Association between serious psychological distress and nonparticipation in cancer screening and the modifying effect of socioeconomic status: Analysis of anonymized data from a national cross-sectional survey in Japan
We thank Ms. Shoko Yoshimoto, who always supported our study with logistics assistance, and Rachel Baron, PhD, from Edanz Group (www.edanzediting.com/ac) for editing a draft of this article.
Abstract
BACKGROUND
It is unclear whether individuals who have serious psychological distress (SPD) are less likely to participate in screening tests for gastric cancer, lung cancer, and other types of cancer. Of the few studies that have examined the association between SPD and participation in cancer screening, none have reported modifying effects of educational, marital, or employment status.
METHODS
The authors analyzed a national representative data set from the 2010 Comprehensive Survey of Living Conditions of Japan., including individuals aged <69 years who met the national program criteria for each type of cancer screening (colorectal, gastric, and lung cancers, n = 29,926; breast cancer, n = 15,423; and cervical cancer, n = 24,735). SPD was defined as a score of 13 or greater on the Kessler 6 scale. Logistic regression analyses were conducted to examine the association between SPD and participation in cancer screening, and multivariate analyses stratified by socioeconomic status also were conducted.
RESULTS
SPD was significantly associated with a lower odds ratio (OR) for participation in screening for colorectal cancer (OR, 0.743; 95% confidence interval [CI], 0.638-0.866), gastric cancer (OR, 0.823; 95% CI, 0.717-0.946), and lung cancer (OR, 0.691; 95% CI, 0.592-0.807). Only educational status significantly modified the effect of SPD on participation in these 3 types of cancer screening (P < .05).
CONCLUSIONS
Individuals with SPD, especially those with lower education levels, were less likely to participate in screening for colorectal, gastric, and lung cancers. Individuals with SPD should be encouraged and supported to participate in cancer screening tests. Cancer 2018;124:555-62. © 2017 American Cancer Society.
INTRODUCTION
Cancer screening programs are an important part of preventive health care. Several studies have reported that individuals experiencing psychological distress (PD), such as depression or anxiety, are less likely to participate in cancer screening. According to our in-house, preliminary, systematic review, most population-based studies examining the associations between PD and screening for colorectal,1-14 breast,1-6, 10, 14-21 or cervical cancers1, 3, 10, 15, 17-19, 22-24 have been conducted in North America or Europe. One meta-analysis reported a significant reduction in mammography screening rates among women with mood disorders.25 However, subsequently published studies reported a nonsignificant association between PD and participation in mammography screening.10, 14, 18-21 Findings for colorectal and cervical cancer screening have been controversial. Moreover, because the majority of previous studies were conducted in countries with relatively high cancer screening rates, their findings may not be applicable to countries with lower cancer screening rates, such as Japan.26 Only a few studies have examined the association between PD and cancer screening behavior in Asia.7, 27 We identified only 1 study on gastric cancer screening, which is more important in Eastern Asia, where the incidence and mortality rates of gastric cancer are relatively high.27
Low socioeconomic status (SES) reportedly is associated with both high PD28 and nonparticipation in cancer screening.29 However, only a few studies have considered the modifying effect of SES on the relation between PD and participation in cancer screening.4, 8, 11, 20 The objectives of the current study were: 1) to investigate the association between serious PD (SPD) and participation in cancer screening in the general population, and 2) to explore the modifying effect of SES on the association between SPD and participation in cancer screening.
MATERIALS AND METHODS
Setting
In Japan, all residents are required to join the public health insurance program, and individuals' medical expenses are kept below specified limits according to their income. Residents can access cancer screening provided by municipalities or opportunistic screening, which may be free or include a low copayment.
Data Source
We used an anonymized data set from the 2010 Comprehensive Survey of Living Conditions (CSLC) conducted by the Ministry of Health, Labor, and Welfare (MHLW) of Japan,26 which is a national cross-sectional survey. The 2010 Household and Health questionnaire covered 289,363 households (comprising approximately 750,000 individuals) in Japan who were randomly sampled from 5510 districts included in the 2005 National Census. The response rate for the 2010 Household and Health questionnaire was 79.1% (228,864 responses). We used an anonymized subset of the data comprising 93,730 responses, which was resampled by the MHLW while maintaining representativeness.30 Because our analysis of the anonymized data set was conducted independently, our statistical data may differ from those published by the MHLW.
