This cross-sectional study found an inverse association between an 8-item CLI and migraine prevalence in Hong Kong Chinese women. The results were consistent in leave-one-out analyses. Among the eight components, migraine was significantly associated with sleep, stress, fatigue, and diet; but was not related to smoking, physical activity, BMI, and alcohol. Furthermore, combining the four lifestyles that showed no individual association with migraine still maintained the inverse relationship between CLI and migraine. Subgroup analyses based on migraine subtypes also generated similar results. Among women with migraine, CLI was inversely associated with migraine attack frequency.

There were limited studies exploring the associations between combined lifestyles and migraine. A Norway cross-sectional study of 5588 students (12–19 years) found that, compared to participants with all three health lifestyles (high physical activity, non-smoking, and normal weight), subjects with two, one, and zero lifestyles had increased migraine prevalence, with ORs (95% CIs) of 1.5 (1.2–1.9), 2.1 (1.5–2.8), and 3.7 (1.9–7.1), respectively [19]. Conversely, another German cross-sectional study of 6309 participants (35–75 years) showed that a 4-item health index (smoking, physical activity, alcohol, and BMI) was not associated with migraine [20]. There are several possible reasons for the different results. First, the two studies recruited participants with different ages (12–19 vs. 35–75 years). Second, the difference in lifestyle components might contribute to the discrepancies. Third, the sample sizes were different between the studies. Specifically, although the German study had 6309 participants, it was actually performed separately by three databases [20]. Therefore, the sample size of each dataset might not be sufficient.

Unlike the limited studies focusing on combined lifestyles, many prior studies have investigated the relationships between individual lifestyles and migraine. For example, an 11-year cohort research in Norway found that, among 15,276 individuals without a baseline headache, the migraine incidence was higher in smokers than in never smokers, with a relative risk (RR) (95% CI) of 1.30 (1.11–1.52) [4]. Meanwhile, that study also indicated that, compared to subjects being physically inactive, those with light (1–3 h/week) and vigorous physical activity (1–2 h/week) had reduced migraine incidence, with RRs (95% CIs) of 0.78 (0.62–0.99) and 0.71 (0.54–0.94), respectively [4]. Likewise, a Swedish cross-sectional study of 43,770 subjects (18–79 years) revealed that smoking and physical inactivity were positively associated with migraine [38]. However, a German cross-sectional study with 6309 subjects reported no associations of smoking and physical activity with migraine [20]. While the relatively small sample size might partially explain the difference.

Regarding sleep, a meta-analysis of 23 case-control studies showed that the PSQI score of adults with migraine was higher than that of controls [5]. Additionally, a randomized controlled trial (RCT) of 31 US adults with chronic migraine and insomnia indicated that the insomnia cognitive-behavioral therapy group had a lower headache event rate than the control group (38.8% vs. 48.1%; OR = 0.40; 95% CI = 0.17–0.91; p = 0.028) [39]. Another RCT of 43 US women with transformed migraine also found that the behavioral sleep modification could reduce headache frequency and intensity [40]. For stress, a meta-analysis of 2 RCTs with 43 patients found that mindfulness-based stress reduction could decrease migraine pain intensity (standardized mean difference = − 0.84; 95% CI = -1.48 to − 0.19; p = 0.01) [6]. Additionally, a prospective cohort study of 1125 US individuals with episodic migraine suggested that tiredness/fatigue was positively correlated with migraine [8]. Moreover, a meta-analysis of 11 studies found that, compared to normal-weight subjects, underweight people and obese women had an increased migraine risk, with ORs (95% CIs) of 1.21 (1.07–1.37) (p = 0.002) and 1.44 (1.05–1.97) (p = 0.023), respectively [7].

