Several ecologic studies reported positive associations between different indicators of salt consumption and stomach cancer mortality at the population level[17–21]. In an ecologic study of 24 countries, median urinary sodium levels, ascertained on randomly selected samples from each country, were significantly correlated with stomach cancer mortality (r = 0.70 in men; 0.74 in women) (both P < 0.001)[17]. Another study evaluated correlations between both salt intake and 24 h urinary sodium excretion and stomach cancer mortality among men in four geographic regions of Japan, and reported a strong correlation between stomach cancer mortality and salt excretion, but not with dietary salt intake[21]. An ecologic study of stomach cancer mortality in 65 Chinese counties observed significant, yet modest, correlations for intake of salt-preserved vegetables (r = 0.26 in men and 0.36 in women)[18]. A Japanese study administered a 38-item food frequency questionnaire to a sample of 634 men and the wives of 373 of these men from five districts in Japan. The rank correlation coefficient between gastric cancer mortality and pickled vegetable consumption was 0.36[19]. In a similar study of 207 Japanese men and the wives of 165 of the men, average daily sodium consumption, estimated using a 3 d weighed food record, was correlated with stomach cancer mortality rates (partial rank r = 0.45)[20].

In summary, the majority of ecological studies indicated that the average salt intake in each population was closely correlated with gastric cancer mortality. However, employing dietary assessment methods in these studies has some limitations, such as variation of the questionnaire validity in different population, and use of the same composition table calculating salt intake in diverse dietary cultures. One validated questionnaire used by one population may not be appropriate or valid for another population; thus in population based studies, dietary estimates may not be highly accurate and the association between dietary factors and disease may be much compromised. Applying the same composition table to calculate the same food in different dietary culture may cause bias, since the same food in different areas may have different salt content. For example, the regional average salt content of miso in 39 regions of 20 prefectures across Japan ranged from 9.1% to 18.2%[22]. Furthermore, as with all ecologic studies, diet and stomach cancer were neither measured nor analyzed on the individual level. Rather, the diets of sampled individuals were used to represent entire populations or geographic regions. Thus, misclassification is an obvious concern. Furthermore, associations observed at the population level cannot be assumed to hold at the individual level, and causality cannot be inferred from this type of study.

Forty two case-control studies on salt consumption in relation to stomach cancer are presented in Table 1[23–64]. Of these, twenty two studied overall salt, table salt or sodium[2324–44], twelve estimated salted/dried fish and/or salted fish gut or cod roe[253945–54], six investigated salted or pickled vegetables[253252–55], three focused on salty snacks[415657], and twelve studied salted foods in general[293032525458–64].

Among the sixteen studies that estimated overall dietary salt or sodium intake, eight in Puerto Rico, Spain, Korea, Italy, Mexico, China (two) and USA have shown strong statistically significant increases in risk (OR = 1.5-5.0 for the highest intake levels)[28–343643]. Seven of them reported statistically non-significant OR of 1.1 to 1.5 for consumption in the upper half of overall salt or sodium intake[24–27323544], and the remaining study reported no association[23]. Six of the studies specifically examined the use of table salt, with three studies (in Belgium, England, and Poland) reporting statistically significant increases in risk with OR = 1.6, 1.8, 6.2, respectively[384042]. Two other studies each reported statistically non-significant odds ratios of 1.5 for consumption of the upper half of the table salt intake distribution[3739]; the remaining study reported no association[41].

Of the twelve studies that estimated salted fish intake, four found strong statistically significant increases in risk (OR = 1.4-5.5 for the highest intake levels)[39454748]. One Japanese study reported a statistically significant increase in risk for high consumption of salted fish gut and cod roe in males, but not females, and no significant association for salted/dried fish for both genders[48]. Seven other studies reported statistically non-significant correlations[254649–53]; the remaining study reported a statistically significant inverse association with odds ratios of 0.7 for consumption in the upper half of the salted fish intake distribution[54].

Six studies in Table 1 examined salted vegetables; of these, three reported statistically significant increases in risk with higher intakes of salted vegetables[325355], the remaining three studies in China, Korea and Lithuania showed no relationship to stomach cancer risk[255254]. Additionally, three studies reported on salted snacks. Of them, only one study in Mexico reported a statistically significant relationship to stomach cancer[56], with the other two reporting no substantial associations[4157].

Twelve studies examined consumption of salted soya paste, salted side dishes and salty foods in general[293032525458–64]. Nine of them observed a moderate increase in risk with higher consumption (OR = 1.6-4.5)[293032546061–64], the remaining three reported no association[525859].

In summary, many case-control studies found similar results, indicating a moderate to high increase in risk for the highest level of salt or salted food consumption. Given the large number of studies that reported data on salt, sodium and salty foods consumption, some inconsistent results were to be expected. The inconsistencies may be due, at least in part, to the retrospective assessment of salt exposure, which might have changed after the diagnosis of stomach cancer. Furthermore, the degree to which each of these measures reflects total salt intake varies, and it is therefore not surprising that results would vary.

