February 2008, Vol 98, No. 2 | American Journal of Public Health Krieger | Peer Reviewed | Government, Politics, and Law | 221

 GOVERNMENT, POLITICS, AND LAW 

Proximal, Distal, and the Politics of Causation: What’s Level Got to Do With It? | Nancy Krieger, PhD

Causal thinking in public health, and especially in the growing literature on social de- terminants of health, routinely employs the terminology of proximal (or downstream) and distal (or upstream).

I argue that the use of these terms is problematic and ad- versely affects public health research, practice, and causal accountability. At issue are distortions created by conflat- ing measures of space, time, level, and causal strength.

To make this case, I draw on an ecosocial perspective to show how public health got caught in the middle of the problematic proximal–distal divide—surprisingly embraced by both biomedical and social determinist frameworks—and propose replacing the terms proximal and distal with ex- plicit language about levels, pathways, and power. (Am J

Public Health. 2008;98:221–230. doi:10.2105/AJPH.2007.111278)

PROXIMAL. DISTAL. UPSTREAM. Downstream. Risk factor. Determi- nant. Level. Multilevel. These terms feature prominently in current discussions of causal pathways and public health, especially in work on the social determinants of health. A central focus is on how “upstream” societal influ- ences—typically referred to as distal—shape “downstream,” or proximal, exposures, thereby affecting population health.1–16

Exemplifying this line of thought are recent reports issued by the World Health Organization Com- mission on Social Determinants on Health2 and the World Health Organization Regional Office for Europe.3 Common assumptions are that (1) diseases

are attributable to many causes, located outside and within the body; (2) the social lies in the realm of the distal; (3) the bio- logical belongs to the proximal; and (4) the distal and proximal are connected by levels, e.g., soci- etal, institutional, household, in- dividual, which can be conceptu- alized as near to or far from the causes under consideration. For example, as discussed in both re- ports, “distal” societal factors drive the risk of smoking; how smoking harms health involves “proximal” biology.2,3 What could be more obvious?

Yet what seems clear-cut can be deceiving. I argue that al- though notions of proximal, distal, and level all matter for elu- cidating causal pathways, clear thinking—and, hence, public health research, practice, and causal accountability—is distorted

by conflating measures of space, time, level, and causal strength. When it comes to causation, it is one thing to think about near and far in relation to space and time; it is another matter entirely to do so for levels. To make this intellectual argument, I draw on an ecosocial perspective1,17–21 to show how public health got caught in the middle of the prob- lematic proximal–distal divide— surprisingly embraced by both biomedical and social determinist frameworks—and propose replac- ing the terms proximal and distal with explicit language about lev- els, pathways, and power.

PROXIMAL AND DISTAL IN PUBLIC HEALTH THOUGHT

The idea that disease etiology and distribution are attributable to causes deemed “far” from and

American Journal of Public Health | February 2008, Vol 98, No. 2222 | Government, Politics, and Law | Peer Reviewed | Krieger

 GOVERNMENT, POLITICS, AND LAW 

“near” (including within) the body is ancient22–27; Hippocratic tradition, in the 5th century BCE, famously invoked both atmos- phere and individual constitution as explanations for epidemic dis- ease.22,25 By contrast, the idea that there is a causal etiological hierarchy, spanning from distal to proximal, is relatively new. It be- came a core part of the public health canon only in the mid–20th century CE. How this change happened and its public health implications have been lit- tle discussed.

Strand 1. 19th Century Emergence of Proximal and Distal as Scientific Terms for Spatiotemporal Scale

Only in the early 19th century CE did the terms proximal and distal enter the scientific dis- course.28 Invented to describe anatomical location and distance, as measured on a spatial scale, these words were coined by biol- ogists at a time when compara- tive anatomy occupied a key place in debates over the classifi- cation and nature of species.29,30

Proximal, derived from the Latin noun proximus (“nearest”), took on the meaning “situated toward the center of the body, or the point of origin or attachment of a limb, bone, or other structure.”28

Its antonym, distal, derived from distant, was intended to echo 2 other widely used biological con- cepts: ventral and dorsal.28 Soon other natural sciences adopted the terms, albeit with some criti- cal modifications. In geology, for example, the terms took on a temporal as well as a spatial di- mension, reflecting how adjacent

geological strata typically are “close” in time as well as in space.28

The moment time entered the picture, however, the terms prox- imal and distal were primed to develop new meanings. This is because of the ubiquitous metaphorical linkage of time, space, and causal reason- ing.31(pp133–138) In all known lan- guages, temporal events are de- scribed in spatial terms: Time moves through space.31(p134) This metaphorical relationship, as ar- gued by the linguist Deutscher, is essential to causal reasoning, be- cause it enables us to “talk freely about one thing coming ‘from’ another, ‘out of’ another, or hap- pening ‘through’ another, to ex- press abstract chains of cause and event.”31(p137)

New European scientific dis- coveries of powerful physical laws for gravity, electricity, and magnetism32–34 further affected scientific thinking about causa- tion. These inverse square laws, expressed as pithy equations, clarified that force depends on distance: The more proximal the mass or the charge of the inter- acting objects, the greater the force—and the more powerful the effect. It was a short step from here to equate distance with causal strength, in not only the physical but also the life sciences.