Meanwhile, previous studies identified several dietary factors and patterns associated with migraine. A cross-sectional study of 25,755 US women indicated that, compared to participants with migraine without aura, those with migraine with aura had a low intake of chocolate, cheese, ice cream, hot dogs, and processed meats [10]. In an Iran cross-sectional study with 224 women with migraine (20–50 years), inverse associations were observed between the Mediterranean diet and migraine headache index score, headache frequency, headache duration, and headache impact test-6 [12]. Another Iran cross-sectional study of 285 women with migraine also found negative relationships of the Dietary Approaches to Stop Hypertension (DASH) diet with migraine index score and attack frequency [13]. For alcohol, a recent meta-analysis of 19 studies with 126,173 participants showed that alcohol drinkers had a lower migraine risk than non-drinkers (RR = 0.71; 95% CI = 0.57–0.89) [41]. One potential reason for the inverse alcohol-migraine relationship could be that patients with migraine might abstain from alcohol due to its capacity to trigger headache [38].

The advantage of combining lifestyle factors lies in considering their potential interactions. Individuals often adopt multiple lifestyles simultaneously, and these factors may interact, yielding different associations with migraine compared to individual analyses. Our aim was to evaluate not only individual lifestyle factors but also the overall lifestyle pattern in relation to migraine. Our results showed that, compared to the lowest CLI group, the ORs of the other four groups seemed to be lower than those of significant individual lifestyle factors. Although the differences in the ORs did not achieve statistical significance. Additionally, when we devised a new CLI that incorporated the four components exhibiting no individual association with migraine in their respective analyses, the new index consistently exhibited an inverse relationship with migraine. These findings underscored the importance of a comprehensive healthy lifestyle pattern. However, combining lifestyles does not mean neglecting individual factors. Hence, we presented results for both overall CLI and each component individually. Moreover, to discern whether any individual lifestyle factor exerted a significant influence on the results of the CLI, we conducted leave-one-out analyses by excluding each item from the index one at a time. The results of leave-one-out analyses were consistent with the main results, suggesting the robustness of the index.

The correlations between lifestyles and migraine could be partially explained by some mechanisms. For example, the analgesic property of tobacco might influence the central nervous system and then cause headache [4]. However, it is also possible that patients with migraine might smoke more than non-headache subjects due to headaches [4]. The mechanisms of physical activity might involve elevated plasma levels of beta-endorphin, endocannabinoids, and brain-derived neurotrophic factor [4, 42]. Additionally, physical activity might interact with other lifestyles like BMI [4, 42]. By contrast, it is also reported that those with migraine might avoid physical activity [4, 42].

Furthermore, the sleep-migraine relationship might be associated with their shared brain regions [43]. Rapid eye movement sleep was regulated by cells in certain brain regions, such as the ventrolateral periaqueductal gray (vPAG) which was supplied by orexinergic inputs from the hypothalamus [43]. However, the vPAG also exerted an inhibitory impact on the nociception in the trigeminal nucleus caudalis (TNC), the principal region in the brainstem accountable for the perception of head pain [43]. Hence, disturbances in sleep might interfere with the signaling from hypothalamus to vPAG, consequently impacting its ability to inhibit pain perception in the TNC, thereby leading to headache attack [43]. Conversely, migraine might also lead to poor sleep via these shared brain regions [43].

Additionally, stress could directly affect the autonomic nervous and neuroendocrine systems, potentially leading to a sensitization of nociceptors [44]. Prolonged exposure to stress might imped the brain’s capacity to maintain allostasis [44]. Furthermore, stress might also indirectly cause headaches by contributing to other unhealthy lifestyles, such as poor diet, sleep, and fatigue [44]. In contrast, subjects with migraine were also reported to experience more stress than non-headache individuals [44].

For diet, the food categories included in this study might partially explain the inverse correlation. These food categories contained high amounts of fiber, vitamin B, vitamin C, coenzyme Q10, and magnesium; and low amounts of fat and sodium [13, 45]. Some of these constituents exhibited anti-inflammatory and antioxidant properties, potentially inhibiting the generation of inflammatory cytokines and mitigating neurogenic inflammation associated with migraine [13]. For example, high-fiber diets were linked to inflammation reduction by impeding glucose absorption and modifying gut microflora [45]. The low levels of riboflavin, coenzyme Q10, and magnesium in migraineurs might contribute to the pathophysiology, as these nutrients were essential for energy generation within mitochondria [13]. Meanwhile, magnesium could prevent migraine by blocking N-Methyl-D-aspartate receptors, curtailing serotonin-dependent vascular spasms, and hindering platelet aggregation [13].

