TLDR summary of what I found:
- diet-wise brassica vegetables, in particular broccoli seems to have this effect (but other vegetables that share P450 inhibitors in common with citrus fruits have the opposite effect)
- medication-wise, the OTC "heartburn" drug Omeprazole has been fairly consistently reported to have this effect (but many, many more drugs of various classes have the opposite effect)
- out of the cigarette smoke components, it's the carcinogenic aromatic hydrocarbons and not the nicotine that's responsible for this interaction with caffeine.
There is a more extensive list of caffeine-drug interactions in Pharma. Rev., but few drugs shorten the half-life of caffeine, mainly some (but not all) proton pump inhibitors ("heartburn" drugs) like Omeprazole (−41%) and Cimetidine (−31%). (I think PPIs like Omeprazole are generally available without prescription.)
I would caution on using the claims that source without double-checking with the primary studies because they listed Fluvoxamine (an antidepressant) as −80% "decreases caffeine half-life by sixfold". Most other drugs in this class have the opposite effect on caffeine half-life. And it seems they misinterpreted that study as Fluvoxamine is one of the more potent inhibitors of P450 pathway involved, actually having the opposite effect on caffeine half-life:
The selective serotonin reuptake inhibitor fluvoxamine is a very potent inhibitor of the liver enzyme CYP1A2, which is the major P450 catalysing the biotransformation of caffeine. [...] During fluvoxamine, the median of the total clearance of caffeine decreased from 107 ml min-1 to 21 ml min-1 and the half-life increased from 5 to 31 h
31 / 5 = 6.2 so they almost used this study as a ref, but misinterpreted the direction of the effect.
But their listing of PPIs effect on caffeine metabolism is probably correct; a quick search found a paper tiled "Accelerated caffeine metabolism after omeprazole treatment [...]"
Omeprazole has previously been shown to induce hepatic cytochrome P4501A2 activity, as evidenced by an accelerated N-3-demethylation in the 13C-[N-3-methyl]-caffeine breath test [...]
A review of PPIs (see table 3) confirms this with additional studies, i.e. there is indeed increased/faster caffeine clearance on Omeprazole.
The Pharma Rev source alas also claims that some anti-psoriasis / anti-eczema drugs decrease caffeine half-life, listing Methoxsalen (−70%) and 5-Methoxypsoralen (−31%)... but again that seem to be an error because a primary study I found says:
During the period of maximum inhibition the mean elimination half-life of caffeine increased from 5.6 hours at baseline to 57 hours after administration of methoxsalen
And likewise for 5-methoxypsoralen.
There is a much longer list of drugs that increase the half-life of caffeine though (presumably mostly by competing on the P450 pathway). I would not be surprised if some foods do the same (given the known interactions between e.g. grapefruit and some drugs), but I couldn't find any data specific to foods and caffeine, except for grapefruit, which does increase caffeine half-life too.
From an older (2000) review on the same caffeine-drug interaction issue, it seems indeed that Omeprazole is the only drug with such a significant effect (well included in the review):
In fact that 2000 review lists Cimetidine in the opposite direction than Pharma Rev, which actually seems to be the correct one.
There's also an interesting 2011 study on what exact components of cigarette smoke accelerate the clearance of caffeine. It seems that these components of cigarette smoke are the polycyclic aromatic hydrocarbons.
Cigarette smoking accelerates the metabolism of certain drugs, particularly those primarily metabolized by cytochrome P450 1A2 (CYP1A2) and, to a lesser extent, CYP2E1 and some UDP-glucuronosyltransferases [1, 2]. The induction of CYP1A2 is mediated by binding of polycyclic aromatic hydrocarbons of the tobacco smoke to the aryl hydrocarbon receptor (AHR) with consequent transcriptional activation of the CYP1A2 gene. Furthermore, CYP1A1 and CYP1B1 enzymes are induced by tobacco smoking via AHR in various human tissues such as lung and placenta.
PAHs are carcinogenic and don't seem to have any uses as drugs.
The 2011 study itself looked at whether nicotine itself contributes to this effect on caffeine clearance, but the contribution from nicotine seems to be nonexistent (or at least undetectable after 8 days of dosing with nicotine.)
There was no significant influence of nicotine administration on the pharmacokinetic parameters of caffeine or the formation pharmacokinetics of paraxanthine, theophylline and theobromine. Since caffeine metabolism to paraxanthine is a specific probe reaction for CYP1A2 , it can be concluded that CYP1A2 activity is not affected by 8 days of nicotine dosing. Although previous studies in experimental animals have provided evidence for the role of nicotine in the induction of CYP1A1 and CYP1A2 enzymes [4–10], our study disproves the hypothesis that nicotine induces CYP1A2 activity in humans in vivo. The discrepancy between human and animal data may be explained by tissue and species specific expression patterns. The human caffeine phenotyping probes the hepatic CYP1A2 activity, whereas the animal studies are mainly on extrahepatic CYP1A1 induction or based on methods not capable of differentiating between CYP1A1 and CYP1A2 enzymes.
