safety and Health Administration (OSHA) has
maintained since 1979 the Integrated Management Information System (IMIS), which
contains measurement results from surveys performed by OSHA to verify compliance to
Permissible Exposure Limits (PELs). OSHA was created as a federal agency in 1971 (US
Congress, 1970). Some states opted out of the federal OSHA agency and created their own

State OSHA agencies, and some states use a combination of federal and State OSHA
agencies. Since 1972, IMIS has served as a data-entry and information retrieval system
associated with enforcement activities of both federal and State OSHA.
Each OSHA inspector is responsible for documenting the outcome of each inspection,
including entering exposure measurements into IMIS. The actual exposure levels measured
during inspections were only entered starting in 1979. Before that, only a severity index was
provided, representing the ratio of the measurement to the PEL. The Salt Lake Technical
Center, created in 1984, processed most of the samples collected by the federal and some of
the samples collected by State OSHA inspectors. The CEHD data made available by the Salt
Lake Technical Center are analytical sample results of the measurements collected by OSHA
inspectors while assessing compliance.
SUMMARY
This chapter was designed to identify the purpose of hazard surveillance, describe its
benefits and some of its limitations, and to offer a variety of examples in which it has already
provided useful public health information. It would be a mistake, however to suggest that
hazard surveillance was a panacea and should replace disease surveillance for noninfectious
disease. In 1997 a NIOSH task force summarized the relative importance and
interdependence of the two major types of surveillance.
The surveillance of hazards and disease cannot proceed in isolation from each other.
The successful characterization of the hazards associated with different industries or
occupations, in conjunction with toxicological and medical information relating to the
hazards, can suggest industries or occupational groups appropriate for epidemiologic
surveillance.
Conversely, unusual health patterns in certain industries or occupations elucidated by
surveillance of health effects will be more fully explained by surveillance of the potentially
causative agents. A few disease entities, e.g., mesothelioma, are sufficiently cause specific to
diminish the need for hazard surveillance. Some agents are sufficiently effect specific to
make the task of illness surveillance relatively straightforward.
There remains, however, a vast middle ground where exposures are complex and
symptoms diverse, which will yield only to the combined efforts of hazard and disease
surveillance.

Notes
1. Centers for Disease Control. Draft Policy Statement
2. Benenson, A.S., ed.1990. Control of Communicable Disease in Man. Washington, D.C.:
American Public Health Association.
3. Rutstein, D.D., W. Berenberg, T.C. Chalmers, C.G. Child, A.P. Fishman, and E.B.
Perrin. 1976. Measuring the quality of medical care : a clinical method. NEJM 294:
582-588
4. Rutstein DD, Mullan RJ, Frazier TM, Halperin WE, Melius JM, Sestito JP. Sentinel
Health Events (occupational): a basis for physician recognition and public health
surveillance. Am J Public Health.1983 Sep;73(9):10541062
5. Froines JR, Dellenbaugh CA, Wegman DH. Occupational health surveillance: a means
to identify work-related risks. Am J Public Health. 1986 Sep;76(9):10891096.
6. Landrigan PJ, Markowitz S: Current magnitude of occupational disease in the United
States: Estimates from New York State. Ann NY Acad Sci 1989; 572: 2745.
7. Landrigan PJ, Selikoff IJ (editors). Occupational Health in 1990s: Developing a
Platform for Disease Prevention. Annals NY Academy of Sciences: 572 1-296,
1989. ISBN 0-89766-523-6
8. Sundin

DS,

Pedersen

DH,

Frazier

TM.

Occupational

surveillance. Am J Public Health. 1986 Sep;76(9):10831084

hazard

and

health