Epidemiological triangle/triad

This model makes the agent a component of causation along with the host and environment (Fig. 1). The model implies that all components are equally important in disease causation and that a change in any one of them would change the frequency of disease. The model applies to both infectious or non-infectious diseases. For instance, in Johnne's disease the agent would be the bacterium, MAP; host factors include non-immune, weakened resistance, poor nutrition, age, gender; and environmental factors include animal stocking density, poor environmental conditions (such as temperature, humidity, wind velocity, precipitation, poor housing as in crowded conditions, poor ventilation, and bad sanitation).

Fig. 1.

Epidemiological triangle/triad [Reference Torrence77].

Wheel of causation

The wheel model places genetic factors in the core of the wheel and varies the size of the host and environmental components depending on their influence in the particular disease process [Reference Krieger78]. Surrounding the host is the total environment divided into the biological, physical, and social environments (Fig. 2). These divisions, of course, are not true divisions – there are considerable interactions among the environment types. Although it is a general model, the wheel of causation does illustrate the multiple aetiological factors of human infectious diseases [Reference Mausner and Bahn79]. According to Jantchou et al. [Reference Jantchou, Monnet and Carbonnel80], many environmental factors for IBD have been investigated, including infectious agents, diet, drugs, stress and social status. Among these factors, MAP, oral contraceptives and antibiotics could play a role in Crohn's disease [Reference Jantchou, Monnet and Carbonnel80–Reference El Zaatari84]. Sicilia et al. [Reference Sicilia81] reported that the pathogenesis of IBD probably involves an interaction between genetic and environmental factors: cigarette smoking, appendectomy and oral contraceptives are the factors most frequently linked to its aetiology.

Fig. 2.

Wheel of causation [Reference Mausner and Bahn79].

Web of causation

This model refers to the ‘web’ of interconnected factors which lead to disease. The web of causation merely reflects the fact that there is a complex mixture or a ‘web’ of factors that can cause disease [Reference Krieger78]. Many aetiological factors for Crohn's disease have been suggested, including autoimmune, genetic, dietary components plus various infectious agents including MAP [Reference Chiodini and Rossiter82–Reference El Zaatari84].

Causal pie model

The main causal model used by epidemiologists today is Rothman's ‘pies’ [Reference Rothman and Greenland75]. The idea is that a sufficient causal complex (a pie) is represented by the combination of several component causes (slices of the pie) (Fig. 3). A set of component causes occurring together may complete the ‘pie’, creating a sufficient cause and thus initiating the disease process. Rothman & Greenland [Reference Rothman and Greenland75] define both ‘necessary’ cause and ‘sufficient’ cause. A ‘sufficient’ cause is one that always results in disease [Reference Rothman and Greenland75], while a ‘necessary’ cause is one that must be present but might not be the cause of a disease to develop. In other words, a necessary cause is a component cause that is a member of every sufficient cause [Reference Timmreck85–Reference Kleinbaum, Kupper and Morgenstern91].

Fig. 3.

Causal pie model [Reference Rothman and Greenland92]. This illustration shows a disease that has three sufficient causal complexes, each having five component causes. A is a necessary cause since it appears as a member of each sufficient cause. B, C, and F are not necessary since they fail to appear in all three sufficient causes.

Rothman [Reference Rothman, Schottenfield and Fraumeni93] defines ‘a cause of a disease event as an event, condition, or characteristic that preceded the disease event and without which the disease event either would not have occurred at all or would not have occurred until a later time’. If disease does not develop without the factor being present, then the causative factor is termed ‘necessary’. If the disease always results from the factor, then the causative factor is termed ‘sufficient’. In reference to Rothman's causal pie model, the possibility exists that MAP is a ‘necessary’ but not a ‘sufficient’ cause of Crohn's disease. As a necessary cause, MAP is required to be present to trigger the inflammatory reaction seen in Crohn's disease. However, not being a sufficient cause means that MAP cannot cause Crohn's disease alone but acts in concert with immune dysfunction and genetic susceptibility in order for Crohn's disease to occur [Reference Chiodini and Rossiter82,Reference El Zaatari84]. Therefore, not every one with the presence of MAP in the intestine would suffer from Crohn's disease. Moreover, the failure to detect MAP in some cases of Crohn's disease may not necessarily indicate that MAP is absent. Low specificity and sensitivity of the test, among other factors, may explain the inability to detect MAP in some patients [Reference Jeyenathan94]. It has been reported that the sensitivity of nucleic-acid based tests is influenced by the bacterial burden, as exemplified by the compromised sensitivity of PCR-based assays for sputum smear-negative tuberculosis [Reference Forbes95–Reference Noordhoek97]. By extension, assays that reliably detect abundant MAP organisms in livestock with Johne's disease may not provide sufficient sensitivity to study human Crohn's disease [Reference Jeyenathan94].