Strand 2. From Spatiotemporal Scale to Causal Hierarchies and Levels

Not until the later 19th cen- tury, however, did the scientific meanings of proximal and distal

leap from referring only to spa- tiotemporal scale to also describ- ing levels and causal hierarchies. In their new usage, the “closeness”— or “distance”—of levels defined a new type of proximity, one that could be measured only concep- tually, not in meters or minutes.

Initially, this conceptual change occurred within disci- plines focused on a different type of body: that of body politic, i.e., the social sci- ences.35–42 In books with such titles as Social Pathology41 and Organism and Society,42 influen- tial late–19th century sociolo- gists drew parallels between the biologically nested hierarchies of cell–organism–species and the socially nested hierarchies of individuals–families–societies.35,

36,37(pp4–8),38(pp231–323),40–42 In their view, just as organs, com- posed of their constituent cells, must collectively work together for an organism to survive, so too do social groups and their constituent individuals have com- plementary roles they must per- form for society to thrive.35,36,41,42

The intent was counter not only to the ruthless competition of Social Darwinism30(pp87–90),39,

40(pp196–199),42(p10) but also to the contending Marxist view that class conflict determined societies’ structure and development.35,

37(pp4–8),41,42(pp182–186),43,44(pp178–179)

Borrowing biological terminol- ogy, these sociologists newly de- ployed the terms proximal and distal to describe societies’ struc- tural “levels.”41(xxiii) Ranging from individual to institutional, these levels and the “distance” be- tween them became defined by their nested relationships:

Adjacent levels were “close,” and nonadjacent levels were “far.”

Meanwhile, biologists likewise expanded the use of the terms proximal, distal, and level, bring- ing these terms explicitly into their thinking about causal dis- tance. As part of the early 20th century modern evolutionary synthesis, which integrated Dar- winian evolutionary biology, paleontology, and Mendelian ge- netics,29,45,46(pp503–591),47 these bi- ologists newly contrasted what they termed “proximate” (physio- logical) versus “ultimate” or “distal” (evolutionary) causes.29(pp313–321),

35,46(pp1340–1343),47 This distinction recognized that asking how a bi- ological event occurs (e.g., a muscle contraction) is not the same as asking why a biological phenomenon exists (e.g., muscles enable locomotion to find food and flee predators). Drawing on holistic thinking,35,48 they argued that valid explanations could co- exist across levels (e.g., species, organism, cell, molecule) and in- volve the distant past (evolution) and the immediate present (cur- rent stimulus). In the instant of a muscle contraction, both proxi- mal and distal causes were at play.

The Mid–20th Century Public Health Embrace of Proximal and Distal

By mid–20th century, to be close or far could thus refer to space, to time, to lineage, or to location in hierarchical concep- tual levels. The terms proximal and distal thus became widely encompassing terms to express— and contest—causal conceptions in both the social and the natural

February 2008, Vol 98, No. 2 | American Journal of Public Health Krieger | Peer Reviewed | Government, Politics, and Law | 223

 GOVERNMENT, POLITICS, AND LAW 

sciences. Amid these divergent uses, the terms proximal and dis- tal finally entered the public health causal lexicon.

Prompting their adoption was growing recognition that the field of public health, still riding the crest of enormous success against infectious diseases in the 19th and 20th centuries CE, had to move beyond a monocausal to a multifactorial account of disease causation, which involved not only the agent but also the host and the environment.1,17,24,49–51

As exemplified by the findings of the Framingham study of heart disease, rising rates of chronic disease and cardiovascular mor- tality seemed to be attributable not to any one single exposure but instead to a variety of factors,51–53 leading the Framing- ham researchers to coin the term risk factor to describe these partial—i.e., not sufficient, not al- ways necessary, but nonetheless contributing—component causes.53