Participants
We included 1) individuals ages 40 to 69 years for analyses of colorectal, gastric, and lung cancer screening; 2) women ages 40 to 69 years for analyses of breast cancer screening; and 3) women ages 20 to 69 years for analyses of cervical cancer screening. We excluded participants based on the following criteria: 1) unknown age; 2) in hospital or in social welfare facilities at the survey date, or this information was unknown; 3) currently attending outpatient clinics for malignant neoplasm, pregnancy, or postpartum care; and 4) not independently mobile.
Variables
Cancer screening
The main outcome variables were participation in colorectal, gastric, or lung cancer screening within 1 year and breast and cervical cancer screening within 2 years. The MHLW recommended the following types of cancer screening at the time of the survey: fecal occult blood testing for colorectal cancer screening, upper gastrointestinal x-ray for gastric cancer screening, chest x-ray for lung cancer screening for individuals aged ≥40 years every year, mammography for breast cancer screening for women aged ≥40 years every 2 years, and Papanicolaou (Pap) smear testing for cervical cancer screening for women aged ≥20 years every 2 years. The 2010 CSLC questionnaire asked whether respondents had received these recommended screening tests as well as an upper endoscopy test for gastric cancer screening and breast ultrasound for breast cancer screening, which were not included on the MHLW list of recommended tests in 2010.
Psychological distress
The Kessler 6 (K6) scale was used to measure PD in the CSLC. The K6 scale was developed as a brief screening scale for nonspecific PD in adults31, 32 and consists of 6 questions asking about symptoms of PD in the previous month. We classified SPD as a K6 score ≥ 13.31
Covariates
The categorical variables age, sex, educational status, employment status, marital status, type of health insurance, smoking status, the presence or absence of current visits to outpatient clinics for physical illness, and independence of daily living were included as potentially confounding variables.
Statistical Analysis
All analyses were conducted with SPSS version 22 (IBM Corporation, Tokyo, Japan). The characteristics of the 3 groups are described in Table 1. We conducted multivariate logistic regression analyses to assess the associations between SPD and participation in each type of cancer screening. In model 1, we calculated the partially adjusted odds ratio (OR) and 95% confidence interval (CI) of SPD for participation in cancer screening with adjustment for age and sex. In model 2, we calculated the fully adjusted OR of SPD with adjustment for all covariates. In addition, we calculated the fully adjusted OR of PD divided into 4 score categories (no PD, 0-4; mild PD, 5-9; moderate PD, 10-12; and severe: PD, ≥ 13)33 to explore potential associations in more detail (for example, the possible association between moderate levels of anxiety and higher likelihood of participating in cancer screening, compared with individuals with milder or more severe levels of anxiety).
