Table 1.

Hill’s considerations for causality inference in observational associations. Each causality benchmark is examined for cachexia–death association

Benchmark	Definition/comments	Application to cachexia–death association
1. Temporality^a	The cause (exposure) must precede the effect (outcome).	PRO: Wasting and weight loss occurs weeks, months, or years before death
1. Temporality^a	The cause (exposure) must precede the effect (outcome).	CON: Death is inherently the final event. Death may also occur during weight gain, e.g., upon nutritional support.
2. Strength of association	Stronger association may make causality more likely.	PRO: Most studies indicate strong and consistent associations between weight loss (pre-cachexia to cachexia) and imminent death.
2. Strength of association	Stronger association may make causality more likely.	CON: The reported strengths of the associations are not consistent. Disease severity, rather than cachexia, predicts death.
3. Biological gradient (dose–response)	Greater exposure increases the incidence or magnitude of the effect.	PRO: Wasting severity or more rapid weight loss may be associated with higher likelihood of death.
3. Biological gradient (dose–response)		CON: The wasting severity or cachexia progression rapidity has inconsistent and in some studies even weak association with death risk.
4. Consistency	The association can be replicated in studies in different settings using different methods.	PRO: Different weight loss patterns, e.g., anorexia nervosa and cancer cachexia, and different types of wasting, i.e., both fat loss and muscle loss, lead to death
4. Consistency		CON: Some types of weight loss such intentional weight loss or fat loss may be fully reversed without any risk of death (e.g., yoyo dieting).
5. Biologic plausibility	The association is consistent with known biological or pathological processes.^b	PRO: Cachexia may lead to thromboembolic events, arrhythmia, sudden cardiac death, immune system disarrays and higher rates of cardiovascular and infectious disease events and death.
5. Biologic plausibility		CON: There is essentially no confirmed pathophysiologic pathway between cachexia and death.
6. Experimentation	The putative effect can be altered (prevented or mitigated) by an experimental regimen.	PRO: In some animal models, starvation and weight loss can lead to death. Improving cachexia in human subjects improves survival.
6. Experimentation		CON: The current cachexia animal models are scarce and not convincing. Trials of nutritional support or anti-cachectic interventions in human subjects have often not improved survival.
7. Specificity	A single cause produces the effect without other pathways.	PRO: Preceding wasting and weight loss can fully explain death events.
7. Specificity	A single cause produces the effect without other pathways.	CON: Cachexia is only one of the correlates of chronic progressive disease states and likely an epiphenomenon (see Fig. 1, models 2 and 3). Death often happens in various acute and chronic diseases independent of cachexia.
8. Biologic coherence	The association is consistent with the natural history of the disease or laboratory findings.	PRO: A lower risk of death should result from preventing weight loss or by nutritional support.
8. Biologic coherence		CON: Natural history of death in chronic disease states has little to do with wasting and weight loss if any.
9. Analogy	The effect of similar factors may be considered in other populations or under different settings.	PRO: Wasting, fat, and muscle mass loss precede death as seen in any risk factors that precede mortal events.
9. Analogy		CON: There is no biologically plausible analogy in death due to other conditions, such as cardiovascular (atherosclerosis) or cancer death.

^aNote that temporality is the only necessary (but not sufficient) condition of causality

^bHowever, studies that disagree with established understanding of biological processes may force a reevaluation of accepted beliefs