It was through the multifactor- ial perspective that the terms proximal and distal emerged as terms for the discussion of causal- ity in the public health litera- ture.17 Unfortunately, however, their new usage drew on shallow understandings of the terms near and far that impeded rather than deepened multilevel thinking. The essential features of the multifactorial framework remain well-sketched by the still highly influential spiderless17 “web of causation,” first articulated in the 1960s54 and which, as I have previously argued,17 (1) leveled all exposures to a single plane; (2) defined “proximal” factors to

be those operating directly on or within the body, and relegated all other exposures to the murky realm of “distal”; (3) linked causal potency to distance—i.e., the “closer” the cause, the greater the effect (following the logic of the previously described physical in- verse square laws); (4) held that distal causes necessarily exerted their influence through succes- sively more proximal factors; (5) took a studied agnosticism as to what accounted for the array of exposures included in the web and eschewed any discussion of power or injustice; and, hence, (6) adopted a narrow stance of what may best be termed causal pragmatism55,56 that prioritized focusing on what they considered to be “proximal” factors ostensi- bly amenable to control by either individuals or by public health or medical professionals (including by health education) rather than what they termed the more “dis- tal” determinants requiring socie- tal change.

The use of the terms proximal and distal persists to this day. It underlies the 21st century suc- cessor to the web of causation— that is, the “gene–environment interaction” framework,57–60

which posits that the occurrence of common and complex dis- eases reflects the interplay of in- dividual genetic variability with an array of exogenous exposures.57–60 Work in this area is chiefly engaged57–60 (albeit with some exceptions61–65) in the quest to discover genetic deter- minants of biological susceptibil- ity and to develop pharmacologi- cal interventions that can block deleterious gene expression.

The proximal–distal discourse likewise pervades the social de- terminants of health perspec- tive,1–16 which holds that “distal” institutional priorities and prac- tices of government and the pri- vate sector shape people’s cumu- lative exposure, across the life course via intermediary path- ways, to the proximal physical, behavioral, psychosocial, and bio- logical exposures that trigger pathogenic processes (including gene expression), thereby causing disease. Secondarily, once illness occurs, the social determinants of health framework asks how prog- nosis is affected by socially pro- duced inequities in access to needed medical care.1–16

In both cases, causal distance still matters for causal strength: In the gene–environment interac- tion model, “proximal” causes re- main most potent, whereas for the social determinants of health perspective, “distal” causes are decisive. Despite their fundamen- tally different approaches, both frameworks cling to the proxi- mal–distal divide. This little re- marked convergence hints that some causal logic may be askew.

AN ALTERNATIVE ECOSOCIAL APPROACH TO LEVELS, EMBODIMENT, AND ACCOUNTABILITY

I suggest that one reason the proximal–distal terminology can be so readily used by such totally disparate frameworks is their now deeply entrenched confla- tion of relationships among space, time, distance, levels, and causal potency. Three examples, based on arguments offered from

an ecosocial perspective (Figure 1, Table 1),1,17–21 supplemented by the conceptual clarifications pro- vided in Box 2, illustrate the problems that can arise when logics of scale are confused with analysis of levels and when dis- tance is conflated with power.

The basic point is that societal patterns of disease represent the biological consequences of the ways of living and working dif- ferentially afforded to the social groups produced by each soci- ety’s economy and political pri- orities.1,17–21 Class and racial inequality, for example, differen- tially affect the living standards, working conditions, and envi- ronmental exposures of the dominant and subordinated classes and racial/ethnic groups, thereby creating class and ra- cial/ethnic health disparities. Stated more generally, a soci- ety’s economic, political, and so- cial relationships affect both how people live and their eco- logic context, and, in doing so, shape patterns of disease distri- bution. The understanding of the societal distributions of health thus cannot be divorced from considerations of political economy and political ecol- ogy.1,17–21 Driving health in- equities are how power—both power over and power to do,66–68

including constraints on and possibilities for exercising each type—structures people’s engage- ment with the world and their exposures to material and psy- chosocial health hazards. Notably, neither type of power readily maps onto a metric of proximal or distal. Nor do they neatly par- tition across levels. A critical

American Journal of Public Health | February 2008, Vol 98, No. 2224 | Government, Politics, and Law | Peer Reviewed | Krieger

 GOVERNMENT, POLITICS, AND LAW 

Note. To explain current and changing population distributions of disease, including health inequities, and who and what is accountable for the societal patterning of health, it is necessary to consider causal pathways operating at multiple levels and spatiotemporal scales, in historical context and as shaped by the societal power relations, material conditions, and social and biological processes inherent in the political economy and ecology of the populations being analyzed. The embodied consequences of societal and ecologic context are what manifest as population distributions of and inequities in health, disease, and well-being. Source. See references 1,17–21.