Inductivism

This doctrine holds that science proceeds from observation to theory, beginning with observations derived from experiments, and extrapolating from these to general laws [Reference Last76, 98–Reference Bacon102]. Bacon's vision of the ‘true induction’ comprises three interrelated stages: (i): Observation and Experiment (ii) Classification and Concept Formation and (iii) Eliminative Induction and Causal Inference [Reference Bacon102]. The traditional view of science is that induction – the formation of a hypothesis based on observation – is cardinal to the scientific method. However Hume [Reference Hume and Nidditch103], the deductivist and others [Reference Hendel104] argued that a hypothesis that is derived by induction is flawed because it can be refuted by the first observation that proves an exception. Deduction refers to reasoning that proceeds from the general to the particular and relies on general theory to infer particular conclusions [Reference Last76]. A century after Hume, Mill [Reference Mill105] proposed a canon of five methods to infer causes from their effects incorporating some of the ideas that had been proposed earlier by Bacon [Reference Bacon and Jones106]. The canons of Mill [Reference Mill107] have evolved into inferential criteria that are in use today.

Refutationism or falsificationism

This theory is a rival account of the processes involved in scientific research to inductivism. While inductivism holds that science proceeds from observation to theory, beginning with observations derived from experiments, and extrapolating from these to general laws, falsificationism suggests that science proceeds in the opposite direction, beginning with scientific theories or ‘conjectures’, and then conducting experiments and eliminating those theories that are falsified by results [Reference Rothman108–Reference Susser and Rothman111]. Karl Popper, one of the most influential philosophers of science of the twentieth century [Reference Susser112], followed Hume in rejecting induction, claiming that it is always possible to produce a theory to fit any set of observations [Reference Buck113]. According to Karl Popper ‘Our belief in a hypothesis can have no stronger basis than our repeated unsuccessful critical attempts to refute it’ [Reference Popper114]. Popper and other scientists believed that causation is established through a process of conjecture and refutation, and that science advances only by disproofs [Reference Buck113,Reference Page115,Reference Platt116]. Popper insisted strictly on deduction, allowing the sole capability of science to be the falsification of prior hypotheses (the so-called hypothetico-deductive method), rejecting any place for verification [Reference Rothman108].

Deriving causal inferences

The variation among the viewpoints of epidemiologists with regard to causality is rooted in the variation among philosophical viewpoints. However, Hill's criteria provide interpretive guidelines for evaluating epidemiological evidence. Hill established the following classic operational causal criteria: strength of association, consistency, specificity, temporality, biological plausibility, dose–response effect, coherence, experimental evidence, and analogy [Reference Hill121–Reference Phillips and Goodman124]. According to Hill [Reference Hill121], not all of these guidelines will be applicable in all situations and that there may be times when we wish to conclude that a putative cause–effect relationship is real even when some of the criteria are not met. According to Rothman [Reference Rothman125], only the criterion of temporality is a sine qua non for causality. If the putative cause did not precede the effect, that indeed is indisputable evidence that the observed association is not causal. Other than that one condition, there is no necessary or sufficient criterion for determining whether an observed association is causal [Reference Rothman125]. This conclusion is in accordance with the view of Hume, Popper, and others that causal inferences cannot attain the certainty of logical deductions [Reference Hume and Nidditch103,Reference Kuhn110,Reference Susser and Rothman111]. There is no explicit consensus about what constitutes sufficient evidence to establish causation from association.

