Article Text

Download PDFPDF

Socioeconomic position and incidence of gastric cancer: a systematic review and meta-analysis
  1. Olalekan A Uthman1,2,
  2. Elham Jadidi3,
  3. Tahereh Moradi3,4
  1. 1Warwick-Centre for Applied Health Research and Delivery (WCAHRD), Division of Health Sciences, Warwick Medical School, The University of Warwick, Coventry, CV4 7AL, United Kingdom
  2. 2International Health Group, Liverpool School of Tropical Medicine, Liverpool, UK
  3. 3Division of Epidemiology, Institute of Environmental Medicine, Karolinska Institutet Stockholm, Stockholm, Sweden
  4. 4Health Care Services, Centre for Epidemiology and Social Medicine, Stockholm County Council, Sweden
  1. Correspondence to Associate Professor Tahereh Moradi, Division of Epidemiology, Institute of Environmental Medicine, Karolinska Institutet Stockholm, Nobels väg 13, Stockholm SE-171 77, Sweden; Tahereh.moradi{at}ki.se

Abstract

Background Low socioeconomic position (SEP) has been associated with increased risks of morbidity and mortality from many diseases. We investigated the associations between gastric cancer incidence and education, occupation and income as indicators for SEP.

Methods We searched the PubMed and EMBASE databases for studies on SEP and gastric cancer incidence published from 1966 through February 2013. We used a random-effect model to pool the risk estimates from the individual studies. The relative indexes of inequality (RIIs) with their 95% CIs were used as summary estimates. We stratified the analysis by SEP indicators, sex, country's income group, geographical area, level of adjustment for an established risk factor, publication year, study design, type of control and length of follow-up.

Results Of 1549 citations, 36 studies met our inclusion criteria. We observed an increased risk of gastric cancer among the lowest SEP categories in education (RII=2.97; 95% CI 1.923 to 4.58), occupation (RII=4.33; 95% CI 2.57 to 7.29) and combined SEP (RII=2.64; 95% CI 1.05 to 6.63) compared with the highest SEP categories. Although the association between the incidence of gastric cancer and the level of income is evident, it did not reach a statistically significant level (RII=1.25; 95% CI 0.93 to 1.68).

Conclusions We found that the risk of gastric cancer incidence is higher among low SEP groups.

  • CANCER
  • CANCER: OCCUPATIONAL
  • ECONOMICS

Statistics from Altmetric.com

Introduction

Gastric cancer is the second cause of cancer-related deaths in the world1 ,2 and still an important public health problem.3 Risk factors associated with gastric cancer include Helicobacter pylori infection,4 gastro-oesophageal reflux disease,5 family history,6 dietary habits,7 obesity and cigarette smoking.7–9 The other main candidate risk factors are socioeconomic positions (SEPs). Low SEP has been associated with increased risks of morbidity and mortality from many diseases.10–13 Studies from high-income countries have reported a variation in gastric cancer incidence among people with different levels of educational attainment.14–16

Numerous studies have examined the association between the incidence of gastric cancer and SEP. However, there is high variability in magnitude and direction in the reported association between gastric cancer and SEP. There may be several possible reasons for these inconsistencies including study design, source population, sample size and study periods. To our knowledge, no systematic review has examined the association between gastric cancer and SEP. A goal of systematic review is to provide the policymakers and researchers with a balanced assessment of the totality of the evidence. Thus, the overall aim of this study was to identify all studies that examined gastric cancer incidence in relation to SEP and perform a meta-analysis.

Material and methods

Information sources and search strategy

We conducted searches on the PubMed and EMBASE databases from inception to February 2013. We used key words related to socioeconomic determinants such as “socioeconomic position”, “socioeconomic status”, “social class”, “occupational category”, “occupational classification”, “educational level” or “income” versus tumour, neoplasm, cancer, carcinoma, malignancy, metastasis, versus stomach, gastric, cardia, non-cardia to search in these databases. In addition, we manually checked the reference list of the identified studies.

