Changes in the magnitude of social inequality in the uptake of cervical cancer screening in Taiwan, a country implementing a population-based organized screening program

Studies using nationally-representative survey data have demonstrated prominent social inequality in the reported use of cervical cancer screening [1‐13]. In most countries, women in a higher socioeconomic position have higher rates of Pap smear test uptake than their counterparts in a lower socioeconomic position. However, according to a review, very few studies directly compare the trends in socioeconomic inequality in the uptake of cervical cancer screening [14]. An international comparison study further indicated that the magnitude of social inequality in the prevalence of the uptake of cervical cancer screening is lower in countries implementing a population-based organized screening program [11]. Nevertheless, little is known regarding the changes in the magnitude of social inequality in a country implementing population-based organized screening program in an Asian country. In this study, we examine changes in the magnitude of social inequality in the reported uptake of cervical cancer screening between 2001 and 2009 in Taiwan, a country that has been implementing population-based organized screening program since 1995.

Women aged 30–69 years who had not had a hysterectomy were included in this study, with a total of 5,704 women for the year 2001 and 6,420 women in 2009. Table 2 presents the demographic and socioeconomic characteristics of the respondents. A higher percentage of women aged 50–59 years was noted in 2009 compared with those in 2001 (25% vs. 19%). The distributions of respondents by urbanization level and household income in 2001 were similar to those in 2009. However, fewer respondents reported monthly income in 2009. We also found that the 2009 respondents had higher education levels.

The overall prevalence of undertaking a Pap smear was 62% in 2001 and increased to 69% in 2009 (Table 3). The increase in prevalence was different between age group and was more prominent for elder woman; thus in the groups of 50–59 and 60–69 year-olds, the increase between years was by 10%. The increase of prevalence between 2001 and 2009 was less prominent in women living in metropolitan areas (an increase from 63% to 67%), compared with those in suburban areas (an increase from 63% to 71%), and rural areas (from 63% to 72%). Similarly, women with highest education level and highest household income group showed lowest magnitude of increase in prevalence between the two years (from 66% to 69% for the highest education level and from 68% to 75% for the highest income group). Noteworthy women with university or graduate education level had lower prevalence rate (69% in 2009) than those with high school education level (73%).

Table 4 shows correlation matrix among variables in 2001 and 2009. Although the four independent variables were significantly correlated with each other, the VIF for all variables are less than 2, so the threat of colinearity between variables was not so significant (data not shown). We therefore put all variables into the final multivariate logistic regression model. A significant interaction term between urbanization and income were noted, we thus stratified urbanization and income into five categories in the final multivariate regression model.

The results of final multivariate logistic regression model are presented in Table 5. The SII and RII by age moderately increased from 2001 to 2009, and the main increase occurred in women aged 50 years and above. The SII and RII by urbanization level were, respectively, 0.18 (95% CI 0.01-0.27) and 1.20 (95% CI 1.10-1.31) in 2001, and significantly decreased to 0.13 (95% CI 0.04-0.21) and 1.14 (95% CI 1.04-1.24) in 2009 according to the examination of interaction term between urbanization and year. People living in rural areas showed the most prominent decrease in magnitude of inequality, as shown by the aOR which was 1.87 (95% CI 1.38-2.54) in 2001 and decreased to 1.33 (95% CI 1.02-1.74).

With regard to education level, SII and RII were, respectively, 0.14 (95% CI 0.07-0.21) and 1.15 (95% CI 1.08-1.23) in 2001; and significantly decreased to <0.001 (95% CI -0.07-0.08) and 1.00 (95% CI 0.93-1.08) in 2009. The magnitude of decrease in aOR was most prominent among women with a secondary educational level (from 1.63 to 1.23) and women with a high school education level (from 1.68 to 1.32).

In terms of household monthly income, the SII and RII were, respectively, 0.15 (95% CI 0.09-0.22) and 1.16 (95% CI 1.09-1.24) in 2001. These significantly increased to 0.21 (95% CI -0.14-0.28) and 1.24 (95% CI 1.16-1.32) in 2009. The increase in inequality was mainly confined to people with monthly income of 70,000-99,999. The aOR of this group was 1.65 in 2001 and increased 2.06 in 2009. Stratified analysis further suggested that the inequality in prevalence between women who live in suburban and rural areas with highest income level compared with women live in metropolitan areas with lowest income level showed the most significant increase in inequality between years: the aOR was 0.31 (95% CI 0.20-0.50) in 2001 and increased to 1.03 (95% CI 0.64-1.65) in 2009.