Strength of association

Strength of association refers to the extent to which a supposed cause and effect are related and should not be confused with statistical significance [Reference Pearce and Crawford-Brown109]. The most common measure of strength of association is relative risk or rate ratio [Reference Pearce and Crawford-Brown109]. Other measures of association in epidemiology include the odds ratio, a correlation coefficient and attributable risk [Reference Pearce and Crawford-Brown109]. According to Chamberlin et al. [Reference Chamberlin129], technical advances have allowed the identification and/or isolation of MAP from a significantly higher proportion of Crohn's disease tissues than from controls. These methodologies include: (i) improved culture techniques; (ii) development of MAP-specific polymerase chain reaction assays; (iii) development of a novel in situ hybridization method; (iv) efficacy of macrolide and anti-mycobacterial drug therapies; and (v) discovery of Crohn's disease-specific seroreactivity against two specific MAP recombinant antigens [Reference Chamberlin129]. Several studies [Reference Naser130,Reference Mishina131] reported that 50% of Crohn's disease patients and 22% of ulcerative colitis patients were MAP positive and MAP was not cultured from the non-IBD patients. Some researchers suggest that all of IBD may be due to MAP [Reference Naser130,Reference Mishina131]. Chiodini et al. [Reference Chiodini132], described the isolation of a slow-growing, mycobactin-dependent Mycobacteria species from the intestinal mucosa of Crohn's disease patients but not from control tissue.

Consistency of effect

This epidemiological criteria refers to the fact that an association is found in many studies despite different circumstances, research designs, or time-periods [Reference Krieger78]. Relationships that are demonstrated in multiple studies are more likely to be causal than those that are not. Several studies conducted at different times by different research methods have reported on the isolation of MAP from patients with Crohn's disease [Reference Chamberlin129, Reference Hermon-Taylor133–Reference Sechi138]. MAP has been found in Crohn's disease patients by genetic probes (including both DNA, and RNA) [Reference Jeyenathan94]. The insertion element IS900, found at 14 to 18 copies per genome has been shown to be genomically specific for MAP [Reference Sechi138] and, has been widely used as a target for PCR [Reference Chamberlin129, Reference Mishina131, Reference Turenne139–Reference Ryan146].

Specificity of effect

Specificity describes the precision with which a factor will predict the occurrence of a specific disease; it adds plausibility to the causal claim but, if absent, does not detract from it [Reference Susser and Rothman111]. Routine culture of MAP from Crohn's disease patients' tissues is difficult because when present MAP is commonly in spheroplast form (cell wall deficient), which does not thrive in standard culture conditions [Reference Chamberlin129,Reference Chiodini143]. It has also proved difficult to detect MAP in Crohn's disease tissues by other methods: the mycobacterial cell wall Ziehl–Neelsen (ZN) staining techniques, first described in 1882 [Reference Ziehl147,Reference Neelsen148] have not shown MAP in humans because MAP exists in the cell-wall-deficient form [Reference Greenstein21]; serology studies have been beset by problems of non-specificity because of antigen cross reactivity [Reference Fiocchi149], although more recent studies have reported a specific high immune reactivity to recombinant MAP antigens in Crohn's patients [Reference Naser, Shafran and El-Zaatari150,Reference El-Zaatari151]. These difficulties reflect the fact that MAP microorganisms when present in Crohn's disease are few in number, relative to bovine cases of MAP infection (Johne's disease) [Reference Sanderson152]. An assay for MAP in Crohn's disease must be able to specifically detect small numbers of organsisms with tissue, near or below the threshold of microscopic detection [Reference Jeyenathan94]. Molecular methods have been used to determine the prevalence of MAP in cases of Crohn's disease [Reference McDonald153]. An important limitation of studies looking for novel pathogens is that information about the sensitivity and specificity of assays applied is generally lacking [Reference Jeyenathan94]. In separate studies, it has been shown that IS900 element is genomically specific for MAP [Reference Turenne139] and that IS900 sequences from a heterogenous collection of MAP are invariant [Reference Semret154]. According to Sechi et al. [Reference Sechi138], MAP has been identified by in situ hybridization to the MAP-specific IS900 gene in tissue specimens of Crohn's disease. However, despite these favourable considerations, the IS900-based in situ probe was prone to non-specific hybridization, compromising the utility of IS900-based in situ hybridization and indirect in situ PCR [Reference St. Amand155]. Jeyanathan et al. [Reference Jeyenathan94], reported that the alternative means of increasing specificity and sensitivity involves the use of rRNA-specific oligonucleotide probe in situ hybridization. Probes targeting rRNA provided excellent specificity resulting in forms that were morphologically consistent with ZN-positive organisms on adjacent sections. Ryan et al. [Reference Ryan146], reported the detection of MAP DNA in 40% of Crohn's cases where microdissected granulomas were examined. However, only half of the granuloma-positive cases had corresponding whole tissue sections that were positive for MAP. The greater detection rate of MAP in laser capture microdissection (LCM) isolated granulomas compared with whole tissue sections may have been attributable to better targeting of MAP DNA in granulomas – PCR may suffer loss of sensitivity because of the potential dilutional effect of the large quantities of non-target DNA found in whole tissue sections. Failure to detect MAP in some studies may have been attributable to inefficient amplification of long sequences (>250 bp) [Reference Riggio156].