Eleven cohort studies, investigating salt or salted food consumption and stomach cancer risk in the US, Japan, Sweden, and the Netherlands have produced inconsistent results (Table 2)[65–75]. When viewed separately in the Table, the results were inconsistent for both salt intake and intake of salty foods. Four Japanese studies reported statistically significant associations (range of RR = 2.2-5.4 for the highest intake level)[67687375], including one study that reported significantly elevated risks in both men and women after 10 years follow up of 20 300 men and 21 812 women[73]. Another study, conducted in rural Japan with 8035 subjects and 76 stomach cancer deaths, reported a significantly elevated relative risk for the most frequent intake of highly salted foods compared with the least frequent intake (RR = 5.4; 1.8-16.3; P for trend < 0.01)[75]. In a study that examined 18 684 men and 20 381 women and included 486 histologically confirmed stomach cancer cases (358 men and 128 women), there was a dose-dependent association between salt consumption and stomach cancer risk in men (P for trend < 0.001), but not in women (P for trend = 0.48)[67]. Shikata et al[68] categorized 2476 subjects into four groups according to daily salt intake. After 14 years of follow-up, the age- and sex-adjusted incidence was significantly higher in the second to fourth groups than in the first group (RR = 2.4, 95% CI: 1.2-4.7; RR = 2.1, 95% CI: 1.0-4.3; RR = 3.0, 95% CI: 1.5-5.8, respectively). With the exception of these positive findings, the remaining seven cohort studies showed no substantial associations[656669–7274]. It is perhaps noteworthy that four studies with statistically significant positive results were conducted in Japan, which may be related to a potentially higher range of salt intake in that country.

In summary, some cohort studies suggest that a higher intake of salt or of salted food, as estimated by validated food frequency questionnaires, may be directly associated (or at least indirectly linked) with subsequent development of stomach cancer. Although the overall results from cohort studies are not totally consistent, they are suggestive of a moderate direct association.

Most epidemiological data suggest an association between salt intake and the development of stomach cancer. When interpreting these data, several issues must be considered.

Assessment of salt intake is difficult and prone to some potential biases. Many commonly used dietary assessment methods, such as food frequency questionnaires and diet records, have reported only moderate reproducibility in epidemiological studies, and thus some misclassification of dietary intake is inevitable[7980]. This may be particularly true of total salt intake, given its nearly ubiquitous addition to most processed foods. In addition, use of different salt assessment methods may lead to different conclusions. For example, both 24 h urinary excretion and 3 d dietary record methods for salt intake estimation were used in one Japanese ecologic study. Only urinary salt excretion level showed a strong correlation with stomach cancer mortality, while dietary salt was weakly and non-significantly correlated[21]. Although 24 h urine collection may be an optimal method in estimating routine salt intake, it is impracticable for a large-scale population study, especially a cohort study with long-term follow-up.

Another critical issue in interpreting salt consumption in relation to stomach cancer risk is the variation in the salt consumption levels across the population. To date, there is no standard method for salt intake categorization, and several studies have reasonably compared categories, such as quintiles for the application in the specific study population. Because average salt intake varies across different populations, salt consumption levels considered “high” in one study might be considered “low” in another study. For example, in a study conducted in Japan, subjects reporting once per week consumption of salted food were in the lowest exposure category[48]; in contrast, the subjects in a Turkish study reporting once per week were classified into the highest exposure level (defined as ≥1-2 times/week)[60]. Finally, in some studies[75], salted food levels were calculated from the sum of the scores of foods belonging to the food items (pickled vegetable + foods deep-boiled in soy sauce), making it impossible to compare effects at similar levels of consumption across studies.

Because of the complexity of diets, the traditional approach with a single nutrient may potentially be confounded by the interactions between food components that are likely to be interactive or synergistic[81]. It is possible that the increased risks in stomach cancer could be due to compounds other than salt in foods that were produced during the preservation process[56]. In East Asia, salted foods and sauces are also high in NO₃, a chemical carcinogen, which may either be added to the foods or synergize from amino acids during fermentation. Nitrite and salt may work at an early stage[82] in a synergistic fashion on stomach cancer carcinogenesis[17] that might cause the strong associations between highly salted foods and gastric cancer[67]. However, nitrite was not clearly related with stomach cancer risk[83] and its function may be influenced by other factors. For example, when lower salt intake was combined with higher NO₃ intake, stomach cancer mortality rates tended to be lower[17]. However, this might be explained by a higher intake of fresh fruits and vegetables, which are the major source of nitrate and also protect against cancer[8485].

Moreover, it may be difficult to separate the effects of salt from other nutrients that may contribute to stomach cancer risk. The absence of adjustment for confounding factors (such as age, sex, smoking and dietary habit) can hamper the statistical estimation causing over- or underestimation of the real association between salt or salted food and stomach cancer. In Tables 1 and 2, few studies have controlled for dietary factors in their analyses of salt consumption, which makes it difficult to compare the different studies according to the dietary variables adjusted in the analysis. However, the study results that were adjusted by a wide range of potentially confounding variables, such as age, sex, H pylori infection, atrophic gastritis, medical history of peptic ulcer, family history of cancer, body mass index, diabetes, total cholesterol, physical activity, alcohol intake, smoking habit and other dietary factors[68], showed no difference from the crude results. Studies with adjustment for some or most of the above potential confounding factors[66687273] showed no systematically apparent differences from the studies with adjustment for a few or several confounders.