FIGURE 1—A heuristic diagram for guiding ecosocial analyses of disease distribution, population health, and health inequities.

corollary is that, contrary to the logic of the proximal–distal di- vide, within the very phenom- ena of disease occurrence and distribution—just as in a muscle contraction—the distal and the proximal are conjoined.

Example 1. Why Spatiotemporal Scale Is Not the Same as Level

The first example, drawn from ecology, the original multilevel

science, clarifies why population sciences cannot afford to confuse metrics of spatiotemporal scale with the phenomena of levels. The example concerns, literally, the forest and the trees. Forests are levels within ecosystems, which involve not only trees but also the other plants and animals that inhabit them.69–75 Notably, forests can be large or small (a spatial metric), as well as old or young (a temporal metric). Indeed,

one key issue in conservation ecology today, spurred by intensi- fied commercially driven logging and deforestation, forest fragmen- tation, habitat degradation, and spread of zoonoses (e.g., Lyme dis- ease), is just what size, spatially, an expanse of woods needs to be— and how close it needs to be to other such expanses—to function as a particular type of forest.69–75

Too small, with the ratio of edge- to-interior too high, or too spatially

isolated, without connecting corri- dors, and its species composition will change, often losing diversity, including to the point of outright extinction.69–75

The phenomenon of a forest (a level), and interactions among both the entities that constitute it and also between the forest and its environs, is affected by, but not identical to, the forest’s size (spatiotemporal scale). Similarly, for measles to become endemic

February 2008, Vol 98, No. 2 | American Journal of Public Health Krieger | Peer Reviewed | Government, Politics, and Law | 225

 GOVERNMENT, POLITICS, AND LAW 

TABLE 1—Core Constructs of Ecosocial Theory—an Epidemiological Theory of Disease Distribution—and Some Predictions

Construct Elaboration

Embodiment A concept that refers to how we literally incorporate, biologically, the material and social world in

which we live, from in utero to death; a corollary is that no aspect of our biology can be

understood absent knowledge of history and individual and societal ways of living.

Epidemiologically, “embodiment” is thus best understood:

(1) As a construct, process, and reality, contingent upon bodily existence;

(2) As a multilevel phenomenon, integrating soma, psyche, and society, within historical and

ecological context, and, hence, an antonym to disembodied genes, minds, and behaviors;

(3) As a clue to life histories, hidden and revealed; and

(4) As a reminder of entangled consequences of diverse forms of social inequality.

Pathways of embodiment Causal pathways that involve exposure, susceptibility, and resistance (as both social and biological

phenomena), structured simultaneously by (1) societal arrangements of power, property, and

contingent patterns of production, consumption, and reproduction, and (2) constraints and

possibilities of our biology, as shaped by our species’ evolutionary history, our ecologic context,

and individual histories, that is, trajectories of biological and social development, and that

involve gene expression, not just gene frequency.

Cumulative interplay among exposure, Expressed in pathways of embodiment, with each factor and its distribution conceptualized at multiple

susceptibility, and resistance levels (individual, neighborhood, regional or political jurisdiction, national, international, or

supranational) and in multiple domains (e.g., home, work, school, other public settings), in relation

to relevant ecologic niches, and manifested in processes at multiple scales of time and space.

Accountability and agency Refers to who and what is responsible for social inequalities in health and for rectifying them, as well

as for the overall current and changing contours of population health, as expressed in pathways

of and knowledge about embodiment. At issue are the accountability and agency of not only

institutions (government, business, and public sector), communities, households, and individuals,

but also of epidemiologists and other scientists for theories used and ignored to explain social

inequalities in health. A corollary is that, given likely complementary causal explanations at

different scales and levels, epidemiological studies should explicitly name and consider the

benefits and limitations of their particular scale and level of analysis.

Analytic implications and predictions Determinants of disease distribution (a population-level phenomenon) presume but are not reducible

to mechanisms of disease causation (which occur within individuals’ bodies). Key contingent

hypotheses are: (1) population patterns of health and disease constitute the embodied biological

expression of ways of living and working differentially afforded by each society’s political economy

and political ecology, and (2) policies and practices that benefit and preserve the economic and

social privileges of dominant groups simultaneously structure and constrain the living and working

conditions they impose on everyone else, thereby shaping particular pathways of embodiment.

Source. See references 1,17–21.

with a community (a level), community size (a scale) must exceed 250000 people.76,77

Hence, argument 1: Confuse scale and level—or consider only one, not both—and understand- ing of population phenomena will be undermined.