Now on the ARH angle, it seems that Omeprazole also works this way with regard to caffeine (being listed in a review on the receptor), but the evidence is pretty muddy (or at least was in the year 2003 when that review was written):
Interestingly, some chemicals have
been identified that can induce AhR-dependent gene expression, yet they reportedly
fail to competitively bind to the AhR (Table 1 [first entry there is Omeprazole]). It has been proposed that
these chemicals are not AhR ligands themselves, but that they can activate AhR-dependent
gene expression indirectly, either via metabolic conversion into a ligand
or by their ability to affect some cellular pathway that results in AhR activation.
These conclusions are difficult to reconcile, especially given what is known about
the AhR-dependent mechanism of gene activation. Although these weak inducers
have not been observed to competitively bind to the AhR, they may still be AhR
ligands, albeit ligands that bind with relatively low affinity. Demonstration of the
ability of weak ligands (Kd in the uM range) to competitively bind to the AhR in
standard binding assays is technically challenging, especially given the extremely
high AhR binding affinity of TCDD [Kd in the pM range (143)]. Recent modifications
of the AhR ligand binding assay that favor competitive binding by weak
AhR ligands (i.e., reduction of [3H]TCDD and increased competitor concentration)
have been used to demonstrate that carbaryl, previously reported to not bind to
the AhR (144, 145), is actually a weak AhR ligand (146). The 300,000-fold lower
potency of carbaryl, as compared to TCDD (146), likely explains its inability to
competitively displace [3H]TCDD from the AhR ligand binding domain using the
ligand binding assay conditions described by those investigators. In addition, the
competitive binding of Trp metabolites (namely tryptamine and indole acetic
acid) and several benzimidazoles (omeprazole, thiabendazole, albendazole, and fenbendazole) has also been observed (87, 147), even though these chemicals were
previously reported to induce AhR-dependent gene expression in a ligand independent
manner (148–151). Not only are these binding results consistent with what is
known about other CYP1A1 inducers and AhR activators, but it is likely that many
(if not all) of the other chemicals that reportedly induce in an AhR-independent
manner (Table 1) are actually weak AhR ligands. The absolute requirement for
the AhR in the chemical-inducible response, combined with the demonstration
that some “ligand independent” inducers actually bind to the AhR, raises questions
as to the existence of these proposed alternative induction pathways. Unlike
other ligand-dependent receptors (i.e., steroid hormone receptors) that can be
activated in a ligand-independent manner by processes such as phosphorylation
(152, 153), ligand-independent activation for the AhR remains to be confirmed.
And on other sources of ArH ligands, particularly dietary, it notes that:
of relatively potent AhR ligands from precursors that have little or no AhR
agonist activity is significant, especially considering that most dietary ligands are
themselves relatively weak AhR ligands/agonists. Flavonoids, including flavones,
flavanols, flavanones, and isoflavones, represent the largest group of naturally
occurring dietary AhR ligands. Although the majority of these natural plant products
are AhR antagonists (43–47), numerous agonists, such as quercetin (48),
diosmin (49), tangeritin (50), and tamarixetin (43), have also been identified. In
addition to interacting with the AhR, many of these flavonoids are also substrates
for CYP1A1 (51). These chemicals are widely distributed in dietary vegetables,
fruits, and teas (52–55), and flavonoid levels in human blood have been reported
to be in the low uM-range (56–58), concentrations sufficient to inhibit/activate the
AhR. Thus, it is not surprising that crude extracts of a large number of different
vegetables, teas, fruits, and natural herbal products have AhR agonist and/or antagonist
activity (59, 60). Thus, plant-derived materials appear to commonly contain
AhR ligands or products that can readily be converted into AhR ligands, and as
such, they are perhaps the largest class of natural AhR ligands to which humans
and animals are exposed.
So it seems somewhat plausible that some dietary sources of such ligands might have a similar effect to PAHs on caffeine clearance, but insofar I haven't found some specific studies on the interaction in the direction of increased clearance. Actually, there's one study in the opposite direction that we might have hoped for:
Quercetin significantly inhibits the caffeine metabolism
There is actually one year 2000 study on diets and CYP1A2 activity, as
measured by excreted caffeine metabolites. It found that adding brassica
vegetables (specifically, they added broccoli, cabbage, radish sprouts, and cauliflower) to a diet increased caffeine
metabolism (by some 18-37%), while apiaceous vegetables (carrots, parsnips,
celery, dill and parsley) had the opposite effect by 13-25%. (Exact measure
depends on the metabolite.)
It also notes that furanocoumarins are probably responsible for the latter
effect, in common with that observed with citrus fruits and caffeine, while for the increased caffeine clearance on brassica vegetables, they suggest glucosinolates are responsible. (There's an older study adding just broccoli, which found the same effect. The broccoli-only study was also replicated.)