For obesity, adipose tissues can release proinflammatory cytokines (e.g., tumor necrosis factor alpha, interleukin-1, interleukin-6, and adiponectin), which could activate the nitric oxide pathway in the brain and then cause headaches [7]. For underweight, factors like psychiatric comorbidities (e.g., anxiety, depression, stress, and fatigue) might serve as potential contributors [7]. Furthermore, potential mechanisms behind alcohol-induced migraine involved the vasodilation of cerebral blood vessels after drinking and the receptors located in the cortex or brainstem [41].

Overall, there were some common mechanisms for the association between lifestyle factors and migraine, involving the influence of both brain function and structure, as well as eliciting inflammatory responses. Additionally, these lifestyles demonstrated significant interconnections, potentially culminating in a synergistic impact on migraine. On the contrary, individuals with migraine might also have some unhealthy lifestyles due to headaches.

Although this might be the first cross-sectional study to explore the correlation between CLI and migraine in Chinese people, there are several limitations. First, due to its cross-sectional nature, this study cannot establish causal associations. Further cohort studies involving the CLI or RCTs based on some behavior change interventions should be performed. For instance, Robblee J et al. proposed that primary care physicians can help patients with migraine reduce attack likelihood and symptom severity through lifestyle counselling related to the SEEDS intervention (sleep, exercise, eat, diary, and stress) [46]. The SEEDS program included adherence to standard sleep hygiene for optimal sleep quantity and quality (S); engagement of 30–60 minutes of physical activity 3–5 times/week (E); regular and nutritious meals with controlled caffeine intake (E); the utilization of a headache diary for the follow-up of headache (D); and stress management such as cognitive behavioral therapy, mindfulness, relaxation, or biofeedback (S) [46]. Thus, the application of our CLI in RCTs may be similar to the SEEDS program, which includes a comprehensive set of interventions targeting each lifestyle factor within the CLI, along with specific methods for the management of migraine such as a headache diary.

Second, recall bias might exist due to the use of questionnaires. Nevertheless, we mitigated this bias by using face-to-face interviews rather than self-administered questionnaires. Third, the definitions of “healthy” for some lifestyle factors were relatively arbitrary, which introduced the inconsistencies in defining “healthy” across CLI components. For instance, a healthy diet was delineated as ≥ the median value of a healthy diet index. Whereas the dietary data only contained the intake frequency of several food items. This method, however, lacked consideration for actual intake quantities and overlooked other key food categories like whole grains. Future studies should incorporate more dietary data. In addition, a healthy status of fatigue was established as below the median value of an 11-degree self-perceived fatigue scale, rather than some structured tools like the fatigue severity scale. The two arbitrary definitions were different with the definitions of other components, which were derived from previous studies or recognized standards. Therefore, further studies are recommended to include more detailed data on diet and fatigue with the use of some systematic tools.

Fourth, in this study, BMI was dichotomously classified as falling within the range of 18.5–23 kg/m^2 or not, thus grouping underweight and overweight individuals together. This classification approach raises concerns regarding potential misclassification, given that the associations of migraine with underweight, normal weight, and overweight might be different [7]. Whereas the selection of this threshold was based on a meta-analysis, which demonstrated that both underweight individuals and obese women exhibited an increased risk of migraine compared to those within the normal weight range [7]. Thus, employing this cut-off value for BMI may have minimal misclassification biases.

Finally, the study exclusively involved Hong Kong Chinese women. The main advantage of such a target population was the reduction of potential confounders associated with sex-specific physiological and hormonal differences, since migraine prevalence has been observed to be higher in women than men [2]. However, it should be noted that, by limiting the population to Hong Kong Chinese women, the generalizability of the findings to men and other ethnicities is limited. Moreover, by only focusing on women, potential interactions between sex and other variables that related to the lifestyle-migraine association might be overlooked. Future studies that include both sexes are needed to provide a more comprehensive understanding of the relationship between lifestyle and migraine.