| No. of Study Participants (%) | ||||||
|---|---|---|---|---|---|---|
| Colorectal, Lung, and Gastric Cancer | Breast Cancer | Cervical Cancer | ||||
| Characteristic | K6 < 13 | K6 ≥ 13 | K6 < 13 | K6 ≥ 13 | K6 < 13 | K6 ≥ 13 |
| Total no. of participants | 28,814 | 1112 | 14,813 | 610 | 23,556 | 1179 |
| Sex | ||||||
| Women | 14,715 (51.1) | 611 (54.9) | 14,813 (100) | 610 (100) | 23,556 (100) | 1,179 (100) |
| Age group, y | ||||||
| 20-29 | ― | ― | ― | ― | 3597 (15.3) | 240 (20.4) |
| 30-39 | ― | ― | ― | ― | 5146 (21.8) | 329 (27.9) |
| 40-49 | 9738 (33.8) | 492 (44.2) | 5076 (34.3) | 273 (44.8) | 5076 (21.5) | 273 (23.2) |
| 50-59 | 9731 (33.8) | 384 (34.5) | 5045 (34.1) | 202 (33.1) | 5045 (21.4) | 202 (17.1) |
| 60-69 | 9345 (32.4) | 236 (21.2) | 4692 (31.7) | 135 (22.1) | 4692 (19.9) | 135 (11.5) |
| Education level | ||||||
| ≤High school | 15,576 (54.1) | 591 (53.1) | 8287 (55.9) | 342 (56.1) | 11,138 (47.3) | 562 (47.7) |
| >High school but ≤ junior or vocational college | 4855 (16.8) | 197 (17.7 | 3658 (24.7) | 133 (21.8) | 6688 (28.4) | 316 (26.8) |
| ≥University or college | 5819 (20.2) | 190 (17.1) | 1508 (10.2) | 58 (9.5) | 3718 (15.8) | 191 (16.2) |
| Unknown | 2564 (8.9) | 134 (12.1) | 1360 (9.2) | 77 (12.6) | 2012 (8.5) | 110 (9.3) |
| Employment status | ||||||
| Employed | 20,606 (71.5) | 703 (63.2) | 8911 (60.2) | 329 (53.9) | 14,922 (63.3) | 699 (59.3) |
| Unemployed, homemaker | 5344 (18.5) | 219 (19.7) | 5170 (34.9) | 209 (34.3) | 7232 (30.7) | 321 (27.2) |
| Unemployed, student | 24 (0.1) | 2 (0.2) | 15 (0.1) | 1 (0.2) | 324 (1.4) | 23 (2.0) |
| Unemployed, others | 2532 (8.8) | 168 (15.1) | 633 (4.3) | 62 (10.2) | 973 (4.1) | 125 (10.6) |
| Unknown | 308 (1.1) | 20 (1.8) | 84 (0.6) | 9 (1.5) | 105 (0.4) | 11 (0.9) |
| Marital status | ||||||
| Married | 23,438 (81.3) | 783 (70.4) | 11,876 (80.2) | 433 (71.0) | 16,338 (69.4) | 662 (56.1) |
| Never married | 2615 (9.1) | 149 (13.4) | 999 (6.7) | 56 (9.2) | 4914 (20.9) | 347 (29.4) |
| Bereaved | 1120 (3.9) | 51 (4.6) | 891 (6.0) | 39 (6.4) | 906 (3.8) | 40 (3.4) |
| Divorced | 1641 (5.7) | 129 (11.6) | 1047 (7.1) | 82 (13.4) | 1398 (5.9) | 130 (11.0) |
| Health insurance | ||||||
| NHI municipalities | 9486 (32.9) | 421 (37.9) | 5078 (34.3) | 226 (37.0) | 6540 (27.8) | 360 (30.5) |
| NHI associations | 528 (1.8) | 13 (1.2) | 241 (1.6) | 9 (1.5) | 397 (1.7) | 18 (1.5) |
| EHI, insured | 12,918 (44.8) | 424 (38.1) | 4175 (28.2) | 167 (27.4) | 8153 (34.6) | 391 (33.2) |
| EHI, dependent | 5222 (18.1) | 188 (16.9) | 5023 (33.9) | 178 (29.2) | 7992 (33.9) | 359 (30.4) |
| Long-life medical care system | 30 (0.1) | 2 (0.2) | 17 (0.1) | 1 (0.2) | 17 (0.1) | 1 (0.1) |
| Others | 382 (1.3) | 47 (4.2) | 149 (1.0) | 19 (3.1) | 246 (1.0) | 35 (3.0) |
| Unknown | 248 (0.9) | 17 (1.5) | 130 (0.9) | 10 (1.6) | 211 (0.9) | 15 (1.3) |
| Smoking status | ||||||
| Nonsmoker | 20,143 (69.9) | 700 (62.9) | 12,739 (86.0) | 460 (75.4) | 19,761 (83.9) | 852 (72.3) |
| Past smoker | 1041 (3.6) | 39 (3.5) | 205 (1.4) | 16 (2.6) | 487 (2.1) | 38 (3.2) |
| Smoker, ≤ 20 cigarettes/d | 5360 (18.6) | 223 (20.1) | 1502 (10.1) | 94 (15.4) | 2773 (11.8) | 225 (19.1) |
| Smoker, ≥ 21 cigarettes/d | 1770 (6.1) | 113 (10.2) | 167 (1.1) | 24 (3.9) | 260 (1.1) | 40 (3.4) |
| Smoker, no. per d uncertain | 9 (0.0) | 1 (0.1) | 3 (0.0) | 0 (0.0) | 9 (0.0) | 0 (0.0) |