Figure 1 illustrates changes in both rate ratio and rate difference in the uptake of Pap smear tests by the three socioeconomic variables. The pattern of change according to rate ratio and rate difference was similar.

The findings of this study suggest that the overall prevalence of cervical cancer screening uptake increased from 2001 to 2009. However, we observed mixed results regarding the changes in relationship between socioeconomic position and the uptake of Pap smear tests. While the magnitude of social inequality by urbanization and education level decreased, the magnitude of social inequality by income level increased. Specifically, the greatest increase of inequality between 2001 and 2009 in screening prevalence occurred between women living in suburban and rural areas with highest income level and women living in metropolitan areas with lowest income level.

A review study that examined the impact of interventions to improve attendance in female cancer screening among lower socioeconomic groups concluded that while organized population screening programs are successful in increasing overall participation rates, they may not necessarily be able to substantially reduce social inequalities [14]. We found the following two studies that specifically examined changes in the magnitude of social inequality in the uptake of cancer screening practices conducted since the publication of that review [22, 23]. An organized breast cancer screening program was implemented in 2001 in Belgium. The prevalence of mammography use in women aged 50–69 years increased from 2001 to 2004; however, the RII by education level also increased [22]. In Korea, the magnitude of social inequality by education level in undertaking gastric cancer screening decreased from 2005 to 2009, but the magnitude of social inequality by income increased [23].

Our findings in Taiwan are similar to those in Korea: we found a decrease in magnitude of social inequality by education level and an increase in magnitude of social inequality by income level. The most often cited hypothesis is the “inverse equity hypothesis” proposed by Victora, which suggests that effective new interventions will initially reach those of a higher socioeconomic position and will only later filter down to those of poorer status [24]. Inequalities in coverage, morbidity, and mortality therefore increase first and then reduced later after lower socioeconomic class have goo access to the intervention. If this hypothesis applies for cervical cancer screenings in Taiwan between the years of 2001 and 2009, it implies that the availability of screening is still limited to those with greater incomes. Yet this inequality is not likely due to financial barriers because the Bureau of Health Promotion covers the fee for Pap smear examination in all medical settings under the National Health Insurance scheme (which has been in place since 1995). The findings of this study also suggest that there were interaction effects between urbanization and income level. Further studies are needed to explore the possible mechanisms resulting in the increase in magnitude of social inequality by income level.

The findings of this study further suggest that different indicators of socioeconomic position show alternate pictures of changes in magnitude of social inequality. The magnitude of social inequality as shown by urbanization level in both study years was smaller than those based on education and income level differences. One possible explanation was the launch of outreach community-based multiple disease screening program in many rural areas in some cities/counties, which resulted in the increase of prevalence cervical cancer screening [25]. The multiple disease screening program used the Pap smear screening program as a base to integrate other screening regimens encompassing four other neoplastic diseases (liver, breast, colon and oral cancer) and three chronic diseases (hypertension, hyperlipidemia, hyperglycemia). The physicians and public health nurses would outreach to many rural areas in which the screening rates were relatively low.

The decrease in magnitude of social inequality by education level was mainly due to the prominent increase in prevalence of the uptake of cervical cancer screening among women with primary or lower education level. This, again, may have been due to the launch of community-based, multiple disease screening outreach program in many rural areas. One evaluation study indicated that the outreach programs are most beneficial to elderly, widowed and less-educated women in rural areas [26]. This would also help explain why there was such a large increase of prevalence of screening in the older populations between years as well. As the demographic characteristics of respondents in 2009 were older and with higher education level than those in 2001, which would be one possible explanation of decrease in the magnitude of social inequality by educational level, as women with the highest education level have lower prevalence rate. The multiple disease screening program is a kind of intervention targeting socioeconomic disadvantages, which could effectively tackle the social inequality in health [27].

Several limitations are worthy of note. First, similar to other studies using National Health Interview Survey datasets, that there might be some recall bias in the self-reporting of the uptake of Pap smear screening. However, as the main objective of this study was to compare the social inequality in the uptake of Pap smear between 2001 and 2009, it is very unlikely that the recall bias changed markedly between the two study years. Another limitation was we could not obtain three waves of data for comparison (the interim 2005 National Health Interview Survey cervical cancer screening questions were not comparable). The third limitation was the higher non-respondent rate and missing rate in providing income information in 2009 wave, which would affect the interpretations of the results.

We conclude that despite the increase in the prevalence of the uptake of cervical cancer screening between 2001 and 2009, the changes in magnitude of social inequality in the uptake of cervical cancer screening differed by indicators of socioeconomic position. Further studies are needed to explore the different mechanisms resulting in social inequality by different indicator of socioeconomic position.