Selection criteria

We evaluated each identified study against the following predetermined selection criteria:

  • Study population: all patients with diagnosed gastric cancer, regardless of location.

  • Study design: case–control or cohort studies.

  • Outcomes: gastric cancer incidence and present risk estimates with 95% CI on the association between incident cases of gastric cancer and at least one measure of SEP, or sufficient information to compute these for men, women or both. Income, educational attainment, occupational categories or a combination of these were considered as SEP indicators.

Data abstraction

For each identified study that met the selection criteria, we extracted the following data from each publication: first author's last name, year of publication, country where the study was performed, study design, years of data collection, type of controls (population-based, hospital-based) in case–control studies, duration of follow-up in cohort studies, sample size, measure of exposure (indicators of SEP), source of SEP (individual-level measure, area-level measure), age, sex, risk estimate with corresponding 95% CIs and variables controlled for. The information on the country where the study was performed was then classified according to the geographical area (the USA/Canada, Europe, Asia and Latin America) and the country's income level (high-income and middle-income countries). We classified occupation according to the Erikson-Goldthorpe scheme in five groups of employment: (1) class I high-level employees, (2) class II medium-level employees, (3) class III low-level employees, (4) class V/VI skilled workers and (5) class VII unskilled workers.17

Study selection

Two of the authors (OU and EJ) evaluated the eligibility of the studies obtained from the literature search. In cases of discrepancy, the third author (TM) reviewed the studies until agreement was reached by consensus. One reviewer (OU) extracted the data and the others checked the extracted data.

Data synthesis

We used the relative index of inequality (RII) to assess the magnitude of the association between gastric cancer incidence and SEP.18 RII is a regression-based summary measure for social inequality that takes into account the size of all the categories in a socioeconomic hierarchy.19 ,20 A value between 0 and 1 was calculated according to the proportion of participants with a higher SEP than the midpoint of each SEP group starting with the best-off group. The measures of the gastric cancer rate of the SEP groups are then regressed on this measure of their relative position. As proposed in the literature, we used log-linear models, with a logarithmic function to calculate RIIs.21 ,22 RII can be interpreted as the ratio of the gastric cancer incidence of the most disadvantaged to the most advantaged. Thus, if the index is 2.5, then the gastric incidence of the most disadvantaged is 2.5 times as high as that of the most advantaged, and RII of 1.00 would specify equal morbidity across the socioeconomic hierarchy. The meta-analyses were performed using a random-effects model of DerSimonian and Laird,23 which incorporates within-study and between-study variability, since we anticipated between-study heterogeneity.

Following the overall analyses, a number of subgroup analyses were performed with respect to sex (by pooling the risk estimates from all studies where separate data for men and women were originally reported), study design, geographical area and income level of the country. To take into account the quality of the studies, the analyses were also stratified by the type of controls in case–control studies and the length of follow-up in cohort studies (10 years and less, more than 10 years).

To evaluate the stability of the results and to test whether one study had an excessive influence on the meta-analysis, leave-one-study-out sensitivity analysis was performed.24 The scope of this analysis was to evaluate the influence of individual studies, by estimating a pooled estimate in the absence of each study. We assessed heterogeneity among trials by inspecting the forest plots and using the χ2 test for heterogeneity with a 10% level of statistical significance, and using the I2 statistic with a value of 50% representing moderate heterogeneity.25 ,26 We assessed the possibility of publication bias by evaluating a funnel plot for asymmetry. As graphical evaluation can be subjective, we also conducted Begg's adjusted rank correlation test27 and Egger's regression asymmetry test28 as formal statistical tests for publication bias. Random-effect meta-regression was performed to investigate the source of heterogeneity. The independent variable was log (RII) and the explanatory factors included sex, study design, geographical area, country-income level, publication year, type of adjustment (none, minimal or maximal), type of controls in case–control studies, length of follow-up in cohort studies and source of SEP measures. All tests were two-tailed. For all tests, a probability level less than 0.05 was considered significant. This review was performed according to the PRISMA recommendations for meta-analyses.29 Stata V.11 (Stata Corporation, College Station, Texas, USA) software was used for statistical analyses. All statistical tests were two-sided.