| Unknown | 491 (1.7) | 36 (3.2) | 197 (1.3) | 16 (2.6) | 266 (1.1) | 24 (2.0) |
| Current outpatient visit | ||||||
| No | 16,373 (56.8) | 500 (45.0) | 8331 (56.2) | 252 (41.3) | 15,226 (64.6) | 561 (47.6) |
| Yes | 12,096 (42.0) | 600 (54.0) | 6305 (42.6) | 350 (57.4) | 7978 (33.9) | 595 (50.5) |
| Unknown | 345 (1.2) | 12 (1.1) | 177 (1.2) | 8 (1.3) | 352 (1.5) | 23 (2.0) |
| Independence of daily living | ||||||
| Independent | 28,618 (99.3) | 1084 (97.5) | 14,728 (99.4) | 594 (97.4) | 23,444 (99.5) | 1154 (97.9) |
| Almost independent | 158 (0.5) | 22 (2.0) | 67 (0.5) | 13 (2.1) | 93 (0.4) | 21 (1.8) |
| Unknown | 38 (0.1) | 6 (0.5) | 18 (0.1) | 3 (0.5) | 19 (0.1) | 4 (0.3) |
- Abbreviations: EHI, Employees' Health Insurance; K6, the Kessler 6 scale; NHI, National Health Insurance.
We performed multivariate analyses (model 2) stratified by each SES category to test for the possible modifying effect of SES on the associations between SPD and participation in each type of cancer screening. Marital status and employment status were reclassified into 2 categories to simplify the models a priori: married and nonmarried and employed (regardless of type of employment) and unemployed. In the analysis stratified by employment status, we used the subsample of respondents younger than the general compulsory retirement age of 65 years.
RESULTS
A sample flowchart is provided in Figure 1A-C. Table 1 provides the individual characteristics of the 3 groups. The cancer screening rates of the samples were 29% for colorectal cancer screening, 36% for gastric cancer screening, 28% for lung cancer screening, 42% for breast cancer screening, and 41% for cervical cancer screening.
Flow charts illustrate (A) colorectal, gastric, and lung cancer screening samples; (B) breast cancer screening samples; and (C) cervical cancer screening samples. K6 indicates the 6-item Kessler scale.
Table 2 provides the ORs for participation in each type of cancer screening for SPD. SPD was significantly associated with nonparticipation in screening for colorectal, gastric, and lung cancers. The association between SPD and participation in breast cancer screening was not significant in model 2, and SPD was not significantly associated with cervical cancer screening in either model. In the additional analyses using PD divided into 4 categories, compared with the reference group (“no PD”), none of the ORs for other PD categories were >1.0 for any type of cancer screening (data not shown).
| Screening Typea | OR (95% CI)b | P |
|---|---|---|
| Colorectal cancer, n = 29,926 | ||
| Model 1c | 0.671 (0.579-0.779) | < .001 |
| Model 2d | 0.743 (0.638-0.866) | < .001 |
| Gastric cancer, n = 29,926 | ||
| Model 1 | 0.727 (0.636-0.830) | < .001 |
| Model 2 | 0.823 (0.717-0.946) | .006 |
| Lung cancer, n = 29,926 | ||
| Model 1 | 0.625 (0.537-0.728) | < .001 |
| Model 2 | 0.691 (0.592-0.807) | < .001 |
| Breast cancer, n = 15,423 | ||
| Model 1 | 0.829 (0.702-0.979) | .027 |
| Model 2 | 0.922 (0.775-1.096) | .356 |
| Cervical cancer, n = 24,735 | ||
| Model 1 | 0.906 (0.802-1.025) | .116 |
| Model 2 | 0.980 (0.862-1.113) | .751 |
- Abbreviations: CI, confidence interval; K6, the Kessler 6 scale; OR, odds ratio; SPD, serious psychological distress; CI, confidence interval.