Temporality

Temporality refers to the necessity that the cause precedes the effect in time [Reference Fletcher, Fletcher and Wagner128]. Causation is not possible without the cause occurring before the effect [Reference Rothman, Schottenfield and Fraumeni93]. Data exist that indicate that temporal sequence criteria have been fulfilled for the association between Crohn's disease and MAP [Reference Van Kruiningen157,Reference Van Kruiningen, Ruiz and Gumprecht158]. In a study by Van Kruiningen et al. [Reference Van Kruiningen157], a goat was infected with MAP organism taken from a human patient with Crohn's disease and showed progression to Johne's disease. A 1991 report found that 24-day-old specific pathogen-free Leghorn-Cochin chicks could be infected by multiple exposure routes using the same MAP strain (‘Linda’) [Reference Van Kruiningen, Ruiz and Gumprecht158].

Dose–response relationship

A dose–response effect is present when the effect increases with the dose or level of exposure. In a study conducted by Schwartz et al. [Reference Schwartz159], the intestinal mucosal layer from patients with IBD had high numbers of bacteria compared with people without Crohn's disease, however, there was no correlation between the numbers of bacteria present and either the degree of inflammation or the use of anti-inflammatory agents or sulfasalazine compounds [Reference Schwartz159]. This study suggests that a demonstration of dose–response criterion may not be applicable to a relationship between MAP and Crohn's disease. The pivotal event that convinced a totally sceptical gastroenterological community to accept that Helico bacter pylori was the aetiological factor in peptic ulcers was the cure rate that was achieved when the putative H. pylori infection was treated with appropriate antibiotics [Reference Greenstein21]. Similarly, Greenstein [Reference Greenstein21] suggested that the failure to cure IBD with anti-MAP antibiotics is the main impediment to convincing a sceptical gastroenterological community that MAP is zoonotic. Possible reasons that could account for this inability to cure patients with Crohn's disease include, the use of the wrong antibiotics and lack of satisfactory performed studies that are prospective, randomized, double blinded and placebo controlled, that have been performed using acknowledged satisfactory anti-MAP antibiotics [Reference Greenstein21]. Recently, Greenstein et al. [Reference Greenstein160], demonstrated that methotrexate and 6-mercaptopurine inhibit MAP growth in vitro. However, the dosages of methotrexate and 6-mercaptopurine in clinical use have not been titrated according to standard antibiotic conventions [Reference Greenstein160].

Biological plausibility

A hypothesized effect is biologically plausible if it makes sense in the context of current biological knowledge [Reference Rothman and Greenland75]. By the 1930s, Johne's disease was found to be caused by an odd bacteria named Mycobacteria paratuberculosis. This organism is from the same family of bacteria which cause tuberculosis and leprosy. Current concepts regarding the cause of Crohn's disease emphasize a dysfunction of the immune system resulting in a prolonged and intense process of inflammation [Reference Danze161–Reference Van Hogezand and Verspaget165]. The damage to the bowel appears to be due to this inflammatory process [Reference Sartor163–Reference Van Hogezand and Verspaget165]. MAP is thought to produce disease by over-stimulating the immune system. The bacterium lives inside the cells of the host, where it divides only once about every 2–12 h. By way of contrast, other bacteria in the gut such as Escherichia coli, Salmonella spp., Shigella spp., divide about once every 20 min. There are no toxins or poisons produced by MAP. Disease happens when the immune system recognizes the ‘foreign’ proteins of the bacteria, even inside a living cell and mounts a furious attack [Reference Danze161–Reference Sartor163]. The immune ‘attack’ focuses on the infected cells in the mucosal layer of the digestive system and results in massive inflammation, as well as ulcers, diarrhoea and weight loss [Reference Schwartz159, Reference Danze161–Reference Sartor163].