Results

Study characteristics

Figure 1 shows the process of study identification and selection. The literature search yielded 1549 citations. After review of the title and abstract, 150 full-text articles were selected for critical reading. A total of 36 studies met the inclusion criteria. Table 1 shows the characteristics of the included studies. These studies were published between 1976 and 2012. The USA (n=5) and Italy (n=5) contributed the highest number of studies, followed by China (n=3) and Korea (n=3). Most of the studies were from high-income countries (n=25). Most of the studies were case-control studies (n=13) and 13 were cohort studies. Among the 23 case–control studies, 16 used hospital-based controls and the remaining 7 used population-based controls. Most of the studies (n=26, 72%) assessed the incidence of gastric cancer and education attainment. Only seven studies (19%) examined the association between the incidence of gastric cancer and combined SEP.

Table 1

Pooled estimates for the lowest versus the highest socioeconomic category and incidence of gastric cancer in a series of subgroup analyses

Figure 1

Flow diagram of search strategy and study selection process.

Overall summary of the meta-analyses

The random-effect meta-analysis yielded a pooled RII of 2.97 (95% CI 1.93 to 4.58, n=26), such that the risk of developing gastric cancer in patients with the lowest education was 2.97 times as high as that of patients with the highest education (figure 2). There was evidence of substantial statistical heterogeneity between the study results (χ2=1886; df=25; p<0.0001), with the degree of heterogeneity quantified by the I2 as 98.7%.

Figure 2

Relative index of inequality and 95% CI for gastric cancer incidence and education attainment categories.

The association between the incidence of gastric cancer and the level of income, although evident, did not reach a statistically significant level (pooled RII=1.25; 95% CI 0.93 to 1.68, I2=80.3%, n=10; figure 3). There was a statistically significant association between gastric cancer incidence and occupation (pooled RII=4.33; 95% CI 2.57 to 7.29, I2=28.4%, n=4; figure 4) as well as combined SEP (pooled RII=2.64; 95% CI 1.05 to 6.63, I2=66.4%, n=4; figure 5).

Figure 3

Relative index of inequality and 95% CI for gastric cancer incidence and income categories.

Figure 4

Relative index of inequality and 95% CI for gastric cancer incidence and occupation categories.

Figure 5

Relative index of inequality and 95% CI for gastric cancer incidence and combined socioeconomic position.

Additional analyses

We found no evidence of publication bias as indicated by the relatively symmetrical funnel plot of studies precision against RII (in logarithm scale). This was confirmed when formally tested using the Egger method (weighted regression) and the Begg method (rank correlation) for all the four measures of SEP (see figure 1 for test results).

The results of leave-one-study-out sensitivity analyses showed that no study had an excessive influence on the pooled relative risks of the association between gastric cancer incidence, educational attainment, income, occupation and combined SEP, thus confirming the stability of the results.

We found no evidence of statistically significant differentials in the socioeconomic inequalities across various study-level factors (table 1). Socioeconomic inequality due to education attainment tended to be higher in middle-income countries than in high-income countries (RII=5.11 vs 2.65); however, this difference did not reach a statistically significant level (p=0.208). In a series of meta-regression analyses, none of the study-level characteristics were associated with logs of relative risk of the association between the incidence of gastric cancer and all four indicators of SEP.

Discussion

The results from this meta-analysis indicate an overall increased risk in gastric cancer incidence among individuals with all studied SEP indicators (education, occupation, income and combined SEP). These negative social gradients were consistent for most of the possible sets of pooled estimates obtained in all possible subgroup analyses including sex, country's income group, geographical area, level of adjustment for established risk factor, publication year, study design, type of control and length of follow-up.