- a The reference group was those with Kessler 6 scores < 13.
- b Crude ORs (95% CIs) of colorectal, gastric, lung, breast, and cervical cancer screenings were 0.639 (0.551-0.741), 0.710 (0.622-0.810), 0.612 (0.526-0.713), 0.866 (0.734-1.022), and 0.920 (0.816-1.037), respectively.
- c Model 1 was adjusted for age and sex.
- d Model 2 was adjusted for age, sex, educational level, employment status, marital status, types of health insurance, smoking status, the presence or absence of current visits to outpatient clinics for physical illness, and independence of daily living.
Table 3 indicates that, among respondents who had a high school level of education or lower and those who had a junior college/vocational college level of education, respondents with SPD were significantly less likely to participate in colorectal, gastric, and lung cancer screening than those without SPD. In contrast, SPD was not significantly associated with cancer screening among respondents who had a college or higher level of education. We conducted post hoc, multivariate logistic regression analyses to test the potential modifying effect of educational status on the relation between SPD and participation in cancer screening (including the interaction term of SPD × dichotomous educational status [“college or higher” vs “other lower education group”] in model 2). The interaction term was significant for all 3 types of screening (colorectal cancer, P = .003; gastric cancer, P = .004; lung cancer, P = .041). For breast and cervical cancer screening, the point estimates of ORs were lower in the “other lower education” group than in the “college or higher” group, although the interactions were not significant in the post hoc analyses (breast cancer, P = .055; cervical cancer, P = .473). For the analyses stratified by marital status, similar ORs were obtained in the “married” and “nonmarried” groups for all cancer sites, and the interactions were also nonsignificant in the post hoc analysis (data not shown). Similarly, for all cancer sites, employment status (“employed” vs “unemployed”) did not significantly modify the relation between SPD and participation in screening (data not shown).
| Screening Type | No. of Participants | OR (95% CI) | P |
|---|---|---|---|
| Stratified by educational status | |||
| ≤High school | |||
| Colorectal | 16,167 | 0.628 (0.501-0.787) | < .001 |
| Gastric | 0.718 (0.587-0.877) | .001 | |
| Lung | 0.709 (0.570-0.882) | .002 | |
| Breast | 8629 | 0.810 (0.640-1.026) | .081 |
| Cervical | 11,700 | 0.955 (0.794-1.148) | .622 |
| >High school but ≤ college or vocational college | |||
| Colorectal | 5052 | 0.633 (0.436-0.919) | .016 |
| Gastric | 0.691 (0.491-0.972) | .034 | |
| Lung | 0.509 (0.344-0.753) | .001 | |
| Breast | 3791 | 0.876 (0.610-1.259) | .475 |
| Cervical | 7004 | 0.935 (0.733-1.192) | .587 |
| ≥University or college | |||
| Colorectal | 6009 | 1.063 (0.781-1.447) | .697 |
| Gastric | 1.156 (0.854-1.565) | .348 | |
| Lung | 0.945 (0.694-1.288) | .721 | |
| Breast | 1566 | 1.395 (0.780-2.497) | .262 |
| Cervical | 3909 | 1.130 (0.811-1.573) | .470 |
| Stratified by marital status | |||
| Married | |||
| Colorectal | 24,221 | 0.734 (0.617-0.872) | < .001 |
| Gastric | 0.809 (0.690-0.950) | .009 | |
| Lung | 0.671 (0.563-0.801) | < .001 | |
| Breast | 12,309 | 0.934 (0.765-1.140) | .502 |
| Cervical | 17,000 | 0.972 (0.827-1.141) | .727 |
| Not married | |||
| Colorectal | 5705 | 0.807 (0.584-1.115) | .193 |
| Gastric | 0.897 (0.676-1.190) | .451 | |
| Lung | 0.764 (0.551-1.058) | .105 | |
| Breast | 3114 | 0.910 (0.640-1.292) | .597 |
| Cervical | 7735 | 1.011 (0.820-1.246) | .919 |
| Stratified by employment status | |||
| Employed | |||
| Colorectal | 19,722 | 0.765 (0.633-0.925) | .006 |
| Gastric | 0.820 (0.690-0.975) | .025 | |
| Lung | 0.670 (0.553-0.813) | < .001 | |
| Breast | 8654 | 1.009 (0.797-1.278) | .938 |
| Cervical | 15,035 | 1.077 (0.913-1.270) | .379 |
| Unemployed | |||
| Colorectal | 5894 | 0.695 (0.504-0.959) | .027 |
| Gastric | 0.867 (0.650-1.156) | .331 | |
| Lung | 0.699 (0.505-0.968) | .031 | |
| Breast | 4680 | 0.827 (0.618-1.106) | .200 |
| Cervical | 7558 | 0.927 (0.744-1.155) | .500 |
- Abbreviations: CI, confidence interval; OR, odds ratio.
- a The model was adjusted for age, sex, educational level, employment status, marital status, types of health insurance, smoking status, the presence or absence of current visits to outpatient clinics for physical illness, and independence degree of daily living, excluding the variable used to stratify. The reference group was those with Kessler 6 scores < 13.
DISCUSSION
This nationally representative survey demonstrated an association between SPD and participation in colorectal, gastric, and lung cancer screening, but not breast or cervical cancer screening. The results were confirmed by additional analyses using a variable of PD divided into 4 categories. The current findings provide new knowledge relevant to Asia and other countries with low cancer screening rates. In addition, educational status modified the effect of SPD on participation in colorectal, gastric, and lung cancer screening.
Three previous US population-based studies using the K6 scale demonstrated a nonsignificant association between PD and participation in colorectal cancer screening.1, 10, 12 This inconsistency may be caused by differences in cancer screening systems, medical systems, or cultures. Further studies in other countries are needed to clarify the association. We observed a significant association between SPD and nonparticipation in gastric cancer screening. This finding is inconsistent with results from a Korean study reporting a nonsignificant association between depression, as assessed by a single question, and participation in gastric cancer screening.27 Korea has higher cancer screening rates than Japan, perhaps because of differences in its national health care systems, such as a call-recall system and large financial incentives.34 The inconsistent findings also may be related to the use of different PD measures (the K6 scale vs a single question). In addition, the association between SPD and lung cancer screening in the current study is a novel finding. To our knowledge, no previous studies have reported such as association, possibly because very few countries have lung cancer screening programs.
The current study did not reveal a significant association between SPD and nonparticipation in breast cancer screening in the fully adjusted model. However, 3 of 4 previous population-based studies reported a significant association using the K6 scale,10, 17, 18 and the other reported a nonsignificant association,1 whereas a meta-analysis demonstrated a significant association between mood disorders and nonparticipation in mammography.25 In addition, our current post hoc analysis of the association between SPD and participation in breast cancer screening within 1 year revealed an OR that was nearly significant in the fully adjusted model (OR, 0.834; 95% CI, 0.693-0.1.003; P = .054). Thus, a 1-point assessment using the K6 scale cannot measure 2-year PD, and the association may be diluted.
Our study did not indicate a significant association between SPD and nonparticipation in cervical cancer screening in either model. Of 4 previous population-based studies using the K6 scale, 2 studies demonstrated a significant association,10, 18 and 2 indicated a nonsignificant association.1, 17 Our post hoc analysis also was nonsignificant. Therefore, overall, the association seems to be nonsignificant. However, although previous studies reported that the relation differed by age group,17, 22 our post hoc analysis demonstrated no significant difference in the OR between the groups ages 20 to 39 years and 40 to 69 years (data not shown). Compared with other cancers, cervical cancer is characterized by relatively young age of onset and screening, which might lead to differences in individual's attitudes toward screening or in the effectiveness of public activities to increase the cervical cancer screening rate.
Educational status modified the association between SPD and participation in colorectal, gastric, and lung cancer screening. To our knowledge, only 1 study has reported an interaction between PD and educational status on receiving cancer screening.8 None of the reported interactions between depression and covariates, including educational level, were significantly associated with participation in colorectal cancer screening with colonoscopy.8 In contrast, we examined the interaction with a different colorectal cancer screening method, fecal occult blood testing, which may have contributed to the inconsistency. In the highly educated group (university and above), cancer screening rates did not differ between individuals with and without SPD. High health literacy, reflected by higher education,35 may attenuate the influence of SPD. Thus, less educated individuals with SPD may be at high risk of nonparticipation in colorectal, gastric, and lung cancer screening.
Our findings have significant implications for public policy makers. Individuals with SPD, especially those who have less education, might not benefit equitably from public activities to promote cancer screening. Therefore, an effective public-awareness campaign specifically targeted to this population is needed. An individual approach for individuals with SPD could also be worth considering. Given that almost one-half of those with serious mental health disorders receive treatment,36 primary care and psychiatric clinics may be ideal settings in which to approach individuals who have SPD—particularly those with chronic mental illness—to encourage and support them to undergo cancer screening. It may be also useful to increase health literacy using an educational approach in community mental health centers and clinical settings. Further studies are needed to develop and examine the effects of such interventions.
The current study involved several limitations that should be considered. First, the self-report data could have overestimated the level of participation in cancer screening. Second, the presence of SPD was self-reported using the K6 scale, which is just a screening tool for psychiatric disorders. Therefore, it is unknown which kind of PD or type of psychiatric disorder most strongly affected cancer screening behavior. Third, there was a timing gap between answering the K6 and participation in cancer screening, which may have been as much as 1 year for colorectal, gastric, and lung cancer screening and 2 years for breast and cervical cancer screening. Fourth, there may have been a selection bias. Individuals with more serious psychological conditions might have declined to participate in the survey or to answer the K6 item, which might have led to underestimation of the effect of SPD on cancer screening participation. Fifth, we did not adjust for all potential confounding factors, such as personality traits, household income, and residential area, which were not included in our anonymized data set. Sixth, the applicability of our findings to countries with different health care systems is unknown. Finally, the true medical endpoint was the influence of SPD on cancer mortality, which was not clarified by the current study.
Conclusion
This large-scale, cross-sectional study of nationally representative data revealed that individuals with SPD are less likely to participate in screening for colorectal, gastric, and lung cancers. This association is stronger for individuals with lower levels of education. Therefore, policy makers and health care providers should address disparities in the rates of cancer screening among this population.
FUNDING SUPPORT
This work was supported by the Japan Society for the Promotion of Science (Japan Society for the Promotion of Science [JSPS] Grant-in-Aid for Scientific Research [KAKENHI] JP17K09112).
CONFLICT OF INTEREST DISCLOSURES
Masaki Fujiwara reports personal fees from Mochida, Sentan Igaku-Sha, Igaku-shoin, and Seiwa Shoten outside the submitted work. Masatoshi Inagaki reports grants from Novartis and personal fees from Mochida, Shionogi, Sumitomo Dainippon, Daiichi-Sankyo, Meiji Seika, Takeda, Nippon Hyoron-Sha, Nanzando, Seiwa Shoten, Igaku-shoin, and Technomics outside the submitted work. Maiko Fujimori reports personal fees from Mochida, Igaku-Shoin, Chugai-Igakusha, Nankodo, and Kongo-Shuppan outside the submitted work. Kyoko Kakeda reports personal fees from Mochida outside the submitted work. Yosuke Uchitomi reports personal fees from Mochida and Eisai and other fees from QOL Company outside the submitted work. Norihito Yamada reports grants from Daiichi Sankyo, Eisai, Otsuka, Astellas, MSD, and Pfizer; and personal fees from UCB Japan, Tsumura, Sumitomo Dainippon, Daiichi-Sankyo, MSD, Pfizer, Eisai, Meiji Seika, and Mochida outside the submitted work. Naoki Nakaya and Yuji Higuchi made no disclosures.
AUTHOR CONTRIBUTIONS
Masaki Fujiwara: Study design, data interpretation, literature review, data analysis, and writing of the initial draft. Masatoshi Inagaki, Naoki Nakaya, Maiko Fujimori, Yuji Higuchi, Kyoko Kakeda, Yosuke Uchitomi, and Norihito Yamada: Study design, data interpretation, and writing of the manuscript.