Example 2. On Nonlinear Causal Pathways, With Immediate and Long-Term Effects

The second example illus- trates that levels need not play by the proximal–distal schema that the path from what is

considered “far” to “near” nec- essarily travels through what is termed “intermediate.” This is because events at one level can directly and profoundly affect nonadjacent levels, instantly and persistently, without intermediaries.29,46–48,66,78,79

Consider, for example, the 1973 US Supreme Court ruling that legalized abortion, on the grounds of individuals’ rights to privacy.80 Here, the levels at issue were defined jurisdiction- ally, with the federal judicial rul- ing on individual constitutional rights overturning federal and state laws that interfered with in- dividual privacy by prohibiting abortion. In this case, the so- called distal determinant (1) di- rectly affected individual girls’ and women’s reproductive rights and (2) reverberated up to other levels, by requiring changes in state laws and by expanding the permitted range of services that could be provided by health pro- fessionals and health facilities.

The positive health conse- quences were both immediate and long-term: US girls and women alike no longer were forced, by law, to face the risk of having an unsafe illegal abortion and they were also less likely to bear unwanted children, thereby reducing risks of adverse mater- nal and birth outcomes.81–84

More recent US Supreme Court decisions restricting the right to abortion likewise illustrate this principle of skipping levels, with contrary effects.85,86

Analogous examples can read- ily be drawn from the health and human rights literature, whereby state obligations to respect, pro- tect, and fulfill individuals’ human rights affect policies and interven- tions at multiple levels.87 The im- plication, argument 2, is that non- adjacent levels can have direct causal relationships, an insight obscured by the proximal–distal logic.

American Journal of Public Health | February 2008, Vol 98, No. 2226 | Government, Politics, and Law | Peer Reviewed | Krieger

 GOVERNMENT, POLITICS, AND LAW 

TABLE 2—Proximal and Distal, Spatiotemporal Scale Versus Level—Meanings, Contrasts, and Causal Implications

Spatiotemporal Scale

Category Space Time Level

Metric of distance Units of spatial distance, measured in nested increments; Units of temporal distance, measured in nested Adjacency of levels, which can be organized—theoretically,

examples include: kilometer–meter–millimeter– increments; examples include: millennium– conceptually, or structurally—as nested or nonnested

micron; or mile–foot–inch century–year–day–hour–minute–second– hierarchies; examples include:

millisecond (1) nested: nation–region–city–neighborhood–household;

or ecosystem–species–organism–organ

system–organ–cell;

(2) nonnested: school | workplace | neighborhood–individual

“Near” Proximal, near in space, close Proximal, near in time, recent Conceptual or structural nonscalar relationship: adjacent levels

“Far” Distal, distant in space, far away Distal, distant in time, long ago Conceptual or structural nonscalar relationship: nonadjacent levels

Strength of effect Usually inverse relationship of spatial distance and force: Usually inverse relationship of temporal distance Cannot predict “strength” of “effect” based solely on level: a

closer = stronger, hence and force: given phenomenon at any given level potentially can powerfully

proximal = powerful; closer = stronger, hence or weakly affect or be affected by phenomena at the same

farther = weaker, hence proximal = powerful; level, adjacent levels, and nonadjacent levels

distal = dilute farther = weaker, hence

distal = dilute

Typical causal Proximal = stronger cause Proximal = stronger cause Causal inference depends on level of question being asked: There

inference Distal = weaker cause Distal = weaker cause may be different explanations for phenomena at different

levels, and explanations for events observed within any given

level may involve solely phenomena within that level or also

interactions between levels; adjacency of levels may or may not

predict causal strength of cause–effect relationship

Relationship to Physical distance is a spatial dimension distinct from Chronological distance is a temporal dimension Level is not a spatiotemporal phenomenon. It is, instead, a conceptual

space and time, but space and time can be related distinct from space, but time and space can nonspatiotemporal relational construct that organizes and

time mathematically, e.g., distance = speed × time be related mathematically, e.g., time = distinguishes (conceptually or structurally) different orders

(and the length of a meter is now defined in distance/speed (and initial time units were of hierarchically linked systems and processes (including

relation to time and the speed of light32[p537]) based on the earth’s rotation, involving both nested and nonnested hierarchies). “Distance” for

spatial distance32[pp3–5]) levels does not involve spatiotemporal separation: For any

phenomenon at any given point in space and time, all levels

co-occur simultaneously, even though some levels may be

more causally relevant than others to phenomena occurring

at any given level. Space and time nevertheless do matter

for levels in the case of nested hierarchies, whereby units

within lower-order levels typically are smaller and involve

faster processes than units in higher-order levels.

Example 3. On Levels and the Perils of Commodity Fetishism—the Simultaneity of Material Properties and Social Relations