In order to fully understand the association between SEP and gastric cancer incidence, we included studies that used different SEP measures. It has been recommended that studies should not rely on any single socioeconomic indicator and ignore the others and that these SEP measures are not interchangeable as they represent different causal processes and pathways.30 We found that educational attainment was the most extensively used measure of SEP. Although the pathway through which SEP increases the risk of gastric cancer is yet to be established, higher educational attainment could be associated with more favourable working conditions and income and thus a healthier lifestyle and access to good healthcare.30 Risk factors associated with gastric cancer such as H pylori infection, genetic inheritance and lifestyle factors such as dietary habits, obesity and cigarette smoking have been shown to be associated with low SEP.31–36 The increased risk of gastric cancer among people with lower educational attainment and lower income may therefore be mediated partly through these factors. Low SEP may increase the risk of H pylori, which may in turn increase the risk of gastric cancer. The variation in environmental and host factors influencing gastric cancer risk between different populations is well documented.37 ,38 The biological mechanism by which H pylori contributes to gastric carcinogenesis varies across geographic regions. The observed heterogeneity in the association reported across studies may be due to the striking geographical patterns and biological mechanisms in gastric cancer incidence.37 ,38 It has been reported that there are different molecular pathways of gastric carcinogenesis.37 ,38

The results of this meta-analysis should be interpreted with caution. The observational nature of the data limits the ability to draw causal inferences. We found statistically significant heterogeneity across studies, thus suggesting that the percentage of the variability in effect estimates that is due to heterogeneity rather than sampling error (chance) is important. The heterogeneity may be due to differences in study follow-up, geographical location, reporting of SEP and variability in handling confounding and mediating factors. The study can be criticised for combining or pooling studies that used a different definition and classification of SEP. In addition, RII is sensitive to the average health situation of the population and can yield unreliable results when applied to small samples with aggregated data.17 ,18 Despite these limitations, the study's strengths are important. We conducted comprehensive searches of databases to ensure that all relevant publications were identified. The results of tests for publication bias provided evidence that we are unlikely to have missed studies that could have altered the meta-analyses results. We also reduced potential bias in the conduct of this review by having the authors independently scan through the search output and extract the data. We used a sophisticated measure of social inequality that has the advantage of taking into account the size of the population and the relative SEP of groups.17 ,18

In conclusion, we found that gastric cancer incidence was inversely associated with educational attainment, income, occupation and combined SEP regardless of gender, country of origin, type of study design, length of follow-up, country's income group and the year in which the study was published.

What is already known about this subject

  • Risk factors associated with gastric cancer include Helicobacter pylori infection, gastro-oesophageal reflux disease, family history, genetic inheritance, dietary habits, obesity and cigarette smoking.

  • An inconsistent association between gastric cancer incidence and low socioeconomic position (SEP) has been reported in the literature.

What are the new findings?

  • An increased risk for gastric cancer was associated with low levels of education attainment, income, occupation and combined socioeconomic position (SEP) regardless of sex, country of origin, type of study design, length of follow-up, country's income group and the year in which the study was published.

How might it impact on clinical practice in the foreseeable future?

  • Strategies for tackling socioeconomic inequality in gastric cancer incidence are needed.

References

Footnotes

  • Contributors TM designed the research. TM, OAU and EJ searched the publications, extracted the data and wrote the materials and methods, results and discussion sections. OAU checked all data. TM, OAU and EJ were responsible for data synthesis and helped design the study’s analytic strategy. OAU wrote the introduction. TM and EJ edited the manuscript. All authors read and approved the final manuscript.

  • Funding This work was supported by a grant from the Swedish Council for Working Life and Social Research (grant number: FAS 2006-0230 to TM), Stockholm, Sweden. The study founder had no role in the design, collection, analysis or interpretation of the data or in the writing or in the decision to submit the article.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles